1
2
NORTH PACIFIC RESEARCH BOARD
3
BERING SEA INTEGRATED ECOSYSTEM RESEARCH PROGRAM
4
5
6
FINAL REPORT
7
8
9
Retrospective analysis of patterns in productivity of fish, seabirds, and
10
marine mammals in the eastern Bering Sea ecosystem
11
12
13
NPRB BSIERP Project B68 Final Report
14
15
16
Franz J. Mueter and Gordon Kruse
17
18
19
University of Alaska Fairbanks, School of Fisheries and Ocean Sciences, Fisheries Division
20
17101 Point Lena Loop Road, Juneau, AK 99801 (907) 796-5448, fmueter@alaska.edu
21
22
23
24
January 2014
1
25
Abstract
26
Through retrospective analyses of physical and biological time series for the eastern Bering Sea
27
ecosystem, as well as the broader Northeast Pacific, we documented covariation between large-scale
28
climate drivers and biological variability, synchrony and asynchrony among different biological
29
components, and linkages between important climate drivers and the productivity of individual
30
populations. At an ecosystem level, our results highlight the regime-shift like behavior of Northeast
31
Pacific physical and biological systems and provide evidence for a directional trend in the biological
32
system that cannot be accounted for by natural climate variability alone, but is consistent with a climate
33
change effect. At the level of communities of interacting species, including fish, seabirds and marine
34
mammals on the eastern Bering Sea shelf we found moderate to strong covariation among the
35
productivity of some components that reflects similarities and differences in the mechanisms driving
36
productivity and in the spatial scales at which the populations are distributed. For individual stocks, such
37
as walleye pollock and snow crab, we identified empirical relationships linking temperature variability to
38
recruitment. In the case of walleye pollock, a dome-shaped relationship between pollock recruitment and
39
temperature is supported by a mechanistic understanding of the importance of temperature in determining
40
prey conditions for juvenile pollock and their effect on subsequent survival. This empirical relationship
41
was used along with projections of future climate variability to generate plausible forecasts of recruitment
42
under different climate scenarios through 2050. Results suggested a long-term decline in the eastern
43
Bering Sea pollock population under a warming climate.
44
45
Key Words
46
Eastern Bering Sea, walleye pollock, Pacific cod, flatfish, snow crab, seabirds, fur seals, climate change,
47
climate variability, recruitment, productivity, synchrony
48
49
50
51
52
Citation
53
Mueter, F.J. and G.H. Kruse. Retrospective analysis of patterns in productivity of fish, seabirds and
54
marine mammals in the eastern Bering Sea ecosystem. NPRB BSIERP Project B68 Final Report, xx p.
2
55
Table of Contents
56
Study Chronology ..............................................................................................................5
57
Introduction ........................................................................................................................5
58
Overall Objectives ..............................................................................................................7
59
Chapter 1: Indicators of variability in the Bering Sea ecosystem ...............................10
60
Chapter 2: Patterns of covariation among fish, seabirds, and marine mammals in the
61
eastern Bering Sea reflect bottom-up controls and spatial scales of distribution ......29
62
Chapter 3: Climate-biology covariation ........................................................................62
63
Chapter 4: Ecosystem Considerations contributions ...................................................65
64
Chapter 4a: Aggregated catch-per-unit-effort of fish and invertebrates in bottom trawl surveys .......... 66
65
Chapter 4b: Average local species richness and diversity of the eastern Bering Sea groundfish
66
community .............................................................................................................................................. 70
67
Chapter 4c: Spatial distribution of groundfish stocks in the Bering Sea ................................................ 73
68
Chapter 4d: Total annual surplus production and overall exploitation rate of groundfish...................... 78
69
Chapter 4e: Combined Standardized Indices of recruitment and survival rate ....................................... 82
70
Chapter 4f: Indicators of basin-scale and Alaska-wide community regime shifts .................................. 85
71
Chapter 5: Zooplankton prey, growth and energy density of larval pollock, and
72
recruitment .......................................................................................................................91
73
Chapter 6: Effects of temperature and gadid predation on snow crab recruitment:
74
Comparisons between the Bering Sea and Atlantic Canada .......................................94
75
Chapter 7: Patterns of change in diets of two piscivorous seabird species during 35 years in
76
the Pribilof Islands .........................................................................................................121
3
77
Chapter 8: Climate Change Brings Uncertain Future for Subarctic Marine Ecosystems and
78
Fisheries ..........................................................................................................................123
79
Chapter 9: Spatial match-mismatch between juvenile fish and prey provides a mechanism
80
for recruitment variability across contrasting climate conditions in the eastern Bering Sea
81
..........................................................................................................................................158
82
Chapter 10: Expected declines in recruitment of walleye pollock (Theragra chalcogramma)
83
in the eastern Bering Sea under future climate change..............................................202
84
Chapter 11: Evaluating management strategies for eastern Bering Sea walleye pollock
85
(Theragra chalcogramma) in a changing environment ...............................................232
86
Conclusions .....................................................................................................................234
87
BSIERP and Bering Sea Project connections ..............................................................238
88
Management or policy implications .............................................................................240
89
Publications ....................................................................................................................241
90
Poster and oral presentations (chronological order) ..................................................243
91
Scientific Conferences ....................................................................................................245
92
Community Meetings.....................................................................................................247
93
Workshops ......................................................................................................................247
94
Outreach .........................................................................................................................249
95
Press Articles ..................................................................................................................250
96
Radio/Broadcast Interviews ..........................................................................................251
4
97
Acknowledgements ........................................................................................................252
98
Literature cited...............................................................................................................252
99
100
Study Chronology
101
This was a new project and was the first NPRB-funded project for PI Franz Mueter. Some of the proposed
102
work was an extension of prior work funded by NOAA NPCREP on the changes in the distribution of the
103
Bering Sea groundfish community (Mueter and Litzow 2008), and on general patterns of covariation
104
among Bering Sea and Gulf of Alaska fish stocks (Mueter et al. 2007). The project consisted of separate
105
awards to Sigma Plus (PI: Franz Mueter) and the University of Alaska Fairbanks (PI: Gordon Kruse). It
106
began on October 1, 2007 and ended on September 30, 2013. This project is related to another BSIERP
107
component, B75 (Correlative Biomass Dynamics model, PIs Kruse, Mueter) because some of the results
108
from the current project were used to inform the structure of the multi-species model adopted for B75. As
109
well, this project was related to NPRB project 1024 (Four decades of climate-biology covariation in
110
Alaskan and North Pacific ecosystems) as many of the annual indices compiled under the current project
111
were also used in that analysis and Mueter was a collaborator on project 1024. A notable development
112
that affected some of the time lines and deliverables was that, in early 2008, Franz Mueter accepted an
113
Assistant Professor position at the University of Alaska Fairbanks in Juneau with teaching and service
114
responsibilities. A portion of the grant was transferred from Sigma Plus to UAF with NPRB approval, the
115
remaining funds stayed with Sigma Plus for contract work during the summer outside the academic year.
116
Semi-annual progress reports for the project were submitted in September 2008 and every April and
117
October from 2009 to 2013, covering reporting periods from October 1 to March 31, and from April 1 to
118
September 30, respectively.
119
120
Introduction
121
The productivity of upper trophic level species in the eastern Bering Sea varies in response to climate
122
variability and human forcing (NRC 1996), although the relative contribution of these drivers and the
123
underlying mechanisms remain poorly understood. Human forcing includes fishing as a major driver of
124
the dynamics of commercial fish and shellfish populations in the Bering Sea, as well as anthropogenic
125
climate forcing associated with increasing CO2 levels in the atmosphere. Effects of anthropogenic
126
warming on biological communities include effects on distribution, growth, reproduction, recruitment,
5
127
and mortality (Drinkwater et al. 2010), but will be difficult to distinguish from the effects of natural
128
environmental variability on these same attributes.
129
There is increasing interest in incorporating the effects of environmental variability into stock assessment
130
and management advice as part of a broader effort to implement ecosystem-based fisheries management
131
(Essington and Punt 2011). Effective implementation of an ecosystem approach requires a better
132
understanding of the effects of natural and anthropogenic forcing on individual populations and on the
133
ecosystem in order to evaluate the effectiveness of different management strategies (Punt et al. 2013).
134
Environmental variability, including long-term trends, decadal-scale variability, and abrupt regime shifts
135
is often a dominant driver of the recruitment and abundance of fish populations (Vert-pre et al. 2013) and
136
has long been recognized as a major driver of Northeast Pacific fish populations (Francis et al. 1998, Hare
137
and Francis 1995, Hollowed and Wooster 1992, 1995).
138
Empirical analyses of environment-recruitment relationships often turn out to be spurious (Myers 1998)
139
and may change over time. Several approaches can be used to guard against identifying spurious
140
relationships. First, meta-analyses across multiple species or geographic areas can help identify important
141
drivers, where species or areas serve as "replicates" for measuring environmental influences (Mueter et al.
142
2002, Myers and Mertz 1998) Second, short-term process studies such as those conducted during the
143
BEST-BSIERP years can be used to identify plausible mechanisms that can be tested using empirical
144
analyses of longer-term data series and can be explored through ecological models! The combination of
145
these approaches – process studies, retrospective analyses, and modeling – can provide a powerful
146
approach to improve our understanding of the Bering Sea ecosystem and was the basis for some of the
147
results presented here.
148
This project component utilized existing data on productivity, including measures of recruitment, survival
149
and growth or condition, of selected upper trophic level species to identify major drivers of variability in
150
the productivity of fish, seabirds, and marine mammals. A database of over 160 indices of environmental
151
and biological variability in the Bering Sea was compiled and used in retrospective analyses to examine
152
historical variability in the system and its key components. Selected results, combined with a mechanistic
153
understanding from the BEST/BSIERP field studies, were used in a case study to quantify the impacts of
154
climate change on future population trends of walleye pollock (Theragra chalcogramma) Results further
155
contributed to the overall research program by providing a set of indicators for other researchers to use,
156
by estimating parameters linking the productivity of individual species to climate variability and by
157
identifying some relationships between climate and productivity that can be incorporated into existing
158
stock assessment models and can help inform an ecosystem-based approach to fisheries management.
6
159
Finally, we contributed to retrospective analyses of several other BSIERP components.
160
161
Overall Objectives
162
The objectives as listed below were only partially met because the focus of the project changed from a
163
strictly retrospective analysis of patterns in the variability of fish, seabird, and mammal productivity, as
164
originally intended, to one that included future projections of fish productivity based on known or
165
hypothesized climate effects. During 2009 and 2010, we increasingly participated in and contributed to
166
the vertically integrated modeling effort at the request of the EMC and NPRB. Because of the daunting
167
task of getting a fully operational end-to-end model of the Bering Sea to run, there was increasing
168
emphasis on producing alternative projections of the possible responses of fish populations to future
169
climate variability using simpler single-species models. In discussions with other Co-PIs and NPRB, we
170
therefore increased our focus on projecting future variability in walleye pollock recruitment and
171
abundance, based on what we learned about the survival of juvenile pollock from field observations,
172
laboratory analyses, and retrospective analyses. This was clearly beyond the scope of retrospective
173
analyses and came at the expense of some of the other objectives. Nevertheless, we were able to address
174
all of the objectives to varying degrees, with much of the focus being on walleye pollock and less
175
emphasis on Pacific cod, flatfishes, and crab.
176
177
Objective 1: Quantify past patterns of variability and covariation among time series of productivity of
178
selected fish, seabird, and marine mammal species.
179
To meet this objective we first compiled or computed approximately 160 indices characterizing
180
interannual variability in important environmental and biological attributes of the Bering Sea ecosystem,
181
including time series of productivity, abundance, growth, and condition. These indices are listed and
182
derived indices are briefly described in Chapter 1. Selected indices, reflecting measures of productivity
183
for six important fish stocks, two crab stocks, four bird species nesting at the Pribilof Islands, and Pribilof
184
Island fur seals, were used to examine if these species display synchronous patterns of productivity,
185
identify which species display similar or opposite patterns of variability, and examine the observed
186
variability relative to environmental variability (Chapter 2). Additional analysis of a broader set of
187
indices, reflecting physical and biological variability throughout the Northeast Pacific, were examined in
188
collaboration with Mike Litzow (NPRB Project 1024), resulting in two papers that characterize large
189
scale variability and identify significant correlations between physical and biological variability in the
7
190
Northeast Pacific and document recent unknown shifts in Alaskan ecosystems. Abstracts of these papers
191
are included here for reference (Chapter 3), while drafts of the full papers are included in the NPRB
192
Project 1024 Final Report "Four decades of climate-biology covariation in Alaskan and North Pacific
193
ecosystems" (Litzow et al. 2012) and were published in late 2013 (Litzow and Mueter 2013, Litzow et al.
194
2013). A retrospective analysis of lower trophic level variability based on satellite-based observations of
195
chlorophyll-a contributed to another BSIERP paper (Sigler et al. 2014). Finally, a summary of the status
196
and trends of a number of key indicators have been contributed to the Ecosystem Considerations chapter
197
of the annual SAFE (Stock Assessment and Fishery Evaluation) reports to the North Pacific Fishery
198
Management Council (Zador 2013) and to the Bering Sea chapter of the North Pacific Ecosystems Status
199
Report (Hunt et al. 2010). The most recent contributions for six types of indicators for the Ecosystem
200
Considerations are included as Chapter 4.
201
202
Objective 2: Test whether historical patterns and trends in these series are consistent with existing
203
hypotheses
204
At a community-wide level, hypotheses about covariation among different groups of species are
205
addressed through correlation analyses in Chapter 2. Specific hypotheses about what drives variability in
206
productivity were examined through more detailed statistical modeling for two species, walleye pollock
207
and snow crab. For walleye pollock we addressed hypotheses relating to the importance of ice extent,
208
timing of ice retreat, summer stratification, and predation / cannibalism, but ultimately focused on a new,
209
emerging hypothesis regarding the importance of late summer prey conditions. The biological basis for
210
these is laid out in two BEST/BSIERP contributions that synthesize findings from field work and
211
statistical analyses (Hunt et al 2011, Coyle et al 2011). Retrospective analyses of variability in pollock
212
recruitment in these papers were contributed by Franz Mueter and abstracts for both papers are included
213
as Chapter 5. For snow crab, a graduate student (Laurinda Marcello, primarily funded through a separate
214
non-NPRB grant to Franz Mueter) examined the importance of ice and temperature conditions, as well as
215
potential predation by cod, on the recruitment dynamics of snow crab (Chapter 6) in the context of a
216
larger project comparing gadid-crustacean interactions across multiple subarctic ecosystems (Mueter et al.
217
2012). In addition, Franz Mueter collaborated with Heather Renner to conduct a retrospective analysis of
218
long-term variability in the diets of black-legged kittiwakes (Rissa tridactyla) and thick-billed murres
219
(Uria lomvia) (Renner et al 2012, abstract included as Chapter 7). Other species listed in the original
220
workplan (Pacific cod, flatfish) were not examined individually. Finally, in collaboration with a Ph.D.
221
student and Dr. George Hunt, we synthesized some of our current (as of late 2010) understanding of some
8
222
of the effects of climate variability on subarctic and Arctic systems (Chapter 8) for a more popular outlet
223
as a book chapter in North by 2020: Perspectives on Alaska’s Changing Social-Ecological Systems
224
(Lovecraft and Eicken 2011).
225
226
Objective 3: Suggest new hypotheses based on relationships among the productivity of different
227
ecosystem components and relationships between their productivity and observed climate variability
228
Most of the work on this objective focused on walleye pollock for reasons elaborated above. As
229
mentioned under Objective 2, initial results from both the field work and from retrospective analyses
230
suggested that prey conditions during the late summer or early fall period is particularly important for
231
juvenile walleye pollock (Chapter 5) and other recent BEST/BSIERP contributions provided the
232
physiological underpinnings for this hypothesis (Heintz et al. 2013, Siddon et al. 2013a). Bioenergetic
233
modeling work by Ph.D. student Elizabeth Siddon, in collaboration with Trond Kristiansen and several
234
BSIERP PIs (including Mueter, in part supported by this grant), suggested a new hypothesis ("spatial
235
match-mismatch") based on analyses of the energetics of juvenile pollock and the distribution of juveniles
236
and their prey during contrasting warm and cold years (Chapter 9).
237
238
Objective 4: Provide functional forms and parameter estimates (and their uncertainty) that link the
239
productivity of different ecosystem components to climate variability
240
This objective was only partially addressed and was limited to linking variability in walleye pollock
241
recruitment to climate variability due to the aforementioned shift in focus during the early stages of the
242
project. Ultimately this shift in focus proved very successful and produced one of the first credible
243
projections to quantify the impacts of climate change on future population trends of walleye pollock in the
244
Bering Sea (Chapter 10). For the analysis we projected plausible long-term population trajectories based
245
on our current best understanding of pollock life history, in combination with IPCC climate projections.
246
Scenarios are based on an empirical relationship between late summer water temperatures and walleye
247
pollock recruitment, informed by a mechanistic understanding of the importance of later summer
248
conditions for juvenile pollock from the BEST/BSIERP program (Chapter 5). A related publication based
249
on the same empirical relationship between temperature and recruitment explored the use of different
250
harvest control rules in a changing climate (Ianelli et al 2011, Abstract included as Chapter 11).
251
9
252
Chapter 1: Indicators of variability in the Bering Sea ecosystem
253
254
Franz J. Mueter
255
256
University of Alaska Fairbanks, School of Fisheries and Ocean Sciences, Juneau, Alaska 99801, USA
257
258
Introduction
259
This chapter describes the data and data processing steps used to derive a set of monthly and annual
260
indices for use in the retrospective analyses and for use by other investigators. Most indices are from
261
publicly available data sources and were included either unchanged, after aggregating data with high
262
spatial and/or temporal resolution, or after more extensive processing steps as explained below. The time
263
periods covered differ among indicators and range from as early as 1900 through 2012.
264
All indices are included in a spreadsheet that was initially compiled in 2008 and updated or expanded at
265
various times. The most recent version was submitted to the data manager in March 2014. The
266
spreadsheet includes both monthly and annual indices with a brief description of each index, information
267
on the temporal and spatial scale covered by the index, units, and data sources. Indices were grouped into
268
'Environmental', 'Lower trophic level', 'Forage fish', 'Groundfish / crab', 'Seabirds' and 'Marine mammals'.
269
Indices are described below by major group and by variable type (e.g. 'temperature', 'biomass',
270
'productivity', etc.)
271
272
Environmental data: Large-scale Indices
273
PDO.win (winter average of Pacific Decadal Oscillation)
274
PDO.sum (summer average of Pacific Decadal Oscillation)
275
AO (Winter average Arctic Oscillation)
276
Description and rationale: The large-scale PDO captures leading mode of variability of North Pacific
277
Ocean sea-surface temperature variability, which is related to temperature variability in the Bering Sea.
278
We used the winter (Dec-Feb) and summer (June-August) averages following Mantua et al (1997).
279
Monthly standardized values for the PDO index are derived as the leading PC of monthly SST anomalies
280
in the North Pacific Ocean, poleward of 20˚N. The monthly mean global average SST anomalies are
10
281
removed first to separate this pattern of variability from any "global warming" signal that may be present
282
in the data.
283
The AO is related to atmospheric circulation over the Bering Sea and shows a strong shift associated
284
with the 1988/89 regime shift. The daily AO index is constructed by projecting the daily 1000mb height
285
anomalies poleward of 20°N onto the loading pattern of the AO. The year-round monthly mean anomaly
286
data has been used to obtain the loading pattern of the AO. Since the AO has the largest variability
287
during the cold sesaon, the loading pattern primarily captures characteristics of the cold season AO
288
pattern. We used the Jan-Mar means following Hare & Mantua (2000).
289
Raw data: Monthly data sources underlying the PDO index are based on the UKMO Historical SST data
290
set for 1900-81; Reynold's Optimally Interpolated SST (V1) for January 1982-Dec 2001; OI SST Version
291
2 (V2) beginning January 2002. For more details, see Zhang et al (1997) and Mantua et al. (1997).
292
Time period:
293
PDO: Mantua et al. (1997) use monthly SST data since January 1900, hence Dec-Feb averages are
294
available 1901-2012, summer averages from 1900-2012.
295
AO: 1950-2009
296
Source: PDO: http://www.atmos.washington.edu/~mantua/abst.PDO.html
297
AO: http://www.cpc.noaa.gov/products/precip/CWlink/daily_ao_index/
298
monthly.ao.index.b50.current.ascii
299
Summary of processing steps: Compute simple averages of monthly indices available from above
300
sources.
301
302
Environmental data: Air temperatures at St. Paul airport and derived indices
303
(annual index: airT.win; monthly index: airT.StPaul)
304
Description and rationale: Winter air temperature can serve as a longer-term proxy for winter ice
305
conditions, which affect conditions on the shelf during summer through its effects on temperature, bloom
306
timing, and stratification. The index is strongly correlated with PMEL's ice cover index (see "ICI", 1979-
307
2008, r = -0.63), as well as with summer surface and bottom water temperatures. In addition, air
308
temperature data were used to characterize the spring transition to warmer conditions and "growing
309
degree days" as an index of the length and temperature of the growing season.
11
310
Data source: Daily recorded minimum and maximum air temperatures at St. Paul airport (57.15N,
311
170.22W). NOAA-National Environmental Satellite, Data, and Information Service (NESDIS), National
312
Climatic Data Center (NCDC). http://lwf.ncdc.noaa.gov/oa/climate/stationlocator.html
313
Time period: 1950-2013 (season length and growing degree days only computed through 2008)
314
Summary of processing steps: The mid-point between daily minimum and maximum recorded
315
temperature was used as an index of daily average temperature. Daily average temperatures were
316
averaged over the period December 1 to Feb 28 (or 29) to obtain an annual index of average winter
317
temperature. Monthly average temperatures were previously downloaded from
318
http://www.wrcc.dri.edu/summary/climsmak.html (no longer supported) and updated with recent
319
temperatures from NCDC (http://lwf.ncdc.noaa.gov/oa/climate/stationlocator.html). To compute a spring
320
transition index for the Pribilof region daily mean air temperatures at St. Paul were smoothed (loess-
321
smoother with span=0.2) to estimate the Julian day on which mean air temperature first exceeds 3 deg C.
322
Growing Degree Days were defined as the sum of daily average temperatures at St. Paul Island that
323
exceeded 3 °C.
324
325
Environmental data: Summer sea-surface and bottom temperatures, Eastern Bering Sea shelf
326
(RACE.SST, RACE.BT, Pribs.SST, Pribs.BT)
327
Description and rationale: The NMFS bottom trawl survey provides the only long-term dataset of
328
measured sea-surface temperatures over the eastern Bering Sea shelf that covers much of the shelf (Fig.
329
1.1). It therefore represents the best available measurements of summer temperature conditions
330
experienced by pelagic species in the upper water column.
331
Data source: Bob Lauth, Alaska Fisheries Science Center, NOAA-NMFS, Seattle. Data consist of
332
measured sea-surface temperature and bottom temperature at standard survey stations sampled over a
333
period of about 6 weeks in the summer. The total number of stations sampled per year ranged from 334 to
334
405 stations, with surface temperature measurements available at a minimum of 271 stations (1987) and a
335
maximum of 398 stations (2005) and bottom temperatures available at a minimum of 292 stations (1994)
336
and a maximum of 396 stations (2006). Surface and bottom temperatures, as well as water column
337
profiles, are generally taken at each station with a datalogger attached to the headrope of the net.
338
Time period: 1982-2010, summer only
339
Summary of processing steps: Spatially averaged sea-surface temperatures over the entire survey region
340
were adjusted for differences in the timing of the survey (which strongly affects temperatures because of
12
341
seasonal warming). We predicted spatial means of surface temperature from a Generalized Additive
342
Model. The model estimated a smooth spatial trend across years, a seasonal trend that was allowed to
343
vary (smoothly) across years in addition to annual mean temperatures corresponding to the overall mean
344
sampling date and to the center of the survey region (mean latitude & longitude). The resulting model fit
345
the data well (R2 = 0.83) with no indication of systematic biases or violation of regression assumptions.
346
The model was fit to all years simultaneously to better estimate the seasonal trend which is strongly
347
confounded with longitude because sampling generally proceeds from east to west. The timing of the
348
survey had a strong impact on temperatures, which generally increase as the season progresses, but the
349
timing of the temperature increase appeared to be earlier in the 1990s and later in the 1980s and towards
350
the end of the time series. To obtain an annual temperature index that is comparable across years we used
351
predicted temperatures on July 1, the approximate mid-day of the survey, and on the middle shelf at the
352
M2 site (56.19 ˚N, 165 ˚W). It should be noted that the model assumes a consistent spatial pattern in
353
temperatures across years to obtain an index that is representative of the entire shelf rather than a single
354
location.
355
The following indices were derived from the data:
356

357
358
RACE.SST: Annual index of predicted sea-surface temperature at mooring 2 site on July 1, based
on Generalized Additive Model as described above.

RACE.BT: Annual index of predicted bottom temperature at mooring 2 site on July 1, based on a
359
similar GAM with two differences: (1) Only stations shallower than 150 m were included because
360
very few deeper stations were sampled and they were not sampled consistently in all years. (2)
361
the model included bottom depth as additional covariate, considerably improving the model fit.
362

Pribs.SST: Annual index of spatially averaged sea-surface temperature over a restricted region
363
around the Pribilof Islands as an index of surface water characteristics within foraging range of
364
birds and mammals. We computed average SST over all standard survey stations within 100 km
365
of St.Paul or St. George that were over a water depth of less than 100 m (i.e. middle domain). The
366
index was computed as a straight average across all stations sampled in a given year, which
367
ranged from 28 to 33 stations. However, temperature measurements were not always available,
368
particularly in 1987, when only 16 stations had surface temperature measurements, and in 1987
369
and 1994, when only 16 and 17 stations had bottom temperature measurements. The stations with
370
temperature measurements were broadly distributed over the region and had a similar depth range
371
than that of the maximum number of stations, therefore we simply averaged temperature
372
measurements across stations.
373

Pribs.SST: Spatially averaged bottom temperatures over the same areas.
13
374
375
Environmental data: Extended reconstructed sea-surface temperatures (SST.sum, SST.ann) and
376
derived indices of spring transition, season length, and growing degree days (ST.sst, SL.sst, GDD)
377
Description and rationale: As a longer-term measure of SST over the eastern Bering Sea shelf, we used
378
the extended, reconstructed SST data set (v.3) for the approximate area corresponding to the NMFS trawl
379
survey region (Fig. 1.1). Derived indices based on the long-term ERSST data were computed to
380
characterize the spring transition and length of the warm season as an indicator of the onset and length of
381
the growing season for marine organisms.
382
Data source: Monthly data are provided by CDC on a 2˚latitude by 2˚longitude grid. See
383
http://www.cdc.noaa.gov/cdc/data.noaa.ersst.html for a detailed description of the data.
384
Time period: January 1900 – December 2013
385
Summary of processing steps: In addition to the monthly data (Fig. 1.2), two indices of SST condition,
386
averaged over different seasons, were included in the database:
387

388
389
390
Average SST during July - September was used as an index of late summer surface conditions,
which are likely indicative of prey conditions for juvenile walleye pollock (Fig. 1.2).

Average SST during April – September was used as an annual index of spring and summer
surface temperature conditions in the southeastern Bering Sea.
391
A "spring transition” index and total summer season length were indexed by first interpolating monthly
392
temperatures using a cubic spline interpolation (function 'spline' in R) and estimating the days when SST
393
first increases above 4˚C in the spring (= spring transition) and first decreases below 4˚C in the fall.
394
Season length was computed as the total number of days with SST above 4˚C (the approximate long-term
395
mean). The warm season has not moved over time but the length of the warm season has increased (Fig
396
1.3) as evident in significant trends towards both earlier spring transition dates (first day above 4˚C,
397
Linear regression with AR1 errors, t = -3.80, n=108, p <0.0001) and later fall transition dates (first day
398
above 4˚C, Linear regression with AR1 errors, t = 3.82, n=108, p =0.0002). Earlier onset of summer
399
(SST>4˚C) and a later end date translate into a substantial increase in season length (Fig. 1.4, Linear
400
regression with AR1 errors, t = 4.28, n=108, p <0.001). In addition, we computed the number of degree-
401
days above 4˚C for each year, which increased from 400-500 in the early part of the century to over 700
402
in recent years (Fig. 1.5, Linear regression with AR1 errors, t = 4.28, n=108, p <0.001).
403
14
404
Environmental data: Cold pool index (CPI)
405
Description and rationale: The cold pool is generally defined as the water mass on the eastern Bering
406
Sea shelf that has a temperature below 2˚C. The cold pool index quantifies the proportion of the NMFS
407
bottom trawl survey area on the southeastern Bering Sea shelf where the bottom water temperature at the
408
time of the survey is below 2˚C.
409
Data source: Bob Lauth, Alaska Fisheries Science Center, NOAA-NMFS, Seattle. Measured bottom
410
temperature at standard survey stations sampled over a period of about 6 weeks in the summer. The total
411
number of stations sampled per year ranged from 334 to 405 stations, with bottom temperatures available
412
at a minimum of 292 stations (1994) and a maximum of 396 stations (2006). Bottom temperatures were
413
generally taken at each station with a datalogger attached to the headrope of the net.
414
Time period: 1982-2008, summer only
415
Summary of processing steps: Bottom temperatures were estimated as follows:
416
1. The latitude and longitude coordinates were mapped onto an equal-area grid using an Albers
417
equal-area projection.
418
2. A grid of equally spaced stations was overlaid over the survey area (Fig. 1.6).
419
3. Bottom temperatures at these grid points were estimated for each year using a thin-plate
420
regression spline surface (Generalized Additive Model of bottom temperature as a smooth
421
function of x and y coordinates from Albers projection).
422
4. The fraction of the survey area below 2˚C was estimated as the fraction of grid points with a
predicted temperature below 2˚C.
423
424
425
Environmental data: Winds (Wind.NS, Wind.opt, Wind.str, Wind.mixing.JJ, Wind.mix.AS)
426
Description and rationale: Wind speed and direction drive across-shelf and along-shelf fluxes that
427
influence the supply of nutrients to the shelf and advect eggs and larvae from spawning sites to nursery
428
areas.
429
Data source: All wind indices are based on NCEP/NCAR reanalysis data available from NOAA's Earth
430
System Research Laboratory website and were downloaded as daily means of northerly (v-) and easterly
431
(u-) wind speeds (m/s) at 10 m height above the surface for selected locations
432
(http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanalysis.surfaceflux.html).
433
Time period: 1949-2009
434
Summary of processing steps: We derived several indices of mixing, upwelling, and cross-shelf transport
435
based on surface wind data (10m height) from the NCEP/NCAR reanalysis (Kalnay et al. 1996). We
15
436
obtained daily near-surface (10m) data along 56.19˚N (168.75˚W, 166.875˚W, and 165˚W) and derived
437
the following summary indices, similar to the atmospheric indices (Favorable wind, Strong wind, N-S
438
wind) described on the Bering Climate website (http://www.beringclimate.noaa.gov/index.html), but
439
updated through 2009:
440

Wind.NS: As an index of N-S winds along the shelf break during winter, we recomputed and
441
updated the average daily N-S wind speed at 56.19oN, 168.75oW during the period from
442
December through March (year corresponds to Jan-Mar).
443

Wind.opt: We computed an index of wind speeds favorable for successful larval feeding near the
444
Mooring 2 site (56.19 oN, 165 oW). It was originally developed at PMEL and referred to the
445
number of days each year during the period 1 May through 15 July in which the daily average
446
wind speed was in the range 4.8 to 9.5 ms-1 at 57oN, 164oW. Because winds in this region are
447
forced by large-scale pressure gradients, these winds are believed to be representative of the
448
southeastern Bering Sea shelf. It is unclear what the exact location of the original index was as
449
our index was only weakly correlated with the original index from the Bering Climate website
450
for the period of overlap (1950-2006, r = 0.398).
451

Wind.str: This index consists of the number of days during the period May 1 – July 15 on which
452
the daily average wind speed exceeds 9.5 ms-1 near Mooring 2 based on the same location as
453
above. This is an index of excessively strong winds that may disrupt feeding of larval fishes. Our
454
version of this index was strongly correlated with the index on the Bering Climate website (1950-
455
2006: r = 0.775).
456

Wind.mix.JJ, Wind.mix.AS: We computed a mid-summer and a late-summer wind mixing index
457
for the same location near Mooring 2. The index was computed by averaging cubed average daily
458
wind speeds (because wind mixing is proportional to wind speed cubed) over the period June 1 –
459
July 31 and August 1 – September 30, respectively. These provide simple indices that may be
460
considered proxies for the re-supply of nutrients into the surface layer during these two periods.
461
The August / September index captures the period of increased storminess and may be indicative
462
of a late summer bloom.
463
464

Other wind indices were taken directly from the Bering Climate website and include:
o
465
466
467
Wind.UPW.win: Mean winter (previous Nov-Apr) along-peninsula component of wind
stress as index of northward transport through Unimak Pass
o
Wind.UPW.spr: Mean spring (May-Jun) along-peninsula component of wind stress as
index of northward transport through Unimak Pass.
16
468
469
Environmental data: Stratification / Water column stability
470
Description and rationale: The strength of stratification or the stability of the water column over the
471
middle Bering Sea shelf determine the ease with which winds can mix the water column to bring new
472
nutrients from the sub-surface into the surface layer. Strong stratification/stability implies that it takes
473
more wind mixing to deepen the mixed layer. Hence the strength of stratification during summer (after
474
the spring bloom removes surface layer nutrients) is likely to be inversely related to total new production
475
in the water column.
476
Data source: Mooring 2 oceanographic data from Phyllis Stabeno, PMEL
477
Time period: 1995-2009
478
Summary of processing steps: The strength of summer stratification is based on based on a 1-D model of
479
mixed-layer depth by Carol Ladd (PMEL) and reflects the maximum daily temperature gradient, averaged
480
over 29 June-27 Sept (Julian day 180-270) on the eastern Bering Sea shelf at the Mooring 2 location
481
(56.19 oN, 165 oW). See Mueter et al (2006) for a description of the 1-D model.
482
483
Environmental data: Nutrients (Nitrate)
484
Description and rationale: Nitrate concentrations in late winter, prior to the production season, provide a
485
measure of production potential during the upcoming summer as new production in the eastern Bering
486
Sea is typically limited by the availability of macronutrients such as nitrate. Attempts were made to
487
construct a time series of nutrient availability in the deeper source waters off the Bering Sea shelf as well
488
as in near-bottom waters on the shelf in the spring and early summer using available nutrient data from
489
the World Ocean Database and other sources. Spatial and interannual patterns in the available
490
measurements were explored and summarized in a poster. However, early nutrient data proved to be
491
unreliable and estimated annual mean values were judged to not provide a reliable index of interannual
492
variability.
493
494
Lower trophic levels: Chlorophyll concentrations
495
Description and rationale: Satellite-based observations provide a measure of Chlorophyll a
496
concentrations at or near the surface and can serve as an indicator of the strength and timing of the spring
497
and fall phytoplankton blooms. However, surface Chl a concentrations largely miss subsurface blooms
498
and cannot be interpreted as a measure of production in the surface layer because the bloom is grazed by
17
499
zooplankton. Pixel-by-pixel Chl-a concentrations were also used to quantify spatial correlations between
500
mooring-based measures of fluorescence and satellite-based estimates of Chl a (Sigler et al 2014)
501
Raw data: We retrieved eight-day composite Level-3 SeaWiFS and MODIS-Aqua chl-a data at 9-km
502
spatial resolution using the Giovanni online data system, developed and maintained by the NASA
503
Goddard Earth Sciences Data and Information Services Center (Acker and Leptoukh 2007). Data for a
504
rectangular region (54–66°N and 157–180°W) that encompasses the eastern Bering Sea shelf were
505
extracted from the global coverage datasets (units: mg/m3). Open ocean values (offshore of the 500m
506
isobath) and nearshore values (inshore of 10m isobath) were excluded (Fig. 1.7). We compared eight full
507
years of overlap between the two datasets (2003-2010) based on correlations and absolute differences
508
(bias) of the logarithmically transformed data on a pixel-by-pixel basis. Estimates from the two satellites
509
were strongly correlated (r=0.65) and bias was found to be negligible (median bias < 0.01, 88% of 1.4
510
million paired observations had absolute bias < 0.05). In order to produce the longest possible continuous
511
time series of remotely-sensed chl-a data, we combined SeaWiFS data from 1998-2002 with MODIS-
512
Aqua data from 2003-2011.
513
Time period: 1998-2012 for Chl a concentrations; 1998-2007 for bloom timing
514
Summary of processing steps: To construct aggregate indices the 8-day composite Chl a concentrations
515
were averaged over different seasons and regions to compute annual indices as follows:
516

Annual (~March – October, Julian days 57-306) averages of Chlorophyll a concentrations (mg m-
517
3
518
(Chl.ann.inn, Chl.ann.mid, Chl.ann.out). Cloud cover from November through February was
519
extensive and did not provide meaningful estimates of average Chl a concentrations.
520
) were computed for the inner (10-50m), middle (50-100m) and outer domain, respectively

Seasonal Chl a concentrations were computed for each of the three regions for the periods from
521
March to mid-July (Julian days 57-193) and from mid-July to the end of October (Julian days
522
194-306) (Chl.spr.inn, etc; Chl.fall.inn, etc).
523
524
Lower trophic levels: Bloom timing
525
Description and rationale: The Chl a data described in the previous section were used to obtain an index
526
of the onset of the bloom timing as a potential explanatory variable for recruitment success of fish.
527
Time period: 1998-2007
18
528
Summary of processing steps: Bloom timing was estimated by fitting smooth trends to the 8-day
529
averages of Chl a concentrations (see above) for a given region over time and determining the day of the
530
year when the smoothed concentration first exceeded a region-specific threshold (2-3 μg l-1). While other
531
authors have defined bloom timing based on the date when concentrations first exceed some fraction of
532
the median annual concentration (e.g. Henson et al 2006 used 5% of annual median), we considered an
533
absolute cut-off value more appropriate because some minimum threshold is likely to be needed for
534
successful grazing by zooplankton! Indices of the onset of the spring bloom were computed for the entire
535
southeastern Bering Sea shelf (Figure ) (Bloom.EBS), for the inner, middle, and outer domains
536
(Bloom.inner, Bloom.middle, Bloom.outer), and for several smaller subregions as follows:
537

We averaged chl a concentrations over all pixels whose centers were within a 50km radius of
538
Mooring 2 over the middle shelf (Bloom.M2b). Data from this region was used to estimate the
539
timing of the bloom around Mooring 2 and to compare the satellite-based estimate of bloom
540
timing to an estimate of bloom timing based on 1-D model as described below.
541

Average for the Pribilof Islands region (Bloom. Pribs) for analysis pertaining to the Pribilof
542
Island region only, for example comparing bloom timing to the timing of hatching of birds at St.
543
Pau and St. George.
544
An additional index for the timing of the onset of the bloom at the mooring 2 location (Bloom.M2) was
545
constructed based on a simple 1-D model as described in Mueter et al. (2006) and updated by Carol Ladd
546
(PMEL, pers. comm.) to include the years 1960-2009.
547
548
Lower trophic levels: jellyfish
549
Description and rationale: Relative index of jellyfish catch-per-unit-effort during summer bottom trawl
550
surveys conducted over the eastern Bering Sea shelf. Jellyfish are important predators on zooplankton and
551
compete with fish for zooplankton prey. Jellyfish also provide shelter for juvenile pollock.
552
Data source: Bob Lauth, Alaska Fisheries Science Center, NOAA-NMFS, Seattle (available at
553
http://access.afsc.noaa.gov/reem/EcoWeb/EcoChaptDataMainFrame.htm)
554
Time period: 1982-2011
555
556
Forage fish: CPUE of forage fishes in the Eastern Bering Sea
557
Description and rationale: Indices of total abundance of selected forage fishes, who serve as prey for
558
many upper trophic level species and may impact their dynamics.
19
559
Data source: Catch per unit effort data from Bob Lauth, Alaska Fisheries Science Center, NOAA-NMFS,
560
Seattle (available at: http://www.afsc.noaa.gov/RACE/groundfish/survey_data/data.htm)
561
Time period: 1982-2009
562
Summary of processing steps: CPUEs of forage species or species groups captured in the summer bottom
563
trawl survey were computed by fitting a delta-lognormal model of station-specific catch-per-unit effort as
564
a function of various explanatory variables or nuisance variables to estimate mean CPUE across all
565
stations within the survey area where the species was captured in at least one year. Probability of
566
occurrence at each station was predicted based on year, area swept, net width, Julian day, bottom depth,
567
latitude and longitude using a Generalized Additive Model with a logit link as follows:
568
569
log(pi/(1-pi) = α + yt + f1(area) + f2(net width) + f3(x,y)+ f4(day, depth)
570
571
where pi is the probability that species i is observed at a given station, yt is the 'year effect' for year t that
572
corresponds to the mean probability of occurrence when all other covariates are fixed at their mean, the fj
573
are cubic spline smoothers of the predictor variables, and x and y are latitude and longitude, respectively.
574
CPUE-where-present on the log-transformed scale was modeled using the same predictor variables except
575
area swept because catch rates (CPUE) were adjusted for area swept. A combined estimate of average
576
CPUE was computed by multiplying the average probability of occurrence by the average CPUE-where-
577
present for each year. Indices were computed for the following species: pandalid shrimp, Pacific herring,
578
capelin, and eulachon. Indices of abundance for the Pribilof Islands foraging area, defined as the area
579
within 100 km of St. George or St. Paul Island, were computed for the same species as well as for
580
juvenile (< 20cm) pollock based on a simple average across all stations with the foraging area.
581
582
Groundfish & crab: Total biomass, spawning stock biomass, and recruitment
583
Description: Estimates of total biomass, spawning stock biomass, and recruitment were compiled for the
584
following species from the 2012 Stock Assessment and Fisheries evaluation reports for groundfish
585
(NPFMC 2012a) and crab (NPFMC 2012b): walleye pollock, Pacific cod, arrowtooth flounder, yellowfin
586
sole, northern rock sole, flathead sole, snow crab, and red king crab. Recruitment was the abundance of
587
the youngest age class that could reliably be estimated. Total biomass included all ages from the age of
588
recruitment to the oldest age group. Some analyses were conducted with indices from earlier stock
589
assessment reports, which do not have the same values for the overlapping years because estimates of
590
historical levels of biomass and recruitment change with each annual assessment. Earlier assessments are
591
available at http://www.npfmc.org/safe-stock-assessment-and-fishery-evaluation-reports.
20
592
Time period: Variable. All biomass values through 2012, recruitment estimates were only included
593
through the most recent year class that was reliably estimated.
594
595
Groundfish & crab: Survival rate indices
596
Description and rationale: As a measure of environmentally induced variability in recruitment we
597
computed survival rate indices from spawning to recruitment. Indices were computed as the residuals
598
from a best-fit stock-recruitment model relating log-survival (log{recruits / spawning stock biomass}) to
599
spawning stock biomass.
600
601
Groundfish & crab: Production and harvest rate
602
Description:: See chapter 4.
603
604
Groundfish & crab: Combined indices of recruitment and survival
605
Description: See chapter 4.
606
607
Groundfish & crab: Aggregated indices of recruitment
608
Description: Several aggregated indices of recruitment were constructed based on observed covariation
609
among stocks (Chapter 2). Recruitment anomalies for Bering Sea gadids were computed by taking the
610
average of standardized recruitment time series for walleye pollock and Pacific cod (1977-2010).
611
Similarly, a combined recruitment anomaly was computed across four flatfish species whose recruitment
612
showed a moderate level of synchrony (arrowtooth flounder, yellowfin sole, northern rock sole, flathead
613
sole, 1975-2005). In addition, the first principal component of the two gadid and four flatfish species was
614
computed as an index of shared recruitment variability for these six stocks (1977-2005).
615
616
Groundfish & crab: Condition indices
617
Description: Annual anomalies in weight at a given size and size at a given age were computed for five
618
groundfish species (walleye pollock, Pacific cod, arrowtooth flounder, yellowfin sole and flathead sole) as
619
measures of condition.
21
620
Data source: RACEBASE database. Data obtained from Bob Lauth, Alaska Fisheries Science Center,
621
NOAA-NMFS (pers. comm.).
622
Summary of processing steps: Weight-at-size anomalies were computed by fitting a mixed-effects model
623
of the following form to weight-length data collected during summer bottom trawl surveys:
624
log( Wi ,t )      log( Li ,t )  at  bt  log( Li ,t )  f (day )  s(1, 2 )   i ,t
625
where Wi,t and Li,t are the weight and length, respectively of the ith fish in year t, α is the overall intercept,
626
β is the overall slope of the length-weight relationship (allometric coefficient), at and bt are random year-
627
specific intercepts and slopes, respectively, f is a smooth function of day of year to account for seasonal
628
trends, s is a smooth surface fit to latitude (1) and longitude (2) to account for spatial patterns, and εi,t is
629
a random residual. The year-specific anomalies in the length-weight relationships (at + bt * log(Li,t)) were
630
used to compute a weight at size anomaly as the predicted anomaly at the lower quartile of the size
631
distribution for each species. Size-at-age anomalies for the same species were similarly computed by
632
modeling residuals from a van-Bertalanffy growth model fitted to size-at-age data as a function of Year
633
and potential "nuisance variables" using a Generalized Additive Model of the following form:
ri ,t  Yeart  f (day )  s(1 , 2 )   i ,t
634
635
where ri,t is the residual for individual i in year t from a van Bertalanffy growth model fit to all data
636
combined, f is a smooth function of day of year and s is a smooth function of latitude and longitude
637
(spatial smooth) as above. The (fixed) year-specific intercepts (Yeart) were used as a measure of condition
638
(size anomaly) in year t. Age data were available through 2010 for walleye pollock and through 2009 for
639
other species.
640
641
Groundfish & crab: Species richness and Diversity
642
Description: See chapter 4.
643
644
Groundfish & crab: Species richness and Diversity
645
Description: See chapter 4.
646
647
648
22
649
Seabirds and Mammals: Abundance and productivity
650
Description: Measure of abundance, productivity (chicks fledged / nest start), overall fledging success
651
(chicks fledged / nest start), fledging success at index sites and breeding phenology (median hatch date)
652
for red-legged kittiwakes, black-legged kittiwakes, common murres and thick-billed murres at both St.
653
Paul Island and St. George Island were compiled from US Fish & Wildlife Service data (Heather Renner,
654
pers. comm.) as described in Chapter 2. In addition, aggregate indices of productivity were constructed
655
as follows:
656

657
658
kittiwake productivity indices (2 species at 2 islands)

659
660
An overall kittiwake productivity index was constructed by averaging four standardized
An overall murre productivity index was constructed by averaging four standardized murre
productivity indices (2 species at 2 islands)

The first and second principal components of all 8 seabird productivity time series were
661
computes as a measure of overall productivity. The first PC primarily reflects the overall success
662
of kittiwakes, while the second PC contrasts kittiwake productivity and murre productivity.
663

The first and second principal components of all 8 seabird phenology time series were computed
664
as a measure of overall breeding phenology. The first PC primarily reflects a trend in the timing
665
of breeding of kittiwakes and the second PC primarily reflects timing of breeding of murres.
666
Finally, measures of fur seal productivity were obtained as the estimated number of pups born on St. Pau
667
and St. George Island. These show a strong decreasing trend over time and the indices were detrended for
668
analysis (Chapter 2).
669
670
Literature cited
671
Acker, J.G., and Leptoukh, G. 2007. Online analysis enhances use of NASA Earth science data. Eos,
672
673
Transactions of the American Geophysical Union 88(2): 14.
Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L, Iredell M, Saha S, White G,
674
Woollen J, Zhu Y, Leetmaa A, Reynolds B, Chelliah M, Ebisuzaki W, Higgins W, Janowiak J,
675
Mo KC, Ropelewski C, Wang J, Jenne R, Joseph D (1996) The NCEP/NCAR 40-Year
676
Reanalysis Project. Bulletin of the American Meteorological Society 77:437-472.
677
Mantua, N.J. and S.R. Hare, Y. Zhang, J.M. Wallace, and R.C. Francis, 1997: A Pacific interdecadal
678
climate oscillation with impacts on salmon production. Bulletin of the American Meteorological
679
Society, 78, pp. 1069-1079.
23
680
Sigler, M.F., Stabeno, P.J., Eisner, L.B., Napp, J.M., and Mueter, F.J. 2014. Spring and fall phytoplankton
681
blooms in a productive subarctic ecosystem, the eastern Bering Sea, during 1995–2011. Deep Sea
682
Research Part II.
683
684
Zhang, Y., J.M. Wallace, D.S. Battisti, 1997: ENSO-like interdecadal variability: 1900-93. J. Climate, 10,
1004-1020.
24
61
60
59
58
55
56
57
LATITUDE
685
-175
-170
-165
-160
686
Figure 1.1: Map of southeastern Bering Sea showing trawl survey stations (circles) and 2˚ by 2˚ grid for
687
NOAA extended reconstructed SST. Red dots denote center of grid cells used in analysis.
688
689
690
Figure1.2: Standardized sea-surface temperature anomalies by year and month, January 1900-December
691
2009, and late summer (July-September) means, 1900-2009.
25
692
693
Figure 1.3: Estimated periods with above 4˚C sea-surface temperatures by year with linear trends (blue
694
lines) of first and last day with SST above 4˚C, 1900-2010. Red lines denote average begin and end date
695
of above 4˚C period. Dots denote mid-points of range and black line denotes linear trend in mid-points
696
26
697
698
Figure 1.4: Summer season length (defined as number of days with SST above 4˚C) from 1900 to 2010
699
with linear trend line.
700
701
Figure1.5: Degree-days during summer period (defined as days with SSTs above 4˚C) with linear trend
702
line (linear regression with AR1 error, t=5.26, p < 0.001).
27
-0.55
-0.60
-0.65
y
703
-0.05
0.00
0.05
704
Figure 1.6: Grid of equally-spaced stations (red dots) that were used to estimate fraction of survey area
705
below 2 ˚C (see text). Black circles denote 2008 survey stations mapped using Albers equal-area
706
projection.
x
707
708
Figure 1.7: Region of the Eastern Bering Sea over which monthly Chlorophyll a concentrations and
709
primary productivity estimates were averaged for analysis (Example shows August 2007 primary
710
production estimates).
711
28
712
Chapter 2: Patterns of covariation among fish, seabirds, and marine mammals in the eastern
713
Bering Sea reflect bottom-up controls and spatial scales of distribution
714
715
Franz J. Mueter1, Heather M. Renner2, Gordon H. Kruse1
716
717
1
718
Rd., Juneau, AK 99801, USA
719
2
720
Homer, AK 99603, USA
School of Fisheries and Ocean Sciences, University of Alaska, 315 Lena Point, 17101 Pt. Lena Loop
Alaska Maritime National Wildlife Refuge, US Fish and Wildlife Service, 95 Sterling Hwy, Suite 1,
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
Citation: Mueter, F.J., Renner, H.M., Kruse, G.H. Patterns of covariation among fish, seabirds, and
737
marine mammals in the eastern Bering Sea reflect bottom-up controls and spatial scales of distribution.
738
Prepared for submission to Deep-Sea Research II.
29
739
Abstract
740
A first step in understanding the effects of climate variability on fish and upper trophic level communities
741
in the eastern Bering Sea is a characterization of shared patterns to identify likely drivers of population
742
trends and interactions among species. We examined available time series of abundance, productivity,
743
seabird phenology, and fish condition to identify commonalities and differences among some key species
744
or groups. We confirmed previously reported strong covariation (both positive and negative) among the
745
recruitment series of major fish species, suggesting that about 50% of the overall variability in
746
recruitment of major gadid and flatfish populations is accounted for by a single shared trend that contrasts
747
recruitment variability in gadids with that in flatfish. Seabirds showed high variability in productivity, but
748
surface-foraging kittiwakes (black-legged, Rissa tridactyla, and red-legged, R. brevirostris) showed
749
distinctly different patterns from diving murres (common, Uria aalge, and thick-billed, U. lomvia), while
750
abundance trends of all four species differed primarily between St. George and St. Paul Island, as
751
previously reported. We found no significant covariation, positive or negative, between seabird and fish
752
populations in either measures of productivity or abundance. Fur seals showed a well-documented decline
753
in pup counts that was not related to variability in either seabird or fish productivity at any lag. However,
754
de-trended fur seal pup counts were significantly related to shared recruitment variability of fish
755
populations during the same year. Our results suggest that conditions that result in successful gadid
756
recruitment (and poor flatfish recruitment) are associated with higher fur seal productivity, but are
757
unrelated to variability in seabird productivity. We find evidence for bottom-up control of fish
758
recruitment and growth related to variability in springtime chlorophyll-a concentrations, winds, and SST,
759
which may be linked through the advection of warmer, nutrient rich waters onto the shelf during winter.
760
Favorable conditions for fish growth may also benefit foraging fur seals, thereby enhancing pup
761
production.
762
30
763
1. Introduction
764
The eastern Bering Sea is a highly productive subarctic shelf that supports large commercial fisheries as
765
well as large seabird and marine mammal populations. The productivity of both fish populations (Mueter
766
et al., 2007) and seabird populations (Byrd et al., 2008a; Renner et al., 2012) on the eastern Bering Sea
767
shelf display some degree of synchronicity over at least three decades that may reflect shared
768
environmental influences or trophodynamic relationships. For example, Mueter et al (2007) showed that
769
there was both significant positive and negative covariation in recruitment among certain groups of
770
groundfish species, but no evidence of covariation between demersal and pelagic species. Recruitment
771
variability for individual stocks was weakly correlated with environmental drivers, but at an aggregate
772
level productivity varied on decadal scales corresponding to well-known climate regime shifts. Similarly,
773
interannual variations in productivity of four species of murres and kittiwakes showed some degree of
774
synchronicity within species and between colonies on St. George and St. Paul Island(Byrd et al., 2008a),
775
while abundance trends differed primarily between islands with long-term declines at St. Paul and stable
776
or fluctuating trends at St. George (Byrd et al., 2008b). The abundance and productivity of Northern fur
777
seals on both St. George and St. Paul Island have decreased since the mid-1970s, but productivity has
778
fluctuated on shorter time scales around a linearly decreasing trend (Towell et al., 2006). The extent to
779
which populations across these groups co-vary and how variability in the productivity of different groups
780
relates to lower trophic level variability is not known. Nor is it clear to what extent these patterns are
781
driven by bottom-up or top-down processes.
782
To the extent that synchronous variations in productivity among populations reflect shared environmental
783
drivers, such synchronicity may be more apparent during multi-year periods of anomalous conditions.
784
Environmental conditions in the Bering Sea have been characterized by high interannual variability, but
785
the eastern Bering Sea experienced two multi-year periods of contrasting conditions in the most recent
786
decade. A prolonged warm period with winter winds from the Southeast, little ice on the southeastern
787
shelf, and an early ice retreat from 2001 to 2005 (Danielson et al., 2011; Stabeno et al., 2012) was
31
788
followed by an extended cold period beginning in 2006 and continuing through at least 2013. Wind
789
speed and direction over the shelf, which are linked to the strength and position of the Aleutian Low,
790
determine the extent and duration of sea ice during winter (Overland and Pease, 1982) and play an
791
important role in alongshore and cross-shelf advection (Bond et al., 1994; Danielson et al., 2011). These
792
conditions in turn affect the success of fish and seabird populations. Advection has been shown to affect
793
the productivity of several flatfish species, whose recruitment is enhanced when winds favor onshelf
794
transport (Wilderbuer et al., 2013; Wilderbuer et al., 2002). Temperature and ice conditions on the shelf
795
are associated with changes in the distribution (Smart et al., 2012) and in the survival of early stages of
796
walleye pollock (Hunt et al., 2011), whose overwinter survival from age-0 to age-1 was markedly reduced
797
during the warm period from 2001 to 2005. Poor overwinter survival of pollock from 2001 to 2005 has
798
been linked to the absence of large zooplankton prey on the shelf during the warm period (Coyle et al.,
799
2011; Hunt et al., 2011; Mueter et al., 2011), which resulted in poor condition of age-0 pollock prior to
800
the winter season (Heintz et al., 2013). The productivity of murres and kittiwakes at the Pribilof Islands
801
has also been linked to sea-surface temperature and ice conditions(Byrd et al., 2008b), but recent work
802
suggests that reproductive success is more closely linked to conditions in prior years (Zador et al., 2013)
803
and to adult condition carried over from previous years (Renner et al., In review). Marked differences in
804
temperature and ice conditions, as well as in wind forcing, over the past decade offer an opportunity to
805
compare and contrast conditions at multiple trophic levels across contrasting regimes.
806
While the recent period has highlighted contrasts in the response of some populations to environmental
807
variability, longer time series are required to quantify such responses, to identify synchronicity in the
808
responses, and to provide a context for how productivity at upper trophic levels has changed over time.
809
Therefore, in addition to comparing two contrasting periods, we examine long-term variability and trends
810
in the productivity of fish, seabird, and marine mammal communities on the eastern Bering Sea shelf
811
from the mid-1970s to the present. We identify linkages across groups and relate variability in these
32
812
upper trophic level groups to temperature and wind forcing and to more recent variability in satellite-
813
derived Chlorophyll-a (chl-a) concentrations.
814
815
2. Methods
816
Data
817
We compiled time series of abundance or biomass, productivity, and condition for major commercial fish
818
and shellfish species on the southeastern Bering Sea shelf and for seabirds and fur seals at the Pribilof
819
Islands (Table 1). As measures of productivity we used different measures for fish, seabirds, and fur seals.
820
The estimated number of recruits at ages ranging from approximately 2 to 6 years, lagged back to
821
correspond to brood year was used to measure the productivity of fish and crab populations. For seabirds
822
we used annual average reproductive success from nest building (kittiwakes) or egg laying (murres) to
823
fledging, and for fur seals we used detrended pup counts.
824
Time series of total biomass and recruitment were obtained from recent stock assessments (NPFMC,
825
2012a, b) for two gadid stocks (walleye pollock, Theragra chalcogramma, and Pacific cod, Gadus
826
morhua), four flatfish stocks (Northern rock sole, Lepidopsetta polyxystra, yellowfin sole, Limanda
827
aspera, flathead sole, Hippoglossoides elassodon, and arrowtooth flounder, Atheresthes stomias), and two
828
crab stocks (Bristol Bay red king crab, Paralithodes camtschaticus, and snow crab, Chionoecetes opilio).
829
Biomass time series for all stocks were available from 1979-2012 and reliable recruitment estimates for
830
all stocks were available for 1977-2005. Longer time series were available for the two gadid stocks and
831
were used to characterize combined variability in these species. Length and age data to compute indices
832
of condition were available for 1999-2012 for walleye pollock, Pacific cod, yellowfin sole, and flathead
833
sole and were obtained from the Alaska Fisheries Science Center RACE Division (Robert Lauth, AFSC,
834
Seattle, pers. comm.).
33
835
Indices of abundance and reproductive success for black-legged kittiwakes (Rissa tridactyla), red-legged
836
kittiwakes (R. brevirostris), thick-billed murres (Uria lomvia), and common murres (U. aalge) at St.
837
George island and at St. Paul island were obtained from Klostermann et al. (2011) and Thomson and
838
Drummond (2011), respectively . Since not all index plots that contribute to the overall abundance index
839
were sampled in 1982 and 1984 at St. Paul or in 1984 and 1985 at St. George, values for these years were
840
estimated by a linear regression on counts obtained at a slightly restricted number of plots that were
841
sampled in all years. Two indices of seabird reproductive success were considered in the analysis.
842
Fledging success (number of chicks fledged per chick hatched) was used to examine variability during the
843
chick rearing period, while an overall index of reproductive success was based on the number of chicks
844
fledged per nest start for kittiwakes and the number of successful nests (at least one chick fledged) per
845
nest site with an egg for murres (Renner et al., In review).
846
The Eastern Pacific stock of northern fur seals (Callorhinus ursinus) is primarily found on St. Paul and St.
847
George Islands during the summer months and we used the estimated numbers of northern fur seal pups at
848
each of these islands as indices of fur seal abundance (Allen and Angliss (2010); Rod Towell, NMML,
849
NOAA, Seattle, pers. comm.), assuming the number of pups is approximately proportional to adult
850
abundance. Summer pup counts were available annually from 1972-1990 and every other year thereafter.
851
The fur seal population has declined with a strong near-linear trend since 1972, hence we removed the
852
linearly decreasing trend from the time series and used the resulting anomalies as an index of shorter-term
853
variability in pup production for comparisons with fish and seabird productivity series. Although seabird
854
time series also had trends, these were not consistent between islands and species and detrending had little
855
effect on results, hence we used the actual series.
856
As a measure of lower trophic level variability, we quantified variability in chl-a concentrations over the
857
southeastern Bering Sea shelf based on satellite-based measurements. We retrieved eight-day composite
858
Level-3 SeaWiFS and MODIS-Aqua chl-a data at 9-km spatial resolution using the Giovanni online data
859
system, developed and maintained by the NASA Goddard Earth Sciences Data and Information Services
34
860
Center (Acker and Leptoukh, 2007). Data for a rectangular region (54–60°N and 157–180°W) that
861
encompasses the southeastern Bering Sea shelf were extracted from the global coverage datasets (units:
862
mg/m3). Open ocean values (offshore of the 500m isobaths) and nearshore values (inshore of 10m
863
isobaths) were excluded. We compared eight full years of overlap between the two datasets (2003-2010)
864
based on correlations and absolute differences (bias) of the logarithmically transformed data on a pixel-
865
by-pixel basis. Estimates from the two satellites were strongly correlated (r=0.65) and bias was found to
866
be negligible (median bias < 0.01, 88% of 1.4 million paired observations had absolute bias < 0.05). In
867
order to produce the longest possible continuous time series of remotely-sensed chl-a data, we combined
868
SeaWiFS data from 1998-2002 with MODIS-Aqua data from 2003-2012. Variability in chl-a standing
869
stocks over the shelf south of 60˚N was summarized by computing mean chl-a levels for three regions
870
(inner shelf: 10-50 m; middle shelf: 50 – 100 m; outer shelf: 100-500 m) and two time periods (spring
871
bloom period: approximately March - June; late summer / fall bloom period: approximately July -
872
October). Separate indices were developed because of pronounced differences in the magnitude and
873
timing of the spring and fall blooms (Fig. 1). In addition, to capture overall variability in chl-a over the
874
southeast Bering Sea shelf, we obtained a combined annual chl-a index by computing the first Principle
875
Component from a Principal Components Analysis (PCA) of the six individual chl-a time series. None of
876
the other Principal Components contained interpretable information based on the broken-stick model
877
(Jackson, 1993).
878
To relate biological variability to potential physical drivers, we focused on two measures of
879
environmental variability: annual mean sea-surface temperatures as a measure of surface layer conditions
880
and mean wind direction over the shelf as a measure of advective forcing during the previous winter.
881
Temperature indices were based on monthly extended reconstructed SSTs (ERSSTv3) (Smith et al., 2008)
882
, which are interpolated values on a 2° latitude by 2° longitude grid, and were averaged over the southeast
883
Bering Sea shelf inshore of the shelf break and extending to 61°N for this analysis. Monthly ERSST data
884
were averaged by year to obtain a measure of annual mean SST. Wind direction was taken from the
35
885
NCEP North American Reanalysis Project 1 (Kalnay et al., 1996) and was computed as the mean
886
October-April wind direction computed at 60°N, 170°W following Danielson et al (2011). Winds at this
887
location are broadly representative of the wind field over much of the eastern Bering Sea shelf because
888
winds at a series of moorings along the 70 m isobaths on the eastern Bering Sea shelf tend to show strong
889
coherence (Stabeno et al., 2010). The mean wind direction at this location is toward 225 °N,
890
corresponding to cross-shelf winds from the Northeast, and anomalies from 225 °N are a proxy for cross-
891
shelf Ekman transport with positive anomalies (> 225°N) corresponding to enhanced on-shelf flow over
892
the southeast Bering Sea shelf (Danielson et al 2011).
893
Data analyses
894
To identify patterns of covariation among fish, seabird and mammal populations in the eastern Bering Sea
895
we examined variability in time series of abundance and productivity. Different methods were used for
896
abundance and productivity data because abundance time series display a high degree of autocorrelation,
897
while productivity varies on interannual time scales with little or limited autocorrelation. Major trends in
898
population abundances were identified using a dynamic factor analysis (DFA, Zuur et al., 2003) of the
899
available abundance series. Available time series from 1977 to 2012 were modeled as a linear
900
combination of one of more underlying common trends (or factors), Z:
901
Yt  ΑZ t  ε t
902
where Yt is a k x 1 column vector with the observed abundances of k populations in year t, Zt is a q x 1
903
column vector representing the q common factors (q << k), A is a k x q matrix of factor loadings, and εt is
904
assumed to be normally distributed with general non-diagonal error matrix V (εt ~ N(0, 2V)). For
905
example, a model with two common trends decomposes each of the k time series into a linear
906
combination of two underlying trends and residual variability around the estimated trend:
907
36
908
 Y1,t   a1,1
  
 Y2,t   a1, 2
 
   
 Y   a1,k
 k ,t  
a2,1 
  1,t 
 

a2, 2  z1,t    2,t 
  
  z 2,t   
 
a2,k 
 k ,t 
909
where both the trends (zi,t) as well as the factor loadings (ai,k) are parameters to be estimated via maximum
910
likelihood. All data series were normalized prior to analysis, so the magnitude of factor loadings is
911
directly comparable across populations to assess the relative contribution of each individual trend to the
912
overall fitted trends. We fit models with one to three common trends and selected the best model based on
913
the small-sample Akaike Information Criterion (AICc) (Akaike, 1973). We were unable to fit trends to all
914
series simultaneously, hence we fit separate trends to seabird abundance series.
915
916
Synchronicities and differences in productivity time series were identified using a cluster analysis of all
917
available productivity time series, using either fledging success or overall reproductive success for
918
seabirds. Because the 1976/77 climate regime shift was associated with a pronounced change in the
919
productivity of numerous fish species (Mueter et al., 2007) and because several series only became
920
available after the shift, we restricted our analyses to the period from 1977 to 2011. All fish recruitment
921
series were strongly right-skewed and were log-transformed prior to analysis, resulting in approximate
922
normality (Shapiro-Wilks test, p > 0.1, except yellowfin sole). Time series of fledging success and overall
923
reproductive success of seabirds were either close to normal or could not be transformed to normality
924
through a simple transformation. Both time series of (detrended) fur seal productivity were approximately
925
normally distributed. While normality is not assumed in the cluster analysis or when computing
926
correlations, transformations to normality reduced the influence of more extreme values and resulted in
927
more readily interpretable patterns. All series were standardized to have mean zero and standard deviation
928
one prior to clustering, where the mean and standard deviation were computed over all available years
929
since 1977. Groupings were identified based on two approaches: one approach used only years with non37
930
missing data (11 years) and a second approach used all available data (ranging from 24 years of data for
931
fur seals to 34 years for the two gadid species). We used a hierarchical cluster analysis based on pairwise
932
Euclidean distances between the standardized time series to identify possible groupings that minimize
933
within-cluster variance via Ward's method (Gordon, 1999). Where missing values were present,
934
Euclidean distances were computed based on the available years of data scaled up proportionally to the
935
number of years with data.
936
Based on results from the cluster analyses, we examined trends and variability in the productivity of
937
individual groups both graphically and statistically. In particular, recruitment trends for gadid and flatfish
938
species were examined and trends in seabird productivity, which were largely uncorrelated with trends in
939
fish recruitment, were compared between the two species and islands. Major modes of variability across
940
productivity time series within selected groups that showed evidence of positive or negative covariation
941
were examined using a Principal Component Analysis (PCA).
942
In addition to abundance and productivity time series, we examined annual condition indices based on
943
weight-at-length anomalies for four fish species: walleye pollock, Pacific cod, yellowfin sole, and
944
flathead sole. We first estimated annual anomalies (1999-2012) for each species by fitting linear mixed
945
effects models of log(Weight) as a simple linear function of log(Length) but with random deviations for
946
the intercept and slope among years. The predicted average weight at the smallest observed length in a
947
given year was then used as an index of condition for each year from 1999-2012. Possible synchronicity
948
among species was examined using a simple correlation analysis. Because of strong positive covariation
949
in condition among species, we computed an overall annual condition index by averaging normalized
950
indices across species.
951
The observed trends and patterns of variability in productivity of upper trophic level groups were
952
compared to variability in the environment and variability at lower tropic levels. First, we examined
953
differences in physical conditions and in productivity between recent warm and cold years. Specifically,
38
954
we compared temperature and ice conditions, Chlorophyll-a concentrations, zooplankton abundances, as
955
well as fish, seabird, and mammal productivity during the 2001-2005 warm period to conditions in
956
subsequent cold years (2007-2012, where available). Second, we related variability in fish, seabird, and
957
mammal productivity to variability in chl-a and environmental drivers using correlation analyses as well
958
as simple and multiple linear regressions. Finally, we modeled mean chl-a concentrations from 1998-
959
2012, in particular on the middle shelf, as a function of sea-surface temperature and wind conditions using
960
multiple linear regression analysis. Because autocorrelation was present in many of the time series,
961
correlation analyses were adjusted for autocorrelation using the modified Chelton method of Pyper and
962
Peterman (1998). If regression residuals showed significant autocorrelation (Durbin-Watson test, p <
963
0.05), we used generalized least-squares regression with a first-order autoregressive correlation structure.
964
All analyses were conducted using R version 2.15.1 (R Core Team, 2012).
965
3. Results
966
The Dynamic Factor Analysis revealed two significant shared trends in the fish and fur seal survey time
967
series, both of which suggest a discontinuity in abundance trends in the early to mid-1990s. Models with
968
only one or with more than two trends had AICc values that were at least 8.5 penalized likelihood units
969
larger than the model with 2 trends. The first trend reflects a decreasing trend in gadid, snow crab, and fur
970
seal abundances beginning in the early 1990s (Trend 1 in Fig. 2) and a concomitant increase in red king
971
crab survey biomass in Bristol Bay (negative loading). There was some indication that this trend reversed
972
after 2008. A second trend was associated with increases in three flatfish species (arrowthooth flounder,
973
rock sole and flathead sole) from at least 1982 through the mid-1990s and less pronounced decreases in
974
several other species. Species that were associated with both trends (St. George fur seals and cod) showed
975
a continuous decrease over the entire time period. The two shared trends capture much of the variability
976
in abundance of the component species although several species show short term variability that was not
977
captured by these trends (Fig. 2).
39
978
The best model for seabird abundances had a single significant trend (delta AICc = 14.0) reflecting a near-
979
continuous decreasing trend for all four seabird species at St. Paul Island (Fig. 3). In contrast, abundances
980
at St. George Island increased for all four species except common murres. The increases at St. George
981
Island began around 1990 and followed initial declines, in particular for red-legged kittiwakes. Inverse
982
trends in abundance at the two islands were particularly pronounced for thick-billed murres .
983
Time trends in productivity showed distinct differences among certain groups of species (Fig. 4). These
984
groupings were identical regardless of whether years with any missing values were deleted or not.
985
Moreover, patterns were very similar using either fledging success or overall reproductive success as a
986
measure of seabird productivity, hence we show only results based on the latter and using all available
987
data. Four distinct clusters with strong positive within-group correlations were identified and can be
988
characterized as a kittiwake group, a murre group, a flatfish group, and a gadid group. Reproductive
989
success of murres and kittiwakes was largely uncorrelated or weakly correlated with the productivity of
990
fish and shellfish in the flatfish and gadid groups. Species within the gadid group showed strong positive
991
covariation in productivity, but the productivity of all species in this group was negatively correlated with
992
the productivity of the four flatfish species, indicating pronounced asynchrony in the productivity of
993
gadids and flatfishes (Fig. 5). While snow crab clustered with the flatfish group, its recruitment was only
994
weakly correlated with either flatfish or gadid recruitment and was negatively correlated with flathead
995
sole recruitment (r = -0.223, p = 0.229). Anomalies in fur seal pup counts at St. Paul were most strongly
996
correlated with kittiwake reproductive success, while fur seals at St. George clustered with the gadid
997
group (Fig. 4, 6).
998
Based on the observed patterns of covariation, we developed three separate indices of variability for
999
kittiwakes, murres, and other species, respectively. An overall index of kittiwake reproductive success
1000
was constructed by computing normalized anomalies of each of the four kittiwake time series and
1001
averaging anomalies for each year (Fig. 5). Similarly, the four murre time series were combined into an
1002
overall index of murre productivity.
40
1003
Because the flatfish and gadid groups showed both positive and negative covariation, we summarized
1004
patterns of variability using a PCA of productivity time series, excluding the incomplete time series for
1005
St. George fur seals. The first principal component captured 40% of the overall variability in the log-
1006
recruitment of 8 major fish and shellfish species (Fig. 7). This mode was positively associated with the
1007
recruitment of pollock, cod, and king crab and was negatively associated with the recruitment of all four
1008
flatfish species, further supporting opposite patterns of variability in the two groups. Snow crab was
1009
uncorrelated with the first PC. The PC showed high interannual variability with a period of above average
1010
values, indicating good gadid and poor flatfish recruitment, in the late 1970, and two periods of below
1011
average values, associated with poor gadid and good flatfish recruitment, in the mid 1980s and from 2001
1012
through 2005. The first PC was strongly and significantly correlated with overall chl-a trends (N = 8
1013
years, r = 0.78, p = 0.037), in particular with spring time chl-a (inner shelf: r = 0.94, p = 0.001; middle
1014
shelf: r = 0.91, p = 0.013; outer shelf: r = 0.78, p =0.038).
1015
The strong pattern of asynchrony between gadid and flatfish species is readily apparent in time series of
1016
combined recruitment anomalies (averaged across fish species within each of the two groups), which are
1017
strongly negatively correlated (r = -0.67, p < 0.0001, Figs. 5, 8). Indeed, there has not been a single year
1018
in the 1977-2005 time series where both gadid and flatfish recruitment were above average (Fig. 8), a
1019
highly unlikely pattern based on a randomization test that simulated time series of the same length and
1020
with the same covariance structure as the observed time series (p = 0.0083). In contrast, there have been
1021
several years, in particular 1988, 1993, '94, '97, and 2004, where both groups experienced poor
1022
recruitment.
1023
1024
Annual condition indices (anomalies in weight at a given size) for pollock, cod, yellowfin sole, and
1025
flathead sole (1999-2008) were significantly and positively correlated with each other (Table 2).
1026
Moreover, each of the indices was strongly and positively correlated with summer bottom temperatures
41
1027
measured during the summer bottom trawl survey (r = 0.93, 0.62, 0.68, 0.34 for pollock, cod, yellowfin,
1028
and flathead sole, respectively) and with annual SST (Table 3). The overall combined condition index
1029
was positively correlated with both SST (r = 0.81, p = 0.0046) and, to a lesser extent, with the overall chl-
1030
a index (r = 0.67, p = 0.070), but was strongly correlated with spring time chl-a on the inner (r = 0.84, p =
1031
0.0093) and middle shelf (r = 0.69, r = 0.087, Fig. 5). However, the effects of SST and chl-a are difficult
1032
to separate with the available time series because of the strong confounding between the two variables (r
1033
=0.64).
1034
Spring chl-a on the inner and middle shelf was positively and significantly correlated with SST, while
1035
wind direction was positively correlated with spring chl-a on the middle and outer shelf only (Table 3).
1036
Correlations with fall chl-a were much weaker and only fall chl-a on the middle shelf was significantly
1037
correlated with wind direction (r = 0.660). In general, productivity and condition indices were much more
1038
strongly correlated with SST and with spring chl-a concentrations than with wind direction and fall chl-a.
1039
However, chl-a was significantly and positively correlated with both SST and wind direction. A multiple
1040
linear regression of chl-a on SST and wind direction was highly significant and dropping either SST or
1041
wind from the model resulted in a poorer model fit with a higher AIC, which implies that higher chl-a
1042
levels are associated with warmer surface temperatures and more northeasterly to easterly winds during
1043
the previous winter (Fig. 9).
1044
Most of the variables examined here and other environmental and biological variables show a marked
1045
contrast between the recent warm and cold period as summarized in Table 4. The warm period was
1046
characterized by little or no ice on the SE Bering Sea shelf and an early ice retreat if ice was present.
1047
Winds were generally to the northwest during the warm period, favoring onshore transport; however,
1048
there was no apparent relationship between SST and wind direction over the longer term (1980-2010).
1049
Chlorophyll-a concentrations were higher during the warm years, particularly in the spring as well as
1050
during fall on the middle shelf. These conditions were associated with high abundances of small
1051
zooplankton but low abundances of large zooplankton such as Calanus marshallae that provide important
42
1052
food for upper trophic level predators such as pollock (Table 4). Consequently, pollock were in poor
1053
energetic condition in the warm years and the recruitment of the 2001-2005 gadid year classes were very
1054
poor. In contrast, flatfish recruitment was relatively high in the warm years. The condition of older fishes,
1055
both gadids and flatfishes, was high in the warm years and low in the cold years. Finally, seabird
1056
productivity was higher in the warm period, particularly for kittiwakes, but decreased prior to the end of
1057
the warm period (Fig. 5) and there was no significant correlation with SST over the full period of record
1058
(Table 3).
1059
4. Discussion
1060
We documented both synchronous and asynchronous fluctuations among time series of fish and shellfish
1061
productivity on the eastern Bering Sea shelf, as well as between fish productivity and detrended northern
1062
fur seal pup counts. In general, there was stronger synchrony within taxonomic groups (gadids and
1063
flatfish, respectively), with weak or opposite synchrony among groups. However, productivity and
1064
abundance of two crab species, snow crab and red king crab, tended to vary asynchronously. In contrast,
1065
fluctuations in the productivity of central-place foraging seabirds were not related to the productivity of
1066
fishes, probably reflecting differences in the spatial scales of foraging.
1067
The reasons for the observed negative covariation between recruitment fluctuations of gadids and
1068
flatfishes remain poorly understood. They could reflect a possible trade-off between the two groups,
1069
possibly related to different pathways of production (benthic versus pelagic) under different climate
1070
conditions. However, variability in benthic-pelagic coupling is generally assumed to favor the pelagic
1071
environment during warm years and the benthic environment during cold years (Hunt and Stabeno, 2002;
1072
Walsh and Peter McRoy, 1986). However, the larvae of both gadids and flatfish share similar pelagic
1073
environments during the spring and early summer. Moreover, while juvenile flatfish may benefit from
1074
enhanced fluxes to the benthos after settlement, our results suggest that flatfish survival was enhanced
43
1075
during recent warm years (Fig. 5), contrary to the hypothesis that less production reaches the benthos
1076
during these years.
1077
An alternative and more likely explanation is that flatfish and gadid recruitment is related to different
1078
mechanisms that are associated with different but correlated environmental drivers. In fact, mechanisms
1079
have been established linking reduced walleye pollock recruitment to very high temperatures on the shelf
1080
(Hunt et al., 2011; Mueter et al., 2011), while the recruitment of several flatfish species is enhanced by
1081
winds that favor stronger on-shelf advection (Wilderbuer et al., 2013; Wilderbuer et al., 2002). While
1082
these relationships reflect very different mechanisms, enhanced onshelf transport (favoring flatfish
1083
recruitment) tends to be associated with warmer temperatures on the shelf (reducing pollock recruitment).
1084
While we did not thoroughly examine potential environmental drivers that may be related to variability in
1085
productivity measures, we find evidence of bottom-up forcing connecting variability in winds and SST to
1086
chlorophyll-a concentrations, particularly during spring, and to measures of fish recruitment and
1087
condition, but no or weaker correlations with measures of seabird productivity. We hypothesize that
1088
winds from the east and northeast during winter, which are associated with enhanced onshelf flows
1089
(Danielson et al., 2012), promote the transport of warmer, nutrient-rich waters onto the shelf, thereby
1090
increasing primary production in the spring as evident in higher chlorophyll-a concentrations. This
1091
enhanced spring production is associated with above-average body condition of all fishes, reflecting good
1092
feeding conditions, as well as enhanced flatfish recruitment associated with stronger onshelf transport.
1093
However, larval walleye pollock do not benefit from the higher spring production due to a lack of large
1094
zooplankton prey in the fall during warm years (Coyle et al., 2011; Hunt et al., 2011).
1095
1096
5. Acknowledgements
1097
NOAA ERSST v3b data and NCEP Reanalysis data were provided by the NOAA/OAR/ESRL PSD,
1098
Boulder, Colorado, USA, from their website at http://www.esrl.noaa.gov/psd/. We thank our many
44
1099
colleagues in the BEST-BSIERP Project, which is supported by NSF and NPRB, for many valuable
1100
discussions and access to data and manuscripts. This is BEST-BSIERP contribution number XXX.
1101
1102
6. References
1103
Acker, J.G., Leptoukh, G., 2007. Online analysis enhances use of NASA Earth science data. Eos,
1104
1105
Transactions of the American Geophysical Union 88, 14.
Akaike, H., 1973. Information theory as an extension of the maximum likelihood principle, in: Petrov,
1106
B.N., Csaki, F. (Eds.), Second International Symposium on Information Theory. Akademiai Kiado,
1107
Budapest.
1108
1109
1110
1111
1112
Allen, B.M., Angliss, R.P., 2010. Northern fur seal (Callorhinus ursinus): Eastern Pacific Stock. Alaska
Marine Mammal Stock Assessments, 2010 NOAA-TM-AFSC-223, 23-31.
Bond, N.A., Overland, J.E., Turet, P., 1994. Spatial and temporal characteristics of the wind forcing of the
Bering Sea. Journal of Climate 7, 1119-1130.
Byrd, G.V., Schmutz, J.A., Renner, H.M., 2008a. Contrasting population trends of piscivorous seabirds in
1113
the Pribilof Islands: A 30-year perspective. Deep Sea Research Part II: Topical Studies in
1114
Oceanography 55, 1846-1855.
1115
Byrd, G.V., Sydeman, W.J., Renner, H.M., Minobe, S., 2008b. Responses of piscivorous seabirds at the
1116
Pribilof Islands to ocean climate. Deep Sea Research Part II: Topical Studies in Oceanography 55,
1117
1856-1867.
1118
Coyle, K.O., Eisner, L.B., Mueter, F.J., Pinchuk, A., Janout, M.A., Cieciel, K., Farley, E.V., Andrews,
1119
A.G., 2011. Climate change in the southeastern Bering Sea: impacts on pollock stocks and
1120
implications for the Oscillating Control Hypothesis. Fisheries Oceanography 20, 139-156.
1121
Danielson, S., Eisner, L., Weingartner, T., Aagaard, K., 2011. Thermal and haline variability over the
1122
central Bering Sea shelf: Seasonal and interannual perspectives. Continental Shelf Research 31, 539-
1123
554.
45
1124
1125
Danielson, S., Hedstrom, K., Aagaard, K., Weingartner, T., Curchitser, E., 2012. Wind-induced
reorganization of the Bering shelf circulation. Geophysical Research Letters 39.
1126
Gordon, A.D., 1999. Classification. Chapman and Hall / CRC, London.
1127
Heintz, R.A., Siddon, E.C., Farley, E.V., Napp, J.M., 2013. Correlation between recruitment and fall
1128
condition of age-0 pollock (Theragra chalcogramma) from the eastern Bering Sea under varying
1129
climate conditions. Deep Sea Research Part II: Topical Studies in Oceanography 94, 150-156.
1130
Hunt, G.L., Coyle, K.O., Eisner, L.B., Farley, E.V., Heintz, R.A., Mueter, F., Napp, J.M., Overland, J.E.,
1131
Ressler, P.H., Salo, S., Stabeno, P.J., 2011. Climate impacts on eastern Bering Sea foodwebs: a
1132
synthesis of new data and an assessment of the Oscillating Control Hypothesis. ICES Journal of
1133
Marine Science: Journal du Conseil 68, 1230-1243.
1134
1135
1136
1137
1138
Hunt, G.L., Jr., Stabeno, P.J., 2002. Climate change and the control of energy flow in the southeastern
Bering Sea. Progress in Oceanography 55, 5-22.
Jackson, D.A., 1993. Stopping Rules in Principal Components Analysis: A Comparison of Heuristical and
Statistical Approaches. Ecology 74, 2204-2214.
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., Saha, S., White,
1139
G., Woollen, J., Zhu, Y., Leetmaa, A., Reynolds, B., Chelliah, M., Ebisuzaki, W., Higgins, W.,
1140
Janowiak, J., Mo, K.C., Ropelewski, C., Wang, J., Jenne, R., Joseph, D., 1996. The NCEP/NCAR 40-
1141
Year Reanalysis Project. Bulletin of the American Meteorological Society 77, 437-472.
1142
1143
1144
Klostermann, M.R., Scopel, L.C., Drummond, B.A., 2011. Biological monitoring at St. George Island,
Alaska in 2011, U.S. Fish and Wildl. Serv. Rep., AMNWR 2011/11, Homer, Alaska.
Mueter, F.J., Boldt, J., Megrey, B.A., Peterman, R.M., 2007. Recruitment and survival of Northeast
1145
Pacific Ocean fish stocks: temporal trends, covariation, and regime shifts. Canadian Journal of
1146
Fisheries and Aquatic Sciences 64, 911-927.
46
1147
Mueter, F.J., Bond, N.A., Ianelli, J.N., Hollowed, A.B., 2011. Expected declines in recruitment of walleye
1148
pollock (Theragra chalcogramma) in the eastern Bering Sea under future climate change. ICES
1149
Journal of Marine Science: Journal du Conseil 68, 1284-1296.
1150
NPFMC, 2012a. Stock Assessment and Fishery Evaluation report for the groundfish resources of the
1151
Bering Sea/Aleutian Islands regions. North Pacific Fishery Management Council, 605 W. 4th Ave.,
1152
Suite 306, Anchorage, AK 99501.
1153
NPFMC, 2012b. Stock Assessment and Fishery Evaluation Report for the King and Tanner Crab
1154
Fisheries of the Bering Sea and Aleutian Islands Regions. North Pacific Fisheries Management
1155
Council, 605 West 4th Ave., Suite 306, Anchorage, AK 99501.
1156
1157
1158
1159
1160
Overland, J.E., Pease, C.H., 1982. Cyclone climatology of the Bering Sea and its relation to sea ice
extent. Monthly Weather Review 110, 5-13.
R Core Team, 2012. R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Australia.
Renner, H.M., Drummond, B.A., Benson, A.-M., Paredes, R., In review. Reproductive success of
1161
kittiwakes and murres in sequential stages of the nesting period: relationships with diet and
1162
oceanography. Deep Sea Research II.
1163
Renner, H.M., Mueter, F., Drummond, B.A., Warzybok, J.A., Sinclair, E.H., 2012. Patterns of change in
1164
diets of two piscivorous seabird species during 35 years in the Pribilof Islands. Deep Sea Research
1165
Part II: Topical Studies in Oceanography 65-70, 273-291.
1166
Smart, T.I., Duffy-Anderson, J.T., Horne, J.K., Farley, E.V., Wilson, C.D., Napp, J.M., 2012. Influence of
1167
environment on walleye pollock eggs, larvae, and juveniles in the southeastern Bering Sea. Deep Sea
1168
Research Part II: Topical Studies in Oceanography 65–70, 196-207.
1169
Smith, T.M., Reynolds, R.W., Peterson, T.C., Lawrimore, J., 2008. Improvements to NOAA's historical
1170
merged land-ocean surface temperature analysis (1880-2006). Journal of Climate 21, 2283-2296.
47
1171
Stabeno, P., Napp, J., Mordy, C., Whitledge, T., 2010. Factors influencing physical structure and lower
1172
trophic levels of the eastern Bering Sea shelf in 2005: Sea ice, tides and winds. Progress In
1173
Oceanography 85, 180-196.
1174
Stabeno, P.J., Kachel, N.B., Moore, S.E., Napp, J.M., Sigler, M., Yamaguchi, A., Zerbini, A.N., 2012.
1175
Comparison of warm and cold years on the southeastern Bering Sea shelf and some implications for
1176
the ecosystem. Deep Sea Research Part II: Topical Studies in Oceanography 65–70, 31-45.
1177
1178
1179
1180
1181
1182
Thomson, G., Drummond, B., 2011. Biological monitoring at St. Paul Island, Alaska in 2011, U.S. Fish
and Wildl. Serv. Rep., AMNWR 2011/14, Homer, Alaska.
Towell, R.G., Ream, R.R., York, A.E., 2006. Decline in northern fur seal (Callorhinus ursinus) pup
production on the Pribilof Islands. Marine Mammal Science 22, 486-491.
Walsh, J.J., Peter McRoy, C., 1986. Ecosystem analysis in the southeastern Bering sea. Continental Shelf
Research 5, 259-288.
1183
Wilderbuer, T., Stockhausen, W., Bond, N., 2013. Updated analysis of flatfish recruitment response to
1184
climate variability and ocean conditions in the Eastern Bering Sea. Deep Sea Research Part II:
1185
Topical Studies in Oceanography 94, 157-164.
1186
Wilderbuer, T.K., Hollowed, A.B., Ingraham Jr., W.J., Spencer, P.D., Conners, M.E., Bond, N.A.,
1187
Walters, G.E., 2002. Flatfish recruitment response to decadal climatic variability and ocean
1188
conditions in the eastern Bering Sea. Progress in Oceanography 55, 235-247.
1189
Zador, S., Hunt, G.L., TenBrink, T., Aydin, K., 2013. Combined seabird indices show lagged
1190
relationships between environmental conditions and breeding activity. Marine Ecology Progress
1191
Series 485, 245-258.
1192
1193
Zuur, A.F., Tuck, I.D., Bailey, N., 2003. Dynamic factor analysis to estimate common trends in fisheries
time series. canadian Journal of Fisheries and Aquatic Sciences 60, 542–552.
1194
48
1195
Table 1: Time series of abundance, productivity and condition for fish, seabirds and northern fur seals in
1196
the eastern Bering Sea used in analysis with sources. Names of species in parentheses refer to
1197
abbreviations used in text and figures.
Fish
Species
Walleye pollock (pollock), Pacific cod (cod), arrowtooth
flounder (ATF), yellowfin sole (YFS), flathead sole
(FHS), rock sole (RS), snow crab (opilio), and red king
crab (RKC)
Abundance
Biomass series from stock assessments (NPFMC 2011)
Productivity
Estimates of recruitment from stock assessment model
(NPFMC 2012)
Condition
Average annual anomaly in weight-at-size (Data from
NMFS bottom trawl survey, Robert Lauth, AFSC)
Birds
Species
Common (COMU) and thick-billed murre (TBMU),
red-legged (RLKI) and black-legged kittiwakes at St.
Paul Island (St.P.) and at St. George Island (St.G.)
Abundance
Estimated abundance at index sites from Kellermann et al.
(2011) and Thomson and Drummond (2011)
Fledging
Number of chicks fledged per eggs hatched, from Renner
success
et al (In Press)
Reproductive
success
Number of chicks fledged per nest start for kittiwakes;
number of successful nests (at least one chick fledged) per
nest site with an egg for murres
49
Mammals Species
Northern Fur Seals (NFS) on St. Paul and St. George
Island
Productivity
Pup production (estimated number of pups born) (Rod
Towell, NMML, Seattle)
1198
1199
1200
Table 2: Pairwise Pearson's product-moment correlations between annual time series of condition (1999-
1201
2010) for four commercial fish species on the eastern Bering Sea shelf. Correlations exceeding 0.5 were
1202
significant at p = 0.05.
cod
pollock
yellowfin flathead
0.59
cod
yellowfin
0.63
0.48
0.48
0.73
0.65
1203
1204
50
1205
Table 3: Correlations between environmental drivers, chl-a indices in two seasons and three areas (inner,
1206
middle, and outer shelf), an overall chl-a index (PC1) and various measures of fish, seabird and marine
1207
mammal productivity and condition (CI = condition index). See table 1 for species abbreviations. PC1
1208
(fish) refers to the first Principal Component from a PCA of fish and shellfish productivity series as
1209
explained in text. Correlations significant at α=0.05 (adjusted for autocorrelation) are shown in bold and
1210
italics.
chl-a
SST
Spring chl-a
Wind
PC 1
inner
middle
1
0.166
0.639
0.724
0.166
1
0.486
PC1 (fish)
-0.414
-0.259
gadids
-0.383
flatfish
kittiwakes
Fall chl-a
outer
inner
middle
0.644
0.531
0.141
0.214
0.451
0.170
0.657
0.606
-0.279
0.660
0.191
-0.788
-0.944
-0.914
-0.805
-0.431
0.190
-0.297
-0.188
-0.539
-0.492
-0.649
-0.505
-0.234
-0.369
-0.138
0.382
0.303
0.596
0.671
0.796
0.617
0.145
0.121
0.157
0.028
-0.132
0.583
0.203
0.493
0.479
0.293
0.568
0.701
murres
-0.024
0.020
0.193
-0.185
0.081
0.207
0.197
0.372
0.342
StP.NFS
-0.228
-0.205
0.270
-0.227
0.337
0.133
-0.150
0.812
0.125
StG.NFS
-0.243
-0.230
-0.710
-0.106
-0.627
-0.455
-0.395
-0.977
-0.744
CI pollock
0.847
0.339
0.621
0.634
0.719
0.467
0.225
0.237
0.380
CI cod
0.808
0.257
0.691
0.800
0.612
0.592
0.390
-0.018
0.556
CI YFS
0.546
-0.129
0.508
0.733
0.542
0.309
0.608
-0.133
0.165
CI FHS
0.498
-0.192
0.408
0.629
0.425
0.299
0.389
-0.392
0.319
CI (total)
0.809
0.083
0.668
0.838
0.689
0.499
0.483
-0.092
0.426
SST
Wind
1211
1212
51
outer
1213
Table 4: Contrasting environmental and biological conditions during the warm period from 2001 to 2005
1214
and the subsequent cold period from 2007-2010 on the eastern Bering Sea shelf. Cold and warm periods
1215
are as defined by Stabeno et al (2012).
Variable
Warm (2001-2005)
Cold (2007-10)
Source
Ice extent in SE Bering Sea
little / no ice
extensive ice
Stabeno et al (2012)
Ice retreat
early
late
Stabeno et al (2012)
Winds
southeasterly
northwesterly
Danielson et al (2011),
Stabeno et al (2011), this
study
Chl. a - spring
high
low
this study
Chl. a – summer/fall
high
low
Small zooplankton abundance
high
low
Coyle et al. (2011)
Large zooplankton abundance
low
high
Coyle et al. (2011)
age-0 pollock energy density
low
high
Heintz et al. (2013)
gadid recruitment
poor
average to good
NPFMC (2011), this study
flatfish recruitment
high
N/A
NPFMC (2011), this study
fish condition (age 1+)
above average
below average
this study
seabird productivity
higher
lower
Renner et al (in review)
this study
1216
52
1217
Figures
1218
1219
Figure 1: Average seasonal trends in satellite-derived 8-day Chlorophyll-a concentrations for the inner
1220
(blue), middle (green), and outer (red) shelf, 1998-2012, and for 8-day aggregates of estimated chl-a
1221
values from a fluorometer moored at 11 m below the surface over the middle shelf (black). Values for
1222
mooring were adjusted to have the same mean as satellite-based estimates for the middle shelf. Vertical
1223
dashed line denotes separation between spring and summer/fall seasons.
1224
53
(C)
2005
St.P NFS
St.G NFS
RS
FHS
St.G NFS
St.P NFS
RKC
opilio
ATF
pollock
0.4
1985 2000
ATF
YFS
RS
FHS
opilio
RKC
St.G NFS
1995
0.10
-0.10
-0.2 0.1
Trend 1
1985
Normalized trend
Loadings 1
-4 -2
0
2
Trend 2
Loadings 2
4
(B)
-6
Contribution to fitted value
(A)
2
1
0
-1
-2
opilio
RKC
2
1
0
-1
-2
pollock
cod
ATF
YFS
2
1
0
-1
-2
1985 2000
1985 2000
1225
1226
Figure 2: Common trends from a dynamic factor analysis of eight fish species and two fur seal
1227
populations based on time series of survey biomass and pup counts from 1982-2010 (A), corresponding
1228
factor loadings (B, only loadings exceeding 0.1 are shown), and observed time series (circles) with model
1229
fits (lines) for each species or population (C).
54
0.2
Loadings
2
BLKI
RLKI
COMU
TBMU
BLKW
RLKW
COMU
TBMU
0
-0.2 0.0
4
(B)
-2
Contribution to fitted value
(A)
1985
1995
2005
St. Paul St. George
(C)
1985 2000
1985 2000
Normalized trend
St.G BLKW St.G RLKW St.G COMU St.G TBMU
St.P BLKI
St.P RLKI
St.P COMU St.P TBMU
2
1
0
-1
-2
2
1
0
-1
-2
1985 2000
1985 2000
1230
1231
Figure 3: Common trends from a dynamic factor analysis of four seabird species at two islands based on
1232
time series of abundance at index sites monitored from 1982-2008 (A), corresponding factor loadings (B),
1233
and observed time series (circles) with model fits (lines) for each species and colony (C).
1234
55
15
10
St.G NFS
Pacific cod
red king crab
snow crab
pollock
arrowtooth
rock sole
flathead sole
yellowfin sole
St.P TBMU
St.G COMU
St.G TBMU
St.P COMU
St.G RLKI
St.P NFS
St.P BLKI
St.G BLKI
St.P RLKI
5
Distance
0
1235
1236
Figure 4: Correlogram showing pairwise correlations among time series of productivity for 18 fish,
1237
shellfish, seabird, and fur seal populations in the eastern Bering Sea and corresponding cluster
1238
dendrogram. Blue and red cells denote positive and negative correlations, respectively. Shading is
1239
proportional to magnitude of correlations. Rows from top to bottom correspond to columns from left to
1240
right. Horizontal and vertical lines separate four major clusters indicated in the denrdogram by dashed
1241
horizontal line.
56
1.5
0.0
2 -1.5
1
0
-1
2 -2
1
0
-1.0 0.0
1.0
x
-0.5
1
0
-1
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
-2
Fish condition
-2.0
murres
1.0
kittiwakes
2.0 -2
-1
SST
Wind direction
Fishes
flatfish
gadids
1242
1243
Figure 5: Normalized anomalies of sea-surface temperature (SST), wind direction at 60 °N, 170°W,
1244
average recruitment of two groups of fishes (gadids and flatfishes), kittiwake fledging success, murre
1245
fledging success, and fish condition indices for 4 groundfish species in the eastern Bering Sea.
1246
Normalized chl-a concentrations (green) over the middle shelf during spring are superimposed on SST,
1247
wind direction, and condition indices for comparison. Recent warm and cold periods are shaded in red
1248
and blue, respectively.
57
(B)
1
0
0
-1
-1
-2
Anomalies
1
2
2
(A)
1975
1249
1250
1985
1995
2005
1975
1985
1995
2005
1251
Figure 6: (A) Anomalies in detrended pup counts at St. Paul Island (solid line and triangles) and kittiwake
1252
productivity (dashed line and circles) and (B) pup count anomalies at St. George Island (solid line and
1253
triangles), gadid recruitment anomalies (dashed line and circles) and red king crab recruitment (heavy
1254
black line).
58
3
2
1
0
-1
-2
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
-3
First PC, fish
1255
1256
Figure 7: First Principal Component from a Principal Component Analysis of the species in the flatfish
1257
and gadid groups in Figure 4 (bars) with chl-a concentrations on the middle shelf during spring (black
1258
line).
1259
59
83
0.5
1.0
0381
86
02
01
05
80
85
9891
95
0.0
-0.5
Flatfish recruitment anomaly
87
88
04
93
84
9099
00
96
79
97
89
92
94
78
77
82
-1.0 -0.5
1260
0.0
0.5
1.0
1.5
Gadid recruitment anomaly
1261
Figure 8: Scatterplot of normalized, combined flatfish recruitment anomalies against normalized,
1262
combined gadid recruitment anomalies. Numbers denote last two digits of the corresponding brood year.
1263
60
0.0 0.1 0.2
0.0 0.1 0.2
-0.2
-0.2
Partial effect on Chl a
5.0
1264
5.5
6.0
6.5
7.0
190
Surface temperature
210
230
Wind direction
1265
Figure 9: Partial fits of chl-a concentration from a multiple liner regression of chl-a on sea-surface
1266
temperature and wind direction. The variables are uncorrelated (r =0.166) and a regression with both
1267
variables included represents the best fit.
1268
61
1269
Chapter 3: Climate-biology covariation
1270
A number of the indices described in Chapter 1 were used along with others to examine patterns of
1271
variability in the ecosystems of the Northeast Pacific. Two publications resulted from this work that was
1272
primarily supported by a separate NPRB grant (Project 10240 to Mike Litzow). Mueter's contribution to
1273
these papers was supported through the current grant (BSIERP B68). Abstracts for the final report and the
1274
published papers are included here for reference. Drafts of the full papers are included in the Final Report
1275
for NPRB Project 1024 "Four decades of climate-biology covariation in Alaskan and North Pacific
1276
ecosystems" (Litzow et al. 2012) and final versions were published in late 2013 (Litzow and Mueter
1277
2013, Litzow et al. 2013).
1278
1279
Litzow, M.A., Mueter, F.J., and Hobday, A.J. 2012. Four decades of climate-biology covariation in
1280
Alaskan and North Pacific ecosystems, North Pacific Research Board Final Report 1024.
1281
Abstract: We updated a previously-published set of 31 climate and 64 biology time series (1965-2011 for
1282
climate time series, 1965-2008 for biology) to address outstanding questions about climate-biology
1283
covariation in Alaskan and northeast Pacific ecosystems. We found that a comprehensive set of natural
1284
climate modes was inadequate for explaining variability in Alaskan climate parameters during 1965-2008,
1285
as indicated by a linear trend in residual values of the leading principal component (PC) for sea surface
1286
temperature, sea level pressure, freshwater discharge and ice cover. This residual trend in Alaskan climate
1287
in turn explained a proportion of variability in PC1 of Alaskan biology time series similar to that
1288
explained by the Pacific Decadal Oscillation (PDO). We also demonstrate a possible abrupt shift in basin-
1289
wide biological variability following a 2007/2008 change to a PDO negative/North Pacific Gyre
1290
Oscillation (NPGO) positive state, and we propose a new framework for assessing both the size and
1291
persistence of putative climate regime shifts at the ends of time series. Finally, we demonstrate that the
1292
apparent 1988/89 transition in relative importance of the PDO and NPGO as leading modes of internal
1293
climate variability led to non-additive forcing-response relationships for PC1 and 2 of Alaskan biology
1294
time series (i.e., different sets of ecologically important forcing parameters before and after 1988/89). The
1295
apparent change in the leading mode of North Pacific climate had implications for ecosystem forecasting,
1296
as predictive skill for forecasts from pre-shift observations were significantly more accurate across the
1297
1976/77 PDO shift than across the 1988/89 PDO-NPGO transition.
1298
62
1299
1300
Litzow, M.A., and Mueter, F.J. 2013. Assessing the ecological importance of climate regime shifts: an
approach from the North Pacific Ocean. Progress in Oceanography.
1301
Abstract: We used an indicator approach to address outstanding questions concerning the ecological
1302
importance of low-frequency climate variability in the northeast Pacific Ocean. Our data consist of a
1303
previously-published set of 33 climate and 64 biology time series, updated by us for the years 1965–2011
1304
(for climate data) and 1965–2008 (for biology data). A model-selection approach showed that the first
1305
axis of variability in large-scale climate indices (PC1ci), the first and second axes for local climate
1306
parameters (PC1cp and PC2cp) and the second axis for biological variability (PC2biol) all showed
1307
temporal variability best described by abrupt shifts. In contrast, PC1biol showed gradual, rather than
1308
abrupt, temporal variability, suggesting that the leading axis of biological variability was not dominated
1309
by abrupt transitions following climate regime shifts. The leading mode of variability in detrended North
1310
Pacific sea surface temperature, the Pacific Decadal Oscillation, showed reduced amplitude from the late
1311
1980s until the mid-2000s, and we found that this change in PDO behavior was associated with a decline
1312
in the strength of the leading pattern of basin-scale biological variability (PC1biol). A reversion to a
1313
PDO-negative state in the winter of 2007/08 was associated with the largest observed annual change in
1314
the PC1bio–PC2bio phase space, suggesting renewed ecological importance of the PDO. However, a
1315
subset of biology time series (n = 23) for which more recent data were available did not show persistent
1316
change in PC1bio or PC2bio during 2008–2011, thus failing to support the hypothesis of widespread
1317
ecological response to the putative 2007/08 shift. To further assess the possible ecological importance of
1318
low-frequency climate variability in recent years, we compared changes in the PDO-North Pacific Gyre
1319
Oscillation (NPGO) phase space for 2007/08 with ecologically important (1976/77) and less important
1320
(1988/89) climate regime shifts of the past. We found that all three shifts involved PDO-NPGO
1321
variability of similar magnitude (i.e., similar pulse disturbances), but that the 1976/77 shift was followed
1322
by a period of stability in a new climate state (i.e., strong press disturbance), while the 1988/89 shift was
1323
not followed by a period of stability (weak press disturbance). Data through 2013 suggest that the press
1324
disturbance following 2007/08 is similar to that following 1976/77, implying that the putative 2007/08
1325
shift may eventually prove to be ecologically important. Our ‘‘pulse-press’’ approach provides a formal
1326
framework for distinguishing transient and persistent climate perturbations at the ends of time series.
1327
63
1328
Litzow, M.A., Mueter, F.J., and Hobday, A.J. 2013. Reassessing regime shifts in the North Pacific:
1329
incremental climate change and commercial fishing are necessary for explaining decadal-scale
1330
biological variability. Global Change Biology.
1331
Abstract: In areas of the North Pacific that are largely free of overfishing, climate regime shifts – abrupt
1332
changes in modes of low-frequency climate variability – are seen as the dominant drivers of decadal-scale
1333
ecological variability. We assessed the ability of leading modes of climate variability (PDO, NPGO, AO,
1334
PNA, NPI, ENSO) to explain decadal-scale (1965-2008) patterns of climatic and biological variability
1335
across two North Pacific ecosystems (Gulf of Alaska and Bering Sea). Our response variables were the
1336
first principle component (PC1) of four regional climate parameters (SST, SLP, freshwater input, ice
1337
cover) and PCs 1-2 of 36 biological time series (production or abundance for populations of salmon
1338
[Oncorhynchus spp.], groundfish, herring [Clupea pallasii], shrimp and jellyfish). We found that the
1339
climate modes alone could not explain observed ecological variability in the study region. Both linear
1340
models (for climate PC1) and generalized additive models (for biology PC1-2) invoking only the natural
1341
climate modes produced residuals with significant temporal trends, indicating that the models failed to
1342
capture coherent patterns of ecological variability. However, when the residual climate trend and a time
1343
series of commercial fishery catches were used as additional candidate variables, resulting models of
1344
biology PC1-2 satisfied assumptions of independent residuals and out-performed models constructed
1345
from the climate modes alone in terms of predictive power. As measured by effect size and Akaike
1346
weights, the residual climate trend was the most important variable for explaining biology PC1
1347
variability, and commercial catch the most important variable for biology PC2. Patterns of climate
1348
sensitivity and exploitation history for taxa strongly associated with biology PC1-2 suggest plausible
1349
mechanistic explanations for these modeling results. Our findings suggest that, even in the absence of
1350
overfishing and in areas strongly influenced by internal climate variability, climate regime shift effects
1351
can only be understood in the context of other ecosystem perturbations.
64
1352
Chapter 4: Ecosystem Considerations contributions
1353
This chapter combines the most up-to-date contributions to the Ecosystem Considerations Chapter of the
1354
Stock Assessment and Fishery Evaluation (SAFE) reports to the North Pacific Fishery Management
1355
Council (Zador and Gaichas 2011, 2012, Zador 2013) that were supported in whole or in part by the
1356
current project. The report is published annually and provides advice to the North Pacific Fishery
1357
Management Council (NPFMC) in support of an ecosystem-based approach to management.
1358
1359
Literature cited
1360
Zador, S., and Gaichas, S. 2011. Ecosystem considerations 2011. North Pacific Fishery Management 644
1361
1362
1363
1364
1365
Council, 605 W 4th Ave, Suite 306, Anchorage, AK 99501.
Zador, S., and Gaichas, S. 2012. Ecosystem considerations 2012. North Pacific Fishery Management 644
Council, 605 W 4th Ave, Suite 306, Anchorage, AK 99501.
Zador, S. 2013. Ecosystem considerations 2013. North Pacific Fishery Management 644 Council, 605 W
4th Ave, Suite 306, Anchorage, AK 99501.
1366
1367
Individual Contributions:
1368
Chapter 4a: Aggregated catch-per-unit-effort of fish and invertebrates in bottom trawl surveys
1369
Chapter 4b: Average local species richness and diversity of the eastern Bering Sea groundfish community
1370
Chapter 4c: Spatial distribution of groundfish stocks in the Bering Sea
1371
Chapter 4d: Total annual surplus production and overall exploitation rate of groundfish
1372
Chapter 4e: Combined Standardized Indices of recruitment and survival rate
1373
Chapter 4f: Indicators of basin-scale and Alaska-wide community regime shifts
1374
65
1375
Chapter 4a: Aggregated catch-per-unit-effort of fish and invertebrates in bottom trawl surveys
1376
Contributed by Franz Mueter1 and Robert Lauth2
1377
1
1378
1379
2
1380
Last updated: October 2012
University of Alaska Fairbanks, 17101 Point Lena Road, Juneau, Alaska, 99801, fmueter@alaska.edu
Alaska Fisheries Science Center, National Marine Fisheries Service, NOAA, 7600 Sand Point Way
NE, Seattle, WA, U.S.A.
1381
1382
Description of index: The index provides a measure of the overall biomass of demersal and benthic fish
1383
and invertebrate species. We computed catch-per-unit-effort (CPUE in kg km-2) of fish and major
1384
invertebrate taxa for each successful haul completed during standardized bottom trawl surveys on the
1385
eastern Bering Sea shelf (EBS, 1982-2012) and on the Gulf of Alaska shelf (GoA, 1990-2011). Total
1386
CPUE for each haul was estimated as the sum of the CPUEs of all fish and invertebrate taxa. To obtain an
1387
index of average CPUE by year across the survey region, we modeled log-transformed total CPUE (N =
1388
11548, 5782, and 1529 hauls in the EBS, western GoA, and eastern GoA, respectively) as smooth
1389
functions of depth, net width, and location (latitude / longitude in the EBS, alongshore distance, sampling
1390
stratum, and depth in the GoA) using Generalized Additive Models following Mueter & Norcross (2002).
1391
Hauls were weighted based on the area represented by each station. Although catches were standardized
1392
to account for the area swept by each haul we included net width in the model for the Bering Sea because
1393
of differences in catchability of certain taxa with changes in net width (von Szalay & Somerton 2005) and
1394
because there was strong evidence that total CPUE tends to decrease with net width, all other factors
1395
being constant. The CPUE index does not account for gear or vessel differences, which are strongly
1396
confounded with interannual differences and may affect results prior to 1988 in the Bering Sea.
1397
Status and trends: Total log(CPUE) in the western GoA varied over time with lowest abundances
1398
observed in 1999 and 2001 (Fig.4a.1). Mean CPUE ranged from 101 kg/ha in 2001 to 138 kg/ha in 2003.
1399
The eastern GoA shows a significantly increasing trend (p 0.0139) from 55 kg/ha in 1990 to 70 kg/ha in
1400
2011. Total log(CPUE) in the EBS shows an apparent long-term increase from 1982-2005, followed by a
1401
decrease from 2005 to 2009 and an increase in 2010 (Fig.4a.2). Estimates of total mean CPUE ranged
1402
from 180 kg/ha in 1985 to over 370 kg/ha in 2003, decreasing to 225 kg/ha in 2005. Estimated means
1403
prior to 1988 may be biased due to unknown gear effects and because annual differences are confounded
1404
with changes in mean sampling date, which varied from as early as June 15 in 1999 to as late as July 16 in
1405
1985. On average, sampling occurred about a week earlier in the 2000s compared to the 1980s. Recent
1406
changes in CPUE in the EBS have been most pronounced on the middle-shelf, which is occupied by the
66
1407
cold pool during cold years. Higher CPUEs on the middle shelf during the 2001-2005 warm period
1408
appeared to be related to the increasing colonization of this area by subarctic demersal species (Mueter &
1409
Litzow, 2008).
1410
Factors causing observed trends: Commercially harvested species account for over 70% of survey
1411
catches. Fishing is expected to be a major factor determining trends in survey CPUE, but environmental
1412
variability is likely to account for a substantial proportion of the observed variability in CPUE through
1413
variations in recruitment, growth, and distribution. The increase in survey CPUE in the EBS in the early
1414
2000s primarily resulted from increased abundances of walleye pollock and a number of flatfish species
1415
(arrowtooth flounder, yellowfin sole, rock sole, and Alaska plaice) due to strong recruitments in the
1416
1990s. Decreases in 2006-2009 are largely a result of decreases in walleye pollock abundance. Increases
1417
in pollock and Pacific cod biomass in 2010 resulted in the observed increase in log(CPUE). In addition,
1418
models including bottom temperature suggest that, in the EBS, CPUE is greatly reduced at low
1419
temperatures (< 1˚C) as evident in reduced CPUEs in 1999 and 2006-2009, when the cold pool covered a
1420
substantial portion of the shelf. This reduction is likely due to a combination of actual changes in
1421
abundance, temperature-dependent changes in catchability of certain species (e.g. flatfish, crab), and
1422
changes in distribution as a result of the extensive cold pool displacing species into shallower (e.g. red
1423
king crab) or deeper (e.g. arrowtooth flounder) waters. Increases in CPUE in the GoA between 1999/2001
1424
and 2003 were largely due to a substantial increase in the abundance of arrowtooth flounder, which
1425
accounted for 43% of the total survey biomass in 2003 in the western GOA. The significant increase in
1426
total CPUE in the eastern GoA was associated with increases in arrowtooth flounder (particularly 1990-
1427
93), several rockfish species, Pacific hake, and spriny dogfish.
1428
Implications: This indicator can help address concerns about maintaining adequate prey for upper trophic
1429
level species and other ecosystem components. Relatively stable or increasing trends in the total biomass
1430
of demersal fish and invertebrates, together with a relatively constant size composition of commercial
1431
species, suggest that the prey base has remained stable or has increased over recent decades. Decreasing
1432
CPUE in the eastern Bering Sea in the early 2000s was a concern, but biomass has increased as a result of
1433
several strong year classes of walleye pollock entering the survey.
1434
References
1435
Mueter, F.J., and B.L. Norcross. 2002. Spatial and temporal patterns in the demersal fish community on
1436
1437
1438
the shelf and upper slope regions of the Gulf of Alaska. Fish. Bull. 100(3): 559-581.
Mueter, F.J., and M.A. Litzow. 2008. Sea ice retreat alters the biogeography of the Bering Sea continental
shelf. Ecol. Appl. 18: 309-320.
67
1439
1440
von Szalay, P.G., and D.A. Somerton. 2005. The effect of net spread on the capture efficiency of a
demersal survey trawl used in the eastern Bering Sea. Fish. Res. 74: 86–95.
1441
1442
1443
1444
1445
1446
Eastern GoA
4.4
4.2
5.0
4.0
4.8
4.6
1447
3.8
4.4
log(CPUE)
5.2
Western GoA
1990
1995
2000
2005
2010
1990
1995
2000
2005
2010
1448
Figure 4a.1: Model-based estimates of total log(CPUE) for major fish and invertebrate taxa captured in
1449
bottom trawl surveys from in the western Gulf of Alaska (west of 147˚ W) by survey year with
1450
approximate 95% confidence intervals. Estimates were adjusted for differences in depth and sampling
1451
locations (alongshore distance) among years. Linear trend in eastern GOA based on generalized least
1452
squares regression assuming 1st order auto-correlated residuals (t = 3.258, p = 0.014).
1453
68
5.8
5.6
5.4
5.2
5.0
log(CPUE)
1985
1990
1995
2000
2005
2010
1454
1455
Figure 4a.2: Model-based estimates of total log(CPUE) for major fish and invertebrate taxa captured in
1456
bottom trawl surveys from 1982 to 2012 in the Bering Sea with approximate pointwise 95% confidence
1457
intervals and linear time trend. Estimates were adjusted for differences in depth, day of sampling, net
1458
width and sampling location among years. Gear differences prior to 1988 were not accounted for. A linear
1459
time trend based on generalized least squares regression assuming 1st order auto-correlated residuals was
1460
not significant (t = 1.221, p = 0.232).
1461
1462
69
1463
Chapter 4b: Average local species richness and diversity of the eastern Bering Sea groundfish
1464
community
1465
Contributed by Franz Mueter1, Jason Waite1, and Robert Lauth2
1466
1
University of Alaska Fairbanks, 17101 Point Lena Road, Juneau, Alaska, 99801, fmueter@alaska.edu
1467
2
Alaska Fisheries Science Center, National Marine Fisheries Service, NOAA, 7600 Sand Point Way
1468
NE, Seattle, WA, U.S.A.
1469
Last updated: October 2012
1470
1471
Description of indices: This section provides indices of local species richness and diversity based on
1472
standard bottom trawl surveys in the eastern Bering Sea (EBS). We computed the average number of fish
1473
and major invertebrate taxa per haul (richness) and the average Shannon index of diversity (Magurran
1474
1988) by haul based on CPUE (by weight) of each taxon. Indices were based on 45 fish and invertebrate
1475
taxa that were consistently identified throughout all surveys since 1982 (Table 1 in Mueter & Litzow,
1476
2008, excluding Arctic cod because of unreliable identification in early years). Indices were computed
1477
following Mueter & Norcross (2002). Briefly, annual average indices of local richness and diversity were
1478
estimated by first computing each index on a per-haul basis, then estimating annual averages with
1479
confidence intervals across the survey area using a Generalized Additive Model that accounted for the
1480
effects of variability in geographic location (latitude/longitude), depth, date of sampling, and area swept.
1481
In addition to trends in the indices over time, we mapped average spatial patterns for each index across
1482
the survey region.
1483
Status and trends: Species richness and diversity on the Eastern Bering Sea shelf have undergone
1484
significant variations from 1982 to 2012 (Fig. 4b.1). The average number of species per haul increased by
1485
one to two species from 1995 to 2004 and has remained relatively high since then. The Shannon Index
1486
increased from 1985 through 1998, decreased sharply in 1999, and has been highly variable since then.
1487
Diversity was low in 2002/03, increased substantially in 2004, decreased through 2010, but was high in
1488
the last two years.
1489
Factors causing observed trends: The average number of species per haul depends on the spatial
1490
distribution of individual species (or taxa). If species are, on average, more widely distributed in the
1491
sampling area the number of species per haul increases. Spatial shifts in distribution from year to year can
1492
cause high variability in local species richness in certain areas, for example along the 100m contour in the
1493
Eastern Bering Sea. These shifts appear to be the primary drivers of changes in species richness. Local
70
1494
species diversity is a function of how many species are caught in a haul and how evenly CPUE is
1495
distributed among the species. Both time trends (Fig. 4b.1) and spatial patterns in species diversity (Fig.
1496
4b.2) differed markedly from those in species richness. For example, low species diversity in 2003 in the
1497
EBS occurred in spite of high average richness, primarily because of the high dominance of walleye
1498
pollock, which increased from an average of 18% of the catch per haul in 1995-98 to 30% in 2003, but
1499
decreased again to an average of 21% in 2004. The increase in species richness in the EBS, which was
1500
particularly pronounced on the middle shelf, has been attributed to subarctic species spreading into the
1501
former cold pool area as the extent of the cold pool decreased from 1982 to 2005 (Mueter & Litzow,
1502
2008). Spatially, species richness tends to be highest along the 100 m contour in the EBS, whereas
1503
species diversity is highest on the middle shelf because the middle shelf region is less dominated by a few
1504
abundant species.
1505
Implications: The effect of fishing on species richness and diversity are poorly understood at present and
1506
this index likely reflects changes in spatial distribution and species composition that can only be
1507
interpreted in the context of environmental variability in the system. In the EBS, local species richness
1508
may be particularly sensitive to long-term trends in bottom temperature as the cold pool extent changes
1509
(Mueter and Litzow 2008) and may provide a useful index for monitoring responses of the groundfish
1510
community to projected climate warming. However, neither richness nor diversity were significantly
1511
correlated with bottom temperatures; richness was relatively high since 2004 spanning both a warm and
1512
cold period, while diversity varied greatly between years in the most recent cold period (2009/2010 vs.
1513
2011/12).
1514
References:
1515
Magurran, A.E. 1988. Ecological diversity and its measurement. Princeton, New Jersey: Princeton
1516
1517
1518
1519
1520
University Press.
Mueter, F.J., and B.L. Norcross. 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.
Mueter, F.J., and M.A. Litzow. 2008. Sea ice retreat alters the biogeography of the Bering Sea continental
shelf. Ecol. Appl. 18: 309-320.
1521
71
1985
1990
1995
2000
2005
3.2
3.0
2.8
2.6
Shannon index (H')
3.4
14.5
14.0
13.5
13.0
12.5
Species per haul
1522
2010
1985
1990
1995
2000
2005
2010
1523
Figure 4b.1: Model-based annual averages of species richness (average number of species per haul, dots),
1524
and species diversity (Shannon index) in the Eastern Bering Sea, 1982-2010, based on 45 fish and
1525
invertebrate taxa collected by standard bottom trawl surveys with pointwise 95% confidence intervals
1526
(bars) and loess smoother with 95% confidence band (dashed/dotted lines). Model means were adjusted
1527
for differences in area swept, depth, date of sampling, and geographic location.
1528
10
62
Alaska
14
62
Alaska
2.
5
2. 5
60
16
3
14
3
1
3.
5
60
12
2
58
58
5
2
16
14
3
2.5
-175
-170
3
56
14
56
1529
2
2.
12
-165
-160
-175
-170
-165
-160
1530
1531
Figure 4b.2: Average spatial patterns in local species richness (left, number of taxa per haul) and Shannon
1532
diversity in the Eastern Bering Sea. The 50m, 100m, and 200 m depth contours are shown as black lines.
1533
Note highest richness along 100 m contour, highest diversity on middle shelf
72
1534
Chapter 4c: Spatial distribution of groundfish stocks in the Bering Sea
1535
Contributed by Franz Mueter, Michael Litzow2, and Robert Lauth3
1536
1
1537
2
1538
1539
3
1540
Last updated: October 2012
University of Alaska Fairbanks, 17101 Point Lena Road, Juneau, Alaska, 99801, fmueter@alaska.edu
Blue World Research, 2710 E. 20th Ave., Anchorage, AK 99508
Alaska Fisheries Science Center, National Marine Fisheries Service, NOAA, 7600 Sand Point Way
NE, Seattle, WA, U.S.A.
1541
1542
Description of indices: We provide indices of changes in the spatial distribution of groundfish on the
1543
eastern Bering Sea shelf. The first index provides a simple measure of the average North-South
1544
displacement of major fish and invertebrate taxa from their respective centers of gravity (e.g. Woillez et al
1545
2009) based on AFSC-RACE bottom trawl surveys for the 1982-2012 period. Annual centers of gravity
1546
for each taxon were computed as the CPUE-weighted mean latitude across 285 standard survey stations
1547
that were sampled each year and an additional 58 stations sampled in 26 of the 27 survey years. Each
1548
station (N=343) was also weighted by the approximate area that it represents. Initially, we selected 46
1549
taxa as in Table 1 of Mueter and Litzow (2008). Taxa that were not caught at any of the selected stations
1550
in one or more years were not included, resulting in a total of 39 taxa for analysis. In addition to
1551
quantifying N-S shifts in distribution, we computed CPUE and area-weighted averages of depth to
1552
quantify changes in depth distribution. Because much of the variability in distribution may be related to
1553
temperature variability, we removed linear relationships between changes in distribution and temperature
1554
by regressing distributional shifts on annual mean bottom temperatures. Residuals from these regressions
1555
are provided as an index of temperature-adjusted shifts in distribution.
1556
Status and trends: Both the latitudinal and depth distribution of the demersal community on the eastern
1557
Bering Sea shelf show strong directional trends over the last three decades, indicating significant
1558
distributional shifts to the North and into shallower waters (Fig. 4c.1). This distribution was largely
1559
maintained through the recent cold years. Strong shifts in distribution over the 31 year time series remain
1560
evident even after adjusting for linear temperature effects (Fig. 4c.1). Average spatial displacements
1561
across all species by year suggest that most interannual shifts in distribution occur along a NW-SE axis
1562
(i.e. along the main shelf/slope axis), but that a pronounced shift to the Northeast and onto the shelf
1563
occurred between the 1990s and 2000s (Fig. 4c.2). On average, there was a gradual shift to the north from
1564
2001 to 2005, which reversed as temperatures cooled after 2006. In 2009, the average center of gravity
73
1565
temporarily shifted back to deeper waters but has been relatively shallow with little change in latitude
1566
since 2010.
1567
Factors causing trends: Many populations shift their distribution in response to temperature variability.
1568
Such shifts may be the most obvious response of animal populations to global warming (Parmesan and
1569
Yohe 2003). However, distributional shifts of demersal populations in the Bering Sea are not a simple
1570
linear response to temperature variability (Mueter and Litzow 2008, Fig. 4c.1). The reasons for residual
1571
shifts in distribution that are not related to temperature changes remain unclear but could be related to
1572
density-dependent responses (Spencer 2008) in combination with internal community dynamics (Mueter
1573
and Litzow 2008). Unlike groundfish in the North Sea, which shifted to deeper waters in response to
1574
warming (Dulvy et al 2008), the Bering Sea groundfish community shifted to shallower waters during the
1575
recent warm period (Fig. 4c.1). Surprisingly, the summer distribution has remained relatively shallow
1576
despite very cold temperatures on the shelf.
1577
Implications: Changes in distribution have important implications for the entire demersal community, for
1578
other populations dependent on these communities, and for the fishing industry. The demersal community
1579
is affected because distributional shifts change the relative spatial overlap of different species, thereby
1580
affecting trophic interactions among species and, ultimately, the relative abundances of different species.
1581
Upper trophic level predators, for example fur seals and seabirds on the Pribilof Islands and at other fixed
1582
locations, are affected because the distribution and hence availability of their prey changes. Finally,
1583
fisheries are directly affected by changes in the distribution of commercial species, which alters the
1584
economics of harvesting because fishing success within established fishing grounds may decline and
1585
travel distances to new fishing grounds may increase. A better understanding of the observed trends and
1586
their causes is needed to evaluate the extent to which fishing may have contributed to these trends and to
1587
help management and fishers adapt to apparent directional changes in distribution that are likely to be
1588
further exacerbated by anticipated warming trends associated with increasing CO2 concentrations.
1589
1590
1591
References
1592
Dulvy, N. K., Rogers, S. I., Jennings, S., Stelzenmüller, V., Dye, S. R., and Skjoldal, H. R. 2008. Climate
1593
change and deepening of the North Sea fish assemblage: a biotic indicator of warming seas.
1594
Journal of Applied Ecology, 45: 1029-1039.
74
1595
1596
1597
1598
1599
1600
1601
1602
Mueter, F.J. and M.A. Litzow. 2008. Sea ice retreat alters the biogeography of the Bering Sea continental
shelf. Ecol. Appl. 18: 309-320.
Parmesan, C. and G. Yohe. 2003. A globally coherent fingerprint of climate change impacts across
natural systems. Nature 421(6918): 37-42.
Spencer, P.D. 2008. Density-independent and density-dependent factors affecting temporal changes in
spatial distributions of eastern Bering Sea flatfish. Fish. Oceanogr. 17(5): 396 - 410.
Woillez, M., J. Rivoirard, and P. Petitgas. 2009. Notes on survey-based spatial indicators for monitoring
fish populations. Aquat. Living Resour. 22: 155-164.
1603
1604
75
15
10
5
0
-5
-10
Residual displacement (km)
15
10
5
0
-5
-10
Northward displacement (km)
-15
1986
1990
1994
1998
2002
2006
2010
1982
1986
1990
1994
1998
2002
2006
2010
1985
1990
1995
2000
2005
2010
1985
1990
1995
2000
2005
2010
1605
2
1
0
-2
-1
Residual deepening (m)
2
1
0
-1
-2
Average deepening (m)
3
3
1982
1606
Figure 4c.1: Left: Distributional shifts in latitude (average northward displacement in km from species-
1607
specific mean latitudes) and shifts in depth distribution (average vertical displacement in m from species-
1608
specific mean depth, positive indices indicate deeper distribution). Right: Residual displacement from
1609
species-specific mean latitude (top) and species-specific mean depth (bottom) after adjusting the indices
1610
on the left for linear effects of mean annual bottom temperature on distribution. Residuals were obtained
1611
by linear weighted least-squares regression on annual average temperature with first-order auto-correlated
1612
residuals over time (Northward displacement: R2 = 0.24, t = 3.75, p < 0.001; depth displacement: R2 =
1613
0.25, t = -3.60, p = 0.001). Solid lines denote linear regressions of residual variability over time (top: R2 =
1614
0.57, t = 4.34, p < 0.001; bottom: R2 = 0.61, t = -6.68, p < 0.001).
76
0.15
06 08
95
83
03
94 93
0711
10
00
12
09
0.00
0.05
0.10
05
96
88
89
98
-0.05
01
97
-0.10
92
85
90
84
-0.4
1615
02
87
91
-0.15
Relative displacement (degrees latitude)
04
-0.3
-0.2
-0.1
0.0
0.1
82
99
0.2
86
0.3
Relative displacement (degrees longitude)
1616
Figure 4c.2: Average North-South and East-West displacement across 39 taxa on the eastern Bering Sea
1617
shelf relative to species-specific centers of distribution.
1618
1619
77
1620
Chapter 4d: Total annual surplus production and overall exploitation rate of groundfish
1621
Contributed by Franz Mueter
1622
University of Alaska Fairbanks, 17101 Point Lena Road, Juneau, Alaska, 99801. fmueter@alaska.edu
1623
Last updated: July 2010
1624
1625
Description of indices: Total annual surplus production (ASP) of groundfish on the Eastern Bering Sea
1626
shelf (EBS) and on the Gulf of Alaska (GOA) shelf from 1977-2009 was estimated by summing annual
1627
production across major commercial groundfish stocks for which assessments were available (Table
1628
4d.1). These species represent at least 70-80% of the total catch in bottom trawl surveys. Annual surplus
1629
production in year t can be estimated as the change in total adult groundfish biomass across species from
1630
year t (Bt) to year t+1 (Bt+1) plus total catches in year t (Ct):
1631
ASPt = Bt + Ct = Bt+1 – Bt + Ct
1632
All estimates of B and C are based on 2009 stock assessments. An index of total exploitation rate within
1633
each region was obtained by dividing the total groundfish catch across the major commercial species by
1634
the estimated combined biomass at the beginning of the year:
1635
ut = Ct / Bt
1636
1637
Status and trends: The resulting indices suggest high variability in groundfish production in the eastern
1638
Bering Sea (Fig. 4d.1) and a decrease in production between 1977 and 2007 (slope = - 45,500 mt / year, t
1639
= -1.922, p = 0.064). Annual surplus production in the GoA was much lower on average, less variable,
1640
and did not show a significant trend over time (slope = - 1,508 mt/ year, t = -0.109, p = 0.914). Total
1641
exploitation rates for the groundfish complex are generally much higher in the EBS than in the GoA and
1642
were highest in the early part of the time series due to high exploitation rates of walleye pollock (Fig.
1643
4d.1). The overall exploitation rate in the EBS reached a low of 6.9% in 1999, increased to 12% by 2006,
1644
and decreased in 2008/2009 due to a reduction in walleye pollock harvest rates. The overall exploitation
1645
rate in the Gulf of Alaska has generally been less than 6% except in 1984/85.
1646
1647
Because trends in annual surplus production in the Eastern Bering Sea are almost entirely driven by
1648
variability in walleye pollock, ASPt for the Bering Sea was also computed after excluding walleye
78
1649
pollock (Fig. 4d.2). The results suggest a pronounced decrease in aggregate surplus production of all non-
1650
pollock species from a high of over 1 million tons in 1979/1980, due to strong recruitment of a number of
1651
species, to lows of around 300,000 t in the late 1990s. In 2008, annual surplus production increased
1652
substantially due to an estimated increase in the biomass of Alaska plaice.
1653
Factors causing trends: Annual Surplus Production is an estimate of the sum of new growth and
1654
recruitment minus deaths from natural mortality (i.e. mortality from all non-fishery sources) during a
1655
given year. It is highest during periods of increasing total biomass (e.g. 1991-92 in the EBS) and lowest
1656
during periods of decreasing biomass (e.g. 1982-1984 in the GoA and 2004-2007 in the EBS). In the
1657
absence of a long-term trend in total biomass, ASP is equal to the long-term average catch. Theory
1658
suggests that surplus production of a population will decrease as biomass increases much above BMSY,
1659
which has been the case for a number of flatfish species (e.g. rock sole, flathead sole) and rockfish species
1660
(Pacific ocean perch, northern rockfish).
1661
Exploitation rates are primarily determined by management and reflect a relatively precautionary
1662
management regime with rates that have averaged less than 10% across species in the EBS. Exploitation
1663
rates are much lower in the GoA because of the very limited exploitation of arrowtooth flounder, which
1664
currently make up the majority of the biomass in the GoA. If arrowtooth flounder are excluded, rates are
1665
comparable to those in the EBS.
1666
Implications: Under certain assumptions, aggregate surplus production can provide an estimate of the
1667
long-term maximum sustainable yield of these groundfish complexes (Mueter and Megrey 2006, Fig.
1668
4d.3). Although there is little contrast in total biomass over time, it appears that biomass was generally
1669
above the level that would be expected to yield maximum surplus production under a Graham-Schaefer
1670
model fit to aggregate ASP (Fig. 4d.3). The recent decrease in aggregate biomass in the EBS from 2004
1671
through 2007 (largely due to decreases in walleye pollock abundance) to levels last seen in the late 1970s
1672
was associated with an increase in aggregate ASP, suggesting that compensatory mechanisms may be
1673
starting to increase production through increases in growth or reduced predation mortality. If projected
1674
increases in biomass materialize, aggregate ASP will increase further in 2010.
1675
References:
1676
Mueter, F.J., and B.A. Megrey. 2006. Maximum productivity estimates for the groundfish complexes of
1677
the Gulf of Alaska and Eastern Bering Sea / Aleutian Islands. Fish. Res. 81: 189-201.
1678
1679
79
1680
Table 4d.1: Species included in computing annual surplus production in the Bering Sea and Gulf of
1681
Alaska.
Bering Sea
Gulf of Alaska
walleye pollock (Theragra chalcogramma)
walleye pollock (Theragra chalcogramma)
Pacific cod (Gadus macrocephalus)
Pacific cod (Gadus macrocephalus)
yellowfin sole (Limanda aspera)
Sablefish (Anoplopoma fimbria)
Greenland turbot (Reinhardtius hippoglossoides)
Pacific halibut (Hippoglossus stenolepis)
arrowtooth flounder (Atheresthes stomias)
arrowtooth flounder (Atheresthes stomias)
Northern rock sole (Lepidopsetta polyxystra)
flathead sole (Hippoglossoides spp.)
flathead sole (Hippoglossoides spp.)
rex sole (Glyptocephalus zachirus)
Alaska plaice (Pleuronectes quadrituberculatus)
Dover sole (Microstomus pacificus)
Pacific ocean perch (Sebastes alutus)
Northern rockfish (Sebastes polyspinus)
dusky rockfish (Sebastes variabilis)
rougheye (Sebastes aleutianus)
& blackspotted rockfish (S. melanostictus)
1682
2000
0.08
3000
Harvest rate (%)
0.06
4000
0.10
0.12
Bering Sea
Gulf of Alaska
0
0.04
1000
Production (1000 t)
5000
Bering Sea
Gulf of Alaska
1980
1683
1985
1990
1995
2000
2005
2010
1980
1985
1990
1995
2000
2005
2010
1684
Figure 4d.1: Total annual surplus production (change in biomass plus catch) across all major groundfish
1685
species in the Gulf of Alaska and Bering Sea with estimated linear trends and total harvest rate (total
1686
catch / beginning-of-year biomass) across all major groundfish species in the Gulf of Alaska and Bering
1687
Sea.
80
1688
700
600
500
400
300
Production (1000 t)
800
900
1689
1980
1690
1985
1990
1995
2000
2005
2010
1691
Figure 4d.2: Total annual surplus production (change in biomass plus catch) in the Bering Sea across all
1692
major groundfish species, excluding walleye pollock.
Bering Sea
Gulf of Alaska
9383
90
81
80
85
78
01
9297
07
91
02
00
77
82
96
08
94
03
99 95
98 84
86
06
04
87
05 89
88
09
0
1693
79
Total annual surplus production
-200
0
200 400 600 800
Total annual surplus production
0 1000
3000
5000
79
5000
10000 15000
Total biomass
78
80
01
77 08
9109
00 02
89
07
06
85
90
96
99 05
86
9788
03
98 0492
9587
93
94
81
84
82
83
20000
0
1000
3000
5000
Total biomass
1694
Figure 4d.3: Estimated annual aggregated surplus production against total biomass of major commercial
1695
species with fitted Graham-Schaefer curve.
81
1696
Chapter 4e: Combined Standardized Indices of recruitment and survival rate
1697
Contributed by Franz Mueter
1698
University of Alaska Fairbanks, 17101 Point Lena Loop Road, Juneau, Alaska. fmueter@alaska.edu
1699
Last updated: August 2010
1700
1701
Description of indices: Indices of overall recruitment and survival rate (adjusted for spawner abundance)
1702
across the major commercial groundfish species in the Eastern Bering Sea (EBS, 8 stocks) and Gulf of
1703
Alaska (GoA, 11 stocks) are provided. Time series of recruitment and spawning biomass for demersal
1704
fish stocks were obtained from the 2009 SAFE reports to update results of Mueter et al (2007). Only
1705
recruitment estimates for age classes that are largely or fully recruited to the fishery were included.
1706
Survival rate (SR) indices for each stock were computed as residuals from a spawner-recruit model. A
1707
Ricker and Beverton-Holt model (with or without first-order autocorrelated errors) were fit to each
1708
stock’s recruitment and female spawning biomass data and the model with the best fit (based on the
1709
small-sample Akaike Information Criterion) was used to compute the SR index. Each time series of log-
1710
transformed recruitment (logR) or SR indices was standardized to have a mean of 0 and a standard
1711
deviation of 1 (hence giving equal weight to each stock in the combined index). Time series were lined up
1712
by year-class for the period 1977-2006, resulting in matrices of logR or SR indices by year with missing
1713
values at the beginning and end of some series. A combined standardized index of recruitment (CSIR) and
1714
survival (CSISR) was computed by simply averaging indices within a given year across stocks. Prior to
1715
standardizing the series, missing values were estimated by imputation using additive regression,
1716
bootstrapping, and predictive mean matching as implemented in the “Hmisc” package for R (Frank
1717
Harrell, Univ. of Virginia, available at StatLib at http://lib.stat.cmu.edu/). Multiple imputations were
1718
obtained through bootstrap resampling to estimate the variability in the averaged index that results from
1719
filling in missing values. Because missing values are not missing at random, it is assumed that
1720
correlations between time series did not change over the period 1977-2006. Uncertainty in the stock-
1721
specific estimates of logR and SR indices was not accounted for; therefore the most recent estimates of
1722
the combined indices should be interpreted with caution.
1723
Status and trends: The CSIR and CSISR suggest that survival and recruitment of demersal species in the
1724
GoA and BSAI followed a similar pattern with below-average survival / recruitments during the early
1725
1990s (GoA) or most of the 1990s (BSAI) and above-average indices across stocks in the late 1990s /
1726
early 2000s (Fig. 4e.1). Because estimates at the end of the series were based on only a few stocks and are
82
1727
highly uncertain, we show the index through 2006 only, the last year for which reasonable estimates for
1728
the majority of stocks were available in each region. There is strong indication for above-average survival
1729
and recruitment in the GoA from 1994-2000 (with the exception of 1996, which had very low indices)
1730
and below- or near-average survival / recruitment since 2001. In the eastern Bering Sea there was no
1731
strong indication of below average recruitment across multiple stocks until 2004, when all 7 stocks with
1732
recruitment estimates had below average recruitment and stock-recruit indices (P < 0.001).
1733
Factors causing trends: Trends in recruitment are a function of both spawner biomass and
1734
environmental variability. Trends in survival rate indices, which are adjusted for differences in spawner
1735
biomass, are presumably driven by environmental variability, but are even more uncertain than
1736
recruitment trends. Typically, spawner biomass accounted for only a small proportion of the overall
1737
variability in estimated recruitment. The observed patterns in recruitment and survival since the 1976/77
1738
regime shift suggest continuing decadal-scale variations in overall groundfish productivity across multiple
1739
stocks in the Gulf of Alaska and Bering Sea. Unlike earlier analyses including longer time series that
1740
spanned the 1976/77 regime shift, the recruitment and survival series are un-correlated between the two
1741
regions (CSIR: r = 0.014; CSISR: r = 0.165). However, indices in the Bering Sea appear to lag the
1742
corresponding indices in the Gulf of Alaska by approximately 2 years with statistically significant
1743
correlations when adjusted for autocorrelation (r = 0.487, p = 0.034 and r = 0.547, p = 0.019 for CSIR and
1744
CSISR, respectively). While longer time series of the indices (1970-2004) were positively correlated with
1745
the PDO, the post-regime shift indices were not significantly correlated with either the PDO or with
1746
regional SST indices.
1747
References
1748
Mueter, F.J., Boldt, J., Megrey, B.A., and Peterman, R.M. 2007. Recruitment and survival of Northeast
1749
Pacific Ocean fish stocks: temporal trends, covariation, and regime shifts. Can. J. Fish. Aquat.
1750
Sci. 64(6): 911-927.
1751
83
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
1.0
0.5
0.0
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
-0. 5
Survival Rate Index
0.5
0.0
-0. 5
-1. 0
Survival Rate Index
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
-0. 5
0.0
0.5
Gulf of Alaska
Recruitment Index
0.5
0.0
-0. 5
-1. 0
Recruitment Index
Bering Sea
1752
1753
Figure 4e.1: Combined Standardized Indices of recruitment (top) and survival rate (stock-recruit
1754
residuals, bottom) by year class across demersal stocks in the eastern Bering Sea (8 stocks) and in the
1755
Gulf of Alaska (11 stocks). Solid blue bars represent years with data for all stocks or stock groups.
1756
Lighter shading corresponds to years with more missing stocks. Series were truncated in 1970 and only
1757
years with data for at least 6 stocks were included. Bootstrap confidence intervals (95%) depict
1758
uncertainty resulting from filling in missing values, but assume that survival and recruitment are
1759
estimated without error.
84
1760
Chapter 4f: Indicators of basin-scale and Alaska-wide community regime shifts
1761
Contributed by Mike Litzow1,2 and Franz Mueter3
1762
1
1763
2
1764
3
1765
Contact: malitzow@utas.edu.au
1766
Last updated: August 2013
Blue World Research, 2710 E. 20th Ave., Anchorage, AK 99508
University of Tasmania, Private Bag 129, Hobart, TAS, 7001, Australia
University of Alaska Fairbanks, 17101 Pt. Lena Rd., Juneau, AK 99801
1767
1768
Description of indices: The first and second principal components (PCs) for 64 biology time series from
1769
Baja California to the Bering Sea allow basin-scale patterns of biological variability to be monitored
1770
(Hare and Mantua 2000). These data include 36 Alaskan time series (19 from the Gulf of Alaska and 17
1771
from the Bering Sea). Alaskan time series include recruitment estimates for groundfish (n = 15) and
1772
herring (n = 3) populations, log-transformed and lagged to cohort year; commercial salmon catches (n =
1773
16), log-transformed and lagged to year of ocean entry; and measures of invertebrate abundance (n = 2).
1774
These indices are useful for monitoring possible biological responses to the negative Pacific Decadal
1775
Oscillation (PDO)/positive North Pacific Gyre Oscillation (NPGO) conditions that have persisted since
1776
2007/08 (Fig. 4f.1). We updated the Hare and Mantua biology time series for 1965-2008 (for the
1777
northeast Pacific) and 1965-2009 (for the Alaskan time series). Lags inherent in many time series meant
1778
that too many values were missing after 2008 (for the full data set) or after 2009 (for the Alaskan data) for
1779
PC analysis to be conducted. However, subsets of time series that could be updated at least through 2010
1780
(n = 23 for the northeast Pacific; n = 13 for Alaska) allowed PC scores to be estimated through 2011.
1781
Status and trends: Basin-scale – There was some evidence of an abrupt change in leading axes of basin-
1782
scale biological variability in 2008. Change in the PC1- 2 phase space for all 64 northeast Pacific time
1783
series from 2007 to 2008 was significantly greater than the mean for all other year-to-year changes since
1784
1965-66 (t41 = 22.69, p < 0.0001, Fig. 4f.2). While the PC scores for more recent years cannot be
1785
estimated to assess the persistence of this apparent 2007/08 change in the full data set, PC1 from the
1786
reduced data set did not show continuing increases during 2009-11, and PC2 from the reduced data set
1787
showed a single anomalous value in 2008, with a return to negative values during 2009-11 (Fig. 4f.2).
1788
STARS (sequential t-tests for analysis of regime shifts) found no evidence of statistically significant
85
1789
shifts in either of the reduced basin-wide PC time series during 2008-11 (L = 15 years, H = 6 SD,
1790
autocorrelation accounted for with IP4N method, p > 0.05).
1791
Alaska-scale –The estimated 2011 value of PC1 for the reduced Alaska-wide data set was above
1792
0, the first positive value in the time series since 1979 (Fig. 4f.3a). However, STARS showed no
1793
indication of a statistically-significant shift (p > 0.05), so these data do not show support for a recent
1794
change in this axis of variability. PC1 from the reduced data set is strongly correlated with PC1 from the
1795
full data set for the period of overlap (1965-2008, r = 0.97), so this result suggests that PC1 for the full
1796
data set is likely also not experiencing dramatic change since 2008. PC2 scores from the reduced data set
1797
did show a significant shift to more negative values in 2010 (STARS, P = 0.002, Fig. 4f.3b). However,
1798
values of PC2 from the full and reduced data sets are poorly correlated for the years of overlap (r = 0.48),
1799
so the observed 2010 shift provides weak inference concerning possible change in the second axis of
1800
variability across the full community.
1801
Factors causing trends: For the full set of 36 Alaskan time series over 1965-2008, PC1 shows strongest
1802
statistical relationships with regional climate change that is independent of basin-scale climate modes, and
1803
a weaker relationship with the PDO; PC2 shows strongest statistical relationships with the size of state-
1804
wide commercial catches and the NPGO (Litzow et al. 2013). The possibility of a biological response to
1805
persistent PDO-negative/NPGO-positive conditions since 2007/08 has received recent attention in the
1806
literature (Zwolinski and Demer 2012, Hatch 2013, Litzow and Mueter 2013). Based on historical
1807
precedents (e.g., the 1940s and 1970s PDO shifts), the consistent sign in both of these climate modes has
1808
the potential to produce abrupt community-level change at basin-wide or Alaskan-wide spatial scales,
1809
though at this time only PC2 of the reduced Alaskan data set is showing evidence of a recent shift.
1810
Implications: The apparent absence of any recent abrupt shifts in leading axes of basin-wide biological
1811
variability (Fig. 4f.2), indicates a continuation of the northeast Pacific ecosystem states that have existed
1812
over recent decades (Hare and Mantua 2000, Litzow and Mueter 2013). PC1 for Alaskan data tracks the
1813
change from abundant crustaceans to abundant salmon and groundfish that occurred in the 1980s, and
1814
there is currently no indication of abrupt change in the community state tracked by this PC (Fig. 4f.3a).
1815
The shift to more negative values for PC2 of the restricted Alaskan data suggests a trend of increases in
1816
Bering Sea jellyfish abundance and Pacific cod recruitment, increasing pink salmon catches in central and
1817
southeast Alaska and increasing coho salmon catches in southeast; and decreases Gulf of Alaska shrimp
1818
catches and decreases in the catch of coho salmon in western and central Alaska and sockeye salmon in
1819
southeast. Determining the persistence of the apparent change in PC2, and whether it indicates change in
86
1820
the second axis of variability for the larger community, as tracked by the full set of Alaskan time series,
1821
will require further years of observation.
1822
References
1823
Hare, S. R. and N. J. Mantua. 2000. Empirical evidence for North Pacific regime shifts in 1977 and 1989.
1824
1825
1826
1827
1828
1829
Progress in Oceanography 47:103-145.
Hatch, S. A. 2013. Kittiwake diets and chick production signal a 2008 regime shift in the Northeast
Pacific. Marine Ecology Progress Series 477:271-284.
Litzow, M. A. and F. J. Mueter. 2013. Assessing the ecological importance of climate regime shifts: an
approach from the North Pacific Ocean. Progress in Oceanography.
Litzow, M. A., F. J. Mueter, and A. J. Hobday. 2013. Reassessing regime shifts in the North Pacific:
1830
incremental climate change and commercial fishing are necessary for explaining decadal-scale
1831
biological variability. Global Change Biology.
1832
Zwolinski, J. P. and D. A. Demer. 2012. A cold oceanographic regime with high exploitation rates in the
1833
Northeast Pacific forecasts a collapse of the sardine stock. Proceedings of the National Academy
1834
of Sciences of the United States of America 109:4175-4180.
87
1835
1836
1837
Figure 4f.1. Winter (NDJFM) PDO-NPGO phase space, 1965-2013. Colors highlight recent years
1838
(2008-13) and two historical periods of strong PDO influence in the ecosystem (1965-77 and 1978-88).
1839
Plotted values are 3-year running means, except for 2013, which is a 2-year mean. Error bars for 2012-13
1840
are 95% CI, reflecting uncertainty associated with estimating 2013 NPGO value.
88
1841
1842
Figure 4f.2. Assessing the evidence for post-2007/08 community-level biological change at the scale of
1843
the northeast Pacific. Time series for (a) PC1, and (b) PC2 from complete data set and from a subset of
1844
time series that could be updated at least through 2010, which allowed PC scores to be estimated through
1845
2011. Error bars for PC scores calculated from partial data set = 95% CI, and reflect uncertainty
1846
associated with estimating missing values. Error bars for PC scores from full data set are omitted for
1847
clarity. Dashed vertical lines indicate 1976/77 climate regime shift and possible 2007/08 shift. Redrawn
1848
from Litzow and Mueter (in press).
89
1849
1850
Figure 4f.3. Assessing the evidence for post-2007/08 community-level biological change at the scale of
1851
Alaska (Bering Sea and Gulf of Alaska combined). Time series for (a) PC1 and (b) PC2, from complete
1852
data set and from a subset of time series that could be updated at least through 2010, which allowed PC
1853
scores to be estimated through 2011. Error bars for PC scores calculated from partial data set = 95% CI,
1854
and reflect uncertainty associated with estimating missing values. Error bars for PC scores from full data
1855
set are omitted for clarity. Dashed vertical lines indicate 1976/77 climate regime shift and possible
1856
2007/08 shift.
90
1857
Chapter 5: Zooplankton prey, growth and energy density of larval pollock, and recruitment
1858
Four abstracts are included here for papers that form the basis for a new understanding of zooplankton
1859
and pollock dynamics in the southeastern Bering Sea. The temperature – recruitment relationship
1860
presented in Coyle et al (2011) and Hunt et al (2011) was a contribution of this project and is a key
1861
relationship for projecting future trajectories of pollock recruitment and abundance (Chapter 10). In
1862
addition, two recent BEST/BSIERP contributions by Siddon et al (2013) and Heintz et al (2013) lay out
1863
the physiological foundation to support the connection between prey conditions in the summer and fall,
1864
the size and energy density of pollock in the fall, and subsequent recruitment of young pollock to age-1.
1865
1866
Coyle, K.O., Eisner, L.B., Mueter, F.J., Pinchuk, A., Janout, M.A., Cieciel, K., Farley, E.V., and
1867
Andrews, A.G. 2011. Climate change in the southeastern Bering Sea: impacts on pollock stocks
1868
and implications for the Oscillating Control Hypothesis. Fisheries Oceanography 20(2): 139-156.
1869
Abstract: Concern about impacts of climate change in the Bering Sea prompted several research programs
1870
to elucidate mechanistic links between climate and ecosystem responses. Following a detailed literature
1871
review, Hunt et al. (2011) (Deep-Sea Res. II, 49, 2002, 5821) developed a conceptual framework, the
1872
Oscillating Control Hypothesis (OCH), linking climaterelated changes in physical oceanographic
1873
conditions to stock recruitment using walleye pollock (Theragra chalcogramma) as a model. The OCH
1874
conceptual model treats zooplankton as a single box, with reduced zooplankton production during cold
1875
conditions, producing bottom-up control of apex predators and elevated zooplankton production during
1876
warm periods leading to top-down control by apex predators. A recent warming trend followed by rapid
1877
cooling on the Bering Sea shelf permitted testing of the OCH. During warm years (2003–06), euphausiid
1878
and Calanus marshallae populations declined, post-larval pollock diets shifted from a mixture of large
1879
zooplankton and small copepods to almost exclusively small copepods, and juvenile pollock dominated
1880
the diets of large predators. With cooling from 2006–09, populations of large zooplankton increased,
1881
post-larval pollock consumed greater proportions of C. marshallae and other large zooplankton, and
1882
juvenile pollock virtually disappeared from the diets of large pollock and salmon. These shifts in energy
1883
flow were accompanied by large declines in pollock stocks attributed to poor recruitment between 2001
1884
and 2005. Observations presented here indicate the need for revision of the OCH to account for shifts in
1885
energy flow through differing food-web pathways due to warming and cooling on the southeastern Bering
1886
Sea shelf.
1887
91
1888
Hunt, G.L., Coyle, K.O., Eisner, L.B., Farley, E.V., Heintz, R.A., Mueter, F., Napp, J.M., Overland, J.E.,
1889
Ressler, P.H., Salo, S., and Stabeno, P.J. 2011. Climate impacts on eastern Bering Sea foodwebs:
1890
a synthesis of new data and an assessment of the Oscillating Control Hypothesis. ICES Journal of
1891
Marine Science: Journal du Conseil 68(6): 1230-1243.
1892
Abstract: Walleye pollock (Theragra chalcogramma) is an important component of the eastern Bering Sea
1893
ecosystem and subject to major fisheries. The Oscillating Control Hypothesis (OCH) predicted that
1894
recruitment of pollock year classes should be greatest in years with early ice retreat and late blooms in
1895
warm water, because more energy would flow into the pelagic (vs. benthic) community. The OCH further
1896
predicted that, with pollock population growth, there should be a shift from bottom-up to top-down
1897
regulation. New data support the predictions that in those years with early ice retreat, more primary
1898
production accrues to the pelagic compartment and that large numbers of age-0 pollock survive to
1899
summer. However, in these years, production of large crustacean zooplankton is reduced, depriving age-0
1900
pollock of lipid-rich prey in summer and autumn. Consequently, age-0 pollock energy reserves (depot
1901
lipids) are low and predation on them is increased as fish switch to age-0 pollock from zooplankton. The
1902
result is weak recruitment of age-1 recruits the following year. A revised OCH indicates bottom-up
1903
constraints on pollock recruitment in very warm periods. Prolonged warm periods with decreased ice
1904
cover will likely cause diminished pollock recruitment and catches relative to recent values.
1905
1906
Heintz, R.A., Siddon, E.C., Farley, E.V., and Napp, J.M. 2013. Correlation between recruitment and fall
1907
condition of age-0 pollock (Theragra chalcogramma) from the eastern Bering Sea under varying
1908
climate conditions. Deep Sea Research Part II: Topical Studies in Oceanography 94: 150-156.
1909
Abstract: Fishery managers require an understanding of how climate influences recruitment if they are to
1910
separate the effects of fishing and climate on production. The southeastern Bering Sea offers
1911
opportunities to understand climate effects on recruitment because inter-annual oscillations in ice
1912
coverage set up warm or cold conditions for juvenile fish production. Depth-averaged temperature
1913
anomalies in the Bering Sea indicate the past nine years have included three warm (2003–2005), an
1914
average (2006), and five cold (2007–2011) years. We examined how these climatic states influenced the
1915
diet quality and condition (size, energy density and total energy) of young-of-the-year (YOY) pollock
1916
(Theragra chalcogramma) in fall. The implications of fall condition were further examined by relating
1917
condition prior to winter to the number of age-1 recruits-per-spawner the following summer (R/S). The
1918
percentage of lipid in pollock diets was threefold higher in cold years compared with warm years, but
1919
stomach fullness did not vary. Consequently, fish energy densities were 33% higher in cold years
92
1920
(P<0.001) than in warm years. In contrast, neither fish size (P=0.666), nor total energy (P=0.197) varied
1921
with climatic condition. However, total energy was significantly (P=0.007) and positively correlated with
1922
R/S (R2=0.736). We conclude that recruitment to age-1 in the southeastern Bering Sea is improved under
1923
environmental conditions that produce large, energy dense YOY pollock in fall.
1924
1925
Siddon, E.C., Heintz, R.A., and Mueter, F.J. 2013. Conceptual model of energy allocation in walleye
1926
pollock (Theragra chalcogramma) from larvae to age-1 in the southeastern Bering Sea. Deep Sea
1927
Research Part II: Topical Studies in Oceanography 94: 140-149.
1928
Abstract: Walleye pollock (Theragra chalcogramma) support the largest commercial fishery in the United
1929
States and are an ecologically important component of the southeastern Bering Sea (SEBS) pelagic
1930
ecosystem. Alternating climate states influence the survival of walleye pollock through bottom-up control
1931
of zooplankton communities and possible top-down control of predator abundance. Quantifying the
1932
seasonal progression and spatial trends in energy content of walleye pollock provides critical information
1933
for predicting overwinter survival and recruitment to age-1 because age-0 walleye pollock rely on energy
1934
reserves to survive their first winter. Age-0 and age-1 walleye pollock were collected in the SEBS from
1935
May to September 2008-2010. Energetic status was determined through quantification of energy density
1936
(kJ/g) and proximate composition (i.e., % lipid, % moisture) with variation in energy density primarily
1937
driven by variability in % lipid. Energy densities remained relatively low during the larval phase in
1938
spring, consistent with energy allocation to somatic growth and development. Lipid acquisition rates
1939
increased rapidly after transformation to the juvenile form (25-40 mm standard length), with energy
1940
allocation to lipid storage leading to higher energy densities in late summer. This transition in energy
1941
allocation strategies is a physiological manifestation of survival constraints associated with distinct
1942
ontogenetic stages; a strategy favoring growth to escape size-dependent predation appears limited to
1943
larval development while juvenile fish allocate proportionally more mass to lipid storage in late summer.
1944
We propose that the time after the end of larval development and before the onset of winter represents a
1945
short critical period for energy storage in age-0 walleye pollock, and that overwinter survival depends on
1946
accumulating sufficient stores the previous growing season and consequently may be an important
1947
determinant of recruitment success.
93
1948
Chapter 6: Effects of temperature and gadid predation on snow crab recruitment: Comparisons
1949
between the Bering Sea and Atlantic Canada
1950
Laurinda A. Marcello1, Franz J. Mueter1, Earl G. Dawe2, Mikio Moriyasu3
1951
1952
1
1953
States
1954
2
Fisheries and Oceans Canada, Northwest Atlantic Fisheries Centre, St John’s, NL A1C 5X1, Canada
1955
3
Fisheries and Oceans Canada, Gulf Fisheries Centre, Moncton, NB E1C 9B6, Canada
School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, Juneau, AK 99801, United
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
Citation: Marcello, L.A., Mueter, F.J., Dawe, E.G., and Moriyasu, M. 2012. Effects of temperature and
1978
gadid predation on snow crab recruitment: comparisons between the Bering Sea and Atlantic Canada.
1979
Marine Ecology Progress Series 469: 249-261.
94
1980
Abstract:
1981
Snow crab (Chionoecetes opilio) are found in many subarctic ecosystems where they are important
1982
components of marine food webs and support large commercial fisheries. Snow crab abundance is highly
1983
variable, but the causes of large changes in year-class strength are poorly known. We used a regression
1984
approach to examine the effects of snow crab spawning stock biomass, bottom water temperature, cold
1985
area or sea ice extent, and predation by Pacific cod (Gadus macrocephalus) or Atlantic cod (Gadus
1986
morhua) on snow crab recruitment in each of three ecosystems: the eastern Bering Sea, the
1987
Newfoundland-Labrador Shelf, and the southern Gulf of St. Lawrence. Comparing results across systems
1988
showed that cold ocean conditions during early life history were associated with increased snow crab
1989
recruitment or recruitment indices in all three ecosystems. However, we found no consistent evidence
1990
that spawning stock or gadid biomasses were significantly related to subsequent snow crab recruitment.
1991
Our results underscore the value of comparing multiple ecosystems and demonstrate the importance of
1992
ocean conditions in driving variability in snow crab populations.
1993
1994
Keywords: snow crab, recruitment, environment, predation, spawning stock biomass, eastern Bering Sea,
1995
Newfoundland-Labrador Shelf, southern Gulf of St. Lawrence
1996
1997
1998
1 Introduction
Snow crab (Chionoecetes opilio) are found in shelf and slope areas of several subarctic
1999
ecosystems in the Pacific and Atlantic Oceans (Sainte-Marie et al. 2008). They serve an important
2000
trophic role in these systems and support large male-only commercial fisheries. Rational management of
2001
these important resources requires a sound understanding of factors that affect their population dynamics.
2002
Snow crab abundance can be highly variable and is believed to be driven largely by changes in
2003
the strength of incoming year classes (Zheng & Kruse 2006). However, the drivers of fluctuations in
2004
production and survival of these early life history stages, before snow crab are first detected by survey
2005
gear, are poorly known. Spawning stock biomass influences production levels (Zheng & Kruse 2003,
2006
2006) but relationships with recruitment are unclear because of large variations in spawner to recruit
2007
survival. Cannibalism on newly settled juveniles by previously settled year classes may influence
2008
survival and contribute to cyclical recruitment (Sainte-Marie et al. 1996, Sainte-Marie & Lafrance 2002).
2009
Bottom-up effects from ocean-climate variability may affect recruitment and abundance in snow crab
2010
(Zheng & Kruse 2006, Dawe et al. 2008, Boudreau et al. 2011). Snow crab generally inhabit regions of
2011
very cold water (Sainte-Marie et al. 2008), especially during early benthic and juvenile stages (Dawe &
2012
Colbourne 2002, Orensanz et al. 2004), and are energetically confined to cold areas (Foyle et al. 1989).
95
2013
Hence, their distribution and abundance, as well as survival, may be limited by the spatial extent of cold
2014
bottom waters. Alternatively, snow crab population dynamics and subarctic ecosystem structure may also
2015
be regulated by predation pressure from upper trophic levels (‘top-down’ effects) (Frank et al. 2005, Link
2016
et al. 2009). In particular, recent increases in snow crab biomass in Atlantic Canada have been attributed
2017
to declines in biomass of groundfish predators (Frank et al. 2005). Changes in snow crab recruitment and
2018
abundance may be influenced by the above factors or others such as disease (Morado et al. 2010,
2019
Mullowney et al. 2011) and resource competition.
2020
By comparing snow crab dynamics across similar ecosystems we may be able to gain a better
2021
understanding of what drives ecosystem processes (Murawski et al. 2010). Subarctic ecosystems in both
2022
the North Pacific and North Atlantic Oceans share many commonalities including the presence of
2023
commercially important snow crab and large gadid predators. This study compares population dynamics
2024
of snow crab in the eastern Bering Sea (EBS), the southern Gulf of St. Lawrence (SGSL), and two regions
2025
in the Newfoundland-Labrador Shelf (NL) ecosystem to address the following question: Is snow crab
2026
abundance governed by analogous factors across subarctic ecosystems or do snow crab in each ecosystem
2027
respond to important drivers in different ways?
2028
Specifically, we investigate how the spawning stock biomass, the environment (water
2029
temperature or cold area extent), and the biomass of gadoid fishes impact snow crab recruitment or
2030
recruitment indices. Three hypotheses were tested for each of three ecosystems: HA1) snow crab
2031
recruitment shows an increasing or compensatory (e.g. asymptotic or dome-shaped) response to spawning
2032
stock biomass (density dependence); HA2) colder conditions, represented by lower temperatures, a larger
2033
spatial extent of cold bottom temperatures, or more extensive ice cover will result in increased
2034
recruitment; and HA3) increased biomass of potential predators will be associated with decreased snow
2035
crab recruitment. These hypotheses were tested in a statistical modeling framework relating estimates or
2036
indices of snow crab recruitment to variability in predation, spawning stock biomass, and the
2037
environment.
2038
2039
2 Methods
2040
2041
2042
2.1 Study areas
We examined snow crab dynamics in three subarctic ecosystems that are characterized by a high
2043
degree of seasonality in ice cover and sunlight and by large influxes of freshwater (Hunt & Drinkwater
2044
2007). Snow crab generally inhabit shelf or slope areas and are typically associated with a cold pool of
96
2045
water (Dawe & Colbourne 2002, Sainte-Marie et al. 2008). Relationships between snow crab recruitment
2046
or recruitment indices and potential explanatory variables were modeled for four regions in three large
2047
marine ecosystems: (1) EBS, (2) SGSL, Northwest Atlantic Fisheries Organization (NAFO) Division 4T,
2048
and two regions in NL, (3) NAFO Division 3K (NAFO 3K) and (4) NAFO Division 3L (NAFO 3L) (Fig.
2049
6.1). The two NL regions were modeled separately because snow crab abundance trends and thermal
2050
regimes differed between the regions. Data are aggregated at the division level before being combined
2051
for the NL system as a whole (DFO 2010); area NAFO 3K corresponds to the relatively deep and warm
2052
northeast Newfoundland Shelf while NAFO 3L is over the shallow colder Grand Bank.
2053
2054
2.2 Data
2055
2056
2057
2.2.1 Snow crab data
For the EBS and SGSL time series of estimated recruitment were used as the primary response
2058
variable. Hereafter, we use the general term “recruitment” to refer to the estimated or modeled abundance
2059
of a specific size-class, which is smaller than the smallest age group recruiting to the fishery. For the
2060
EBS, recruits become vulnerable to survey gear over the size range of 30 to 60 mm carapace width
2061
(Orensanz et al. 2004). We used the number of individuals, in millions, between 25 and 40 mm carapace
2062
width for the years 1984-2007 as estimated from a statistical, size-structured model based on survey and
2063
fishery data (Turnock & Rugolo 2010). This represents the best available index of recruitment, although
2064
it does not fully account for small crab occurring to the north of the standard survey area. Crabs of this
2065
size class are thought to be 4 years from hatching and 5 years from fertilization (Turnock & Rugolo
2066
2010). Recruitment of instar VIII snow crab in the SGSL was modeled based on survey data from 1989
2067
to 2009 (DFO 2010) and indexed on a scale from 0 to 100 for this study. The SGSL trawl survey was
2068
conducted during fall in fishing area 12, which is the largest fishing area in the SGSL. No survey was
2069
conducted in 1996 and a model-derived value was substituted for this year (Hébert et al. 1997). SGSL
2070
instar VIII snow crab are 30.7 to 34.5 mm carapace width and have been estimated to be 4.3 to 5 years
2071
from hatching (Sainte-Marie et al. 1995, Hébert et al. 2002). Although the SGSL recruits cover a
2072
narrower size range than in the EBS, the two size classes correspond to approxately the same age range.
2073
Snow crab can reach sexual maturity over several different instars, beginning from 35 (Ernst et al. 2005)
2074
to 40 mm carapace width (Hébert et al. 2002).
2075
For the two NL regions, available trawl survey time series were too short for modeling purposes.
2076
However, fluctuations in the exploitable biomass and abundance are largely dependent on the strength of
97
2077
incoming recruitment to the fishery (males larger than 94 mm carapace width) and should reflect the
2078
abundance of the youngest year classes entering the fishery (approximately age 10). Therefore, we used
2079
catch-per-unit-effort (CPUE in kg trap-1 for snow crab > 94 mm carapace width) as a crude recruitment
2080
index in these regions (1977-2008 for NAFO 3K, 1974-2008 for NAFO 3L). To validate this approach
2081
we compared fishery CPUE to trawl survey abundance indices for various limited size groups of new-
2082
shelled crabs. Convincing direct relationships were found between the CPUE and trawl survey
2083
abundance indices at appropriate lags for both NL regions (Dawe, unpubl. data). Hereafter, we refer to
2084
these CPUE series as “recruitment index.”
2085
The reproductive potential of snow crab influences larval production and this may be reflected in
2086
variations in recruitment. To address our first hypothesis linking spawner biomass to later recruitment,
2087
we used proxies for spawning stock biomass as measures of reproductive potential (Fig. 6.2). In the EBS
2088
we used an estimate of total mature snow crab biomass (morphometrically mature males plus mature
2089
females) based on survey size composition and maturity at size estimated in the stock assessment
2090
(Turnock & Rugolo 2010). In all other areas (SGSL, NAFO 3K and 3L), CPUE of male snow crab 95
2091
mm carapace width or larger was used as a proxy for spawning stock biomass, hereafter called “spawning
2092
stock biomass index”. These measures were incorporated at appropriate lags in regression models to
2093
assess potential impacts of spawner biomass on recruitment as described below.
2094
2095
2096
2.2.2 Environmental data
To capture the potential effects of ocean climate variability on snow crab recruitment we selected
2097
bottom temperatures and cold area extent (measured either as the extent of the cold layer of water that
2098
forms as a result of winter cooling and ice cover or as the spatial extent of the sea ice itself) as
2099
environmental variables for modeling (Fig. 6.2).
2100
For the EBS, a long-term proxy for bottom temperature was constructed using the NOAA
2101
extended reconstructed sea surface temperature data series (Smith et al. 2008, NOAA 2011). The proxy
2102
uses sea surface temperatures averaged over the spring and late summer periods (March, April, May, and
2103
August) when water column temperatures were most strongly correlated with average bottom
2104
temperatures over the shelf (Pearson's product moment correlation r = 0.88) due to vertical mixing.
2105
Bottom temperatures in the SGSL were measured annually at Magdalen Shallows during September
2106
(Chassé & Pettipas 2010). For both NL areas (NAFO 3K and NAFO 3L) a time series of annual mean
2107
bottom temperatures at Station 27, an oceanographic monitoring station located 7 km from St. John’s
2108
Harbour (Newfoundland), was selected as a measure of temperature variability in these regions (Dawe et
2109
al. 2008).
98
2110
Seasonal ice cover is an important feature of all the study areas examined here. It determines the
2111
spatial extent of the cold pool in the EBS and is related to the area of the cold intermediate layer in
2112
Atlantic Canada. Because the spatial extent of these cold waters has important effects on the distribution
2113
of snow crab (Dionne et al. 2003, Orensanz et al. 2004), we examined the influence of cold area extent in
2114
each region. For the EBS, cold area extent represents the fraction of the National Marine Fisheries
2115
Service summer bottom trawl survey area with water less than 2°C (F. Mueter, unpubl. data). The SGSL
2116
cold area variable represents the area (km2) of Magdalen Shallows covered by -1 to 3°C bottom water
2117
during September (Chassé & Pettipas 2010). For NAFO 3K and NAFO 3L the annual ice cover area
2118
(km2) from 45-55°N on the Newfoundland-Labrador Shelf (Dawe et al. 2008) was used to represent the
2119
spatial extent of cold water.
2120
2121
2122
2.2.3 Predators
To examine the potential effects of key predators on snow crab recruitment in each system, and to
2123
address hypothesis HA3, suitable estimates of predator biomass were compiled (Fig. 6.2). Although there
2124
are many predators of snow crab, this study focused on predation by gadoid fishes. Walleye pollock
2125
(Theragra chalcogramma) dominate gadid biomass in the EBS, but were not included here because snow
2126
crab make up a very small proportion of their diet (Aydin et al. 2007). The major gadid predator on snow
2127
crab in the EBS is Pacific cod (Gadus macrocephalus) (Livingston 1989) and we used a model-based
2128
estimate of the total biomass of Pacific cod age 0+ from the 2007 stock assessment model (Thompson et
2129
al. 2010) to quantify potential gadid predation in this system. In both Atlantic Ocean ecosystems, Atlantic
2130
cod (Gadus morhua) has historically been a predator of snow crab. In the SGSL, Atlantic cod age 3+
2131
biomass estimates from the 2009 stock assessment model (Swain et al. 2009) were used to examine
2132
predation effects. In NL regions NAFO 3K and 3L, estimates of total Atlantic cod biomass from autumn
2133
surveys (conducted through 2007) were used.
2134
2135
2136
2.3 Analytical approach
A general regression approach was used to examine the effects of snow crab spawning stock
2137
biomass, environmental variability, and predation on snow crab recruitment in each ecosystem. The basic
2138
modeling structure was as follows:
2139
yt = β0 + β1 spawners t-k1 + β2 environment t-k2 + β3 predator t-k3 + εt
(1)
2140
where yt represents snow crab recruitment (or recruitment index) in year t, βs are regression coefficients,
2141
k1, k2, and k3 represent time lags, and the errors εt are assumed to be either independent and normally
99
2142
distributed with mean zero and variance 2 or first-order autocorrelated (εt =  εt-1 +t) with auto-
2143
regressive coefficient  and innovations t that are independent and normally distributed.
2144
Recruitment data often have a skewed distribution and may require a transformation to achieve
2145
normality in model residuals. A range of Box-Cox transformations (Box & Cox 1964) were explored to
2146
identify the best power transformation for each response variable (yt) to achieve approximate normality of
2147
the residuals. Likelihood profiles indicated that a log transformation was the best possible transformation
2148
for the EBS and SGSL recruitment series, while no transformation was necessary for the NAFO 3K and
2149
NAFO 3L recruitment indices.
2150
Spawning stock biomass affects larval production, while both predation and the environment may
2151
influence survival. Each of these factors would be expected to alter population abundances several years
2152
prior to recruitment and the effects may extend over multiple years, necessitating the use of lags and
2153
moving averages for the explanatory variables. Available diet studies suggest that early life history stages
2154
of snow crab are particularly vulnerable to predation (Livingston 1989, Chabot et al. 2008). For example,
2155
in the Gulf of St. Lawrence four size groups of crab were susceptible to Atlantic cod predation (Chabot et
2156
al. 2008). These sizes roughly correspond to ages 1 to 4 (Hébert et al. 2002) although there can be some
2157
variability in growth to each size class. Similarly, environmental conditions including temperature can
2158
affect larval and juvenile stages of snow crab by influencing development rate (Moriyasu & Lanteigne
2159
1998), hatch timing (Webb et al. 2007), growth rate, and molt frequency (Orensanz et al. 2007), as well as
2160
energy budgets (Foyle et al. 1989). Since effects may range over several life stages and because snow
2161
crab cannot be aged reliably, 3 or 4-year moving averages were applied to time series of both predators
2162
and environmental variables to capture their potential effects on multiple age classes of snow crab.
2163
Likewise, moving averages were applied to spawning stock biomass series because the time period
2164
between fertilization and growth to the recruitment size class varies. The averaged variables were then
2165
lagged by k years relative to the response in year t to correspond to the approximate period when the
2166
effect is presumed to occur (year t-k). For example, the recruitment of EBS snow crab is most likely to be
2167
influenced by the parental stock 4-6 years prior (Table 6.1), hence recruits in 1995 were modeled as a
2168
function of the spawning stock biomass averaged over 1989 to 1991. In some cases, several biologically
2169
reasonable lags were considered such as in NL regions NAFO 3K and NAFO 3L where the recruitment
2170
index was based on snow crab of commercial size (>94 mm carapace width). In those areas it was
2171
reasonable to consider predation effects over several lags (ranging from 5-8 years to 7-10 years prior) due
2172
to uncertainty and variability in the age at which snow crab reach commercial size. When multiple lags
2173
were considered, the lag that resulted in the lowest p-value in a simple linear regression between the
100
2174
recruitment index or the log of recruits and each explanatory variable was selected for use in regression
2175
models. The lags and moving averages used for all variables in each model are listed in Table 6.1.
2176
For some ecosystems the length of overlapping time series was limited once the predator and/or
2177
spawner series were lagged appropriately. For these systems both “short models” including all variables
2178
and “long” models including fewer variables were tested. For example, the Atlantic cod series in NAFO
2179
3L overlapped with the response variable by only 16 years, whereas the spawner and environmental
2180
variables were available over longer time periods (n = 24 and n = 35 years, respectively). Results from
2181
analyses using the short and long series were then compared for consistency. We considered the results to
2182
be consistent when the sign of all explanatory variables remained unchanged regardless of whether the
2183
short or the long series were used.
2184
A species may exhibit non-linear, non-additive, or threshold responses to external pressures (Cury
2185
et al. 1995, Ciannelli et al. 2007, Ciannelli et al. 2008). To explore potential non-linearities and
2186
thresholds we fit exploratory generalized additive models (GAMs) to the data using a cross-validation
2187
approach before constructing a corresponding linear model for further analysis. To avoid over-fitting we
2188
limited parameter smoothness by constraining the estimated degrees of freedom (EDF) for any variable to
2189
3, corresponding to a biologically realistic set of possible responses (approximately linear, asymptotic, or
2190
dome-shaped). Partial fits for each explanatory variable and the estimated degrees of freedom were
2191
examined and a squared term was included in the full linear model for any variable with EDF > 1.5. No
2192
higher-order polynomial terms were considered as they were not deemed biologically reasonable.
2193
For each ecosystem both a temperature variable and cold area extent (or sea ice extent) were
2194
available for modeling. However, bottom temperatures were strongly correlated with cold area cover
2195
and/or sea ice extent in each system (multicollinearity). Therefore, the effects of these variables were
2196
modeled separately and compared for consistency. For each system, the initial full linear model included
2197
one environmental variable, a cod predator, and spawning stock biomass. We tested for interactions
2198
among variables in each full model, and because we found no significant interactions, we did not consider
2199
interactions further. We compared the full model with all three variables to reduced models with one or
2200
two variables using the small-sample Akaike Information Criterion (AICc). The model with the smallest
2201
AICc was selected as the final model when the difference between AIC values, ∆AICc, was larger than 4
2202
(Burnham & Anderson 2002). If differences in AICc were smaller than 4, the most parsimonious model
2203
was selected as the final model.
2204
Residuals from both the initial (full) model and from the final (AICc-best) model were examined
2205
for normality, heteroscedasticity, and autocorrelation. When significant autocorrelation (p < 0.05) was
2206
present in residuals from the full model according to a Durbin-Watson (DW) test, all model comparisons
101
2207
were based on generalized least squares fits that included a first-order auto-regressive error term. The
2208
overall significance of the final (AICc-best) model was tested using the likelihood ratio test (LRT) for
2209
generalized least squares models or F-test for linear models without autocorrelation.
2210
2211
Program R version 2.9 (R Development Core Team 2011) was used for all analyses and a level of
α = 0.05 was chosen to assess significance.
2212
2213
2214
3 Results
In the EBS cold temperatures were related to increased recruitment, consistent with our
2215
hypothesis, but results did not support a predation effect (Tables 6.2, 6.3). The final model included
2216
temperature and spawning stock biomass; both variables were negatively and linearly related to the log of
2217
recruitment (n = 24 years, LRT = 20.75, p < 0.001) and explained approximately 38% of the variability in
2218
recruitment (Fig. 6.3a). The model included a first-order auto-regressive term (φ = 0.66) to account for
2219
significant autocorrelation in the residuals (DW = 0.93, p < 0.001, Fig. 6.4a). When replacing
2220
temperature with the cold area extent (n = 22 years) in the full model, its estimated effect was not
2221
significant (t = 1.45, p = 0.164), but its sign was consistent with a positive effect of cold conditions on
2222
recruitment.
2223
In the SGSL we found a negative linear relationship between temperature and the log of
2224
recruitment, but neither a predation nor a spawner effect (n = 21 years, Tables 6.2, 6.3). The final model
2225
included the temperature variable (LRT = 20.81, p < 0.001, Table 6.2) and explained about 41% of the
2226
recruitment variability (Fig. 6.3b). A first-order auto-regressive term (φ = 0.73) was included in the model
2227
to account for residual autocorrelation (DW = 0.77, p < 0.001, Fig. 6.4b). The full model for SGSL based
2228
on cold area extent rather than temperature was not significant overall (LRT = 4.66, p = 0.32) and
2229
contained no significant variables (p > 0.05). Therefore, models with cold area extent were dropped from
2230
further consideration.
2231
In NL area NAFO 3K sea ice extent was positively related to the snow crab recruitment index 7-9
2232
years later (Table 6.2) meaning sea ice conditions appear to affect snow crab at approximately 1-3 years
2233
of age. There was no evidence of an effect of spawner stock biomass index or temperature on snow crab
2234
recruitment index and the relationships with Atlantic cod biomass were inconsistent (Table 6.3). Models
2235
for NAFO 3K that included the Atlantic cod series (n = 18 years) differed substantially from those
2236
including just other variables (n = 21 for spawners, n = 32 for temperature or ice area). Specifically, when
2237
Atlantic cod biomass was included in these models, the sign and significance of other coefficients
2238
changed drastically, most likely as a result of strong correlations between Atlantic cod and the other
2239
explanatory variables. Therefore, models including Atlantic cod were not considered further. Spawner
102
2240
stock biomass index was neither significant to the model with bottom temperature nor to the model using
2241
sea ice extent as a proxy for the cold area extent. Sea ice extent had a significant positive linear
2242
relationship with the snow crab recruitment index (Table 6.2), though it only explained a small proportion
2243
of overall variability (Fig. 6.3c, n =32 years, LRT = 33.00, p < 0.001, R2 = 0.18). A first-order
2244
autoregressive term (φ = 0.91) was included to account for significant autocorrelation in the residuals
2245
(Fig. 6.4c, DW = 0.45, p < 0.001). When using bottom temperature as the environmental variable, none of
2246
the variables in the model were significant (p > 0.05), but a negative temperature coefficient was
2247
consistent with the observed positive effect of extensive ice on the snow crab recruitment index.
2248
In the other NL region, NAFO 3L, cold conditions and low predator biomasses were associated
2249
with a high recruitment index in following years. Spawner stock biomass index showed a negative
2250
association with the later recruitment index (Tables 6.2, 6.3). Models including Atlantic cod covered a
2251
shorter time period (n = 16 years) than those with spawner stock biomass index (n = 24 years) or
2252
environmental conditions (temperature or sea ice extent, n = 35), but the sign of coefficients for spawner
2253
stock biomass index and temperature or sea ice extent remained consistent among models. Temperature
2254
or sea ice extent were significant in all models, though spawner stock biomass index was not significant
2255
in the longer (n = 24 years) model. In the short models spawner and predator biomasses were both
2256
negatively related to the subsequent snow crab recruitment index. Likewise, colder conditions were
2257
linked with a higher recruitment index. Both model fits explained about 90% of recruitment index
2258
variability over the 16 years modeled (temperature model: F(3,12) = 45.62, p < 0.001; sea ice extent model:
2259
Fig. 6.3d, F(3,12) = 43.91, p < 0.001) and there was no evidence of first-order autocorrelation in the
2260
residuals (temperature model: DW = 2.05, p = 0.22; sea ice-extent model: Fig. 4d. DW = 2.04, p = 0.22).
2261
Significant and pronounced autocorrelation was present in the recruitment or recruitment index
2262
residuals in the final models for three of four regions studied (EBS, SGSL, and NAFO 3K in the NL
2263
ecosystem), as was clearly evident in residual plots (Fig. 6.4). The first-order autoregressive structure of
2264
the generalized least squares models accounted for the observed serial correlation and ensured that
2265
significance tests and model comparisons are valid. However, clear cyclic patterns with several multi-
2266
year runs of alternating positive and negative residuals were evident in both the full models and in the
2267
AICc-best models. This suggests that models including environmental variables, predation and/or
2268
spawner effects, in combination with random measurement errors, do not fully account for the dynamics
2269
of these populations.
2270
2271
4 Discussion
2272
103
2273
2274
4.1 Overview
This study investigated the effects of three factors (spawning stock biomass, environmental
2275
conditions, and gadid biomass) on the strength of subsequent recruitment levels in snow crab and
2276
compared results across three subarctic ecosystems in the Northwest Atlantic and Northeast Pacific
2277
(Table 6.3). Our modeling results showed that ocean climate variability was the only factor that was
2278
consistently associated with recruitment in all ecosystems and regions. This strongly supports our
2279
hypothesis (HA2) that cold conditions during early life history stages promotes subsequent snow crab
2280
recruitment (Fig. 6.5). In contrast, there was little support for a consistent effect of spawning stock or
2281
predator biomasses (hypotheses HA1 or HA3) on recruitment across systems. Spawning stock biomass was
2282
inversely related to recruitment in the EBS and NAFO 3L, possibly indicating a density-dependent effect
2283
on recruitment in those areas. Atlantic cod biomass was inversely related to recruitment, consistent with a
2284
predation effect, but only in NAFO 3L.
2285
Our study shows that bottom-up processes related to ocean climate conditions appear to have a
2286
consistent influence on snow crab recruitment while the importance of gadid biomass or spawning stock
2287
biomass is less clear. This study also highlights the value in comparing multiple ecosystems because
2288
consistent responses across several systems greatly strengthen our conclusions about significant effects of
2289
temperature conditions to early life stages of snow crab.
2290
2291
2292
4.2 Environment, spawning stock biomass, and predation
Our study suggests that colder conditions during early life are associated with better snow crab
2293
recruitment in all of the systems examined. Our indices of ocean climate are closely related, and we did
2294
not address mechanisms linking climate to snow crab life history, so it is unclear whether these indices
2295
reflect a common regulatory mechanism or if they represent different processes. Climate can be linked to
2296
snow crab reproduction and survival in several ways. For example, the hatch timing of larval crab is
2297
linked to temperature (Webb et al. 2007) and ice cover can affect stratification and larval feeding
2298
conditions in the spring (Orensanz et al. 2004). Recently settled juveniles are thought to be very
2299
stenothermic and have low mobility; they could represent the “weakest link” in snow crab life history
2300
because they cannot easily escape unfavorable environmental conditions such as unfavorable
2301
temperatures (Dionne et al. 2003). Therefore, the area of cold water during summer may limit the amount
2302
of suitable habitat and hence the carrying capacity for early benthic stages (Dawe & Colbourne 2002). In
2303
addition, temperature likely affects early survival directly by regulating the energy budget of individual
2304
crabs (Foyle et al. 1989). A laboratory experiment found that for mature male snow crab taken from the
2305
Scotian Shelf, their total metabolic costs exceeded digestible energy intake in waters 7°C or higher and
104
2306
slightly exceeded digestible energy intake in 0-1 °C water (Foyle et al. 1989). Therefore, snow crab may
2307
be excluded from warm waters based on energetic demands (Foyle et al. 1989). Conversely, it is known
2308
that cold conditions promote snow crab to terminally molt at a smaller size, which may reduce the
2309
proportion of commercially harvestable males in the total adult male population (Dawe et al. this volume-
2310
b). Overall, the effects of temperature or cold area extent on early life stages seem to have a dominant
2311
effect on recruitment. However, in this study mean temperatures in each system varied within a narrow
2312
range (roughly 1.5°C) and recruitment responses should not be extrapolated outside of the modeled
2313
temperature range. Although temperatures differed substantially among regions, relationships between
2314
recruitment and temperature were similar, suggesting that snow crab are adapted to local temperature
2315
conditions. Therefore, the effect of a given temperature on recruitment must be evaluated relative to
2316
typical conditions for snow crab in each region.
2317
For the cod predators considered in this study, we found a relationship with recruitment only in
2318
NAFO 3L in the NL ecosystem. Considering the absence of any such effect in other regions, especially in
2319
the adjacent NAFO 3K area, our study found little evidence that cod biomass exhibited top-down control
2320
on larval and juvenile snow crab since the late 1970s. In both the Newfoundland-Labrador Shelf (Lilly
2321
2008) and in the Gulf of St. Lawrence (Ruppert et al. 2010), Atlantic cod populations have crashed from
2322
their historic levels and were at low abundances during most of the time period examined here.
2323
Consequently, there simply may be too few cod at the present time to exert significant population-level
2324
predation effects on young snow crab. The apparent predation effect in NAFO 3L may also result from
2325
opposing responses of snow crab and Atlantic cod to ocean climate variability, with snow crab increasing
2326
and cod decreasing during a prolonged cold period when cod was also excessively exploited (Dawe et al.
2327
2008). However, our results do not imply that predation has no effect on snow crab recruitment.
2328
Previous studies indicate that predation may be important. For example, Livingston (1989) estimated that
2329
in the EBS, Pacific cod remove 27-57 percent of age 1 snow crab from the population. Our results may
2330
differ for several reasons. First, the spatial scale of data may influence statistical analyses (Ciannelli et al.
2331
2008, Windle et al. this volume) and hence our results. Biological data for this study represented large
2332
geographic regions and will fail to detect small-scale predator-prey interactions. Second, predation can
2333
have indirect effects on snow crab recruitment by limiting their geographic range. Climate conditions
2334
may affect predator-prey relationships by altering the spatial distributions of either species (Zheng &
2335
Kruse 2006). Also, the EBS snow crab population has contracted northward in recent decades (Orensanz
2336
et al. 2004). The environmental ratchet hypothesis proposes that Pacific cod predation may help prevent
2337
snow crab from expanding and returning to more southern portions of the eastern Bering Sea shelf that
2338
provide favorable spawning habitat (Orensanz et al. 2004). Such spatial dynamics can lead to important
105
2339
predation effects that may not be evident in a relationship between time series of aggregated predator
2340
biomasses and prey recruitment. Finally, the diet of both cod species changes with age (Livingston 1989,
2341
Chabot et al. 2008) and the age classes of cod which are the most important consumers of snow crab may
2342
not be well-represented in aggregated biomass series.
2343
A recent meta-analysis by Boudreau et al. (2011) looked for evidence of top-down and bottom-up
2344
controls on snow crab in the northwest Atlantic Ocean. Similar to our study, their results linked colder
2345
temperatures during the early years after settlement with higher subsequent snow crab abundances. In
2346
contrast to our study, they found statistical evidence of top-down control by Atlantic cod on snow crab 0
2347
to 5 years later. For crab entering the fishery these lags would correspond to snow crab approximately 29
2348
- 97 mm carapace width or 3.3 – 8.7 years of age. Their result including older crab is unexpected
2349
because cod generally consume snow crab that are younger than 4.5 years post-settlement (Chabot et al.
2350
2008) and predation by cod on crabs in the NL ecosystem has been virtually non-existent for two decades,
2351
due to very low abundance of large cod (Dawe et al. this volume-a). However, at the longer lags, which
2352
correspond to predation on younger crab, their results agree with expectations from available diet studies.
2353
Our predation results may differ from those found in Boudreau et al. (2011) in part due to methodological
2354
approaches. We selected suitable lags corresponding to the crab ages most vulnerable to predation a
2355
priori in order to minimize the chance of finding spurious relationships. Boudreau et al. (2011)
2356
considered a much larger range of lags (25 years) since they were examining both the effects of crab on
2357
cod and of cod on crab. Furthermore, we used a regression approach that allowed us to consider the
2358
combined effects of multiple covariates, including spawning stock biomass, on crab recruitment at once.
2359
In contrast, Boudreau et al. (2001) considered pairwise full and partial correlation coefficients.
2360
The lack of an obvious relationship between cod biomass and snow crab recruitment in our study
2361
may reflect a true absence of top-down control by cod on snow crab. However, in light of the results
2362
from Boudreau et al. (2011) and the aforementioned drawbacks of using biomass as an indicator of
2363
overall predation pressure, we cannot discount top-down controls of cod on snow crab.
2364
Spawning stock biomass did not show a consistently positive or dome-shaped relationship with
2365
recruitment, as we hypothesized. Rather, it was significantly negatively correlated to subsequent
2366
recruitment in two areas (EBS, NAFO 3L in the NL ecosystem). This relationship could occur if our data
2367
represented the right (declining) limb of a dome-shaped curve. However, that would imply that the
2368
spawner-recruit relationships peak at low spawner biomasses. Also, climate variability and numerous
2369
other factors may affect the survival of young crab in the years between fertilization and recruitment. For
2370
example, cannibalism may reduce the abundance of recently settled snow crab (Sainte-Marie & Lafrance
2371
2002) and could perhaps confound or even mimic spawning stock biomass effects, particularly in the EBS
106
2372
and SGSL regions where the lag time between spawners and recruits was relatively short. In addition,
2373
several elements of snow crab reproductive biology may mask any direct link between adult spawning
2374
stock biomass and later recruitment. Female snow crab are able to store sperm and to use these reserves
2375
to fertilize ova (Elner & Beninger 1992), so current male or total biomass may not reflect reproductive
2376
potential in a given year. In addition, the length of embryonic development (Webb et al. 2007) and the
2377
overall reproductive cycle varies and is thought to be influenced by temperature (Kuhn & Choi 2011) and
2378
by whether the female is a primiparous or multiparous spawner (Elner & Beninger 1992). A crab that
2379
remains on a one year reproductive cycle can produce up to twice as many clutches over its reproductive
2380
lifetime than a crab on a two year reproductive cycle (Kuhn & Choi 2011). Therefore, the proportion of
2381
crab reproducing on each cycle can have a large impact on a population's overall fecundity. Finally, the
2382
geographic distribution of female snow crab may affect reproductive success. For example, in the EBS
2383
female snow crab undergo ontogenetic migrations (Ernst et al. 2005). Parada et al. (2010) used an
2384
individual-based model to show that larval retention, and presumably recruitment success, in the EBS
2385
varies greatly depending on the location of larval release. They also hypothesize that primiparous
2386
females, which aggregate in the middle domain of the EBS, may be the largest source of renewal for the
2387
EBS snow crab stock (Parada et al. 2010, Ernst et al. 2012). Hence we may have failed to detect any
2388
relationship between measures of spawning biomass and recruits because spawning biomass is weakly
2389
related to total reproductive output.
2390
Another important finding from this study is that the variables we modeled cannot fully explain
2391
the observed cycles in recruitment that are characteristic of snow crab and other crab populations. This is
2392
evident in the residual patterns for three of the regions modeled here (Fig. 6.4) and is likely to result from
2393
internal community dynamics. Such dynamics may be linked to a stronger contribution of primiparous
2394
than multiparous females to recruitment (Parada et al. 2010, Ernst et al. 2012) or because of cannibalism
2395
among cohorts within settlement grounds (Sainte-Marie et al. 1996). Alternatively, small-scale changes
2396
in climate related to atmospheric circulation like the El Niño could be related to snow recruitment cycles
2397
(Zheng & Kruse 2003). Residual cycles could also be related to changes in predation pressures owing to
2398
the spatial distribution of snow crab or their predators, as discussed earlier.
2399
2400
2401
4.3 Implications for stock assessment and management
Reliable stock assessment provides the foundation for successful fisheries management. Stock
2402
assessment scientists must make assumptions about recruitment levels to project future snow crab
2403
populations, to evaluate the effects of harvesting, and to develop or compare rebuilding plans (NOAA
2404
2001) for depleted stocks. The dominant role of bottom-up climate processes in regulating recruitment
107
2405
and the cyclic patterns we observed in model residuals imply that productivity is not stationary.
2406
Incorporating ocean climate effects and population cycles into projections will lead to more reliable
2407
predictions of future recruitment that would result in improved management advice. Precautionary
2408
reference points, guideline harvest levels, and thresholds for overfishing should be set to reflect variability
2409
in production and subsequent recruitment. Further, the current study indicates that warm conditions are
2410
linked with poor recruitment. Therefore, if climate change causes bottom conditions to continue to warm,
2411
snow crab recruitment will likely decline in our study areas. Studies such as ours, if applied to other
2412
important predatory and forage species, can be very important in developing a better basis for ecosystem-
2413
based management, especially under a scenario of a changing ocean climate.
2414
2415
2416
5 Acknowledgements
This study was part of the Ecosystem Studies of Sub-Arctic Seas (ESSAS) program, which
2417
provided travel funds for LM. The study was made possible through LM’s graduate fellowship from the
2418
University of Alaska Fairbanks Rasmuson Fisheries Research Center and additional funding from the
2419
UAF Dr. H. Richard Carlson Scholarship. Partial funding for FM and LM was provided by the North
2420
Pacific Research Board's Bering Sea Integrated Ecosystem Research Program (BSIERP). We thank
2421
Denis Chabot, Bernard Sainte-Marie, and Joel Webb for valuable discussions of snow crab early life
2422
history. Ginny Eckert and Gordon Kruse reviewed and provided helpful comments on an early draft of
2423
this paper. Special thanks to all the people who have been involved with design, sampling, and data
2424
analyses for research surveys; without their work this project would not be possible.
2425
2426
6 References
2427
Aydin K, Gaichas S, Ortiz I, Kinzey D, Friday N (2007) A comparison of the Bering Sea, Gulf of Alaska,
2428
and Aleutian Islands large marine ecosystems through food web modeling. US Dep Commer,
2429
NOAA Tech Memo. NMFS-AFSC-178
2430
2431
Boudreau SA, Anderson SC, Worm B (2011) Top-down interactions and temperature control of snow
crab abundance in the northwest Atlantic Ocean. Mar Ecol Prog Ser 429:169-183
2432
Box GEP, Cox DR (1964) An analysis of transformations. J R Statist Soc Ser B 26:211-252
2433
Burnham KP, Anderson DR (2002) Model selection and multi-model inference: A practical information-
2434
2435
2436
theoretic approach. Springer, New York
Chabot D, Sainte-Marie B, Briand K, Hanson J (2008) Atlantic cod and snow crab predator-prey size
relationship in the Gulf of St. Lawrence, Canada. Mar Ecol Prog Ser 363:227-240
108
2437
2438
2439
Chassé J, Pettipas RG (2010) Temperature conditions in the southern Gulf of St. Lawrence during 2008
relevant to snow crab. DFO Can Sci Advis Res Doc 2009/087.
Ciannelli L, Dingsor GE, Bogstad B, Ottersen G, Chan K-S, Gjoster H, Stiansen JE, Stenseth NC (2007)
2440
Spatial anatomy of species survival: Effects of predation and climate-driven environmental
2441
variability. Ecology 88:635-646
2442
2443
2444
Ciannelli L, Fauchald P, Chan KS, Agostini VN, Dingsør GE (2008) Spatial fisheries ecology: Recent
progress and future prospects. J Mar Syst 71:223-236
Cury P, Roy C, Mendelssohn R, Bakun A, Husby D, Parrish R (1995) Moderate is better: exploring
2445
nonlinear climatic effects on the Californian northern anchovy (Engraulis mordax). In: Beamish
2446
R (ed) Climate Change and Northern Fish Populations, Can Spec Publ Fish Aquat Sci, 121.
2447
Dawe EG, Colbourne EB (2002) Distribution and demography of snow crab (Chionoecetes opilio) males
2448
on the Newfoundland and Labrador Shelf. In: Paul AJ, Dawe EG, Elner R, Jamieson GS, Kruse
2449
GH, Otto RS, Sainte-Marie B, Shirley TC, Woodby D (eds) Crabs in Cold Water Regions:
2450
Biology, Management, and Economics, Alaska Sea Grant Coll Program, AK-SG-02-01,
2451
Fairbanks, AK, p 577-594
2452
Dawe E, Koen-Alonso M, Stansbury D, Mullowney D, Chabot D (this volume - a) Interactions between
2453
demersal fishes and crustaceans in two Northwest Atlantic ecosystems and effects on ecosystem
2454
structure and function. Mar Ecol Prog Ser
2455
Dawe E, Mullowney D, Moriyasu M, Wade E (this volume - b) Effect of temperature on size-at-terminal
2456
molt and skip-molting in snow crab (Chionoecetes opilio) from the Newfoundland and Labrador
2457
and Southern Gulf of St. Lawrence Regions, Canada. Mar Ecol Prog Ser
2458
Dawe EG, Parsons DG, Colbourne EB (2008) Relationships of sea ice extent and bottom water
2459
temperature with abundance of snow crab (Chionoecetes opilio) on the Newfoundland-Labrador
2460
Shelf. ICES CM Doc 2008/B02
2461
2462
DFO (2010) Assessment of snow crab in the southern Gulf of St. Lawrence (Areas 12, 19, 12E, and 12F).
DFO Can Sci Advis Sec Sci Advis Rep 2010/015
2463
Dionne Ml, Sainte-Marie B, Bourget E, Gilbert D (2003) Distribution and habitat selection of early
2464
benthic stages of snow crab Chionoecetes opilio. Mar Ecol Prog Ser 259:117-128
2465
2466
2467
Elner RW, Beninger PG (1992) The reproductive biology of snow crab, Chionoecetes opilio: A synthesis
of recent contributions. Am Zool 32:524-533
Ernst B, Armstrong DA, Burgos J, Orensanz JM (2012) Life history schedule and periodic recruitment of
2468
female snow crab (Chionoecetes opilio) in the eastern Bering Sea. Can J Fish Aquat Sci 69:532-
2469
550
109
2470
2471
2472
2473
2474
2475
2476
Ernst B, Orensanz JM, Armstrong DA (2005) Spatial dynamics of female snow crab (Chionoecetes
opilio) in the eastern Bering Sea. Can J Fish Aquat Sci 62:250-268
Foyle TP, O'dor RK, Elner RW (1989) Energetically defining the thermal limits of the snow crab. J Exp
Biol 145:371-393
Frank KT, Petrie B, Choi JS, Leggett WC (2005) Trophic cascades in a formerly cod-dominated
ecosystem. Science 308:1621-1623
Hébert M, Benhalima K, Miron G, Moriyasu M (2002) Moulting and growth of male snow crab,
2477
Chionoecetes opilio, (O. fabricius, 1788) (Decapoda, Majidae), in the southern Gulf of St.
2478
Lawrence. Crustaceana 75:671-702
2479
Hébert M, Wade E, DeGrâce P, Biron M, Hébert A, Moriyasu M (1997) 1996 assessment of snow crab
2480
(Chionoecetes opilio) stock in the southern Gulf of St. Lawrence (Areas 12, 18, 19, 25/26 and
2481
zones E and F). Can Stock Ass Res Doc 97/86 63
2482
Hunt GL, Drinkwater K (2007) Introduction to the proceedings of the GLOBEC symposium on effects of
2483
climate variability on sub-arctic marine ecosystems. Deep-Sea Res Part II 54:2453-2455
2484
Kuhn PS, Choi JS (2011) Influence of temperature on embryo developmental cycles and mortality of
2485
female Chionoecetes opilio (snow crab) on the Scotian Shelf, Canada. Fish Res 107:245-252
2486
Lilly GR (2008) The decline, recovery, and collapse of Atlantic cod (Gadus morhua) off Labrador and
2487
Eastern Newfoundland. In: Kruse GH, Drinkwater K, Ianelli JN, Link JS, Stram DL, Wespestad
2488
V, Woodby D (eds) Resiliency of gadid stocks to fishing and climate change. Alaska Sea Grant
2489
Coll Program, AK-SG-08-01, Fairbanks, AK, p 67-88
2490
2491
2492
2493
2494
Link JS, Bogstad B, Sparholt H, Lilly GR (2009) Trophic role of Atlantic cod in the ecosystem. Fish and
Fisheries 10:58-87
Livingston PA (1989) Interannual trends in Pacific cod, Gadus marcocephalus, predation on three
commercially important crab species in the eastern Bering Sea. Fish Bull 87:807-827
Morado JF, Dawe EG, Mullowney DR, Shavey CA, Lowe VC, Cawthorn RJ, Burmeister A, Zisserson B,
2495
Colbourne E (2010) Climate change and the worldwide emergence of hematodinium-associated
2496
disease: Is there evidence for a relationship? In: Kruse GH, Eckert GL, Foy RJ, Lipcius RN,
2497
Sainte-Marie B, Stram DL, Woodby D (eds) Biology and management of exploited crab
2498
populations under climate change. Alaska Sea Grant Coll Program, AK-SG-10-01 Fairbanks, AK,
2499
p 153-174
2500
Moriyasu M, Lanteigne C (1998) Embryo development and reproductive cycle in the snow crab,
2501
Chionoecetes opilio (Crustacea: Majidae), in the southern Gulf of St. Lawrence, Canada. Can J
2502
Zool 76:2040-2048
110
2503
Mullowney DR, Dawe EG, Morado JF, Cawthorn RJ (2011) Sources of variability in prevalence and
2504
distribution of bitter crab disease in snow crab (Chionoecetes opilio) along the northeast coast of
2505
Newfoundland. ICES J Mar Sci 68:463-471
2506
2507
2508
2509
2510
2511
2512
Murawski SA, Steele JH, Taylor P, Fogarty MJ, Sissenwine MP, Ford M, Suchman C (2010) Why
compare marine ecosystems? ICES J Mar Sci 67:1-9
NOAA (2001) Fisheries of the exclusive economic zone off Alaska; rebuilding overfished fisheries.
Federal Register 66:742-746
NOAA (2011) Extended Reconstructed Sea Surface Temperature (ERSST.v3b). Accessed 26 July 2011.
http://www.ncdc.noaa.gov/ersst/
Orensanz JL, Ernst B, Armstrong DA, Stabeno P, Livingston P (2004) Contraction of the geographic
2513
range of distribution of snow crab (Chionoecetes opilio) in the eastern Bering Sea: An
2514
environmental ratchet. CalCOFL Rep 45:65-79
2515
Orensanz JM, Ernst B, Armstrong DA (2007) Variation of female size and state at maturity in snow crab
2516
(Chinoecetes opilio) (Brachyura: Majidae) from the eastern Bering Sea. J Crustac Biol 27:576-
2517
591
2518
Parada C, Armstrong DA, Ernst B, Hinckley S, Orensanz JM (2010) Spatial dynamics of snow crab
2519
(Chionoecetes opilio) in the eastern Bering Sea - putting together the pieces of the puzzle. Bull
2520
Mar Sci 86:413-437
2521
2522
2523
R Development Core Team (2011) R: A language and environment for statistical computing.
http://www.R-project.org
Ruppert JLW, Fortin M-J, Rose GA, Devillers R (2010) Environmental mediation of Atlantic cod on fish
2524
community composition: an application of multivariate regression tree analysis to exploited
2525
marine ecosystems. Mar Ecol Prog Ser 411:189-201
2526
Sainte-Marie B, Gosselin T, Sévigny J-M, Urbani N (2008) The snow crab mating system: Opportunity
2527
for natural and unnatural selection in a changing environment. Bull Mar Sci 83:131-161
2528
Sainte-Marie B, Lafrance M (2002) Growth and survival of recently settled snow crab Chionoecetes
2529
opilio in relation to intra- and intercohort competition and cannibalism: a laboratory study. Mar
2530
Ecol Prog Ser 244:191-203
2531
2532
2533
2534
Sainte-Marie B, Raymond S, Brêthes J-C (1995) Growth and maturation of the benthic stages of male
snow crab, Chionoecetes opilio (Brachyura: Majidae). Can J Fish Aquat Sci 52:903-924
Sainte-Marie B, Sévigny J-M, Smith BD, Lovrich GA (1996) Recruitment variability in snow crab
(Chionoecetes opilio): Pattern, possible causes, and implications for fishery management. In:
111
2535
Baxter B (ed) High latitude crabs: biology, management, and economics. Alaska Sea Grant Coll
2536
Program, AK SG 96-02, Fairbanks, AK, p 451-477
2537
2538
2539
2540
2541
Smith TM, Reynolds RW, Peterson TC, Lawrimore J (2008) Improvements to NOAA’s historical merged
land–ocean surface temperature analysis (1880–2006). J Clim 21:2283-2296
Swain DP, Savoie L, Hurlbut T, Surette T, Daigle D (2009) Assessment of the southern Gulf of St.
Lawrence cod stock, February 2009. DFO Can Sci Advis Sec Res Doc 2009/037
Thompson GG, Ianelli JN, Lauth RR (2010) Assessment of the Pacific Cod Stock in the Eastern Bering
2542
Sea and Aleutian Islands Area. In: Stock assessment and fishery evaluation report for the
2543
groundfish resources of the Bering Sea/Aleutian Islands regions. North Pacific Fishery
2544
Management Council, Anchorage, AK
2545
Turnock BJ, Rugolo LJ (2010) Stock assessment of eastern Bering Sea snow crab. In: Stock assessment
2546
and fishery evaluation report for the king and Tanner crab fisheries of the Bering Sea and
2547
Aleutian Islands regions. North Pacific Fishery Management Council, Anchorage, AK
2548
2549
Webb JB, Eckert GL, Shirley TC, Tamone SL (2007) Changes in embryonic development and hatching in
Chionoecetes opilio (snow crab) with variation in incubation temperature. Biol Bull 213:67-75
2550
Windle MJS, Rose GA, Devillers R, Fortin M-J (this volume) Spatio-temporal variations of invertebrate-
2551
cod-temperature relationships on the Newfoundland Shelf, 1995-2009. Mar Ecol Prog Ser
2552
Zheng J, Kruse GH (2003) Stock–recruitment relationships for three major Alaskan crab stocks. Fish Res
2553
2554
2555
65:103-121
Zheng J, Kruse GH (2006) Recruitment variation of eastern Bering Sea crabs: Climate-forcing or topdown effects? Prog Oceanogr 68:184-204
112
2556
Table 6.1. Explanatory variables used in this study by region with lags considered for each
2557
variable. Lags denote the time range (in years) prior to the year when recruitment (eastern Bering
2558
Sea, EBS; southern Gulf of St. Lawrence, SGSL) or recruitment indices (in Newfoundland-
2559
Labrador NAFO Divisions 3K and 3L) was measured. Ranges include only those years during
2560
which explanatory variables may reasonably be expected to affect snow crab subsequent
2561
recruitment.
Region
Explanatory Variable
Lags (years)
EBS
Spawners
4-6
Temperature
2-4
Cold area cover
2-4
Pacific cod
2-4
Spawners
4-6
Temperature
3-5
Cold area cover
3-5
Atlantic cod
2-4
Spawners
9-11
Temperature
7-9
Sea ice extent
7-9
Atlantic cod
5-8
Spawners
9-11
Temperature
7-9
Sea ice extent
7-9
Atlantic cod
7-10
SGSL
NAFO 3K
NAFO 3L
2562
113
2563
Table 6.2. Regression coefficients with p-values (in parentheses) for final model fits of snow crab
2564
recruitment or fishery recruitment indices in four regions (eastern Bering Sea, EBS; southern Gulf
2565
of St. Lawrence, SGSL; NAFO Division 3K, NAFO 3K; and NAFO Division 3L, NAFO 3L).
2566
Coefficients denote intercept (β0) and estimated effects of spawning stock biomass (β1),
2567
temperature or sea ice extent (β2), predator biomass (β3), and a first-order autoregressive
2568
parameter (φ), if significant. Predators are Pacific cod in the EBS and Atlantic cod elsewhere.
Region
Environmental
β0
β1
β2
16.50
-5.38x10-3
-2.12
(<0.001)
(0.032)
(0.006)
β3
φ
Variable
EBS
SGSL
NAFO 3K
NAFO 3La
temperature
temperature
sea ice extent
temperature
sea ice extent
5.12
-2.85
(<0.001)
(0.020)
6.80
4.72
(0.035)
(0.034)
0.66
0.73
0.91
20.27
-0.65
-2.73
-9.63x10-3
(<0.001)
(<0.001)
(0.008)
(0.001)
20.91
-0.70
2.98
-9.63x10-3
(<0.001)
(<0.001)
(0.011)
(0.001)
2569
a
2570
longer model (n=24 years) that did not include Atlantic cod.
Results based on model fit to 16 years with data for all variables. See text for results from
114
2571
Table 6.3. Summary of estimated effects of spawning stock biomass, the environment, and
2572
predation by cod on snow crab recruitment. For the eastern Bering Sea (EBS) and southern Gulf
2573
of St. Lawrence (SGSL) the response variable was the log of the estimated number of recruits at a
2574
small size. In the two Newfoundland/Labrador systems (NAFO 3K and NAFO 3L) a fishery
2575
recruitment index was the response variable. Predators are Pacific cod (EBS) and Atlantic cod
2576
(SGSL, NAFO 3K, NAFO 3L).Variables that were significant to the final models are shown in
2577
plus or minus signs, while those that are not significant are indicated by NS.
EBS
SGSL
NAFO 3K
NAFO 3L
Spawners
-
NS
NS
-a
Temperature
-
-
NS
-
NS
NS
+
+
NS
NS
(b)
-
Cold Area Cover /
Sea Ice Extent
Predators
2578
a
2579
however they were only significant in the short model. b Results from models including and
2580
excluding Atlantic cod were not consistent.
Spawners were negatively related to the recruitment index in both short and long models;
115
2581
Figure 6.1. Study regions used to model the effects of spawners, the environment, and predation
2582
by gadoid fishes on snow crab recruitment in (a) the eastern Bering Sea (EBS) and (b) the NAFO
2583
Division 4T in the Gulf of St. Lawrence (SGSL), and NAFO Divisions 3K (NAFO 3K) and 3L
2584
(NAFO 3L) off the Newfoundland-Labrador Shelf.
116
2585
Figure 6.2. Normalized anomalies of explanatory variables (spawning stock biomass,
2586
temperature, cold area extent, and Pacific or Atlantic cod biomass) in each region were calculated
2587
from raw data, before lags and moving averages were applied. The year range plotted for each
2588
region includes the first year used in analysis to the most recent year of data available, thus
2589
showing a wider range of data than was used in models.
117
2590
Figure 6.3. Time series of snow crab recruitment or recruitment index (dots) with predicted
2591
values from best-fit models (lines) for (a) the eastern Bering Sea (b) the southern Gulf of St.
2592
Lawrence, (c) NAFO 3K, and (d) NAFO 3L (sea ice extent model). Best-fit models for each
2593
region are described in the text and summarized in Table 6.2.
118
2594
Figure 6.4. Model residuals through time for four geographic regions: (a) the eastern Bering Sea,
2595
(b) the southern Gulf of St. Lawrence, (c) NAFO 3K, and (d) NAFO 3L (sea ice extent model).
2596
Residuals from several models have long runs of negative and positive residuals indicating a
2597
possible cyclical pattern in recruitment.
119
2598
Figure 6.5. Partial fits (solid lines) with 95% pointwise confidence intervals (dashed lines)
2599
illustrating the estimated effects of temperature and ice conditions on recruitment, while holding
2600
other variables constant at their mean values. Plots show the effect of temperature on recruitment
2601
in (a) the eastern Bering Sea and (b) southern Gulf of St. Lawrence and the effect of ice cover on
2602
recruitment indices in (c) NAFO Division 3K and (d) NAFO Division 3L. Note that the axes
2603
differ among panels.
2604
2605
120
2606
Chapter 7: Patterns of change in diets of two piscivorous seabird species during 35 years in
2607
the Pribilof Islands
2608
2609
Heather M. Renner1,*, Franz Mueter2, Brie A. Drummond1, John A. Warzybok1, Elizabeth H.
2610
Sinclair3
2611
2612
1
2613
Suite 1, Homer, AK 99603, USA
2614
2
2615
Loop Rd., Juneau, AK 99801, USA
2616
3
2617
Point Way NE, Seattle, WA 98115, USA
Alaska Maritime National Wildlife Refuge, US Fish and Wildlife Service, 95 Sterling Hwy,
School of Fisheries and Ocean Sciences, University of Alaska, 315 Lena Point, 17101 Pt. Lena
NOAA/National Marine Fisheries Service, National Marine Mammal Laboratory, 7600 Sand
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
Citation: Renner, H.M., Mueter, F., Drummond, B.A., Warzybok, J.A., and Sinclair, E.H. 2012.
2632
Patterns of change in diets of two piscivorous seabird species during 35 years in the Pribilof
2633
Islands. Deep Sea Research Part II: Topical Studies in Oceanography 65-70: 273-291.
121
2634
Abstract only (see B65 final report for full paper): As upper level predators in the marine
2635
ecosystem, seabirds reflect fluctuations in the marine environment that influence their prey
2636
supply. Studies of seabird diets thus provide insight into the physical and biological mechanisms
2637
that potentially drive population changes in both predators and their prey. The eastern Bering Sea
2638
shelf, among the most productive marine ecosystems in the world, has undergone significant
2639
restructuring in recent decades that is likely to continue in light of anticipated climatic change.
2640
Using a dataset spanning 35 years at two of the Pribilof Islands in the Bering Sea, we examined
2641
temporal patterns in diet and their relationships with oceanographic variables for black-legged
2642
kittiwakes (Rissa tridactyla) and thick-billed murres (Uria lomvia), two piscivorous seabirds with
2643
differing foraging strategies. Diets varied significantly among years and between islands and
2644
species. Our substantially expanded dataset supported conclusions found in previous studies of
2645
this system, including the importance of pollock, particularly age-0 class, in kittiwake diets and
2646
the absence of capelin in diets of either species since the late 1970’s. Diets of both species
2647
contained more gadids at St. Paul Island and more squid and euphausiids at St. George Island,
2648
likely reflecting differences in foraging location between islands. We found some relationships
2649
between kittiwake diet and broad-scale oceanographic variables (Arctic Oscillation Index and
2650
regional summer sea surface temperature) but not with local physical variables. Almost no time
2651
series data exist on availability and abundance of zooplankton or forage fish species such as age-0
2652
pollock, myctophids and sandlance in the eastern Bering Sea. Our measure of diet (number of
2653
individuals within each prey type) appears too coarse for detecting complex relationships between
2654
local oceanographic variables and seabird responses, but may provide invaluable information
2655
about changes in forage fish stocks, which are frequently expensive or difficult to otherwise
2656
measure. Future diet analyses should increase emphasis on evaluating caloric input (size and
2657
nutrient composition of each prey type) as well as attempts to measure the diet of murre chicks.
122
2658
Chapter 8: Climate Change Brings Uncertain Future for Subarctic Marine Ecosystems and
2659
Fisheries
2660
Franz J. Mueter1, Elizabeth C. Siddon1, and George L. Hunt Jr.2
2661
2662
1
2663
USA
2664
2
School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, Juneau, AK, 99801,
School of Aquatic and Fishery Sciences, University of Washington, Seattle, WA, 98105, USA
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
Citation: Mueter, F.J., Siddon, E.C., and Hunt Jr., G.L. 2011. Climate change brings uncertain
2682
future for subarctic marine ecosystems and fisheries. In North by 2020: Perspectives on Alaska’s
2683
Changing Social-Ecological Systems. Edited by A.L. Lovecraft and H. Eicken. University of
2684
Alaska Press, Fairbanks, Alaska. pp. 329-357.
123
2685
Introduction
2686
Human-induced climate change resulting from increasing CO2 levels in the atmosphere is
2687
anticipated to be most pronounced at high latitudes and will affect both terrestrial and marine
2688
ecosystems (ACIA 2004; IPCC 2007). Projected changes include a continuing decrease in sea ice
2689
cover and thus reduced habitat for ice-associated marine mammals and birds, increased sea
2690
surface temperatures affecting marine fish and invertebrates, and changes in freshwater runoff
2691
(ACIA 2004).
2692
Scientists are increasingly asked to predict the possible effects of climate change on
2693
ecosystems and the organisms that inhabit them. Forecasting possible effects of increasing CO2
2694
levels on the climate itself is based on state-of-the-art global climate models. Marine scientists
2695
use regional or global ocean circulation models, coupled to these global climate models, to study
2696
and predict the physical properties of the oceans. Such coupled atmosphere-ocean models
2697
currently provide the principal basis for developing future scenarios of climate and ocean
2698
variability (IPCC 2007).
2699
Only relatively recently have such atmosphere-ocean models been linked to biological
2700
models for predicting changes in the biological properties of marine ecosystems that may result
2701
from climate change. Biological properties of interest include, for example, the abundance and
2702
growth rate of phytoplankton—small drifting plants that provide the basis for all other
2703
components of the marine food web (Fig. 8.1). These models typically include only the
2704
immediate consumers of phytoplankton (i.e., small and large zooplankton species) and are often
2705
termed nutrient-phytoplankton-zooplankton (NPZ) models. Few models currently include
2706
consumers at higher trophic levels, that is, those fish or other marine animals that graze on
2707
zooplankton or larger species. Such “end-to-end” models, linking coupled atmosphere-ocean and
2708
NPZ through higher trophic level models, are still in their infancy and do not yet provide reliable
2709
predictions of the effects of climate change on commercial fish and shellfish species.
2710
In this chapter we review some of the documented responses of marine ecosystems in
2711
general, and of commercially important fish populations off Alaska in particular, to the observed
2712
variability in climate over the last several decades. Knowing how climate and fish populations
2713
have varied in the past, coupled with a basic understanding of the functioning of marine
2714
ecosystems, allows us to draw some reasonable conclusions about how these systems may
2715
respond to future climate warming, which is projected to continue and accelerate over the coming
2716
decades (IPCC 2007). Here we focus on the eastern Bering Sea, an example of a subarctic
124
2717
ecosystem, because of its importance to Alaska’s commercial fisheries, its location at the
2718
interface between the Arctic and subarctic, and because it is the system with which we are most
2719
familiar. We present a brief characterization of this large and productive ecosystem, focusing on
2720
those features most relevant to understanding the impacts of climate variability. This is followed
2721
by a description of observed climate variability based on long-term temperature records. We then
2722
discuss climate-related changes in overall productivity, the responses of individual species and
2723
whole communities to past climate variability, and what these historical patterns may imply for
2724
possible biological responses to anticipated future warming.
2725
2726
The Importance of Ocean Temperatures
2727
Temperature is a key variable affecting marine organisms and ecosystems, as well as the
2728
fisheries and fishing communities that depend on them. Climate warming directly affects ocean
2729
temperatures and is therefore likely to affect commercial, recreational, and subsistence fisheries.
2730
In addition to being biologically relevant, temperatures are easy to measure and therefore
2731
scientists have access to longer time series of ocean temperatures than other oceanographic
2732
parameters.
2733
Changes in ocean temperature are important because most marine organisms have a limited
2734
range of temperatures in which they thrive and because these ranges differ among populations
2735
and species (Pörtner et al. 2001). How and why marine organisms are adapted to often narrow
2736
temperature ranges is less clear, but laboratory studies suggest that aerobic performance, that is,
2737
the supply of sufficient oxygen to vital tissues, drops sharply when temperatures are outside an
2738
organism’s thermal range (Pörtner 2002). This drop in oxygen supply reduces growth and
2739
movement and enhances the vulnerability of the organism to starvation and predation. Both field
2740
and modeling studies show that slower growing individuals, particularly at the larval and juvenile
2741
stages, experience lower survival than those that grow faster (Anderson 1988; Clarke 2003;
2742
Houde 1987). Through its effect on physiological rates, temperature also affects activity levels
2743
(e.g., feeding activity), the timing and rate of migrations, reproduction (e.g., egg production rates
2744
and spawning activity), and other behaviors (Brander 2010; Drinkwater et al. 2010; Ottersen et al.
2745
2010). Ecological effects of temperature on marine ecosystems, such as changes in distribution,
2746
abundance, and species composition, are a consequence of these temperature effects on the
2747
physiology and behavior of individuals.
125
2748
Community-level consequences of increasing temperatures arise because different species
2749
have different temperature tolerances. These differences affect the relative distribution and
2750
abundance of populations in two important ways. First, populations that are tied to a specific
2751
location for part or all of their life due to limited mobility or because of specific habitat
2752
requirements for spawning, feeding, or shelter are likely to decrease when temperatures exceed
2753
their physiological optimum for extended periods. Populations or species with a higher
2754
temperature optimum will have a competitive advantage, resulting in a species turnover and
2755
change in community composition. This mechanism is believed to contribute to out-of-phase
2756
oscillations of anchovies and sardines in eastern boundary upwelling systems such as the
2757
Humboldt Current off South America or the California Current (Alheit and Bakun 2010). While
2758
sardines generally dominate in warmer regimes (Chavez and Messié 2009), suitable habitat and
2759
prey availability also influence shifts between sardine- or anchovy-dominated systems (Barange
2760
et al. 2009).
2761
Second, highly mobile species will migrate to stay within their preferred thermal range,
2762
provided that suitable habitat (including sufficient prey) is available elsewhere. However,
2763
movement may be restricted if specific life stages remain linked to a geographic location.
2764
Examples of large-scale changes in the distribution of pelagic species in response to temperature
2765
variability include Pacific hake (Merluccius productus) off the US west coast, where warmer El
2766
Niño conditions lead to changes in flow of the California Current and increased availability of
2767
feeding habitat (Agostini et al. 2008), and blue whiting (Micromesistius poutassou) in the
2768
northeast Atlantic, where spawning habitat increases during warmer regimes, potentially leading
2769
to increases in stock size, wider dispersion of eggs and larvae, and increased growth rates and
2770
survival (Hátún et al. 2009).
2771
To understand the effects of climate variability on marine communities, scientists use a
2772
combination of fieldwork, laboratory studies, analyses of historical data, and modeling studies.
2773
All of these approaches are employed in the Bering Ecosystem Study (BEST) and Bering Sea
2774
Integrated Ecosystem Research Program (BSIERP), a multiyear, multi-investigator integrated
2775
study of the Bering Sea (see http://bsierp.nprb.org) that was designed to improve our ability to
2776
forecast changes in the ecosystem in response to future climate variability. We discuss some of
2777
the findings from this and other studies to illustrate the effects of climate changes on Bering Sea
2778
fish communities and to construct some possible future scenarios. We caution that the direct
2779
responses of fish and other organisms to temperature changes are modulated by simultaneous
2780
changes in food availability and predation, which are much more difficult to predict. Moreover,
126
2781
anthropogenic climate change is associated not only with rising temperatures but also with ocean
2782
acidification and increased oxygen limitation (Pörtner 2008; Pörtner et al. 2005). The combined
2783
effects of these changes are likely to interact in unknown ways and enhance the sensitivity of
2784
marine organisms to environmental variability (Pörtner et al. 2005).
2785
2786
The Eastern Bering Sea: An Ice-Dominated Subarctic Ecosystem
2787
The eastern Bering Sea is characterized by a broad continental shelf (> 500 km) with an
2788
average depth of only about 70 m, resembling a large flat plain covered with sandy and muddy
2789
substrates. These substrates harbor a rich community of bottom-dwelling fishes and invertebrates
2790
(Fig. 8.1). Waters enter the Bering Sea from the south through a number of passes in the Aleutian
2791
Islands, and circulation over the shelf is characterized by diffuse flows to the north, which
2792
eventually exit through Bering Strait into the Arctic Ocean (Schumacher and Stabeno 1998). The
2793
southeastern shelf between the Alaska Peninsula and St. Matthew Island (Fig. 8.2) is where most
2794
of the commercial fishing fleet operates and is also the most studied region of the Bering Sea.
2795
From spring to early fall, persistent oceanographic fronts (regions of rapid changes in water mass
2796
characteristics) separate the shelf into three domains: the inner shelf domain (inside of the 50 m
2797
depth contour), the middle domain (between 50 and 100 m), and the outer domain (between 100
2798
and 200 m) (Iverson et al. 1979). During the summer, the inner domain is well mixed from top to
2799
bottom, while the middle shelf is a stratified system with a tidally mixed bottom layer and wind
2800
mixed surface layer. The outer domain consists of mixed upper and lower layers separated by a
2801
zone of gradually increasing density.
2802
Climatically, the Bering Sea is a transition region between warm maritime air to the south
2803
(subarctic) and cold, dry arctic air masses to the north (Overland 1981). Variability in the Bering
2804
Sea is closely linked to the strength and position of two major weather systems that affect the
2805
path and intensity of storms across the Bering Sea: the Aleutian low and the Siberian high (ibid.).
2806
These systems have a major influence on the formation and distribution of ice, wind mixing,
2807
temperature conditions, and other oceanographic processes, which in turn affect all living
2808
components of the ecosystem. Winter temperatures and ice formation in the Bering Sea are
2809
particularly sensitive to the position of the Aleutian low and associated variations in storm tracks
2810
(Niebauer et al. 1999; Rodionov et al. 2007). If storms originating in the western North Pacific
2811
are steered into the Bering Sea (most often associated with a strong Aleutian low), southerly
2812
winds bring warm and moist maritime air poleward and increase temperatures in the Bering Sea
127
2813
(Wyllie-Echeverria and Wooster 1998). Conversely, if storms move eastward to the south of the
2814
Aleutians and continue northward over Alaska (most often associated with a weak Aleutian low
2815
and a high pressure system over Siberia), cold air moves southward over the Bering Sea between
2816
the two pressure systems and results in cool ocean temperatures and extensive sea ice cover
2817
(Rodionov et al. 2007).
2818
Winter ice cover is a dominant feature of the Bering Sea and is both a source and a
2819
consequence of extreme seasonal and interannual variability. The seasonal advance and retreat of
2820
sea ice averages about 1,700 km, the largest range for any arctic or subarctic region (Minobe
2821
2002; Walsh and Johnson 1979). Even small changes in wind speed and direction can have a
2822
large impact on the timing, extent, and duration of winter sea ice cover (Hunt and Stabeno 2002).
2823
Therefore both the spatial extent of sea ice, which covers anywhere from 20% to 56% of the
2824
Bering Sea at its winter maximum (Niebauer et al. 1999), and the timing of ice retreat in spring
2825
vary considerably from year to year. This variability is particularly pronounced in the
2826
southeastern Bering Sea, whereas the northern and northeastern portions of the continental shelf
2827
are generally covered by sea ice during winter (Fig. 8.2).
2828
Summer water temperatures on the shelf are strongly influenced by ice conditions during the
2829
preceding winter. As ice forms in the northern Bering Sea, salts are extruded and the resulting
2830
cold, salty water sinks to the bottom to form a “cold pool” (often defined as waters below 2°C)
2831
beneath the sea ice. When this ice is blown southward by winter winds, it melts at its southern
2832
edge and the cold melt water is mixed to the bottom by winter and early spring storms, adding to
2833
the southern extent of the cold pool. This cold pool is one of the defining characteristics of the sea
2834
floor habitat and is closely related to the spatial distribution of fish stocks (Mueter and Litzow
2835
2008; Wyllie-Echeverria 1996). Specifically, the extent of the cold pool defines the limit between
2836
arctic and subarctic fish communities, as many subarctic species cannot tolerate waters below
2837
about 2°C (Mueter and Litzow 2008).
2838
2839
Past Temperature Variability in the Southeastern Bering Sea
2840
Untangling past relationships between ocean temperatures and the responses of biological
2841
communities helps us to understand future responses under variable climate states, but it requires
2842
long time series of temperature measurements. Currently there are relatively few consistent, long-
2843
term measurements in the Bering Sea, and most monitoring programs are limited in time or space.
2844
For example, summer bottom trawl surveys conducted by the National Oceanic and Atmospheric
128
2845
Administration (NOAA) National Marine Fisheries Service provide an annual snapshot of
2846
summer bottom temperatures over much of the southeastern shelf (Lauth 2010).1 A moored array
2847
of instruments at a single location on the middle shelf (56.9°N, 164.1°W) has provided almost
2848
continuous records of water temperature and salinity at various depths since the summer of 1995
2849
(Stabeno et al. 2007). For a longer-term perspective, records of sea-surface temperature (SST)
2850
variability are obtained from historical at-sea observations using ships of opportunity, dedicated
2851
oceanographic surveys, data buoys, and satellite observations of SST. These and other data
2852
sources have been merged into a single dataset by NOAA scientists (Smith et al. 2008)2 and are
2853
used here to examine historical SST variability over the southeastern Bering Sea shelf.
2854
The southeastern Bering Sea shelf shows high interannual temperature variability as well as
2855
prolonged periods with below-average or above-average temperatures (Fig. 8.3). Underlying this
2856
high variability is a long-term trend of increasing temperature at a rate of approximately 0.1°C
2857
per decade with the most pronounced increase occurring during summer months (F. Mueter,
2858
unpublished data, Fig. 8.3). Sea-surface temperatures were cool during the first decades of the
2859
century and from the 1950s through the early 1970s. A relatively warm period, with high
2860
interannual variability, occurred between 1925 and the mid- to late 1940s. Temperatures
2861
increased again after the well-known 1976/1977 climate shift, which was associated with
2862
pronounced changes in plankton communities as well as changes in fish and shellfish populations
2863
throughout much of the North Pacific (Hare and Mantua 2000). Temperatures were generally
2864
warm after the climate shift, and the highest summer temperatures in the data record were
2865
observed in 2002 through 2005. However, relatively cold temperatures followed the 1998/1999
2866
La Niña event and the most extensive ice cover and coldest water column temperatures since the
2867
early 1970s have been documented in recent years, beginning in 2006 and continuing through at
2868
least the end of 2009. That said, while temperatures have generally been much lower recently,
2869
average SSTs over the shelf during late summer have stayed relatively high (Fig. 8.3).
2870
1
For an animated map of 1982–2009 bottom temperatures, see
www.afsc.noaa.gov/RACE/groundfish/images/ebs/btemps.gif.
2
Here we use Extended Reconstructed SST version 3, which was provided by the NOAA/OAR/ESRL
PSD, Boulder, Colorado, USA, from their website at www.esrl.noaa.gov/psd/.
129
2871
Observed and Predicted Effects of Temperature Variability on Marine Ecosystems and
2872
Fisheries
2873
Climate change has physiological and behavioral effects on individual fish that are expressed
2874
at both the population and ecosystem level through changes in productivity and mortality, spatial
2875
distribution, population dynamics, and community structure (Graham and Harrod 2009). Here we
2876
focus on the population level and on the ecosystems that fishes depend on, as these are the most
2877
relevant units for fisheries management. From the perspective of fisheries-dependent
2878
communities and fisheries managers, a key question about the effects of future climate variability
2879
is: How will climate change affect the overall productivity of ecosystems and the productivity,
2880
abundance, and distribution of species we depend on?
2881
2882
Changes in Overall Productivity
2883
Climate change may affect fish and fisheries by changing the total production of
2884
phytoplankton (primary production). This is important to fisheries because fish harvests from a
2885
given ecosystem increase with increasing primary production (Chassot et al. 2007; Iverson 1990;
2886
Ware and Thomson 2005). Thus if primary production increases in a given region as a result of
2887
climate change, the biomass of fish that can be sustainably harvested would likely increase; lower
2888
harvests would be predicted if primary production decreased with changing temperatures.
2889
Satellite observations and theoretical considerations suggest that primary production will
2890
decrease at lower latitudes, but effects at higher latitudes are not well understood. As the surface
2891
layer warms and becomes less dense, the water column becomes more strongly stratified and the
2892
mixing of deep, nutrient-rich waters into the surface layer is expected to decrease, thus providing
2893
fewer nutrients to fuel production in nutrient-limited areas (Behrenfeld et al. 2006). However,
2894
production at higher latitudes is limited by light for much of the year and increased stratification
2895
may increase production by keeping phytoplankton in the well-lit surface layer. Although global
2896
model results are inconclusive for the Arctic and subarctic (Steinacher et al. 2009), a regional
2897
model for the Barents Sea—which, like the Bering Sea, is relatively shallow and seasonally ice-
2898
covered—suggests moderate increases in primary production in coming decades (Ellingsen et al.
2899
2008). Annual primary production in the Barents Sea is closely linked to the position of the ice
2900
edge in spring and is substantially higher during warmer years with less ice and a higher inflow of
2901
warm Atlantic water (Wassmann et al. 2006).
130
2902
Direct field measurements of primary production are limited in time and space, but satellite-
2903
derived estimates for both the Barents Sea and the Bering Sea suggest that chlorophyll a
2904
concentrations (a measure of phytoplankton abundance) and estimated annual production are
2905
higher in warm years (Mueter et al. 2009). Warm years with less ice and an early ice retreat may
2906
have higher production due to a longer production season, which seems to be supported by a
2907
strong negative relationship between estimated annual primary production and the timing of ice
2908
retreat from the Bering Sea shelf (Mueter et al. 2009). However, productivity during summer
2909
(after the initial spring bloom) is expected to be lower when strong thermal stratification increases
2910
vertical stability, reducing water column mixing and decreasing surface nutrient levels
2911
(Sambrotto et al. 2008; Stabeno et al. 2002). Supporting evidence indicates that high water-
2912
column stability during several of the recent warm years was associated with lower summer
2913
production (Strom and Fredrickson 2008). This has important consequences for small fish feeding
2914
in the upper water column (see below) and research is ongoing to shed light on the differences in
2915
phytoplankton and zooplankton production between warm and cold years in the Bering Sea and
2916
its consequences for fish, seabirds, and mammals (e.g., www.bsierp.org).
2917
Furthermore, even if overall productivity levels do not change, changes in the pathways and
2918
ultimate fate of production are expected (Walsh and McRoy 1986) and have been observed
2919
(Grebmeier et al. 2006). In the northern part of the Bering Sea and in the Arctic, as well as during
2920
cold years in the southeastern Bering Sea, zooplankton growth is limited by cold temperatures.
2921
Much of the primary production, particularly from ice-associated spring blooms, sinks to the sea
2922
floor where it fuels the benthic system, including crab and bottom-dwelling fishes (Fig. 8.4).
2923
Nevertheless, some larger species of zooplankton such as krill (euphausiids) and large copepods
2924
require this early bloom and cool waters, and they thrive when there is an ice-associated spring
2925
bloom (Baier and Napp 2003). In contrast, during warm years on the southeastern shelf, the ice
2926
retreats early before there is sufficient light to support a phytoplankton bloom. Therefore, the
2927
bloom is delayed until solar heating stratifies the water column and triggers a phytoplankton
2928
bloom within the surface layer (Saitoh et al. 2002). Under these relatively warm conditions, small
2929
species of shelf zooplankton (e.g., small copepods) are able to take advantage of the developing
2930
phytoplankton bloom and rapidly increase in abundance, providing food for larval fishes (e.g.,
2931
walleye pollock, Theragra chalcogramma, and Pacific cod, Gadus macrocephalus) and other
2932
predators feeding in the water column (Hunt and Stabeno 2002) (Fig. 8.4).
2933
Primary production in the Arctic is expected to increase under global warming as an earlier
2934
ice melt and a longer ice-free season will greatly extend the period when sufficient light is
131
2935
available for plankton to grow. However, predicting changes in primary production and
2936
subsequent ecosystem level effects in subarctic regions is complicated by our poor understanding
2937
of the mechanisms that drive productivity. In particular, we lack understanding of the
2938
mechanisms that supply nutrients to these systems (Mueter et al. 2009) and the balance between
2939
nutrient limitation and light limitation. Moreover, the response in production of zooplankton,
2940
which act as a conduit of energy and materials from phytoplankton to fish, adds additional
2941
variability in how ecosystems will respond to changes in climate (Richardson 2008).
2942
2943
Shifts in Distribution
2944
A key consequence of a warming climate is the poleward shift in the distribution of species
2945
that have limited temperature ranges. Temperature-dependent shifts have been documented for a
2946
number of marine fish species in both the North Atlantic (Brander et al. 2003; Perry et al. 2005;
2947
Rose 2005) and North Pacific (Mueter and Litzow 2008). Here we briefly review some of the
2948
observed changes.
2949
Large pelagic fish stocks in the northeast Atlantic have been observed to change their
2950
distribution as the distribution of suitable pelagic habitat shifts. For example, both Atlantic
2951
mackerel (Scomber scombrus) and horse mackerel (Trachurus trachurus) migrated farther north
2952
into the Norwegian Sea during warm summers in the 1980s and 1990s (Iversen 2004; Skjoldal
2953
and Sætre 2004). Similarly, increased advection of warm waters into the Barents Sea provides a
2954
larger nursery area for juvenile Atlantic herring (Clupea harengus) and extends their distribution
2955
northward (Cushing 1982; Holst et al. 2004). Northward shifts in distribution of herring, capelin
2956
(Mallotus villosus), and Atlantic cod (Gadus morhua) were also observed during earlier warm
2957
periods such as 1920–1940 (Brander 2010; Drinkwater 2006; Rose 2005). Changes in the
2958
distribution of forage species such as herring and capelin, which serve as important food for
2959
larger fish, have secondary effects on the entire fish community, as they will affect the degree of
2960
overlap with predators, particularly near the edges of their range. Temperature-related shifts in
2961
distribution have also been reported for copepods, the major zooplankton prey for many fish
2962
species, with some warm-water species expanding northward over 1,000 km in the North Atlantic
2963
(Beaugrand et al. 2002). It is unclear whether the northward shift in the distribution of pelagic
2964
fish species is related to the observed shift in zooplankton, is a direct response to increasing
2965
temperatures, or reflects some other indirect mechanism.
132
2966
While pelagic species may simply follow shifts in water masses, distributional shifts in
2967
response to temperature changes have also been documented for groundfish in the northwest
2968
Atlantic (Murawski 1993), in the North Sea (Perry et al. 2005), and in the Bering Sea (Mueter and
2969
Litzow 2008). In the eastern Bering Sea, numerous subarctic groundfish species increased in
2970
abundance on the central portion of the shelf in response to recent warming (Mueter and Litzow
2971
2008). Between the early 1980s and the early 2000s, the southern edge of the cold pool retreated
2972
northward by over 200 km. This retreat was associated with a northward shift in the summer
2973
distribution of numerous fish and shellfish species, averaging approximately 12 km per decade
2974
(Fig. 8.5). This rate is remarkably similar to rates of northward displacement of groundfish
2975
reported for the northwest Atlantic (Murawski 1993) and for the North Sea (Perry et al. 2005). It
2976
is approximately twice the average rate of northward range extensions estimated for terrestrial
2977
species (Parmesan and Yohe 2003).
2978
While the northward shifts of groundfish in the Bering Sea were strongly related to
2979
temperature, they could not be explained by temperature alone and showed a nonlinear,
2980
accelerating trend over time (Mueter and Litzow 2008). Such nonlinear effects suggest a
2981
reorganization of the fish community in response to a shift in average temperature, providing a
2982
serious challenge to understanding and predicting the effects of warming on fish communities.
2983
This difficulty is highlighted by the lack of a strong response in the distribution of most taxa to
2984
the return of cool conditions on the shelf after 2006 (F. Mueter, unpublished data). While there
2985
was some displacement to the south in 2006–2009, relative to the very warm years of 2001–2005,
2986
this southward displacement was less than would be expected based on average bottom
2987
temperatures alone. However, some species do quickly respond to interannual changes in
2988
temperature. For example, walleye pollock (a “semi-pelagic” species living on the bottom as well
2989
as in the water column) are distributed along the shelf edge away from the cold pool during cold
2990
years but spread out over much of the shelf during warm years (Ianelli et al. 2009).
2991
The boundary between arctic and subarctic conditions in the eastern Bering Sea has always
2992
been fluid, but large areas of the shelf are likely to change from an arctic fish community to a
2993
subarctic fish community as the Bering Sea warms. Shifts in the relative distribution of different
2994
species on the southeastern Bering Sea shelf have already contributed to a reorganization of the
2995
groundfish community during recent warm years. In particular, subarctic species expanded their
2996
distribution to occupy large portions of the middle shelf that were formerly covered by the cold
2997
pool, increasing not only the overall abundance of fishes in this region but also species diversity
2998
and the average trophic level (Mueter and Litzow 2008). Trophic level increased because of
133
2999
decreasing abundances of arctic groundfish species, which are smaller and therefore feed on
3000
smaller prey, and increasing abundance of subarctic species, which tend to be larger and feed at a
3001
higher trophic level. These changes imply increasing top-down control of the ecosystem by
3002
predation from large groundfishes. Similar groundfish invasions may have led to collapses in
3003
crustacean stocks in the Gulf of Alaska in the late 1970s after a climate shift resulted in warming
3004
of near-shore waters (Albers and Anderson 1985; Litzow and Ciannelli 2007). Snow crab
3005
(Chionoecetes opilio, an arctic species) abundances in the Bering Sea could be similarly affected
3006
by groundfish colonization of the area that is typically occupied by a cold pool during cool years
3007
(Orensanz et al. 2004; Zheng and Kruse 2006).
3008
In spite of an increasing dominance of the subarctic fish community on the southeastern
3009
Bering Sea shelf, a large-scale expansion of subarctic species into the northern Bering Sea or
3010
through Bering Strait into the Arctic is unlikely in the coming decades. Cold, dark winters imply
3011
that the formation of seasonal sea ice and the associated cold pool on the shallow shelf north of
3012
St. Matthew Island (Fig. 8.2) will continue into the foreseeable future (P. Stabeno, NOAA-
3013
PMEL, Seattle, p.c.). Thus the northward expansion of subarctic groundfishes into this region will
3014
be limited as long as the cold pool continues to persist on the northern Bering Sea shelf.
3015
Nevertheless, earlier ice melt and warmer summer surface temperatures provide potential habitat
3016
for pelagic species such as walleye pollock and salmon, which have been observed as far north as
3017
the Chukchi and Beaufort Seas (Farley et al. in press; Irvine et al. 2009). These fish are not
3018
adapted to cold arctic winters and it is unclear whether they are able to overwinter in the Arctic or
3019
migrate long distances to overwinter in the Bering Sea or Gulf of Alaska (Irvine et al. 2009).
3020
A northward population shift of commercial species such as walleye pollock has direct
3021
consequences for the fishing fleet. For example, trawlers have to travel farther from the port of
3022
Dutch Harbor to find walleye pollock, requiring longer transit times and higher fuel costs (Karl et
3023
al. 2009). These adverse effects of climate change are not evenly distributed among vessels
3024
targeting Bering Sea pollock; some vessels have the ability to conduct seven to ten day fishing
3025
trips and to burn fish oil, a byproduct of processing fish, in their boilers and generators (Kruse
3026
2007). On the other hand, smaller vessels that deliver to shore-side processors only have capacity
3027
for two to four day trips and cannot produce fish oil to burn onboard. Therefore, the northward
3028
shift of walleye pollock may have relatively small impacts on catcher-processor vessels, but may
3029
have significant adverse impacts on the smaller catcher vessel fleet.
134
3030
Clearly, fish populations on both sides of the North Atlantic and in the northeast Pacific have
3031
responded to temperature fluctuations with a northward shift in distribution during historical and
3032
recent warm periods. While the overwhelming evidence is for northward shifts in distribution as
3033
temperatures increase, the apparent distribution of some species has shifted to the south (Brander
3034
et al. 2003; Mueter and Litzow 2008). Predicting future shifts in distribution is complicated by
3035
large variability in the responses of individual species, subsequent changes in the trophic
3036
interactions of marine food webs, and by apparent nonlinear responses to gradual changes in
3037
temperature. What we can predict with some certainty is that there will be winners and losers,
3038
with some species increasing and others decreasing in abundance. Regardless of who wins and
3039
who loses, changes in relative species composition are likely to be highly disruptive to the fishing
3040
industry and to fishing communities, even if total fish production or its economic value remains
3041
unchanged (Hamilton 2007).
3042
3043
Changes in Productivity of Major Fish Populations
3044
Subarctic marine ecosystems are characterized by a few key species that are often of
3045
commercial importance and invoke attention across interest groups. The resilience of these
3046
populations to fishing and to changes in climate is determined by their productivity, that is, the
3047
rate at which they produce offspring and the rate at which these offspring survive to a harvestable
3048
size. Fisheries scientists refer to the number of fish or shellfish surviving to a harvestable size as
3049
“recruitment.” In many commercial species, recruitment varies interannually by an order of
3050
magnitude or more. Scientists have been on a quest to understand the causes of this recruitment
3051
variability, in particular the influence of climate variability on recruitment, for well over a century
3052
(Beamish and McFarlane 1989; Drinkwater et al. 2005; Hjort 1914). Although many relationships
3053
between climate and recruitment have been described, few provide reliable short-term recruitment
3054
predictions. Therefore, modern fisheries management simply acknowledges the large
3055
uncertainties in future recruitment and focuses on developing harvest strategies that are robust to
3056
such uncertainties. However, a critical assumption of these strategies has been that the long-term
3057
average productivity of the managed stocks remains constant. Under anticipated long-term
3058
directional trends in climate, this assumption is no longer tenable and there is renewed interest in
3059
the effects of climate variability on the productivity of fish stocks.
3060
The direct and indirect effects of climate change on reproductive success and on the survival
3061
of egg, larval, and juvenile stages clearly affects recruitment and therefore future abundances of
135
3062
fish stocks. Given the difference in thermal ranges among species, some populations will benefit
3063
from changing climate conditions while others will not. Conventional wisdom suggests that
3064
species at the northern end of their range will benefit from increased temperatures. Positive
3065
correlations between recruitment and temperature have indeed been documented for northern
3066
populations (Myers 1998), including many groundfish populations in the North Pacific (Hollowed
3067
et al. 2001; Hollowed and Wooster 1992). Similarly, many salmon populations in Alaska produce
3068
stronger year classes when juveniles encounter warm coastal sea surface temperatures during
3069
early marine life (Mueter et al. 2002). Notably, most of the above studies used a meta-analytical
3070
approach that considered evidence from a large number of populations simultaneously, thus
3071
indicating broad applicability of the results.
3072
Predictions at the individual species level are generally much more uncertain because many
3073
factors besides temperature affect growth and survival, particularly food availability and
3074
predation. Regardless of temperature, reduced food availability results in slow growth or even
3075
starvation, and enhanced predation will result in increased mortality. Enhanced predation can
3076
occur when an important predator increases in abundance or when alternative prey is less
3077
available. Both of these mechanisms have been invoked to explain fluctuations in the eastern
3078
Bering Sea population of walleye pollock, which has been the most abundant and commercially
3079
most important fish species in the eastern Bering Sea for at least three decades. A brief
3080
examination of our current understanding of pollock dynamics can serve to illustrate the
3081
complexities of fish population responses in a variable environment and the difficulties of
3082
predicting the response of such populations to future climate variability.
3083
3084
Case Study: Eastern Bering Sea Walleye Pollock
3085
The Bering Sea ecosystem, which responds to atmospheric anomalies on relatively short time
3086
scales (Napp and Hunt 2001), experiences multiyear periods of warm and cool conditions within
3087
longer-term phases of the Pacific Decadal Oscillation and the Arctic Oscillation (Macklin et al.
3088
2002). The Oscillating Control Hypothesis (Hunt et al. 2002; Fig. 8.6) provides a theoretical
3089
framework within which to predict ecosystem responses, and the response of walleye pollock, to
3090
such “warm” and “cold” regimes (several consecutive years of similar temperature conditions).
3091
During a cold regime, the recruitment of walleye pollock is highly variable but is lower on
3092
average, all else being equal (Mueter et al. 2006). Increased sea ice extent and a delay in the ice
3093
retreat lead to an ice-associated phytoplankton bloom in relatively cold waters. Cold temperatures
136
3094
delay zooplankton growth, and thus much of the primary production sinks unused to the bottom.
3095
Therefore, recently hatched fish larvae that feed in the surface layer, including walleye pollock,
3096
may not find sufficient small prey during cold years and fewer pollock larvae survive through
3097
their first summer (Moss et al. 2009). However, the surviving larvae need less food in cold water
3098
due to lower metabolic demands and may encounter larger prey (Fig. 8.6). Thus excess energy
3099
from any available prey can be stored as lipids. This stored energy provides larvae with a buffer
3100
against starvation as they go into their first winter (R. Heintz, NOAA/AFSC, Juneau, unpublished
3101
data), a critical period that is thought to greatly influence year class strength for pollock. Cold
3102
conditions have a different effect on one-year-old juveniles, which are restricted in their
3103
distribution to the outer shelf domain by an extensive cold pool, increasing their overlap with
3104
adult pollock (Wyllie-Echeverria and Wooster 1998). Increased spatial overlap is associated with
3105
increased predation by cannibalistic adults and decreased survival of juvenile pollock (Mueter et
3106
al. 2006). In contrast, the distribution of larvae is less affected by the cold pool because larvae
3107
drift passively in the surface layer. Clearly, cold conditions have different effects on early larvae,
3108
late larvae, and juveniles, but on balance recruitment tends to be lower during cold conditions. In
3109
addition, any bottom-up effects of cold conditions are modified by the abundance of cannibalistic
3110
adults, which tend to be high at the beginning of a cold regime, further contributing to poor
3111
survival and recruitment.
3112
In a warm regime, the ice retreats early while late winter storms mix the phytoplankton deep
3113
into the water column where it is too dark for phytoplankton to grow. Thus the spring bloom is
3114
delayed until solar heating stratifies the water column later in the season. In warm water, the
3115
developing spring bloom is largely consumed within the upper water column by zooplankton,
3116
which in turn provide food for pelagic predators. Thus more of the production is cycled within the
3117
pelagic system to the benefit of larval and juvenile pollock and other pelagic fishes (Hunt and
3118
Stabeno 2002). However, with warmer water comes an increased metabolic demand. If enough
3119
food is available, this demand is met by increased consumption, resulting in higher growth rates.
3120
Faster growth can be both beneficial as larvae out-grow vulnerability to predators (Houde 1987)
3121
and detrimental as larvae exhaust their own food supply. Additionally, in warmer water, juvenile
3122
and adult pollock spread across the Bering Sea shelf (inner to outer domains), and therefore
3123
larvae may be less subject to cannibalism (Wyllie-Echeverria and Wooster 1998). However,
3124
multiple consecutive warm years, with strong year classes of recruiting pollock, begin to exert
3125
top-down control of larvae through cannibalism, eventually reducing survival and recruitment
3126
(Hunt et al. 2002).
137
3127
Until recently, the predominant view was that warm conditions on balance had a positive
3128
effect on the survival and recruitment of walleye pollock. However, the occurrence of four
3129
unusually warm years (2002–2005) with below-average year-class strength of walleye pollock led
3130
to concerns over declining pollock abundances and prompted a new look at what drives
3131
recruitment of walleye pollock in the eastern Bering Sea. The new emerging view suggests that
3132
temperatures during the recent warm years may have exceeded a threshold beyond which warmer
3133
temperatures no longer imply better survival of walleye pollock during the larval and juvenile
3134
stages (Fig. 8.7). The mechanisms that caused poor survival during the unusually warm years are
3135
not fully understood but are likely related to poor prey availability. Larval pollock can benefit
3136
from warmer temperatures only if enough suitable food for successful growth and survival is
3137
available. While small zooplankton were abundant during the warm summers of 2002–2005, the
3138
large copepod Calanus marshallae and the shelf euphausiid Thysanoessa raschii were scarce on
3139
the shelf, which may have resulted in poor feeding conditions for walleye pollock larvae in
3140
summer (Coyle et al. 2008; Hunt et al. 2008). Young walleye pollock, while abundant, consumed
3141
primarily small zooplankton species and had poor body condition with an energy content that was
3142
unlikely to be sufficient for overwinter survival (Moss et al. 2009; R. Heintz, unpublished data).
3143
In contrast, in the cold years of 1999 and 2006–2009, the large copepod species C. marshallae
3144
and the euphausiid T. raschii were abundant during summer. Young walleye pollock, although
3145
not as abundant as in the warmer years, were in good condition with high energy density.
3146
Additionally, in the cold years, predators may have focused on the large copepods and
3147
euphausiids, thus consuming fewer larval pollock. Early indications suggest that at least the 2006
3148
year class will be a strong year class and is likely to increase pollock biomass, and hence catch
3149
quotas, in the coming years (Ianelli et al. 2009).
3150
Our current understanding is that late summer conditions during unusually warm years on the
3151
eastern Bering Sea shelf result in a zooplankton assemblage of largely small species that provide
3152
poor prey for young walleye pollock. The expectation is that such warm periods will become
3153
more frequent under continued warming (N. Bond, NOAA/PMEL, Seattle, p.c.), suggesting that
3154
average recruitment of walleye pollock may decrease in the future. However, while the future
3155
biomass of pollock is primarily determined by the strength of incoming year classes (i.e.,
3156
recruitment), growth conditions for older, post-recruitment pollock also contribute to variability
3157
in biomass. In contrast to poor growth conditions for young pollock, warmer summers may
3158
benefit older pollock. For example, trends in size-at-age and weight-at-length for walleye pollock
3159
and other groundfish in the eastern Bering Sea over a period that spans both the recent warm and
138
3160
cold periods show that both size and weight tended to be above the long-term mean in warmer
3161
years, and below the mean in cold years (Fig. 8.8). Moreover, net primary production at the base
3162
of the food chain was higher in warm years, as were surface chlorophyll a values, although
3163
phytoplankton cells and zooplankton were generally smaller, suggesting changes in the pathways
3164
of energy from lower to higher trophic levels. It remains to be seen whether the transfer of energy
3165
to higher trophic levels, including fish, is more or less efficient under warm conditions.
3166
Nevertheless, these findings suggest that the Bering Sea should be particularly productive in
3167
future warm periods, and post-recruitment fish may enjoy higher growth rates. Thus improved
3168
growth of older pollock may at least partially compensate for reduced recruitment.
3169
The above example illustrates the complexities of predicting the responses of individual
3170
species, let alone entire communities, to future climate changes. Despite the complex
3171
mechanisms, we suggest that reasonable predictions about the response of walleye pollock to
3172
future climate variability can be based on simple observed relationships such as that between
3173
pollock survival and water temperature (Fig. 8.7). The dome-shaped relationship is consistent
3174
with earlier studies that showed higher survival of pollock recruitment during warm years
3175
(Hollowed et al. 2001; Hunt and Stabeno 2002; Mueter et al. 2006), corresponding to the
3176
ascending part of the curve in Figure 8.7, as well as with the more recent findings that survival is
3177
low when average summer surface temperatures exceed 9–9.5°C. Under the BEST/BSIERP
3178
research program, we will be using such simple relationships, combined with temperature
3179
projections from the Intergovernmental Panel on Climate Change (IPCC) climate scenarios, to
3180
forecast pollock recruitment for the coming decades. These forecasts will complement predictions
3181
from an “end-to-end” ecosystem model of the eastern Bering Sea that is currently being
3182
developed as part of BEST/BSIERP.
3183
3184
Summary, Conclusions, and Recommendations
3185
We highlighted two important ecological consequences of changing water temperatures on
3186
the Bering Sea, an ice-dominated subarctic marine ecosystem: (1) changes in the relative
3187
abundance of different species resulting from differential shifts in the distribution of component
3188
species, and (2) changes in abundance of individual species resulting from increases or decreases
3189
in average survival and subsequent recruitment. To examine possible effects on recruitment, we
3190
chose to examine walleye pollock, one of the commercially most important species worldwide, as
3191
a case study.
139
3192
We found that distributional changes under continued warming of the Bering Sea are likely to
3193
profoundly alter the biogeography of large portions of the eastern Bering Sea shelf as subarctic
3194
species expand into warmer and more favorable habitat. This reorganization of the fish
3195
community will affect species differently. Some, such as snow crab, are likely to decline, while
3196
others, such as many flatfish, may expand their distribution and increase in abundance. These
3197
changes will likely be confined to the southeastern Bering Sea shelf because the cold pool will
3198
continue to form on the northern Bering Sea shelf in the foreseeable future, thereby limiting the
3199
expansion of subarctic groundfish species into the northern Bering Sea and Arctic Ocean.
3200
However, pelagic species may take advantage of a longer ice-free season and warmer surface
3201
temperatures to expand their summer feeding range into northern waters, including the Chukchi
3202
and Beaufort Seas.
3203
Changes in recruitment success and abundance are much more difficult to predict than
3204
changes in distribution. There is an expectation that many species at the northern end of their
3205
range, for example, Atlantic cod (Drinkwater 2005) and salmon (Mueter et al. 2002), may benefit
3206
from increasing temperatures. However, the walleye pollock example illustrates that such
3207
generalizations are far from certain. The response of an individual population is complicated by
3208
nonlinearities and thresholds in ecological relationships that provide serious challenges to
3209
predicting future responses of biological communities to climate change. Examining how
3210
populations and ecosystems responded to climate variability in the past provides the best basis we
3211
currently have for predicting future responses. Although observed correlations do not imply
3212
causation, they may nevertheless allow useful predictions. For example, the empirical relationship
3213
between temperature and pollock survival (Fig. 8.7) is a consequence of complex interactions
3214
among temperature conditions, developmental rates and growth of pollock, the availability of
3215
suitable prey, and the abundance and distribution of predators. All of these factors are directly or
3216
indirectly affected by temperature variability. Therefore, as long as the underlying relationships
3217
do not fundamentally change—that is, as long as the “rules of the game” remain the same—we
3218
can expect reduced survival of juvenile pollock during exceptionally warm years in the future.
3219
Some impacts of climate change on marine fish communities are predictable, such as the
3220
competitive disadvantage of cold-adapted arctic species as the subarctic-Arctic boundary shifts to
3221
the north. Other impacts, such as changes in complex food web interactions and their effects on
3222
specific species of interest, are difficult to predict even in simple systems. These difficulties
3223
greatly increase uncertainty about the fate of marine fish communities under a changing climate,
3224
beyond the large uncertainties in population dynamics that scientists and managers have long
140
3225
been accustomed to. Partly in response to these added uncertainties, federal fisheries managers in
3226
Alaska have taken several proactive measures to protect fisheries resources while still being able
3227
to adapt to changing conditions. These measures include closing the northern Bering Sea to
3228
bottom trawling until further research is conducted, establishing a “Northern Bering Sea Research
3229
Area,” and prohibiting commercial fishing in federally managed arctic waters until more
3230
information is available to support sustainable fisheries management (see Criddle et al., Chapter
3231
5.2).
3232
In addition to these measures, new approaches to the management of commercial fisheries in
3233
the areas currently open to fishing are also needed. First, new multi-species approaches to
3234
management are needed because current single-species approaches are unable to deal with
3235
anticipated species turnover as formerly abundant species may dwindle and formerly minor
3236
species may increase in abundance. Such large-scale changes in the relative mix of species are
3237
possible and even likely as the climate continues to change. They are certainly not unprecedented,
3238
as is evident in the decline of crab and shrimp fisheries and the rise of pollock, cod, and flatfish
3239
fisheries after a pronounced climate shift in the late 1970s. Such shifts will always be disruptive,
3240
but the public, scientists, and managers could be better prepared by considering such dramatic,
3241
but realistic, future scenarios when determining and evaluating management strategies. The
3242
question of how hard to fish a stock that may inevitably decline, or when to stop fishing a stock to
3243
allow for the possibility of rebuilding, must be openly confronted in a multi-species context.
3244
Second, anticipated shifts in the distributions of populations under climate change require
3245
more attention to spatial management. Fixed spatial management areas that are designed to
3246
protect specific habitats and life stages (e.g., the Bristol Bay red king crab Paralithodes
3247
camtschaticus savings area [Witherell and Woodby 2005]) may need to be reevaluated in light of
3248
changes in distribution. Similarly, new spatial management measures may be needed where none
3249
existed previously. This is particularly true for species that are fished near the edge of their
3250
current distribution and where harvest rates may be disproportionately high. For example, the
3251
fishery for snow crab in the Bering Sea is concentrated in the southern portion of the snow crab
3252
distributional range, an area that may be disproportionately important for reproduction and
3253
successful future recruitment (Parada et al. in press). This could exacerbate the “environmental
3254
ratchet effect” (Orensanz et al. 2004), which tends to reduce the abundance of snow crab near the
3255
southern end of their distribution.
141
3256
In short, our ability to forecast future changes in fish communities will remain limited.
3257
Therefore, it is essential to employ management strategies that work well under large variability
3258
in abundance, that can deal with possible turnovers in species dominance, and that have the
3259
flexibility to respond to unexpected events.
3260
3261
References
3262
Agostini, V. N., A. N. Hendrix, A. B. Hollowed, C. D. Wilson, S. D. Pierce, and R. C. Francis.
3263
2008. Climate–ocean variability and Pacific hake: A geostatistical modeling approach.
3264
Journal of Marine Systems 71, 237–248.
3265
Albers, W. D., and P. J. Anderson. 1985. Diet of Pacific cod, Gadus macrocephalus, and
3266
predation on the northern pink shrimp, Pandalus borealis, in Pavlof Bay, Alaska. Fishery
3267
Bulletin 83(4), 601–610.
3268
Alheit, J., and A. Bakun. 2010. Population synchronies within and between ocean basins:
3269
Apparent teleconnections and implications as to physical-biological linkage mechanisms.
3270
Journal of Marine Systems 79(3–4), 267–285.
3271
3272
3273
3274
3275
3276
Anderson, J. T. 1988. A review of size dependent survival during pre-recruit stages of fishes in
relation to recruitment. Journal of Northwest Atlantic Fishery Science 8, 55–66.
Arctic Climate Impact Assessment (ACIA). 2004. Impacts of a warming Arctic. Arctic Climate
Impact Assessment. Cambridge, UK: Cambridge University Press.
Baier, C. T., and J. M. Napp. 2003. Climate-induced variability in Calanus marshallae
populations. Journal of Plankton Research 25(7), 771–782.
3277
Barange, M., J. Coetzee, A. Takasuka, K. Hill, M. Gutierrez, Y. Oozeki, C. v. d. Lingen, and V.
3278
Agostini. 2009. Habitat expansion and contraction in anchovy and sardine populations.
3279
Progress in Oceanography 83(1–4), 251–260.
3280
Beamish, R. J., and G. A. McFarlane. 1989. Effects of ocean variability on recruitment and an
3281
evaluation of parameters used in stock assessment models. Canadian Special Publication of
3282
Fisheries and Aquatic Science 108, 379pp.
3283
Beaugrand, G., P. C. Reid, F. Ibanez, J. A. Lindley, and M. Edwards. 2002. Reorganization of
3284
North Atlantic marine copepod biodiversity and climate. Science 296(5573), 1692–1694.
3285
Behrenfeld, M. J., R. O'Malley, T. Siegel, D. A. McClain, C. R. Sarmiento, G. C. Feldman, A. J.
3286
Milligan, P. Falkowski, R. M. Letelier, and E. S. Boss. 2006. Climate driven trends in
3287
contemporary ocean productivity. Nature 444, 752–755.
142
3288
3289
Brander, K. 2010. Impacts of climate change on fisheries. Journal of Marine Systems 79(3–4),
389–402.
3290
Brander, K. M., G. Blom, M. F. Borges, K. Erzini, G. Henderson, B. R. MacKenzie, H. Mendes,
3291
J. Ribeiro, A. M. P. Santos, and R. Toresen. 2003. Changes in fish distribution in the eastern
3292
North Atlantic: Are we seeing a coherent response to changing temperature? ICES Marine
3293
Science Symposia 219, 261–270.
3294
Chassot, E., F. Mélin, O. L. Pape, and D. Gascuel. 2007. Bottom-up control regulates fisheries
3295
production at the scale of eco-regions in European seas. Marine Ecology Progress Series 343,
3296
45–55.
3297
3298
3299
3300
Chavez, F. P., and M. Messié (2009). A comparison of Eastern boundary upwelling ecosystems.
Progress in Oceanography 83(1–4), 80–96.
Clarke, A. 2003. Costs and consequences of evolutionary temperature adaptation. Trends in
Ecology & Evolution 18(11), 573–581.
3301
Coyle, K. O., A. I. Pinchuk, L. B. Eisner, and J. M. Napp. 2008. Zooplankton species
3302
composition, abundance and biomass on the eastern Bering Sea shelf during summer: The
3303
potential role of water-column stability and nutrients in structuring the zooplankton
3304
community. Deep Sea Research I 55(16), 1775–1791.
3305
Cushing, D. 1982. Climate and fisheries. London: Academic Press.
3306
Drinkwater, K. F. 2005. The response of Atlantic cod (Gadus morhua) to future climate change.
3307
3308
3309
ICES Journal of Marine Science 62, 1327–1337.
Drinkwater, K. F. 2006. The regime shift of the 1920s and 1930s in the North Atlantic. Progress
in Oceanography 68, 134–151.
3310
Drinkwater, K. F., G. Beaugrand, M. Kaeriyama, S. Kim, G. Ottersen, R. I. Perry, H.-O. Pörtner,
3311
J. J. Polovina, and A. Takasuka. 2010. On the processes linking climate to ecosystem
3312
changes. Journal of Marine Systems 79(3–4), 374–388.
3313
Drinkwater, K. F., H. Loeng, B .A. Megrey, N. Bailey, and R. M. Cook. 2005. Introduction: ICES
3314
Symposium on the influence of climate change on North Atlantic fish stocks. ICES Journal of
3315
Marine Science 62(7), 1203–1204.
3316
3317
3318
3319
Ellingsen, I. H., P. Dalpadado, D. Slagstad, and H. Loeng. 2008. Impact of climatic change on the
biological production in the Barents Sea. Climatic Change 87, 155–175.
Farley, E. V., Jr., J. Murphy, J. Moss, A. Feldmann, and L. Eisner. In press. Marine ecology of
western Alaska juvenile salmon, American Fisheries Society Symposium Proceedings.
143
3320
3321
Graham, C. T., and C. Harrod. 2009. Implications of climate change for the fishes of the British
Isles. Journal of Fisheries Biology 74(6), 1143–1205.
3322
Grebmeier, J. M., J. E. Overland, S. E. Moore, E. V. Farley Jr., E. C. Carmack, L. W. Cooper, K.
3323
E. Frey, J. H. Helle, F. A. McLaughlin, and S. L. McNutt. 2006. A major ecosystem shift in
3324
the northern Bering Sea. Science 311, 1461–1464.
3325
3326
3327
3328
Hamilton, L. C. 2007. Climate, fishery and society interactions: Observations from the North
Atlantic. Deep Sea Research II. 54, 2958–2969.
Hare, S. R., and N. J. Mantua. 2000. Empirical evidence for North Pacific regime shifts in 1977
and 1989. Progress in Oceanography 47, 103–145.
3329
Hátún, H., M. R. Payne, G. Beaugrand, P. C. Reid, A. B. Sandø, H. Drange, B. Hansen, J. A.
3330
Jacobsen, and D. Bloch. 2009. Large bio-geographical shifts in the north-eastern Atlantic
3331
Ocean: From the subpolar gyre, via plankton, to blue whiting and pilot whales. Progress in
3332
Oceanography 80(3–4), 149–162.
3333
3334
Hjort, J. 1914. Fluctuations in the great fisheries of northern Europe viewed in the light of
biological research. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 20, 1–228.
3335
Hollowed, A. B., S. R. Hare, and W. S. Wooster. 2001. Pacific basin climate variability and
3336
patterns of Northeast Pacific marine fish production. Progress in Oceanography 49(1), 257–
3337
282.
3338
3339
3340
3341
3342
3343
3344
3345
Hollowed, A. B., and W. S. Wooster. 1992. Variability of winter ocean conditions and strong
year classes of northeast Pacific groundfish. ICES Marine Science Symposium 195, 433–444.
Holst, J. C., I. Røttingen, and W. Melle. 2004. The herring, in The Norwegian Sea Ecosystem.
Edited by H. R. Skjoldal. Tapir Academic Press, Trondheim, Norway.
Houde, E. D. 1987. Fish early life dynamics and recruitment variability. American Fisheries
Society Symposium 2, 17–29.
Hunt, G. L., Jr., and P. J. Stabeno. 2002. Climate change and the control of energy flow in the
southeastern Bering Sea. Progress in Oceanography 55, 5–22.
3346
Hunt, G. L., Jr., P. J. Stabeno, S. Strom, and J. M. Napp. 2008. Patterns of spatial and temporal
3347
variation in the marine ecosystem of the southeastern Bering Sea, with special reference to
3348
the Pribilof Domain. Deep Sea Research II 55(16), 1919–1944.
3349
Hunt, G. L., Jr., P. Stabeno, G. Walters, E. Sinclair, R. D. Brodeur, J. M. Napp, and N. A. Bond.
3350
2002. Climate change and control of the southeastern Bering Sea pelagic ecosystem. Deep
3351
Sea Research II 49(26), 5821–5853.
144
3352
Ianelli, J. N., S. Barbeaux, T. Honkalehto, S. Kotwicki, K. Aydin, and N. Williamson. 2009.
3353
Assessment of the Walleye Pollock stock in the Eastern Bering Sea. In Stock assessment and
3354
fishery evaluation report for the groundfish resources of the Bering Sea/Aleutian Islands
3355
regions. North Pacific Fishery Management Council, Anchorage.
3356
Intergovernmental Panel on Climate Change (IPCC). 2007. Climate Change 2007: The physical
3357
science basis. Contribution of Working Group I to the Fourth Assessment Report of the
3358
Intergovernmental Panel on Climate Change. Cambridge, UK, and New York: Cambridge
3359
University Press.
3360
Irvine, J. R., R. W. Macdonald, R. J. Brown, L. Godbout, J. D. Reist, and E. C. Carmack. 2009.
3361
Salmon in the Arctic and how they avoid lethal low temperatures. North Pacific
3362
Anadradromous Fish Commission Bulletin 5, 39–50.
3363
3364
3365
3366
Iversen, S. A. 2004. Mackerel and horse mackerel. In The Norwegian Sea Ecosystem. Edited by
H. R. Skjoldal. Trondheim, Norway: Tapir Academic Press.
Iverson, R. L. 1990. Control of marine fish production. Limnology and Oceanography 35, 1593–
1604.
3367
Iverson, R. L., L. K. Coachman, R. T. Cooney, T. S. English, J. J. Goering, G. L. Hunt Jr., M. C.
3368
Macauley, C. P. McRoy, W. S. Reeburgh, and T. E. Whitledge. 1979. Ecological significance
3369
of fronts in the southeastern Bering Sea. In Ecological Processes in Coastal and Marine
3370
Systems. Edited by R. J. Livingston. New York: Plenum Press.
3371
3372
Karl, T., J. Melillo, and T. Petersen. 2009. Regional climate impacts: Alaska, in Global Climate
Change Impacts in the United States. New York: Cambridge University Press.
3373
Kruse, G. 2007. The Written Testimony of Gordon Kruse, Ph.D., University of Alaska Fairbanks
3374
School of Fisheries and Ocean Sciences, Juneau Center, Juneau, Alaska. Hearing on the
3375
Effects of Climate Change and Ocean Acidification on Living Marine Resources before the
3376
Senate Committee on Commerce, Science, and Transportation, Subcommittee on Oceans,
3377
Atmosphere, Fisheries, and Coast Guard. Washington, DC.
3378
Lauth, R. R. 2010. Results of the 2009 eastern Bering Sea continental shelf bottom trawl survey
3379
of groundfish and invertebrate resources. US Dep. Commer., NOAA Tech. Memo. NMFS-
3380
AFSC-204, 228.
3381
3382
3383
3384
Litzow, M. A., and L. Ciannelli. 2007. Oscillating trophic control induces community
reorganization in a marine ecosystem. Ecology Letters 10, 1–11.
Macklin, S. A., V. I. Radchenko, S. Saitoh, and P. J. Stabeno. 2002. Variability in the Bering Sea
ecosystem. Progress in Oceanography 55, 1–4.
145
3385
3386
Minobe, S. 2002. Interannual to interdecadal changes in the Bering Sea and concurrent 1998/99
changes over the North Pacific. Progress in Oceanography 55, 45–64.
3387
Moss, J. H., E. V. Farley Jr., A. M. Feldman, and J. N. Ianelli. 2009. Spatial distribution,
3388
energetic status, and food habits of Eastern Bering Sea age-0 walleye pollock. Transactions
3389
of the American Fisheries Society 138(3), 497–505.
3390
Mueter, F.-J., R. M. Peterman, and B. J. Pype. 2002. Opposite effects of ocean temperature on
3391
survival rates of 120 stocks of Pacific salmon (Oncorhynchus spp.) in northern and southern
3392
areas. Canandian Jouranl of Fisheries and Aquatic Sciences 59(3), 456–463.
3393
Mueter, F. J., C. Broms, K. F. Drinkwater, K. D. Friedland, S. A. Hare, G. L. Hunt Jr., W. Melle,
3394
and M. Taylor. 2009. Ecosystem responses to recent oceanographic variability in high-
3395
latitude Northern Hemisphere ecosystems. Progress in Oceanography 81, 93–110.
3396
Mueter, F. J., C. Ladd, M. C. Palmer, and B. L. Norcross. 2006. Bottom-up and top-down
3397
controls of walleye pollock (Theragra chalcogramma) on the eastern Bering Sea shelf.
3398
Progress in Oceanography 68(2–4), 152–183.
3399
3400
3401
3402
3403
3404
Mueter, F. J., and M. A. Litzow. 2008. Sea ice retreat alters the biogeography of the Bering Sea
continental shelf. Ecological Applications 18(2), 309–320.
Murawski, S. A. 1993. Climate change and marine fish distribution: Forecasting from historical
analogy. Transactions of the American Fisheries Society 122, 647–658.
Myers, R. A. 1998. When do environment-recruitment correlations work? Reviews in Fish
Biology and Fisheries 8, 285–305.
3405
Napp, J. M., and G. L. Hunt Jr. 2001. Anomalous conditions in the south-eastern Bering Sea,
3406
1997: Linkages among climate, weather, ocean, and biology. Fisheries Oceanography 10(1),
3407
61–68.
3408
Niebauer, H. J., N. A. Bond, L. P. Yakunin, and V. V. Plotnikov. 1999. An update on the
3409
climatology and sea ice of the Bering Sea. In Dynamics of the Bering Sea. Edited by T. R.
3410
Loughlin and K. Ohtani. Fairbanks: University of Alaska Sea Grant.
3411
Orensanz, J. L., B. Ernst, D. A. Armstrong, P. J. Stabeno, and P. Livingston. 2004. Contraction of
3412
the geographic range of distribution of snow crab (Chionoecetes opilio) in the eastern Bering
3413
Sea: An environmental ratchet? CalCOFI Report 45, 65–79.
3414
Ottersen, G., S. Kim, G. Huse, J. J. Polovina, and N. C. Stenseth. 2010. Major pathways by which
3415
climate may force marine fish populations. Journal of Marine Systems 79(3–4), 343–360.
146
3416
Overland, J. E. 1981. Marine climatology of the Bering Sea. In The Eastern Bering Sea Shelf:
3417
Oceanography and resources. Edited by D. W. Hood and J. A. Calder. Seattle: University of
3418
Washington Press.
3419
Parada, C., B. Ernst, S. Hinckley, S., J. M. L. Orensanz, D. A. Armstrong, E. Curchitser, and A.
3420
Hermann. In prep. Patterns of connectivity and potential settlement regions of snow crab
3421
(Chionoecetes opilio) larvae in the eastern Bering Sea.
3422
3423
3424
3425
Parmesan, C., and G. Yohe. 2003. A globally coherent fingerprint of climate change impacts
across natural systems. Nature 421(6918), 37–42.
Perry, A. L., P. J. Low, J. R. Ellis, and J. D. Reynolds. 2005. Climate change and distribution
shifts in marine fishes. Science 308, 1912–1915.
3426
Pörtner, H. O. 2002. Physiological basis of temperature-dependent biogeography: Trade-offs in
3427
muscle design and performance in polar ectotherms. Journal of Experimental Biology
3428
205(15), 2217–2230.
3429
3430
Pörtner, H. O. 2008. Ecosystem effects of ocean acidification in times of ocean warming: A
physiologist’s view. Marine Ecology Progress Series 373, 203–217.
3431
Pörtner, H. O., B. Berdal, R. Blust, O. Brix, A. Colosimo, B. D. Wachter, A. Giuliani, T.
3432
Johansen, T. Fischer, R. Knust, G. Lannig, G. Naevdal, A. Nedenes, G. Nyhammer, F. J.
3433
Sartoris, I. Serendero, P. Sirabella, S. Thorkildsen, and M. Zakhartsev. 2001. Climate induced
3434
temperature effects on growth performance, fecundity and recruitment in marine fish:
3435
Developing a hypothesis for cause and effect relationships in Atlantic cod (Gadus morhua)
3436
and common eelpout (Zoarces viviparus). Continental Shelf Research 21, 1975–1997.
3437
Pörtner, H. O., M. Langenbuch, and B. Michaelidis. 2005. Synergistic effects of temperature
3438
extremes, hypoxia, and increases in CO2 on marine animals: From earth history to global
3439
change. Journal of Geophysical Research 110 C09S10.
3440
3441
3442
3443
3444
3445
Richardson, A. J. 2008. In hot water: Zooplankton and climate change. ICES Journal of Marine
Science 65, 279–295.
Rodionov, S. N., N. A. Bond, and J. E. Overland. 2007. The Aleutian Low, storm tracks, and
winter climate variability in the Bering Sea. Deep Sea Research II 54, 2560–2577.
Rose, G. A. 2005. On distributional responses of North Atlantic fish to climate change. ICES
Journal of Marine Science 62(7), 1360–1374.
3446
Saitoh, S., T. Iida, and K. Sasaoka. 2002. A description of temporal and spatial variability in the
3447
Bering Sea spring phytoplankton blooms (1997–1999) using satellite multi-sensor remote
3448
sensing. Progress in Oceanography 55, 131–146.
147
3449
Sambrotto, R. N., C. Mordy, S. I. Zeeman, P. J. Stabeno, and S. A. Macklin. 2008. Physical
3450
forcing and nutrient conditions associated with patterns of Chl a and phytoplankton
3451
productivity in the southeastern Bering Sea during summer. Deep Sea Research II 55(16),
3452
1745–1760.
3453
Schumacher, J. D., and P. J. Stabeno. 1998. The continental shelf of the Bering Sea. In The sea:
3454
The global coastal ocean: Regional studies and synthesis. Edited by A. R. Robinson and K.
3455
H. Brink. New York: John Wiley and Sons.
3456
3457
Skjoldal, H. R., and R. Sætre. 2004. Climate and ecosystem variability. In The Norwegian Sea
Ecosystem. Edited by H. R. Skjoldal. Trondheim, Norway: Tapir Academic Press.
3458
Smith, T. M., R. W. Reynolds, T. C. Peterson, and J. Lawrimore. 2008. Improvements to
3459
NOAA’s historical merged land-ocean surface temperature analysis (1880–2006). Journal of
3460
Climate 21, 2283–2296.
3461
3462
Stabeno, P. J., N. A. Bond, and S. A. Salo. 2007. On the recent warming of the southeastern
Bering Sea shelf. Deep Sea Research II 54, 2599–2618.
3463
Stabeno, P. J., N. B. Kachel, M. Sullivan, and T. E. Whitledge. 2002. Variability of physical and
3464
chemical characteristics along the 70-m isobath of the southeastern Bering Sea. Deep Sea
3465
Research II 49(26), 5931–5943.
3466
Steinacher, M., F. Joos, T. Froelicher, L. Bopp, P. Cadule, S. Doney, M. Gehlen, B. Schneider,
3467
and J. Segschneider. 2009. Projected 21st century decrease in marine productivity: A multi-
3468
model analys. Biogeosciences Discussions 6(4), 7933–7981.
3469
Strom, S. L., and K. A. Fredrickson. 2008. Intense stratification leads to phytoplankton nutrient
3470
limitation and reduced microzooplankton grazing in the southeastern Bering Sea. Deep Sea
3471
Research II 55(16), 1761–1774.
3472
3473
3474
3475
3476
3477
Walsh, J. E., and C. M. Johnson. 1979. An analysis of Arctic sea ice fluctuations, 1953–77.
Journal of Geophysical Research 9, 580–591.
Walsh, J. J., and C. P. McRoy. 1986. Ecosystem analysis in the southeastern Bering Sea.
Continental Shelf Research 5, 259–288.
Ware, D. M., and R. E. Thomson. 2005. Bottom-up ecosystem trophic dynamics determine fish
production in the Northeast Pacific. Science 308(5726), 1280–1284.
3478
Wassmann, P., D. Slagstad, C. Wexels Riser, and M. Reigstad. 2006. Modelling the ecosystem
3479
dynamics of the Barents Sea including the marginal ice zone II. Carbon flux and interannual
3480
variability. Journal of Marine Systems 59(1–2), 1–24.
148
3481
3482
Witherell, D., and D. Woodby. 2005. Application of Marine Protected Areas for sustainable
production and marine biodiversity off Alaska. Marine Fisheries Review 67(1), 1–27.
3483
Wyllie-Echeverria, T. 1996. The relationship between the distribution of one-year-old walleye
3484
pollock, Theragra chalcogramma, and sea-ice characteristics. NOAA Technical Report
3485
NMFS. 126, pp. 47–56.
3486
3487
3488
3489
Wyllie-Echeverria, T., and W. S. Wooster. 1998. Year-to-year variations in Bering Sea ice cover
and some consequences for fish distributions. Fisheries Oceanography 7(2), 159–170.
Zheng, J., and G. H. Kruse. 2006. Recruitment variation of eastern Bering Sea crabs: Climate
forcing or top-down effects? Progress in Oceanography 68, 184–204.
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
149
Fulmar
Polar bear
Ice seals
Walrus
Shearwater
murres
Gray whale
Arctic cod
jellyfish
Chum salmon
copepods
krill
Pacific herring
polychaete
Walleye pollock
Pacific cod
amphipod
Snow crab
3502
3503
Yellowfin sole
capelin
squid
Myctophid
Primary
production
Pacific halibut
Red king crab
Arrowtooth flounder
3504
3505
Figure 8.1. Schematic food web of some key species of the eastern Bering Sea ecosystem.
150
Russia
St. Matthew I.
March
ice edge
maximum
March
ice edge
minimum
Alaska
St. Paul Is.
3506
3507
Figure 8.2. Map of eastern Bering Sea showing bathymetry and approximate minimum and
3508
maximum observed March ice extent.
3509
151
Nov
Oct
Sep
Aug
Jul
Jun
May
Apr
Mar
Feb
Jan
2
1
0
-1
-2
-3
-4
1940
1960
3510
3511
1980
2000
9
8
Late summer (Jul-Sep) average
7
SST (deg. C)
(b)
1920
10
1900
1900
1920
1940
1960
1980
2000
3512
Figure 8.3. (a) Monthly sea-surface temperature anomalies (averaged over the southeast Bering
3513
Sea shelf inshore of the shelf break extending to 61°N) by month and year from January 1900 to
3514
December 2009 based on NOAA’s extended reconstructed sea-surface temperature series (Smith
3515
et al. 2008). (b) Annual index of average temperature during late summer (July–September).
3516
152
SST anomalies (˚C)
(a) Dec
Bloom in warm water
Bloom in cold water
3517
3518
3519
Figure 8.4. Benthic and pelagic pathways of production in the eastern Bering Sea. The relative
3520
flow of energy entering each pathways changes with variations in sea-surface temperature at the
3521
time of the spring bloom.
3522
153
3523
3524
3525
Figure 8.5. Shifts in the center of distribution of forty-five demersal taxa on the eastern Bering
3526
Sea shelf from 1982 to 2006. Northward shifts are positive. The center of distribution is the
3527
average latitude over the survey area, weighted by catch-per-unit effort. For further details and a
3528
list of taxa see Mueter and Litzow (2008).
3529
154
3530
Early Ice Retreat
Late Bloom, Warm Water – Small Copepods favored
Late Ice Retreat
Early Bloom, Cold Water – Large Calanus favored
February
March
April
May
June
3531
Figure 8.6. Schematic illustration of the relationship between the timing of sea ice retreat in the
3532
spring, the timing of the spring bloom, and zooplankton growth. Top: When the ice retreats in late
3533
winter, there is insufficient light to support a bloom, and the bloom is delayed until late spring
3534
when solar heating has stratified the water column sufficiently to prevent algal cells from sinking.
3535
Growth of small copepods is favored. Bottom: When the ice retreat comes later in the spring, then
3536
there is sufficient light to support an ice-associated bloom. This bloom can start under the ice or
3537
at the ice edge in ice-melt-stabilized water. Growth of large copepods is favored. Modified from
3538
Hunt et al. (2002).
3539
155
0.4
0.2
0.0
-0.4
-0.8
Survival anomaly
8.0
8.5
9.0
9.5
10.0
Average July-September SST
3540
3541
Figure 8.7. Survival anomalies of walleye pollock in relation to late summer sea-surface
3542
temperatures during the early juvenile stage. Survival anomalies are in the number of recruits that
3543
are produced per unit of female spawning biomass. Lines indicate, at each temperature, the
3544
estimated average survival (solid line) and the lower and upper 95% confidence intervals for
3545
average survival (dashed lines). The four points with temperatures over 9.5°C correspond to the
3546
2002–2005 warm period.
3547
3548
156
0.02 0.04 0.06
-0.06
0
2
4
5
6
o
3549
3550
(b)
-0.02
Pollock condition index
0.10
0.05
0.00
-0.05
-0.10
index
condition
Pollock effect
Normalized
on log(Weight)
(a)
6
7
8
o
Temperature
( C)
Annual average
temperature
(oC)
( C)(oC)
LocalTemperature
temperature
3551
Figure 8.8. Temperature effects on growth of immature pollock: Effect of local temperatures at
3552
sampling location on average pollock condition index (anomaly in the weight of fish at 250mm
3553
length) with 95% confidence band (left) and scatterplot of annual average condition index of
3554
pollock on the eastern Bering Sea shelf against annual average water temperature (dots, 1998–
3555
2008) with fitted regression line.
157
3556
Chapter 9: Spatial match-mismatch between juvenile fish and prey provides a mechanism
3557
for recruitment variability across contrasting climate conditions in the eastern Bering Sea
3558
3559
Elizabeth Calvert Siddon1*, Trond Kristiansen2, Franz J. Mueter1, Kirstin K. Holsman3, Ron A.
3560
Heintz4, and Edward V. Farley4
3561
3562
1
3563
USA
3564
2
3565
3
3566
Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric
3567
Administration, 7600 Sand Point Way NE, Seattle, WA 98115 USA
3568
4
3569
Fisheries Service, National Oceanic and Atmospheric Administration, 17109 Pt. Lena Loop
3570
Road, Juneau, AK 99801 USA
University of Alaska Fairbanks, School of Fisheries and Ocean Sciences, Juneau, AK 99801
Institute of Marine Research, Bergen, Norway
Joint Institute for the Study of the Atmosphere and Ocean, University of Washington. Alaska
Ted Stevens Marine Research Institute, Alaska Fisheries Science Center, National Marine
3571
3572
3573
3574
3575
3576
3577
3578
3579
Citation: Siddon, E.C., Kristiansen, T., Mueter, F.J., Holsman, K.K., Heintz, R.A., and Farley,
3580
E.V. 2013. Spatial Match-Mismatch between Juvenile Fish and Prey Provides a Mechanism for
3581
Recruitment Variability across Contrasting Climate Conditions in the Eastern Bering Sea. PLoS
3582
ONE 8(12): e84526.
158
3583
Abstract
3584
Understanding mechanisms behind variability in early life survival of marine fishes through
3585
modeling efforts can improve predictive capabilities for recruitment success under changing
3586
climate conditions. Walleye pollock (Theragra chalcogramma) support the largest single-species
3587
commercial fishery in the United States and represent an ecologically important component of the
3588
Bering Sea ecosystem. Variability in walleye pollock growth and survival is structured in part by
3589
climate-driven bottom-up control of zooplankton composition. We used two modeling
3590
approaches, informed by observations, to understand the roles of prey quality, prey composition,
3591
and water temperature on juvenile walleye pollock growth: (1) a bioenergetics model that
3592
included local predator and prey energy densities, and (2) an individual-based model that
3593
included a mechanistic feeding component dependent on larval development and behavior, local
3594
prey densities and size, and physical oceanographic conditions. Prey composition in late-summer
3595
shifted from predominantly smaller copepod species in the warmer 2005 season to larger species
3596
in the cooler 2010 season, reflecting differences in zooplankton composition between years. In
3597
2010, the main prey of juvenile walleye pollock were more abundant, had greater biomass, and
3598
higher mean energy density, resulting in better growth conditions. Moreover, spatial patterns in
3599
prey composition and water temperature lead to areas of enhanced growth, or growth ‘hot spots’,
3600
for juvenile walleye pollock and survival may be enhanced when fish overlap with these areas.
3601
This study provides evidence that a spatial mismatch between juvenile walleye pollock and
3602
growth ‘hot spots’ in 2005 contributed to poor recruitment while a higher degree of overlap in
3603
2010 resulted in improved recruitment. Our results indicate that climate-driven changes in prey
3604
composition and quality can impact growth of juvenile walleye pollock, potentially severely
3605
affecting recruitment variability.
3606
Introduction
3607
The match-mismatch hypothesis [1] proposes that predator survival is dependent on the temporal
3608
and spatial overlap with prey resources [2]. Factors affecting temporal overlap, such as climate
3609
variability through altered phenology, can lead to changes in survival at critical life stages [3,4].
3610
Temporal variation in spatial patterns of physical or biological conditions may concurrently affect
3611
survival. For example, in temperate and sub-arctic marine ecosystems, the timing of the spring
3612
bloom varies between years, driven by physical oceanographic conditions that change due to
3613
climate variability (e.g., [5]). These conditions, such as the onset of stratification and light
3614
availability, also affect the spatial patterns of zooplankton abundance, which further influences
159
3615
the feeding success of planktivorous fish species. Hence, variability in the spatial overlap of
3616
predator and prey, as well as differences in prey quality [6,7], may directly affect differences in
3617
year-class success of many marine fish species [8,9].
3618
Variability in year-class strength of gadids is often associated with changing physical
3619
conditions [10,11]. The eastern Bering Sea (EBS) has experienced multi-year periods of both
3620
warm and cold conditions since the turn of the 21st century [12], with cold years having much
3621
higher walleye pollock (Theragra chalcogramma) recruitment on average [13]. Changes in
3622
zooplankton composition between these periods have been identified as an important driver of
3623
recruitment success for walleye pollock [9,14], but the mechanistic links remain poorly
3624
understood.
3625
Interannual changes in ocean temperatures [12] and shifts in the spatio-temporal
3626
distribution of prey [14] make walleye pollock in the EBS an ideal case study to better understand
3627
drivers of recruitment success in sub-arctic marine fish. Larger zooplankton taxa, such as lipid-
3628
rich Calanus spp., were less abundant during recent warm years, possibly causing reduced growth
3629
rates and subsequent year-class strength of juvenile walleye pollock (hereafter juvenile pollock).
3630
In contrast, higher abundances of lipid-rich prey, combined with lower metabolic demands in
3631
cold years, may have allowed juvenile pollock to acquire greater lipid reserves by late summer
3632
and experience increased overwinter survival [9]. Although the energetic condition of juvenile
3633
pollock in late summer is recognized as a predictor of age-1 abundance during the following
3634
summer in the EBS [13], the causal mechanism linking differences in prey abundance and quality
3635
to walleye pollock survival remains untested.
3636
The objectives of this study were to better understand the roles of prey quality,
3637
prey composition, and water temperature on juvenile pollock growth through (1)
3638
estimating spatial differences in maximum growth potential of juvenile pollock using a
3639
bioenergetics modeling approach, (2) comparing maximum growth potential to predicted
3640
growth from an individual-based model (IBM), and (3) quantifying the impact of
3641
temperature, prey abundance, and prey quality on spatial variability in growth potential.
3642
Materials and methods
3643
Ethics statement
3644
Collection of physical and biological oceanographic data and fish samples during the US Bering-
3645
Aleutian Salmon International Surveys (BASIS) conducted on the EBS shelf was approved
160
3646
through the National Marine Fisheries Service, Scientific Research Permit numbers 2005-9 and
3647
2010-B1. Collection of biological data in the US Exclusive Economic Zone by federal scientists
3648
to support fishery research is granted by the Magnuson - Stevens Fishery Conservation and
3649
Management Act.
3650
Modeling approaches
3651
Two alternative modeling approaches were parameterized based on samples of juvenile pollock,
3652
zooplankton, and oceanographic data collected during the BASIS surveys conducted on the EBS
3653
shelf from mid-August to October 2005 and 2010 ([15]; Figure 9.S1). We selected 2005 (warm)
3654
and 2010 (cold) for our analyses based on data availability and the pronounced contrast in ocean
3655
conditions between these years (e.g., depth-averaged temperature anomalies over the middle
3656
shelf; [12]). Extensive spatial coverage of the surveys, combined with varying climate conditions
3657
between years, provided ample data with which to inform the models and compare differences in
3658
predicted growth between a representative warm and cold year in the EBS.
3659
Maximum growth potential from a Wisconsin-type bioenergetics model parameterized
3660
for juvenile pollock (modified from [16]) was compared with predicted growth from a
3661
mechanistic IBM [17]. Comparing model-based predictions of growth allowed for a better
3662
understanding of the mechanisms behind temporal and spatial variability in growth patterns and
3663
an evaluation of the importance of different model parameters. Growth (g•g-1•day-1; weight
3664
measures refer to wet weight throughout) was estimated for 65 mm standard length (SL; 2.5 g)
3665
juvenile pollock, corresponding to the average size of age-0 fish (<100 mm total length; TL)
3666
observed in late summer (2005: 64.1 ± 6.7 mm SL [mean ± SD] and 1.97 ± 0.93 g; 2010: 64.3 ±
3667
9.2 mm SL and 2.39 ± 0.94 g; conversion between TL and SL followed [18]).
3668
Field observations
3669
Juvenile pollock abundance
3670
Juvenile pollock were collected from the EBS shelf (inner domain: 0-50 m isobath, middle
3671
domain: 50-100 m isobath, and outer domain: 100-200 m isobath) using a midwater rope trawl
3672
following methods described in [19]. Catch per unit effort (CPUE; #•m-2) was calculated as:
3673
CPUEi 
ni
di  h
Eq. 1
3674
where ni is the number of fish collected in a given haul i, di is the trawl distance (m) calculated
3675
from starting and ending ship position, and h is the horizontal spread of the trawl net (m). Only
161
3676
surface tows at pre-defined stations were used to compute CPUE because midwater tows
3677
specifically targeted acoustic sign of walleye pollock.
3678
Water temperature
3679
Vertical profiles of water temperature were collected at each station sampled for oceanography
3680
using a Sea-Bird Electronics (SBE) conductivity-temperature-depth (CTD) profiler SBE-25
3681
(2005) or SBE-911 (2010). The average temperature in the upper 30 m of the water column was
3682
used in the bioenergetics model, assuming juvenile pollock collected from surface trawls were
3683
concentrated within the upper 30 m [15]. For the IBM, the water column was divided into 1 m
3684
discrete depth bins. For all IBM simulations, the depth of the water column was set to the upper
3685
100 m of all deeper stations (n=9 of 116 in 2005, n=27 of 160 in 2010) because MOCNESS data
3686
used to develop vertical profiles of zooplankton distribution (see ‘Zooplankton vertical profiles’
3687
below) was limited to 100 m. For stations with missing temperature data (n=1 for 2005), data
3688
from the nearest station with similar depth was used. For stations with incomplete temperature
3689
profiles (n=1 for 2005), temperatures were linearly interpolated between depths.
3690
Zooplankton data
3691
Determination of main prey taxa
3692
Juvenile pollock (<100 mm TL) collected from both surface (2005 and 2010) and midwater
3693
(2010) tows were used in the analysis to characterize diets across the EBS shelf in two contrasting
3694
years (n=26 stations in 2005, n=47 stations in 2010 [n=16 surface tows, n=31 midwater tows]).
3695
Stomach content analyses followed standard methods as described in [19] to estimate the
3696
contribution of each prey taxon to the dietary volume of juvenile pollock (% volume). To
3697
compute overall average diet composition, contributions were weighted by the CPUE of juvenile
3698
pollock at each station and averaged across stations. All prey taxa of juvenile pollock that
3699
cumulatively accounted for at least 90% of the diet by volume and individually accounted for at
3700
least 2% of the diet by volume were included in the bioenergetics and IBM models (Table 9.1).
3701
Main prey taxa from either year were included in models for both years for comparing growth
3702
across years.
3703
Zooplankton abundance
3704
Water-column abundances of small and large zooplankton taxa were estimated from Juday and
3705
bongo net samples, respectively, as described in [14]. Small zooplankton representing main prey
3706
taxa included Acartia clausi, Acartia spp. (2010 only), Centropages abdominalis, and
162
3707
Pseudocalanus sp. Large zooplankton included Calanus marshallae, Eucalanus bungii, Limacina
3708
helicina, Neocalanus cristatus, N. plumchrus (2005 only), Oikopleura sp., Thysanoessa inermis,
3709
T. inspinata, and T. raschii.
3710
Zooplankton biomass
3711
Total sample weights (g) of taxa collected from the Juday net were computed from wet weight
3712
tables [20]. Densities (g•m-3) of taxa collected from the bongo net were measured during sample
3713
processing at the University of Alaska Fairbanks (2005; [21]) and NOAA/NMFS/Alaska
3714
Fisheries Science Center (2010). The year-specific average biomass of individuals for the main
3715
prey taxa was calculated by dividing the sum of the biomass of all specimens weighed (i.e.,
3716
subsample) by the total number of specimens subsampled in a given year (Table 9.S1).
3717
Zooplankton energy density
3718
Taxa-specific energy density (ED; kJ•g-1) values obtained from available zooplankton collections
3719
from the EBS during 2004 (warm; no ED data available from 2005) and 2010 (cold) were used to
3720
estimate average ED values during warm and cold conditions for the main prey taxa. For five taxa
3721
lacking sufficient information to estimate separate ED values, a single estimate was used in both
3722
years (Table 9.S1). In these cases, only differences in abundance and biomass contributed to
3723
differences in average prey energy between years in the models. A biomass-weighted mean prey
3724
ED was calculated for each station and used as input to the bioenergetics model. At each station,
3725
the biomass of individual taxa was divided by the total prey biomass, multiplied by the taxa-
3726
specific energy density for each year, and summed across all taxa present at a given station.
3727
Estimates of ED and % lipid were available for several copepod species (C. marshallae,
3728
N. cristatus, and N. plumchrus/flemingeri) from 2010 (see [22] for details on the biochemical
3729
processing). A linear regression was developed to predict species specific ED (  i ) from % lipid
3730
values for other copepod species and/or climate conditions (Table 9.S1), such that:
i    Li   i , where  i ~ N (0,  2 )
3731
Eq. 2
3732
where  and  represent the intercept and slope of the regression, respectively, Li is the lipid
3733
composition (%) of the individual copepod sample i and i is a residual. The residuals, i, are
3734
assumed to be independent and normally distributed with mean 0 and variance 2 ( = 19.3, p =
3735
0.02;  = 0.41, p = 0.07; R2=0.98).
3736
Zooplankton vertical profiles
163
3737
To account for diel vertical migrations, taxa-specific vertical profiles for day and night were
3738
developed for all main prey taxa as input for the IBM. Vertical profiles were based on summer
3739
MOCNESS surveys that provided depth-stratified abundance estimates. MOCNESS data were
3740
available for 2004 (warm) and 2009 (cold); these vertical profiles were applied to late-summer
3741
model runs for 2005 and 2010, respectively, assuming that the vertical behavior of zooplankton
3742
taxa is conserved seasonally and across years within similar oceanographic conditions. To assess
3743
the effect of this assumption, a sensitivity analysis was conducted using constant abundances by
3744
depth (see ‘IBM sensitivity analyses’ below).
3745
In 2004, vertically stratified MOCNESS samples were collected at 5 daytime and 42
3746
nighttime stations over the EBS shelf [21]. In 2009, 7 daytime and 22 nighttime stations were
3747
sampled (A. Pinchuk, unpubl. data). Daytime extended from approximately 07:00 (sunrise) to
3748
23:30 (sunset) Alaska Daylight Savings Time during the sampling periods; stations sampled
3749
during crepuscular periods were excluded from the analysis. The depth increments of the
3750
MOCNESS varied depending on water depth; therefore, data were binned to the finest resolution
3751
available (i.e., 5-20 m increments). Zooplankton abundance was assumed to be uniform within
3752
sampling depths and averaged across all daytime and nighttime tows within a given year to obtain
3753
four vertical profiles for each taxon (day vs. night, 2004 vs. 2009). Centropages abdominalis
3754
were not collected by the MOCNESS and a uniform distribution throughout the water column
3755
was applied for both years because their distribution during the 2005 and 2010 BASIS surveys
3756
was predominantly at shallow, well-mixed stations of the inner domain. Oikopleura sp. did not
3757
occur in daytime tows in 2004; therefore, the 2009 daytime vertical distribution was applied for
3758
both 2005 and 2010 model runs. Thysanoessa inspinata were rarely collected by the MOCNESS
3759
(n=1 for 2005; n=3 for 2010), therefore an average vertical profile based on all Thysanoessa spp.
3760
was applied.
3761
Bioenergetics model
3762
A bioenergetics model was used to estimate spatially explicit maximum growth potential of
3763
juvenile pollock. We used the broadly applied Wisconsin bioenergetics modeling approach
3764
[23,24] that has been adapted and appropriately validated for walleye pollock ([16, 25]; Table
3765
9.S2). The model estimates temperature- and weight-specific maximum daily (d) consumption for
3766
an individual fish at station k in year t ( C maxd ,kt ; g•g-1•d-1) as:
Cmax d,kt  Wd1 f (Td ,kt )
3767

164
Eq. 3
3768
where C max d , kt is parameterized from independent laboratory observations of consumption
3769
rates for the species, absent competitor or predator interference, and is assumed to scale
3770
exponentially with fish weight (W) according to  and  (the allometric intercept and slope of
3771
consumption) and thermal experience according to the temperature scaling function f(T) (Table
3772
9.S2).
3773
Realized individual daily consumption rates ( C d , kt ; g•g-1•d-1) based on in situ fish are
3774
typically much lower than C maxd ,kt because inter- and intra-species competition, mismatched
3775
prey phenology or distributions, and predator avoidance behaviors by prey species often limit
3776
capture and consumption rates [16,26]. The ratio of realized consumption to maximum
3777
consumption (i.e.,
3778
foraging efficiency. The rate
3779
set to a specific value and used to predict daily growth ( Gd,kt ) using the mass balance equation
3780
where growth is the difference between energy consumed ( Cd,kt ) and energy lost to metabolism
3781
and waste ( [ | Cd ,kt , Wd 1 , Td ,kt ] ), such that:
h = Cd,kt C maxd,kt ), or the mean relative foraging rate, is a measure of in situ
h can be estimated using field observations of growth or it can be
Gd ,kt  (Cd ,kt  [ | Cd ,kt ,Wd 1 , Td ,kt ])   kt
3782
Eq. 4
3783
where Gd,kt is the estimated daily specific growth (g•g-1•d-1), Cd,kt is realized consumption (
3784
Cd,kt = h ×C maxd,kt ), Wd 1 is the weight of an individual fish at the start of the simulation day d,
3785
Td,kt is the water temperature on simulation day d, and zkt is the ratio of predator energy to prey
3786
energy density and is used to convert consumed biomass of prey into predator biomass (for more
3787
information see [26]). We used station-specific energy densities for prey ( k ) but annual mean
3788
predator energy density ( v t ) for year t because predator information was not available at all
3789
stations.
3790

Energy density values for the main prey taxa were used to derive mean station-specific
3791
(k) available prey energy density for both years ( w kt); diet composition was assumed to be
3792
proportional to the relative biomass of zooplankton prey at each station. Individual fish energy
3793
density (vi) was determined using biochemical processing (see [22]). At stations where sufficient
3794
numbers of juvenile pollock were collected (n=91 in 2005 and n=12 in 2010), 2-8 fish were
3795
selected to represent the size range at each station. Station-specific mean energy density in a
3796
given year ( v kt ) was weighted by CPUE and the number of fish processed at each station to
165
3797
calculate the average fish energy density by year ( v t ).
3798
We ran the model for a single simulation day (i.e., d=1) using base scenario input
3799
parameter values (Table 9.2; see also [16] Tables I and II) that were kept constant across stations
3800
and years (i.e., W =2.5 and  =1), were constant across stations but varied by year (i.e., vt ), or
3801
varied by station and year (i.e., Tkt and  kt ). Because the model is size-specific, running the
3802
model for a single simulation day minimized compound errors that can accumulate over multiple
3803
simulation days when predicting growth and allowed for a comparative index of growth across
3804
stations. Keeping fish starting weights (W) constant allowed us to evaluate spatial effects of
3805
changes in the other parameters; setting =1 implies that growth was constrained by
3806
physiological processes, but not by prey consumption, hence we evaluated variability in
3807
maximum growth potential. Annual average fish energy density was applied across stations in
3808
each year ( v2005 = 3.92 kJ•g-1; v2010 = 5.29 kJ•g-1).
3809
Bioenergetics sensitivity analyses
3810
Individual input parameters were increased and decreased by 1 standard deviation (SD) and the
3811
change in growth relative to maximum predicted growth under the base scenario was recorded. A
3812
pooled SD was calculated across stations after removing the annual means. Relative foraging rate
3813
(  ) was held at 1 for all sensitivity model runs in order to compare the relative effect of other
3814
parameters on maximum growth potential.
3815
Station-specific parameters (i.e., Tkt and  kt ) were varied to evaluate the relative effect
3816
on predicted growth and to examine resulting changes in spatially explicit growth patterns in each
3817
year. To evaluate the effect of variability in fish starting weight and energy density ( W and vt ,
3818
respectively) on estimated growth in 2005 and 2010, we used Monte Carlo simulations at a
3819
representative station (see Figure 9.S1). A single station was used because mean fish weight and
3820
energy density input values did not vary across stations in the model due to data limitations;
3821
hence the spatial pattern in estimated growth is not affected by varying these values by a constant
3822
amount. The model was run 1000 times using parameter values drawn at random from a normal
3823
distribution with the observed mean and SD for each parameter. The resulting distribution of
3824
predicted maximum growth potential was examined.
3825
Mechanistic individual-based model

166
3826
A mechanistic, depth-stratified IBM was used to predict average growth (g•g-1•d-1) and depth (m)
3827
of 100 simulated juvenile pollock by station. The details of the IBM and model validation are
3828
described in [17,28]. The IBM was reparameterized for juvenile pollock and forced with input
3829
data for water column temperatures and prey availability in 1 m discrete depth bins. Prey
3830
abundance (#•m-3) was allocated into depth bins according to vertical profiles of zooplankton
3831
distribution and scaled to station depth.
3832
The IBM used a mechanistic prey selection component that simulated the feeding
3833
behavior of juvenile pollock on zooplankton. The species composition of main prey taxa was
3834
based on observations; stage-specific length and width estimates were based on literature values
3835
or voucher collections from the EBS (Table 9.S1). Optimal prey size was estimated to be 5-8% of
3836
fish length based on larval Atlantic cod research [29,30]; juvenile pollock are predicted to have
3837
nearly 100% capture success for prey smaller than 5% of fish length, while the probability of
3838
capture success decreases with larger prey [17]. The simulated feeding ecology depended on
3839
juvenile pollock development (e.g., swimming speed, gape width, eye sensitivity) and vertical
3840
migratory behavior, prey densities and size, as well as light and physical oceanographic
3841
conditions (for details see [17]). Gape width was calculated as a function of fish size; conversion
3842
between length and weight followed [31]. Juvenile feeding processes were modeled with light-
3843
dependent prey encounter rates and prey-capture success (see [29]).
3844
Vertical migratory behavior was modeled assuming that juvenile pollock would seek
3845
deeper depths to avoid visual predation risk as long as ingestion rates would sustain metabolism
3846
and growth. If not, juvenile fish would seek the euphotic zone where light enhances feeding
3847
success, but also increases predation risk. Prey distributions switched between daytime to
3848
nighttime profiles when the light level (i.e., irradiance) reached 1 mol•m-2•s-1 [28]. The cost of
3849
vertical migration was included as a maximum of 10% of standard metabolic rates if the fish
3850
swims up or down at its maximum velocity, and scaled proportionally for shorter vertical
3851
displacements. Swimming velocity was a function of juvenile fish size [32].
3852
Gut fullness was estimated based on the amount of prey biomass that was ingested and
3853
digested per time step (1 hour) according to the feeding module. Prey biomass flowing through
3854
the alimentary system supplied growth up to a maximum growth potential (Cmax; [25]), and
3855
standard metabolic cost, egestion, excretion, and specific dynamic action [16] were subtracted.
3856
Both maximum growth and metabolic costs were functions of fish weight and water temperature.
167
3857
For all base model scenarios, the starting weight of the fish was held constant across
3858
stations, while zooplankton abundance and vertical distribution varied according to observations.
3859
Fish starting weight was 2.5 g ± 30% assuming a random uniform distribution around the mean.
3860
Year-specific vertical profiles (day and night) for the main prey taxa and station-specific
3861
temperature and prey abundance profiles were applied. The model scenarios were run for 72
3862
hours, but only the last 24 hours of the simulations were used for the analysis to avoid the early
3863
part of the simulations that may be unduly influenced by random initial conditions.
3864
IBM sensitivity analyses
3865
Fish starting weights and the vertical prey profiles were varied and resulting growth and average
3866
depth predictions were compared to values under the base model scenario (see [28] for sensitivity
3867
of the IBM model to variability in other parameters). To evaluate the effect of fish size separately
3868
from the effects of environmental controls, estimated growth based on fish starting weights of 2.0
3869
g ± 30% was compared to the base scenario (2.5 g ± 30%), encompassing the mean weight of
3870
juvenile pollock from the BASIS surveys in 2005 (1.97 ± 0.93 g, mean ± SD) and 2010 (2.39 ±
3871
0.94 g, mean ± SD). To test the effect of vertical distributions and diel migrations of prey taxa,
3872
model runs assuming a uniform distribution of prey with depth were compared to the base
3873
scenario, highlighting the effects of non-uniform zooplankton distribution and diel vertical
3874
migrations on juvenile pollock prey selection.
3875
Results
3876
Field observations
3877
Juvenile pollock abundance
3878
Juvenile pollock abundance and distribution had distinct spatial patterns in the surface layer
3879
between warm and cold years, with a more northerly distribution in warm years. Specifically,
3880
during warm late-summer conditions of 2005 juvenile pollock were distributed over a broad
3881
extent of the middle and outer domain, while in the cooler late summer of 2010 fish were
3882
concentrated over small regions of the southern shelf and outer domain (Figure 9.1, a and b).
3883
Abundance also varied between years with higher mean CPUE observed in 2005 as compared to
3884
2010 (CPUE = 0.08 fish•m-2 vs. 0.001 fish•m-2, respectively) at positive catch stations.
3885
Water temperature
3886
The average water temperature in the upper 30 m of the water column during the BASIS survey
3887
was 8.8ºC in 2005 and 7.6ºC in 2010 (Figure 9.S2, a and b), while the average temperature below
168
3888
40 m was 4.5ºC in 2005 and 2.9ºC in 2010 (Figure 9.S2, c and d). The warmest surface
3889
temperatures occurred in nearshore waters, although 2005 had warm temperatures over much of
3890
the southern shelf. Bottom temperatures reflected the extent of the cold pool (waters <2ºC), which
3891
was limited to the northern portion of the study area in 2005 and covered much of the shelf in
3892
2010.
3893
Zooplankton
3894
Main prey taxa
3895
Diets of juvenile pollock shifted from smaller copepod species in the warmer 2005 summer
3896
season (e.g., Pseudocalanus sp.) to larger species in the cooler 2010 summer season (e.g., N.
3897
cristatus). Several large zooplankton species were present in the diets across years, including L.
3898
helicina, which was the predominant prey item in both years, as well as C. marshallae and T.
3899
raschii. In 2010, the main prey taxa of juvenile pollock collected in surface tows were similar to
3900
those from midwater tows, with the exception of E. bungii accounting for 0% and 3% of surface
3901
and midwater tows, respectively. Eucalanus bungii was included in further analyses because it
3902
represented approximately 3% of combined diets by volume (Table 9.1).
3903
Zooplankton abundance
3904
Changes in juvenile pollock diet composition reflect spatial and temporal variability in
3905
zooplankton species composition and availability. In 2005, the abundance of available prey was
3906
highest in the inner domain and decreased towards the outer domain and northern Bering Sea.
3907
The abundance of prey in 2010 was greater in the inner domain; in the southern region of the
3908
shelf abundances decreased towards the middle and outer domains (Figure 9.1, c and d). The
3909
lowest abundance of zooplankton occurred in areas corresponding to higher concentrations of
3910
juvenile pollock. The total abundance of zooplankton within the optimal prey size range for 65
3911
mm SL juvenile pollock (species with mean length within 5-8% of fish length) was higher in the
3912
northwest region of the study area and over the southern shelf in the outer domain in 2005, with
3913
lesser overlap with juvenile pollock. In 2010, optimal prey was located across the middle and
3914
outer domains with highest abundances in the southern region, mirroring the distribution of
3915
juvenile pollock (Figure 9.1, c and d). Spatial patterns of zooplankton abundance accounting for
3916
all taxa <8% of fish length (not shown) reflected total abundance patterns in both years,
3917
indicating that areas of highest zooplankton abundance are driven by small (<5% of fish length)
3918
zooplankton taxa.
169
3919
Zooplankton energy density
3920
In 2005, available prey energy was highest in the northwest region of the shelf, with low prey
3921
energy over most of the shelf south of 60 °N (Figure 9.1e) where juvenile pollock abundances
3922
were higher. In contrast, prey energy was very high across much of the southern shelf in 2010
3923
(Figure 9.1f), particularly within the cold pool, where juvenile pollock were more abundant.
3924
Spatial patterns in prey energy were similar to spatial patterns of abundance for optimal prey size
3925
classes (Figure 9.1, c and d) because highest energy prey taxa are within 5-8% of fish length.
3926
Bioenergetics model
3927
Differences in the spatial pattern of maximum growth potential (g•g-1•d-1) of juvenile pollock
3928
occurred between a warm and cold year in the EBS (Figure 9.2, a and b). In 2005, growth
3929
potential was highest in the northwest region of the shelf (north of 60 °N) and lowest in the inner
3930
domain with one station having negative growth. Gradients in growth potential, from low to high,
3931
occurred from the inner to outer domains and from southern to northern regions of the shelf
3932
(Figure 9.2a). In 2010, growth was positive at all stations with highest growth potential over the
3933
southern shelf and lower growth predicted in the northeast region (Figure 9.2b).
3934
Bioenergetics sensitivity analyses
3935
Effect of water temperature and prey energy density
3936
In 2005, increasing temperatures by 1 SD (Figure 9.2c) resulted in areas of decreased predicted
3937
growth at shallow inner domain and southern shelf stations where water temperatures already
3938
approached thermal thresholds. Growth could not be estimated at one inner domain station
3939
because the increased temperature exceeded 15 ºC, the maximum temperature for consumption
3940
(Tcm) in the model. Decreasing water temperatures, resulting in increased growth, had the greatest
3941
effect in the same areas (not shown) because temperature-dependent control of growth is
3942
magnified where temperatures are close to thermal thresholds. In 2010, the effect of increasing
3943
water temperatures was an order of magnitude less than in 2005 (Table 9.3), but the spatial
3944
patterns were similar with shallow stations in the inner domain being most sensitive, as well as a
3945
small area in the outer domain (Figure 9.2d). Increasing available prey energy resulted in
3946
increased predicted growth rates across the region in 2005 (Figure 9.2e), with weaker effects in
3947
the inner domain and northwest region. In 2010, increased prey energy also resulted in elevated
3948
growth rates, but the magnitude of change was much lower than in 2005 and the spatial pattern
170
3949
differed; stronger effects occurred in the inner domain and southern region of the outer domain
3950
(Figure 9.2f).
3951
Predicted maximum growth potential generally increases with temperature and prey
3952
energy until temperature-dependent controls limit growth (Figure 9.3). Predicted growth is
3953
negative when available prey energy cannot meet metabolic demands under increased
3954
temperatures. Water temperatures were warmer in 2005, therefore juvenile pollock experienced
3955
conditions at or near their metabolic threshold at some stations. Colder water temperatures and
3956
higher available prey energy in 2010 resulted in better growing conditions over the shelf.
3957
Effect of fish starting weight and fish energy density
3958
Increasing fish starting weight resulted in lower predicted growth rates in both years because
3959
larger fish have higher metabolic demands (Table 9.3). Increasing fish energy density had a
3960
variable effect across stations in 2005 (not shown). In general, the effect of varying fish energy is
3961
dependent on initial fish energy and the relative available prey energy at each station. In 2010,
3962
increasing fish energy density resulted in lower predicted growth rates across stations when
3963
available prey energy was held constant.
3964
Variability in fish starting weight resulted in a broader distribution of predicted growth
3965
rates (2005: 0.002 – 0.109; 2010: 0.017 – 0.170 g•g-1•d-1) than variability in fish energy (2005:
3966
0.007 – 0.013; 2010: 0.022 – 0.036 g•g-1•d-1) from Monte Carlo simulations, indicating that the
3967
model was more sensitive to inputs of fish weight. The simulated mean predicted growth rates,
3968
when varying fish starting weight or fish energy, were lower and less variable for 2005 (0.012 
3969
0.009 [mean  SD] for varying W; 0.010  0.001 for varying fish energy) than for 2010 (0.029
3970
0.012 for varying W; 0.027  0.002 for varying fish energy).
3971
Mechanistic individual-based model
3972
Predicted mean growth rates from the IBM were 30% (2005) and 46% (2010) lower than
3973
maximum growth potential from the bioenergetics model (Tables 9.3 and 9.4) as foraging rates
3974
are restricted in the IBM based on stomach fullness and the prey selection module (i.e., capture
3975
success). The reduction in growth was greater in 2010, resulting in similar predicted growth rates
3976
from the IBM in 2005 and 2010. In addition, predicted growth rates from the IBM have a
3977
narrower range than maximum growth potential from the bioenergetics model.
3978
3979
In 2005, growth was positive across the region with moderate growth predicted across the
southern shelf. North of 60 °N, predicted growth rates decreased from the inner to outer domain
171
3980
(Figure 9.4a). The average depth (m) of juvenile pollock was 44 m (Table 9.4), with shallower
3981
distributions in the northeast region and deeper distributions in the southern region of the outer
3982
domain (Figure 9.4b). In 2010, growth was positive across the region, with highest predicted
3983
growth in the inner domain and areas of lower growth in the middle domain (Figure 9.4c). The
3984
spatial patterns of average depth of juvenile pollock (Figure 9.4d) mirrored those of 2005 with a
3985
slightly deeper average depth of 47 m (Table 9.4).
3986
IBM sensitivity analyses
3987
The effect of smaller fish starting weights on predicted growth was positive across the region,
3988
with stronger effects in 2005 than 2010 (Table 9.4). Similarly, effect strengths varied spatially in
3989
both years with areas of higher predicted growth in the middle domain (Figure 9.4, e and g). In
3990
2005, smaller starting fish weights resulted in shallower depth distributions across the region
3991
(mean=-2.6 m; Table 9.4), with much shallower depths at two stations in the middle domain
3992
(Figure 9.4f). The average change in depth distribution was similar in 2010 (mean=-2.4 m; Table
3993
9.4), but spatially more variable than in 2005 (Figure 9.4h).
3994
Applying uniform vertical distributions to prey taxa had variable effects on predicted
3995
growth rates in both years, with similarly small effect strengths (Table 9.4). Under uniform prey
3996
distributions, modeled fish may move vertically in response to other cues (i.e., predation risk,
3997
thermal boundaries) regardless of diel patterns. In 2005, uniform distributions resulted in
3998
increased predicted growth rates at several stations in the northern-most region of the shelf
3999
(Figure 9.4i). While the average depth of juvenile pollock was 2.1 m deeper across the region,
4000
fish at some of the northern-most stations had shallower depths (Figure 9.4j). In 2010, strongest
4001
effects on growth were observed in the middle domain of the southern shelf, with high spatial
4002
variability (Figure 9.4k). Changes in the depth of fish in response to uniform prey distributions
4003
mirrored spatial patterns in growth effects; stations showing deeper mean depths also resulted in a
4004
decrease in growth and vice versa (Figure 9.4l).
4005
Spatial comparison of bioenergetics- and IBM-predicted growth
4006
Predicted growth rates from the IBM were within the range of maximum growth potential from
4007
the bioenergetics model, but spatial patterns varied due to differences in input parameters of each
4008
model. In both years, the bioenergetics model predicted higher growth rates than the IBM over
4009
the middle and outer domains. The greatest difference occurred in the northwest region of the
4010
shelf in 2005 (Figure 9.5a) and over the southern region of the middle domain in 2010 (Figure
4011
9.5b). The IBM predicted higher growth in the shallow, well-mixed inner domain in both years.
172
4012
4013
Discussion
4014
This study demonstrates that warm and cold conditions in the EBS lead to spatial differences in
4015
zooplankton species composition, energy content, and abundance, which subsequently affect the
4016
feeding ecology and growth of juvenile pollock. Particularly, prey distribution and quality in
4017
combination with water temperatures create spatial patterns of increased growth potential (‘hot
4018
spots’) that vary with climate conditions. Spatial heterogeneity in growth conditions results from
4019
a combination of prey quality and quantity, water temperature, and metabolic costs, which may
4020
contribute to size-dependent fish survival and subsequent annual variability in recruitment. We
4021
provide evidence that a spatial mismatch between juvenile pollock and growth ‘hot spots’ in 2005
4022
is the mechanism that contributed to poor recruitment to age-1 while a higher degree of overlap in
4023
2010 resulted in 42% greater [33] recruitment to age-1.
4024
In the EBS, changes in oceanographic conditions can impact larval and juvenile fish
4025
distributions through front formation [34] and subsequent changes in drift trajectories [35]. The
4026
resultant variability in fish distributions relative to their prey during late summer and fall may be
4027
particularly important because the time period after the completion of larval development and
4028
before the onset of winter has been identified as a critical period for energy storage in juvenile
4029
pollock [22]. As the spatial distribution of fish, including spawning locations of adult walleye
4030
pollock, and zooplankton vary under alternate climate conditions, so do patterns in juvenile fish
4031
growth and recruitment success (Figure 9.6). Here, we find support for the argument that warm
4032
years produce smaller, less energy-rich prey and that this reduced prey quality, in combination
4033
with higher metabolic demands, results in lower growth of juvenile pollock. Conversely, cold
4034
years produce larger, more energy-rich prey which, when combined with lower metabolic
4035
demands, are favorable for juvenile pollock growth and survival. Thus, mechanisms responsible
4036
for controlling growing conditions during the critical pre-winter period can be linked to
4037
variability in recruitment.
4038
Projected declines in walleye pollock recruitment under changing climate conditions [11]
4039
do not account for adaptive behaviors or changes to phenology that could enable fish to maintain
4040
higher growth rates. The sensitivity analyses helped to identify when and where favorable growth
4041
conditions may occur under alternate climate conditions. In the bioenergetics model, varying fish
4042
size had a stronger effect on growth potential than changes in initial fish energy density. Larger
4043
fish have greater capacity for growth due to increased gape size, which allows them to take
173
4044
advantage of larger, more energy rich prey resources (e.g., euphausiids) prior to winter. The
4045
sensitivity analysis of increasing water temperatures showed weaker effects in the cold year of
4046
2010 because fish had a broader range of temperatures over which growth potential was relatively
4047
high (Figure 9.3), including warmer surface waters and a colder refuge in deeper waters that
4048
allows fish to conserve energy and avoid predation. In 2005, fish were near thermal limits based
4049
on temperature-dependent functions in the bioenergetics model; hence further increases in
4050
temperature are predicted to result in negative growth. Increasing available prey energy also had a
4051
stronger effect in the warm year of 2005 because metabolic demands were greater and mean prey
4052
energy density was lower than in 2010.
4053
The relative foraging rate was held constant at =1 across all bioenergetics model
4054
scenarios, but lower values would better reflect realistic foraging rates and could exacerbate
4055
thermal constraints on growth. To maintain positive growth rates at half of all the stations
4056
required relative foraging rates of =0.71 in 2005 and =0.57 in 2010. These values correspond
4057
to a 29% and 43% reduction in achieved growth relative to maximum growth potential and are
4058
similar to the mean differences between growth rates in the bioenergetics and IBM models (i.e.,
4059
30% in 2005 and 46% in 2010), providing support of model agreement. A higher relative
4060
foraging rate was required in 2005 in order to achieve positive growth at half of all stations,
4061
similar to results based on larger juvenile and adult walleye pollock [25], indicating that juvenile
4062
pollock growth was more prey limited and constrained by temperature in 2005 than in 2010.
4063
Thus, a greater reduction in both achieved growth from the IBM relative to maximum growth
4064
potential and relative foraging rates was observed in 2010 compared to 2005. In 2010,
4065
zooplankton abundance was lowest in areas with higher concentrations of juvenile pollock
4066
predators, potentially indicating prey limitation. While our study was not designed to explicitly
4067
test this question, other research demonstrates that pollock do not have strong top-down control
4068
of euphausiid abundance in the eastern Bering Sea (P. Ressler, pers. comm.).
4069
The vertical behavior of modeled juvenile pollock in the IBM moderated predicted
4070
growth rates leading to differences across domains based on stratification. Smaller (i.e., younger)
4071
fish were predicted to move shallower in the water column to improve prey detection, which is
4072
dependent on eye development and light availability. Moving into the surface layer also exposed
4073
juvenile pollock to higher predation risk because of the stronger light intensity. In the middle and
4074
outer domains, once sufficient growth was attained, fish were predicted to move deeper to seek
4075
refuge from predation. While the models were run at all stations in both years, observed juvenile
174
4076
pollock abundances were concentrated over the middle and outer domains in 2005 and over small
4077
regions of the southern shelf and outer domain in 2010. Few fish were observed in the well-mixed
4078
inner domain, possibly due to reduced growth potential based on available prey energy or lack of
4079
stratification and predation refuge in deeper waters. Additionally, the inner front, which
4080
delineates the stratified middle domain from the well-mixed inner domain [34], may act as a
4081
barrier to juvenile pollock distribution [36].
4082
Spatial patterns in juvenile pollock growth differed between models; these differences
4083
elucidate underlying mechanisms in feeding potential and ultimately the possible causes for
4084
growth ‘hot spots’ and variability in recruitment success between warm and cold climate
4085
conditions. The bioenergetics model uses biomass-weighted mean energy density of available
4086
prey, assuming fish feed proportional to what is available in the environment. The IBM is length-
4087
based and growth is dependent on available prey resources, light conditions, metabolism,
4088
development of the fish, and fish behavior. In the middle and outer domains where the water
4089
column is stratified, the bioenergetics model predicted higher growth than the IBM; the
4090
bioenergetics model allowed fish to feed at maximum consumption while the IBM indicated that
4091
fish moved deeper in the water column to conserve energy or avoid predation. In the inner
4092
domain, the IBM predicted higher growth; here juvenile pollock may opt to take advantage of
4093
available prey and warmer water temperatures to maximize growth because predator avoidance in
4094
deeper waters was not an option.
4095
Comparing the bioenergetics model and the IBM provided insights that could not be
4096
gained by either approach alone. For example, the bioenergetics model highlights the importance
4097
of differences in prey energy, a metric not included in the IBM, in determining spatial patterns of
4098
growth. On the other hand, the mechanistic feeding behavior implemented in the IBM highlights
4099
the role of prey size composition, the vertical distribution of prey, and the tradeoff between
4100
predator avoidance and maximizing growth. In practice, data requirements may limit the
4101
applicability of the IBM, whereas the bioenergetics model can be applied when less information
4102
on prey resources is available. Future research could benefit from including information on prey
4103
energy into IBMs to disentangle not only the importance of species composition, size
4104
composition, spatial distribution and abundance of prey, but also the importance of prey quality.
4105
Warm temperature conditions are predicted to result in reduced prey quality and low
4106
energy density of juvenile pollock in late summer [9,13]. Warmer water temperatures are
4107
associated with decreased growth [this study], resulting in lower overwinter survival and
175
4108
recruitment to age-1 [33]. The warm years of 2002-2005 had 67% lower average recruitment to
4109
age-1 relative to the cold years of 2008-2010, although variability during the cold years was very
4110
high with strong year classes in 2008 and 2010 separated by a weak 2009 cohort [33]. These
4111
findings agree with projected declines in recruitment of age-1 walleye pollock [11] under
4112
increased summer sea surface temperatures of 2°C predicted by 2050 [37]. Our results
4113
corroborate these previous studies and suggest that climate-driven increases in water temperature
4114
will have the largest effect on recruitment during anomalously warm years.
4115
This study provides evidence that climate-driven changes in prey dynamics can have
4116
ecosystem-level consequences via bottom-up control of fish populations in sub-arctic marine
4117
ecosystems. This work has improved our understanding of the mechanisms behind recruitment
4118
variability, in particular the underlying spatial patterns that drive relationships between prey
4119
availability, water temperature, growth, and survival. Our findings inform ongoing discussions of
4120
climate effects on predator-prey interactions and recruitment success of marine fishes.
4121
Acknowledgements
4122
We thank NOAA’s Bering Aleutian Salmon International Survey (BASIS) program for data
4123
collection and database management as well as the officers and crew of the R/V Oscar Dyson and
4124
F/V Sea Storm. The BASIS program is funded in part by the North Pacific Anadromous Fish
4125
Commission. MOCNESS data were provided from the PROBES (2004) and BEST-BSIERP
4126
(2009) surveys. This is NPRB publication #444 and BEST-BSIERP Bering Sea Project
4127
publication #111. T.K. was supported by the Norwegian Research Council project SWIM
4128
#13375.
4129
References
4130
1. Cushing DH (1990) Plankton production and year-class strength in fish populations – an
4131
update of the match mismatch hypothesis. Advances in Marine Biology 26: 249-293.
4132
2. Durant JM, Hjermann DO, Ottersen G, Stenseth NC (2007) Climate and the match or
4133
mismatch between predator requirements and resource availability. Climate Research 33:
4134
271-283.
4135
4136
4137
3. Edwards M, Richardson AJ (2004) Impact of climate change on marine pelagic phenology
and trophic mismatch. Nature 430: 881-884.
4. Kristiansen T, Drinkwater KF, Lough RG, Sundby S (2011) Recruitment variability in North
4138
Atlantic cod and match-mismatch dynamics. PLOS ONE 6(3): e17456.
4139
doi:10.1371/journal.pone.0017456.
176
4140
5. Hunt GL Jr, Stabeno PJ, Walters G, Sinclair E, Brodeur RD, Napp JM, Bond NA (2002)
4141
Climate change and control of the southeastern Bering Sea pelagic ecosystem. Deep-Sea
4142
Research II 49: 5821-5853.
4143
6. Sogard SM, Olla BL (2000) Endurance of simulated winter conditions by age-0 walleye
4144
pollock: effects of body size, water temperature and energy storage. Journal of Fish Biology
4145
56: 1-21.
4146
7. Litzow MA, Bailey KM, Prahl FG, Heintz R (2006) Climate regime shifts and reorganization
4147
of fish communities: the essential fatty acid limitation hypothesis. Marine Ecology Progress
4148
Series 315: 1–11.
4149
8. Paul AJ, Paul JM (1998) Comparisons of whole body energy content of captive fasting age
4150
zero Alaskan Pacific herring (Clupea pallasi Valenciennes) and cohorts over-wintering in
4151
nature. Journal of Experimental Marine Biology and Ecology. 226: 75–86.
4152
9. Hunt GL Jr, Coyle KO, Eisner L, Farley EV, Heintz R, Mueter FJ, Napp JM, Overland JE,
4153
Ressler PH, Salo S, Stabeno PJ (2011) Climate impacts on eastern Bering Sea foodwebs: A
4154
synthesis of new data and an assessment of the Oscillating Control Hypothesis. ICES Journal
4155
of Marine Sciences 68(6): 1230-1243.
4156
4157
10. Beaugrand G, Brander KM, Lindley JA, Souissi S, Reid PC (2003) Plankton effect on cod
recruitment in the North Sea. Nature 426: 661-664.
4158
11. Mueter FJ, Bond NA, Ianelli JN, Hollowed AB (2011) Expected declines in recruitment of
4159
walleye pollock (Theragra chalcogramma) in the eastern Bering Sea under future climate
4160
change. ICES Journal of Marine Science 68: 1284–1296.
4161
12. Stabeno PJ, Kachel NB, Moore SE, Napp JM. Sigler M, Yamaguchi A, Zerbini AN (2012)
4162
Comparison of warm and cold years on the southeastern Bering Sea shelf and some
4163
implications for the ecosystem. Deep-Sea Research II 65-70: 31-45.
4164
13. Heintz RA, Siddon EC, Farley EV Jr, Napp JM (In press) Correlation between recruitment
4165
and fall condition of age-0 walleye pollock (Theragra chalcogramma) from the eastern Bering
4166
Sea under varying climate conditions. Deep-Sea Research II.
4167
14. Coyle KO, Eisner LB, Mueter FJ, Pinchuk AI, Janout MA, Cieciel KD, Farley EV, Andrews
4168
AG (2011) Climate change in the southeastern Bering Sea: impacts on pollock stocks and
4169
implications for the Oscillating Control Hypothesis. Fisheries Oceanography 20(2): 139-156.
4170
15. Farley EV, Murphy JM, Wing BW, Moss JH, Middleton A (2005) Distribution, migration
4171
pathways, and size of western Alaska juvenile salmon along the eastern Bering Sea shelf.
4172
Alaska Fishery Research Bulletin 11: 15-26.
177
4173
16. Ciannelli L, Brodeur RD, Buckley TW (1998) Development and application of a
4174
bioenergetics model for juvenile walleye pollock. Journal of Fish Biology 52: 879-898.
4175
17. Kristiansen T, Lough RG, Werner FE, Broughton EA, Buckley LJ (2009) Individual-based
4176
modeling of feeding ecology and prey selection of larval cod on Georges Bank. Marine
4177
Ecolology Progress Series 376: 227-243.
4178
18. Buchheister A, Wilson MT (2005) Shrinkage correction and length conversion equations for
4179
Theragra chalcogramma, Mallotus villosus, and Thaleichthys pacificus. Journal of Fish
4180
Biology 67: 541-548.
4181
19. Moss JH, Farley EV Jr, Feldmann AM, Ianelli JN (2009) Spatial distribution, energetic
4182
status, and food habits of eastern Bering Sea age-0 walleye pollock. Transactions of the
4183
American Fisheries Society 138: 497-505.
4184
20. Volkov AF, Efimkin AY, Kuznetsova NA (2007) Characteristics of the plankton community
4185
of the Bering Sea and several regions of the North Pacific in 2002 – 2006. Izvestiya
4186
Tikhookeanskiy Nauchno-Issledovatelskiy Institut Rybnogo Khozyaystva I Okeanografiy
4187
151: 338–364 (in Russian).
4188
21. Coyle KO, Pinchuk AI, Eisner LB, Napp JM (2008) Zooplankton species composition,
4189
abundance and biomass on the eastern Bering Sea shelf during summer: the potential role of
4190
water column stability and nutrients in structuring the zooplankton community. Deep Sea
4191
Research II 55: 1755–1791.
4192
22. Siddon EC, Heintz RA, Mueter FJ (2013) Conceptual model of energy allocation in walleye
4193
pollock (Theragra chalcogramma) from age-0 to age-1 in the southeastern Bering Sea. Deep-
4194
Sea Research II. http://dx.doi.org/10.1016/j.dsr2.2012.12.007.
4195
23. Kitchell JF, Stewart DJ, Weininger D (1977) Applications of a bioenergetics model to perch
4196
(Perca flavescens) and walleye (Stizostedion vitreum). Journal of the Fisheries Research
4197
Board of Canada 34: 1922–1935.
4198
4199
4200
24. Ney JJ (1990) Trophic economics in fisheries: assessment of demand–supply relationships
between predators and prey. Reviews in Aquatic Sciences 2:55–81.
25. Holsman KK, Aydin K (In prep) Comparative field and bioenergetics-based methods for
4201
evaluating climate change impacts on the foraging ecology of Alaskan walleye pollock
4202
(Theragra chalcogramma), Pacific cod (Gadus macrocephalus), and arrowtooth flounder
4203
(Atheresthes stomias).
4204
4205
26. Hanson P, Johnson T, Schindler D, Kitchell J (1997) Fish Bioenergetics 3.0. Madison, WI:
University of Wisconsin Sea Grant Institute.
178
4206
27. Chesson J (1978) Measuring preference in selective predation. Ecology 59: 211-215.
4207
28. Kristiansen T, Fiksen Ø, Folkvord A (2007) Modelling feeding, growth, and habitat selection
4208
in larval Atlantic cod (Gadus morhua): observations and model predictions in a macrocosm
4209
environment. Canadian Journal of Fisheries and Aquatic Sciences 64: 136-151.
4210
4211
4212
4213
4214
29. Fiksen O, MacKenzie BR (2002) Process-based models of feeding and prey selection in
larval fish. Marine Ecology Progress Series 243: 151–164.
30. Munk P (1997) Prey size spectra and prey availability of larval and small juvenile cod.
Journal of Fish Biology 51(Suppl A): 340–351.
31. Buchheister A, Wilson MT, Foy RJ, Beauchamp DA (2006) Seasonal and geographic
4215
variation in condition of juvenile walleye pollock in the western Gulf of Alaska. Transactions
4216
of the American Fisheries Society 135: 897-907.
4217
32. Peck MA, Buckley LJ, Bengtson DA (2006) Effects of temperature and body size on the
4218
swimming speed of larval and juvenile Atlantic cod (Gadus morhua): implications for
4219
individual-based modelling. Environmental Biology of Fishes 75: 419–429.
4220
33. Ianelli JN, Honkalehto T, Barbeaux S, Kotwicki S, Aydin K, Williamson N (2012)
4221
Assessment of the walleye pollock stock in the Eastern Bering Sea. In: Stock assessment and
4222
fishery evaluation report for the groundfish resources of the Bering Sea/Aleutian Islands
4223
regions. North Pacific Fishery Management Council, 605 W. 4th Ave., Suite 306, Anchorage,
4224
AK 99501. 106 pp.
4225
34. Kachel NB, Hunt GL Jr, Salo SA, Schumacher JD, Stabeno PJ, Whitledge TE (2002)
4226
Characteristics and variability of the inner front of the southeastern Bering Sea. Deep-Sea
4227
Research II 49: 5889–5909.
4228
35. Duffy-Anderson JT, Busby MS, Mier KL, Deliyanides CM, Stabeno PJ (2006) Spatial and
4229
temporal patterns in summer ichthyoplankton assemblages on the eastern Bering Sea shelf
4230
1996–2000. Fisheries Oceanography 15: 80–94.
4231
36. Logerwell EA, Duffy-Anderson JT, Wilson M, McKelvey D (2010) The influence of pelagic
4232
habitat selection and interspecific competition on productivity of juvenile walleye pollock
4233
(Theragra chalcogramma) and capelin (Mallotus villosus) in the Gulf of Alaska. Fisheries
4234
Oceanography 19: 262-278.
4235
37. Hollowed AB, Bond NA, Wilderbuer TK, Stockhausen WT, A'mar ZT, Beamish RJ,
4236
Overland JE, Schirripa MJ (2009) A framework for modelling fish and shellfish responses to
4237
future climate change. ICES Journal of Marine Science 66: 1584–1594.
4238
179
4239
Table 9.1. Main prey taxa included in the models for 2005 and 2010. Prey items cumulatively
4240
accounting for at least 90% of the diet by % volume and individually accounting for at least 2%
4241
of the diet by % volume were included. Prey taxa common to both years are shown in bold.
2005
Taxa
2010
Individ
Cum
% Vol
% Vol
Limacina helicina
26.33
Pseudocalanus sp.
26.04
Oikopleura sp.
Taxa
Individ
Cum
% Vol
% Vol
Limacina helicina
35.45
52.4
Thysanoessa inermis
27.08
62.5
11.86
64.2
Calanus marshallae
13.87
76.4
Centropages abdominalis
8.98
73.2
Neocalanus cristatus
4.84
81.2
Thysanoessa raschii
8.48
81.7
Thysanoessa inspinata
3.16
84.4
Thysanoessa sp.
4.63
86.3
Thysanoessa raschii
3.09
87.5
Acartia clause
3.40
89.7
Neocalanus plumchrus*
2.98
90.5
Calanus marshallae
1.71
91.4
Eucalanus bungii
2.95
93.4
4242
*Neocalanus plumchrus was not identified in the 2010 bongo data, but did occur in the Juday
4243
data (small-mesh; not quantitative for large zooplankton taxa). Due to the absence in the bongo
4244
data, N. plumchrus was excluded from further analyses.
4245
180
4246
Table 9.2. Parameter definitions and values used in the bioenergetics model to estimate maximum
4247
growth potential (g•g-1•d-1) of juvenile walleye pollock. Parameters were used as inputs to the
4248
bioenergetics model described in [16].
Parameter
Definition (units)
C
Consumption (g•g-1•d-1)

Relative foraging rate
O2 cal
Value
Reference
0-1
a
Activity multiplier; convert g O2  g prey
13560
a

Intercept of the allometric function for C
0.119
a

Slope of the allometric function for C
-0.46
a
Qc
Temperature dependent coefficient
2.6
b
Tco
Optimum temperature for consumption
10
b
Tcm
Maximum temperature for consumption
15
b
R
Respiration (g O2•g-1•day-1)
Ar
Intercept of the allometric function for R
0.0075
b
Br
Slope of the allometric function for R
-0.251
b
Qr
Temperature dependent coefficient
2.6
b
Tro
Optimum temperature for respiration
13
b
Trm
Maximum temperature for respiration
18
b
Ds
Proportion of assimilated energy lost to Specific Dynamic
0.125
b
2
b
0.15
b
0.11
b
Action
4249
Am
Multiplier for active metabolism
F
Egestion
Fa
Proportion of consumed energy
U
Excretion
Ua
Proportion of assimilated energy
a
[25]; b [16]
181
4250
Table 9.3. Summary of sensitivity analyses for the bioenergetics model in 2005 and 2010
4251
showing the minimum (min), mean, and maximum (max) growth potential over all stations. Base
4252
values are predicted maximum growth potential (g•g-1•d-1) of juvenile pollock from the base
4253
model scenarios (W=2.5 g, Temp=average temperature in upper 30 m,
4254
density, v2005 = 3.92 kJ•g-1; v2010 = 5.29 kJ•g-1). All other values denote the change in growth rate
4255
resulting from indicated changes in inputs; therefore (-) effects indicate that varied conditions
4256
resulted in lower predicted growth and vice versa. Pooled standard deviations (SDs) for each
4257
parameter were calculated across stations after removing the annual means. W and vt are constant
4258
values applied across all station, so changes (± 1 SD) act as a scalar and result in similar spatial
4259
patterns across the area. Temperature and  k vary across stations.
h =1.0, k =prey energy
2005
Parameter
SD
Base
2010
min
mean
max
min
mean
max
-0.0056
0.0146
0.0291
0.0069
0.0172
0.0272
W + 1 SD
0.935
-0.0056
-0.0041
-0.0017
-0.0052
-0.0037
-0.0023
W – 1 SD
0.935
0.0034
0.0076
0.0103
0.0041
0.0068
0.0094
Temp + 1 SD
1.75
-0.0227
-0.0053
0.0017
-0.0071
-0.0007
0.0018
Temp – 1 SD
1.75
-0.0028
0.0018
0.0129
-0.0026
0.0008
0.003
k + 1 SD
497.5
0.0046
0.0061
0.0065
0.0032
0.0044
0.0048
k – 1 SD
497.5
-0.0065
-0.0061
-0.0046
-0.0048
-0.0044
-0.0032
vt + 1 SD
395.93
-0.0027
-0.0013
0.0005
-0.0019
-0.0012
-0.0005
vt – 1 SD
395.93
-0.0006
0.0016
0.0033
0.0006
0.0014
0.0022
4260
4261
182
4262
Table 9.4. Summary of sensitivity analyses for the IBM model in 2005 and 2010 showing the
4263
minimum (min), mean, and maximum (max) growth potential and depth (m) over all stations.
4264
Base values are predicted growth (g•g-1•d-1) and depth (m) of juvenile pollock from the base
4265
model scenarios (W=2.5 g, zooplankton prey distributed according to vertical profiles). All other
4266
values are predicted changes in growth and depth. Negative changes in depth indicate a shallower
4267
distribution; positive values indicate a deeper distribution. Weight is a constant value applied
4268
across all station, so varying the parameter acts as a scalar and results in similar spatial patterns
4269
across the area. The effect of applying a uniform distribution of zooplankton prey with depth
4270
varies across stations.
2005
Parameter
Base
W
(2.0 g)
Prey
distribution
2010
min
mean
max
min
mean
max
Growth
0.0062
0.0102
0.0121
0.0055
0.0092
0.0123
Depth
10
44.2
80.9
15
47.4
93
Growth
0.004
0.0184
0.0512
0.002
0.0068
0.0254
Depth
-30.5
-2.6
0.14
-43.3
-2.4
21.7
Growth
-0.0009
0.005
0.0058
-0.0034
0.001
0.0064
Depth
-21.4
2.1
15.8
-42.9
-1.8
35.2
(Uniform)
4271
183
4272
4273
Figure 9.1. Log(CPUE) of juvenile walleye pollock collected in surface trawls in 2005 (a) and
4274
2010 (b). Circle size is proportional to catch (#•m-2) at each station; note difference in scale
4275
between years. Stations with zero catch () are shown on white background. Log of total
4276
zooplankton abundance (#•m-3) for the main prey taxa is shown for 2005 (c) and 2010 (d). Circle
4277
size is proportional to the abundance of zooplankton within the optimal size range for 65 mm SL
4278
juvenile pollock (5-8% of fish length); note difference in scale between years. Biomass-weighted
4279
mean energy density (ED) of available zooplankton prey is shown for 2005 (e) and 2010 (f).
184
4280
4281
Figure 9.2. Predicted growth (g•g-1•d-1) of juvenile walleye pollock from the bioenergetics model.
4282
Top panel (a and b) shows growth under the base model scenarios for 2005 and 2010 (W=2.5 g,
4283
Temp=average temperature in upper 30 m,
4284
v2010 = 5.29 kJ•g-1). Middle panel (c and d) shows changes in predicted growth when temperature
4285
is increased by 1 standard deviation (SD). Predicted growth could not be estimated at one station
4286
in 2005 (c) in the inner domain under increased temperatures because the water temperature in
4287
the upper 30 m was greater than 15 ºC (Tcm = 15 ºC in the model). Lower panel (e and f) shows
4288
changes in predicted growth when prey energy density is increased by 1 SD. Spatial plots of
4289
predicted growth when parameters are decreased by 1 SD are not shown, but can be visualized by
4290
subtracting the anomalies (lower two panels) from the base scenario plots (top panel).
h =1.0, k =prey energy density, v2005 = 3.92 kJ•g-1;
185
4291
4292
Figure 9.3. Predicted growth (g•g-1•d-1) of juvenile walleye pollock interpolated over the range of
4293
observed temperatures and prey energy density values across both 2005 and 2010, providing a
4294
continuous scale of growth over a broad range of possible environmental and biological
4295
scenarios. The observed fish energy density was higher in 2010 (v2010 = 5.29 kJ•g-1; used in plot
4296
shown); therefore this interpolation demonstrates the range of predicted growth for fish with high
4297
energy density. Temperatures included 0-16 °C to show possible range under variable climate
4298
conditions. The dashed rectangle encompasses the range of temperatures and prey energy density
4299
values observed in 2005; solid rectangle encompasses values in 2010. Points are shown for
4300
average temperature and prey energy density conditions in 2005 and 2010. Predicted growth
4301
above 15 °C was not possible (black) because the bioenergetics model has a temperature
4302
threshold of 15 °C.
4303
186
4304
4305
Figure 9.4. Predicted growth (g•g-1•d-1) and average depth (m) of juvenile walleye pollock from
4306
the IBM. Top panel shows growth (a and c) and average depth (b and d) under the base model
4307
scenarios for 2005 and 2010 (W=2.5 g, zooplankton prey distributed according to vertical
4308
profiles). Middle panel shows changes in predicted growth (e and g) and average depth (f and h)
4309
for 2.0 g fish, highlighting the relative importance of fish size (relative to 2.5 g) and water
4310
temperature between years. Lower panel shows changes in predicted growth (i and k) and average
4311
depth (j and l) when uniform vertical distributions of prey are implemented, highlighting the
4312
effect of zooplankton diel vertical distribution and migrations on juvenile walleye pollock prey
4313
selection. Negative changes in depth indicate a shallower distribution; positive values indicate a
4314
deeper distribution.
4315
187
4316
4317
Figure 9.5. Difference in predicted growth (g•g-1•d-1) of juvenile walleye pollock between the
4318
bioenergetics model and the IBM for 2005 (a) and 2010 (b). Areas of positive differences indicate
4319
where maximum growth potential from the bioenergetics model was higher than predicted growth
4320
from the IBM.
4321
188
4322
4323
Figure 9.6. Conceptual figure of the spatial relationship between juvenile fish abundance (yellow)
4324
and zooplankton prey availability (blue). Where these areas overlap (green), juvenile fish are
4325
predicted to have higher growth rates and increased survival. Under warm climate conditions,
4326
there is reduced spatial overlap between juvenile fish and prey availability, resulting in lower
4327
overwinter survival and recruitment success to age-1. In colder conditions, increased spatial
4328
overlap between juvenile fish and prey availability results in increased overwinter survival and
4329
recruitment to age-1.
189
4330
Supporting Information:
4331
Comparison of observed and predicted prey preferences
4332
Methods
4333
Stomach contents of juvenile pollock (<100 mm SL) were identified from selected stations across
4334
the eastern Bering Sea shelf to compare observed diet composition, model-predicted diets, and
4335
available prey. Chesson’s prey preference index [1] was calculated for the main prey taxa in each
4336
year and compared to the individual-based model (IBM) estimates of prey preference at
4337
corresponding stations. Chesson’s index (i) for a given prey taxon i is the ratio between ingested
4338
prey items (ri) and the frequency of their occurrence in the environment (ni), standardized by the
4339
sum of the ratios over all m prey types:
4340
ri
n
ai = m i
rj
S
j =1 n j
4341
The standardization implies that neutral selection (neu) corresponds to 1/m and a specific prey
4342
item or group was actively selected if the index is > a neu , as they appear more frequently in the
4343
diet than their abundance in the environment would suggest. To calculate Chesson’s index for
4344
observed diets, % volume in the diet was used as a proxy for biomass consumed (ri) relative to
4345
prey biomass in the environment (ni).
4346
Chesson’s prey preference indices for main prey taxa in each year, based on observed and
4347
predicted diets, were compared at each station for which observed diet, predicted diet, and
4348
zooplankton composition was available and where at least 90% of the observed prey taxa (by %
4349
volume) were accounted for in the zooplankton data (n=7 in 2005; n=9 in 2010). Zooplankton
4350
samples that did not include a known prey item were not considered for this analysis because the
4351
lack of a known prey item in zooplankton samples collected at the same station suggests that the
4352
sample is not representative of prey availability due to small-scale patchiness, or indicates a
4353
spatial and/or temporal mismatch between where captured juvenile pollock were foraging and
4354
where samples were collected. To compare observed (obs) and predicted (pre) prey preferences,
4355
we computed differences between these prey preferences relative to neutral selection:
4356
 iobs /  neu   ipre /  neu
Eq. 1
Eq. 2
190

4357
and averaged them across all stations within each year, as well as by domain (i.e., inner: 0-50 m
4358
isobath, middle: 50-100 m isobath, and outer: 100-200 m isobath).
4359
Results and Discussion
4360
Modeled diets from the IBM were comparable to observed diets from the 2005 and 2010 surveys
4361
(Fig. 9. S1; most differences overlap zero), indicating that the model may adequately capture
4362
predator-prey dynamics. Relatively small differences between observed and predicted prey
4363
preference were consistent across domains in both years. Limacina helicina, the predominant
4364
component of diets across years, was more prevalent in observed diets (except in the inner
4365
domain south of 60 ºN in 2005), as was Thysanoessa raschii. Modeled diets, however,
4366
consistently overestimated consumption of Calanus marshallae and Eucalanus bungii (2010
4367
only) across domains.
4368
Shifts in C. marshallae abundance between warm and cold years in the EBS have been proposed
4369
as a major contributor to differences in juvenile pollock condition and survival [2]. However,
4370
observed diets of juvenile pollock (<100 mm SL) in 2005 and 2010 do not reflect its relative
4371
importance to growth because C. marshallae was less prevalent than expected. Greater
4372
prevalence of specific prey items in observed diets (e.g., Centropages abdominalis in 2005)
4373
indicates that the IBM model underestimates the ability of juvenile pollock to detect, capture, and
4374
ingest that prey item. Alternatively, the prey could have been more abundant in the areas where
4375
juvenile pollock were feeding than in the area sampled by the bongo and/or Juday net due to
4376
patchiness. Differences between observed and predicted diets may also be explained by prey
4377
escape behaviors or size-selectivity by juvenile pollock that is more complex than the prey
4378
selection component of the IBM [3]. Juvenile pollock collected in late-summer likely feed more
4379
heavily in surface waters during crepuscular or nighttime periods [4], moving deeper during the
4380
daytime, while observed diets for this study were sampled from daytime surface hauls. However,
4381
the spatial and temporal disconnect between where juvenile pollock feed and were collected for
4382
diet analyses likely did not affect our results as previous work has shown that proportional diet
4383
compositions do not vary between day and night [4,5] and the IBM integrates predicted diets over
4384
24 hours, encompassing diel vertical patterns.
4385
4386
4387
191
4388
References
4389
1. Chesson J (1978) Measuring preference in selective predation. Ecology 59: 211-215.
4390
2. Coyle KO, Eisner LB, Mueter FJ, Pinchuk AI, Janout MA, Cieciel KD, Farley EV, Andrews
4391
AG (2011) Climate change in the southeastern Bering Sea: impacts on pollock stocks and
4392
implications for the Oscillating Control Hypothesis. Fisheries Oceanography 20(2): 139-156.
4393
3. Petrik CM, Kristiansen T, Lough RG, Davis CS (2009) Prey selection of larval haddock and
4394
cod on copepods with species-specific behavior: a model-based analysis. Marine Ecology
4395
Progress Series 396: 123-143.
4396
4. Brodeur RD, Wilson MT, Ciannelli L (2000) Spatial and temporal variability in feeding and
4397
condition of age-0 walleye pollock (Theragra chalcogramma) in frontal regions of the Bering
4398
Sea. ICES Journal of Marine Science 57: 256–264.
4399
5. Schabetsberger R, Brodeur RD, Ciannelli L, Napp JM, Swartzman GL (2000) Diel vertical
4400
migration and interaction of zooplankton and juvenile walleye pollock (Theragra
4401
chalcogramma) at a frontal region near the Pribilof Islands, Bering Sea. ICES Journal of
4402
Marine Science 57: 1283–1295.
192
4403
Table 9.S1. Stage, sampling gear, length range, width, biomass (g, wet weight), and energy density (kJ•g-1, wet weight) values for the main prey
4404
items of juvenile walleye pollock in late summer 2005 and 2010. Biomass estimates were obtained during processing of the zooplankton samples
4405
from 2005 (warm) and 2010 (cold) (NA=stage was not collected); energy density values were obtained from zooplankton collected in the eastern
4406
Bering Sea during 2004 (warm) and 2010 (cold). Single estimates of energy density (shown in bold) were used when year-specific information
4407
was not available for individual taxa.
4408
Species
Stage
Gear
Length
range
Width
(mm)
(mm TL)
Warm
Cold
Biomass
(g WW)
Biomass
(g WW)
Warm
Energy
Density
(kJ•g-1 WW)
Acartia clausi
A
Juday
0.25 – 1.4a
0.22b
3.5 E-05
AF
Juday
0.8 – 1.4 a
0.29 b
NA
4.5 E-05
AM
Juday
0.8 – 1.2 a
0.27 b
NA
2.5 E-05
A
Juday
0.25 – 0.93 a
0.16 b
1.85 E-05
1.9 E-05
I
Juday
0.25 – 0.42 a
0.09 b
NA
II
III
Juday
Juday
0.42 – 0.51 a
0.51 – 0.65 a
0.12 b
0.16 b
Energy
Density
(kJ•g-1 WW)
3.816 c
4.1 E-06
Length range for A.
clausi
9.4 E-06
Length range for A.
clausi
1.7 E-05
Length range for A.
clausi
NA
NA
193
Comments
3.5 E-05
3.816c
Acartia sp.
Cold
IV
V
Calanus
marshallae
Centropages
abdominalis
AF
Juday
Juday
Bongo
0.65 – 0.76 a
0.76 – 0.93 a
3.2 – 4.2 a
0.19 b
0.23 b
1.0 E-05
2.5 E-05
Length range for A.
clausi
4.0E-05
Length range for A.
clausi
2.7 E-05
0.78d
0.003
0.002
AM
Bongo
3.5 – 4 a
1.0e
0.0026
0.0017
I
Bongo
0.5 – 0.7 a
0.16 e
5.25 E-05
5.25 E-05
II
Bongo
1.2 – 1.5 a
0.36 e
9.19 E-05
1.33 E-04
III
Bongo
1.6 – 2.3 a
0.68 d
1.9 E-04
2.7 E-04
I-III
Bongo
0.5 – 2.3 a
0.37 e
1.13 E-04
2.2 E-04
IV
Bongo
2.3 – 2.6 a
0.69 d
5.82 E-04
5.24 E-04
V
Bongo
2.8 – 3.8 a
0.73 d
0.002
0.0016
A
Juday
0.3 – 2.1 a
0.36h
7.49 E-05
5.73 E-05
AF
Juday
1.6 – 2.1 a
0.53 h
1.61 E-4
1.36 E-4
AM
Juday
1.5 a
0.5 h
9.14 E-05
1.16 E-4
C
Juday
1.13 a
0.5 h
2.33 E-05
2.33 E-05
I
Juday
0.3 a
0.16 h
NA
2.92 E-05
II
Juday
0.38 a
0.22 h
NA
1.5 E-05
III
Juday
0.5 a
0.29 h
1.32 E-05
2.82 E-05
I-III
Juday
0.33 a
0.17 h
NA
9.1 E-06
194
5.325f
5.732g
3.843i
3.843 i
Eucalanus
bungii
Limacina
helicina
Neocalanus
cristatus
IV
Juday
0.65 a
0.39 h
NA
4.25 E-05
V
Juday
0.85 a
0.5 h
8.55 E-05
6.89 E-05
A
Bongo
4.8 – 8 a
1.7e
0.0023
8.67 E-05
AF
Bongo
6 – 8a
1.86 e
0.0086
0.0086
AM
Bongo
4.8 – 5.5 a
1.37 e
0.0031
0.0041
I
Bongo
1.3 – 1.6 a
0.39 e
4.5 E-05
5.3 E-05
II
Bongo
2 – 2.2 a
0.56 e
1.96 E-4
1.56 E-4
III
Bongo
2.9 – 3 a
0.79 e
4.92 E-4
2.66 E-4
I-III
Bongo
1.3 – 3 a
0.57 e
2.44 E-04
8.41 E-4
IV
Bongo
3.36 – 3.8 a
0.95 e
0.0011
0.0009
V
Bongo
4.5 – 5.2 a
1.29 e
0.0027
0.0034
XS
Bongo
0.1 – 0.5l
0.3 l
NA
6.93 E-05
S
Bongo
0.5 – 2 l
1.25 l
9.40 E-05
1.58 E-4
M
Bongo
2 – 4l
3l
3.71 E-4
8.29 E-4
L
Bongo
4 - 10 l
7l
0.0026
0.0045
AF
Bongo
8.5 – 10.4a
2.5 e
NA
0.0137
III
Bongo
3.2 a
0.85 e
8.83 E-4
0.0013
IV
Bongo
4.9 – 5.3 a
1.36 e
0.0059
0.0025
V
Bongo
7.1 – 8.9 a
2.13 e
0.019
0.015
195
3.916j
4.194k
2.51m
2.766g
3.253n
3.39g
N. plumchrus
V
Bongo
4.1 – 5.2 a
1.24 e
0.00395
0.0041
4.207o
4.676g
Oikopleura sp.
A
Bongo
0.1 – 0.6p
0.35 p
1.73 E-4
1.7 E-4
4.076q
4.025r
Pseudocalanus
spp.
A
Juday
0.65 – 1.2s
0.29 d
t
t
Thysanoessa
inermis
4.44 E-05
3.6 E-05
AF
Juday
1.05 – 2.27 s
0.42 d
8.27 E-05
8.01 E-05
AM
Juday
0.91 – 1.74 s
0.35b
4.13 E-05
5.42 E-05
I
Juday
0.5 – 0.7 s
0.16 b
NA
5.98 E-06
II
Juday
0.65 – 0.8 s
0.19 b
1.01 E-05
1.08 E-05
III
Juday
0.8 – 1 s
0.24 b
1.26 E-05
2.0 E-05
I-III
Juday
0.5 – 1 s
0.2 b
NA
1.01 E-05
IV
Juday
1 – 1.2 s
0.3 d
3.06 E-05
3.19 E-05
V
Juday
1.2 – 1.5 s
0.37 d
8.89 E-05
4.93 E-05
II-V
Juday
0.65 – 1.5 s
0.29 d
3.55 E-05
2.8 E-05
A
Bongo
10.1 – 29.2u
2.4 d
NA
0.083
3.951
3.951
4.99m
4.99 m
AF
Bongo
10.1 – 29.2 u
2.4 d
AM
Bongo
10.1 – 29.2 u
2.4 d
0.069
NA
J
Bongo
8.5 – 13.8 u
1.4 d
NA
0.011
J (L)
Bongo
11.1 – 13.8 u
1.5 d
0.061
0.11
J (S)
Bongo
8.5 – 11.1 u
1.2 d
0.011
0.024
T. inspinata
J
Bongo
12 – 17v
2.2w
0.0012
0.0128
4.99x
4.99x
T. raschii
A
Bongo
7 – 29.1y
3.3 d
0.006
NA
4.308aa
5.231g
0.10
196
Length range for P.
moultoni
AF
Bongo
7 – 29.1 y
3.3 d
0.077
0.0903
AM
Bongo
15.3 – 20.2z
2.7 d
0.046
0.088
J
Bongo
7.4 – 8.4 z
1.45 d
0.005
0.0117
J (L)
Bongo
7.9 – 8.4 z
1.5 d
0.0476
0.0936
J (S)
Bongo
7.4 – 7.9 z
1.4 d
0.0089
0.0059
4409
a
[1]; b Estimated width = 26.7% of length (based on Pseudocalanus sp. relationship); c Energy density estimated from % lipid (2.25% wet weight
4410
assuming 80% moisture [2]) using the regression relationship: ED = (0.4098•% lipid) + 19.287; d E. Fergusson, NOAA/AFSC, unpublished data; e
4411
Estimated width = 26.6% of length (based on C. marshallae relationship); f Energy density estimated from % lipid (10.5%; R. Heintz,
4412
NOAA/AFSC, unpublished data); g R. Heintz, NOAA/AFSC, unpublished data; h [3]; i Energy density estimated from % lipid (2.6% wet weight
4413
[4]); j Energy density estimated from % lipid (3.55%; R. Heintz, NOAA/AFSC, unpublished data); k Energy density estimated as 7.1% higher in
4414
cold years (based on copepod data; [this study]); l C. Stark, UAF, unpublished data; m 2006 collection (R. Heintz, NOAA/AFSC, unpublished
4415
data); n Energy density estimated from % lipid (5.85%; R. Heintz, NOAA/AFSC, unpublished data); o Energy density estimated from % lipid
4416
(6.83%; R. Heintz, NOAA/AFSC, unpublished data); p Trunk length/width [5]; q Energy density estimated from % lipid of Chaetognatha (2.67%;
4417
R. Heintz, NOAA/AFSC, unpublished data); r Energy density estimated from % lipid of Chaetognatha (2.04%; R. Heintz, NOAA/AFSC,
4418
unpublished data); s [6]; t Energy density estimated from % lipid (4% wet weight [7]); u Carapace width from [8]; converted to TL using equations
4419
from [9]; v Length range of ‘spineless’ T. longipes [10]; w Estimated width as 15% of length; x Used energy density of T. inermis; y Minimum size
4420
for T. inermis and maximum size for T. spinifera [8]; converted to TL using equations from [9]; z Minimum size for T. inermis and maximum size
4421
for T. spinifera [11]; converted to TL using equations from [9]; aa Energy density estimated as 17.65% higher in cold years (R. Heintz,
4422
NOAA/AFSC, unpublished data).
197
4423
Table 9.S2. Component equations of the bioenergetics model used to estimate maximum growth potential
4424
(g•g-1•d-1) of juvenile walleye pollock.
Consumption
C = a × W b × f (T)× h
f (T) = V X × e(X ×(1-V ))
V = (Tcm - T)/(Tcm - Tco )
X = (Z 2 × (1+ (1+ 40 /Y)0.5 )2 /400
Z = ln(Qc )× (Tcm - Tco )
Y = ln(Qc )× (Tcm - Tco + 2)
Respiration
R = (Ar × W Br × f (T)× Am × 13560) + (Ds × (C - F))
f (T) = V X × e(X ×(1-V ))
V = (Trm - T)/(Trm - Tro )
X = (Z 2 × (1+ (1+ 40 /Y)0.5 )2 ) /400
Z = ln(Qr )× (Trm - Tro )
Y = ln(Qr )× (Trm - Tro + 2)
Egestion
F = Fa × C
Excretion
U  Ua  (C  F)
4425
4426
4427
References

4428
1. Gardner GA, Szabo I (1982) British Columbia pelagic marine copepoda: an identification manual and
4429
annotated bibliography. Canadian Special Publication of Fisheries and Aquatic Sciences 62. 536 p.
4430
2. Yamamoto T, Teruya K, Hara T, Hokazono H, Hashimoto H, Suzuki N, Iwashita Y, Matsunari H,
4431
Furuita H, Mushiake K (2008) Nutritional evaluation of live food organisms and commercial dry
4432
feeds used for seed production of amberjack Seriola dumerili. Fisheries Science 74: 1096-1108.
198
4433
3. Lough RG, Buckley LJ, Werner FE, Quinlan JA, Edwards KP (2005) A general biophysical model of
4434
larval cod (Gadus morhua) growth applied to populations on Georges Bank. Fisheries Oceanography
4435
14: 241-262.
4436
4437
4438
4439
4440
4441
4442
4443
4. Lee RF, Hagen W, Kattner G (2006) Lipid storage in marine zooplankton. Marine Ecology Progress
Series 307: 273-306.
5. Tomita M, Ikeda T, Shiga N (1999) Production of Oikopleura longicauda (Tunicata: Appendicularia)
in Toyama Bay, southern Japan Sea. Journal of Plankton Research 21(12): 2421-2430.
6. Frost BW (1989) A taxonomy of the marine calanoid copepod genus Pseudocalanus. Canadian
Journal of Zoology 67: 525-551.
7. Peters J (2006) Lipids in key copepod species of the Baltic Sea and North Sea – implications for life
cycles, trophodynamics and food quality. PhD Dissertation. University of Bremen. 177 p.
4444
8. Pinchuk AI, Coyle KO (2008) Distribution, egg production and growth of euphausiids in the vicinity
4445
of the Pribilof Islands, southeastern Bering Sea, August 2004. Deep-Sea Research II 55: 1792-1800.
4446
9. Pinchuk AI, Hopcroft RR (2007) Seasonal variations in the growth rates of euphausiids (Thysanoessa
4447
inermis, T. spinifera, and Euphausia pacifica) from the northern Gulf of Alaska. Marine Biology 151:
4448
257-269.
4449
10. Kathman RD, Austin WC, Saltman JC, Fulton JD (1986) Identification manual to the Mysidacea and
4450
Euphausiacea of the northeast Pacific. Canadian Special Publications of Fisheries and Aquatic
4451
Sciences 93. 411 p.
4452
11. Falk-Petersen S (1985) Growth of the euphausiids Thysanoessa inermis, Thysanoessa raschii, and
4453
Meganyctiphanes norvegica in a subarctic fjord, North Conway. Canadian Journal of Fisheries and
4454
Aquatic Sciences 42: 14-22.
199
4455
4456
Figure 9.S1. Eastern Bering Sea with locations of sampling stations at which the bioenergetics model and
4457
IBM were run in 2005 (•) and 2010 (). The Monte Carlo Station () is the representative station used
4458
for Monte Carlo simulations. Depth contours are shown for the 50 m, 100 m, and 200 m isobaths.
4459
200
4460
4461
Figure 9.S2. Water temperatures interpolated across all stations (•) sampled by the CTD. Top panel shows
4462
the mean temperature in the upper 30 m of the water column in 2005 (a) and 2010 (b). Bottom panel
4463
shows the mean temperature below 40 m in 2005 (c) and 2010 (d).
4464
201
4465
Chapter 10: Expected declines in recruitment of walleye pollock (Theragra chalcogramma) in the
4466
eastern Bering Sea under future climate change
4467
4468
Franz J. Mueter1*, Nicholas A. Bond2, James N. Ianelli3, Anne B. Hollowed3
4469
4470
1
4471
2
4472
98195, USA
4473
3
4474
Science Center, 7600 Sand Point Way NE, Seattle, WA 98115, USA
School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, Juneau, AK, 99801, USA.
Joint Institute for the Study of the Atmosphere and Oceans, University of Washington, Seattle, WA
National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Alaska Fisheries
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
Citation: Mueter, F.J., Bond, N.A., Ianelli, J.N., and Hollowed, A.B. 2011. Expected declines in
4490
recruitment of walleye pollock (Theragra chalcogramma) in the eastern Bering Sea under future climate
4491
change. ICES Journal of Marine Science: Journal du Conseil 68(6): 1284-1296.
202
4492
Abstract
4493
A statistical model is developed to link recruitment of eastern Bering Sea walleye pollock (Theragra
4494
chalcogramma) to variability in late summer sea-surface temperatures and to the biomass of major
4495
predators. The model is based on recent advances in the understanding of pollock recruitment, which
4496
suggest that warm spring conditions enhance the survival of early larvae, but high temperatures in late
4497
summer and fall are associated with poor feeding conditions for young-of-year pollock and reduced
4498
recruitment in the following year. A statistical downscaling approach is used to generate an ensemble of
4499
late summer temperature forecasts through 2050 based on a range of IPCC climate projections. These
4500
forecasts are used to simulate future recruitment within an age-structured stock projection model that
4501
accounts for density-dependent effects (stock-recruitment relationship), the estimated effects of
4502
temperature and predation, and associated uncertainties. On average, recruitment in 2040-2050 is
4503
expected to decline by 32 to 58% relative to a random recruitment scenario, depending on assumptions
4504
about the temperature relationship, the magnitude of density-dependence and future changes in predator
4505
biomass. The approach illustrated here can be used to evaluate the performance of different management
4506
strategies and provide long-term strategic advice to managers confronted with a rapidly changing climate.
4507
Keywords: Bering Sea, climate change, forecasting, recruitment, sea-surface temperature, statistical
4508
downscaling, walleye pollock
4509
Introduction
4510
Climate change is expected to affect marine fish communities and fisheries production through a variety
4511
of direct and indirect effects of predicted changes in temperature, winds, salinity, stratification, oxygen,
4512
pH, and other factors (Brander, 2010). Climate-related changes have been documented in many
4513
ecosystems. Although the mechanisms are often uncertain, changes to recruitment success through
4514
changes in production or survival are believed to be the key process driving these changes (Rijnsdorp et
4515
al., 2009). Forecasting the effects of climate change on a fish population requires (1) an understanding of
4516
the mechanisms linking climate drivers to fish production (recruitment, growth, and distribution), (2)
4517
forecasts of the key climate drivers, and (3) a stock projection model that captures the essential dynamics
4518
of the population of interest (Hollowed et al., 2009). A critical requirement of this approach is that the
4519
known or presumed mechanisms affecting fish production can be quantified through robust statistical
4520
relationships and that these relationships remain valid under anticipated climate changes. Here the
4521
approach outlined by Hollowed et al. (2009) is illustrated with a case study by modeling the possible
4522
responses of walleye pollock in the eastern Bering Sea to future climate variability.
203
4523
Walleye pollock are an important component of the eastern Bering Sea ecosystem and currently support
4524
the second largest single species fishery in the world with landings for the eastern Bering Sea alone
4525
ranging from 800 000 to 1.4 million tonnes over the last three decades. Their geographic range extends
4526
from Japan to the Bering Sea and as far south as northern California. In the eastern Bering Sea, walleye
4527
pollock occupy a central position in the food web and serve as a key forage species for many upper
4528
trophic level species, including fish, seabirds, and marine mammals (Aydin et al., 2007). In addition,
4529
cannibalism is a major source of mortality for juvenile pollock. Adults consume primarily age-0 pollock
4530
during fall and winter on the southeastern portion of the shelf and age-1 pollock during summer, fall and
4531
winter to the northwest of the Pribilof Islands (Dwyer et al., 1987), consistent with the observed
4532
distribution of age-0 and age-1 pollock (Figure 10.1).
4533
Spawning concentrations of walleye pollock in the eastern Bering Sea, as inferred from the distribution of
4534
eggs, occur near Bogoslof Island, north of Unimak Island and the Alaska Peninsula, and around the
4535
Pribilof Islands (Figure 10.1, Bacheler et al., 2010). Spawning occurs in February and March around
4536
Bogoslof, in March and April north of Unimak Island, and from April-August around the Pribilof Islands
4537
(Bacheler et al., 2010). However, little spawning has been evident around Bogoslof Island in recent years
4538
due to a very low spawning biomass (Ianelli et al., 2009). By the fall of their first year, pollock are
4539
primarily distributed over the middle shelf, while age-1 pollock in the following summer primarily
4540
occupy the outer shelf to the northwest of the Pribilof Islands (Figure 10.1). Pollock also undergo
4541
northward and shoreward feeding migrations during spring and summer (Kotwicki et al., 2005). Thus
4542
pollock utilize large portions of the middle and outer shelf, hence our efforts to identify environmental
4543
drivers of recruitment will focus on this area, in particular the middle-shelf region, which is important for
4544
age-0 pollock at a potentially critical stage in their early life (Hunt et al., this volume).
4545
Based on previous studies (as reviewed in Mueter et al., 2006) and recent results from the Bering Sea
4546
Integrated Ecosystem Research Program (Hunt et al., this volume), conditions affecting recruitment of
4547
walleye pollock include: (1) ice and temperature conditions at the time of hatching, which determine early
4548
feeding conditions; (2) summer stratification over the shelf during the first summer (age-0), which affects
4549
feeding conditions of late larval stages as well as vulnerability to predation, (3) the abundance and
4550
distribution of potential predators, including predation of adult pollock, arrowtooth flounder, and other
4551
predators on age-0 and age-1 pollock. Moreover, the magnitude of predation may be affected by the
4552
spatial overlap between larval or juvenile pollock and their predators, including adult walleye pollock
4553
(Mueter et al., 2006; Wespestad et al., 2000; Wyllie-Echeverria, 1996).
4554
Our current understanding of the drivers of walleye pollock recruitment in the eastern Bering Sea is based
4555
on a modified version of the Oscillating Control Hypothesis (OCH) originally proposed by Hunt et al.
204
4556
(2002) and recently revised based on new findings (Hunt et al., this volume). The OCH predicted that
4557
pollock recruitment should be greatest in warm years when ice retreats early and a late bloom occurs in
4558
thermally stratified water. These conditions favor the pelagic community because primary production is
4559
consumed within the surface layer by zooplankton that serves as prey for larval walleye pollock. While
4560
this prediction is supported by high survival of walleye pollock larvae from hatching to summer during
4561
the most recent warm period from 2002-2005, these years failed to produce large, lipid-rich zooplankton,
4562
such as Calanus marshallae, that provide important food for walleye pollock during late summer and fall
4563
(Coyle et al., in press). As a consequence, larval pollock during these warm years had low energy
4564
densities in the fall and may have experienced low overwinter survival due to increased predation or
4565
starvation (Moss et al., 2009). Therefore conditions in late summer and fall may be critical to the overall
4566
survival of pollock from spawning to recruitment at age-1.
4567
In this study, we build on these observations to develop an empirical relationship between late summer
4568
environmental conditions and walleye pollock survival from age-0 to age-1. This relationship is used
4569
within a simplified stock projection model to forecast recruitment of walleye pollock over the next 40
4570
years. Our goal is to provide reasonable projections of future recruitment and abundance under different
4571
warming scenarios and under a plausible harvest scenario. Major sources of uncertainty in climate
4572
projections and in the estimated environmental relationships are accounted for to characterize the full
4573
range of likely population trajectories under the selected harvest scenario. We first examine the empirical
4574
evidence for the proposed recruitment mechanisms to develop a robust functional relationship between
4575
recruitment of walleye pollock in the eastern Bering Sea and key climate variables. A statistical
4576
downscaling approach is then used to forecast an ensemble of likely trajectories (time series) of these key
4577
variables through 2050. These time series, in turn, are used to generate recruitment trajectories, which are
4578
input into the stock projection model to examine likely responses of walleye pollock to climate-driven
4579
changes in recruitment, as well as to possible changes in predation from arrowtooth flounder (Figure
4580
10.2).
4581
Methods
4582
Based on our current understanding of walleye pollock recruitment a conceptual model of the most
4583
important factors determining the survival of early gadid stages from spawning through recruitment at
4584
age-1 was developed, assuming that recruitment variability occurs primarily at the larval and early
4585
juvenile stages. Suitable data series for modeling the relationships between potential explanatory
4586
variables and survival of walleye pollock were selected to test the hypothesized mechanisms. Explanatory
4587
variables included timing of sea ice retreat and a spring transition index to capture conditions during the
4588
early larval stage; summer wind mixing, late summer sea-surface temperatures (SST), and late summer
205
4589
water column stability to capture environmental conditions during the late larval stage; and an index of
4590
predation pressure on juvenile walleye pollock to estimate effects of the major predators, including
4591
cannibalism, on juvenile survival (Table 10.1).
4592
Modeling stock-recruitment residuals
4593
Because there is strong evidence for density-dependence in walleye pollock, residuals from a stock-
4594
recruitment (S-R) relationship were used as the primary response variable to quantify variability in
4595
survival. This index, also referred to as log-survival, removes the impact of spawning biomass on
4596
recruitment and should more accurately reflect the influence of environmental variability on recruitment.
4597
The S-R relationship was estimated for year classes 1963-2008 within the 2009 age-structured assessment
4598
model (Ianelli et al., 2009) using the following parameterization of the Ricker model:
4599
𝑅𝑡 = 𝑆𝑆𝐵𝑡−1⁄∅0 ∙ 𝑒 𝛼∙(1−𝑆𝑆𝐵𝑡−1 ⁄𝐵0 ) ∙ 𝑒 𝜀𝑡
(Eq. 1)
4600
where Rt is recruitment at age 1 in year t at a given level of female spawning stock biomass (SSB) and 𝑒 𝜀𝑡
4601
is a multiplicative error term. Parameters of the relationship were estimated to be α = 2.084, ø0 = 0.2631,
4602
and B0 = 4934 (Ianelli et al., 2009). Although age-1 recruitment in the assessment model was estimated
4603
through 2008, relative year-class strength may be modified after the early juvenile stage, hence the
4604
estimate of the 2008 year-class was deemed unreliable for this analysis.
4605
Because of the large number of explanatory variables relative to the length of the available time series,
4606
and because of potential confounding among the variables, a Principal Components Analysis (PCA) was
4607
used to reduce the number of explanatory variables to a smaller number of independent variables
4608
(Principal Components). To model recruitment or survival as a function of these Principal Components or
4609
as a function of individual variables, a generalized additive modeling approach (GAM) was used in the
4610
exploratory stages (Wood, 2006) and a general linear modeling approach was used for selecting the final
4611
model and for quantifying uncertainty for the projections. Thus, S-R residuals (εt) from Eq. 1 were
4612
modeled as a function of one or more environmental variables acting in year t-1 (Xi,t-1, i = 1,2,..) using an
4613
additive model with non-parametric smooth functions (fi) of the explanatory variables:
4614
𝜀𝑡 = 𝛽′0 + 𝑓1 (𝑋1,𝑡−1 ) + 𝑓2 (𝑋2,𝑡−1 ) + ⋯ + 𝜈′𝑡
4615
Potential interactions among the explanatory variables were also explored by fitting smooth functions (=
4616
smooth surfaces) of two environmental variables {e.g. f(X1,X2)}. These interaction terms were not found
4617
to improve the models significantly and were not further considered. The degree of smoothing was
4618
determined through generalized cross-validation (Wood, 2006) and, for the final models, smooth terms
206
4619
were replaced by linear or polynomial terms with approximately the same degrees of freedom as the
4620
estimated smooth terms:
𝜀𝑡 = 𝛽0 + 𝛽1 𝑋1,𝑡−1 + 𝛽2 𝑋2,𝑡−1 + ⋯ + 𝜈𝑡
4621
(Eq. 2)
4622
where the residuals ν't or νt are assumed to follow a normal distribution with variance 𝜎𝜈2 , β0 and β'0 are
4623
intercepts, β1 and β2 are slope parameters, and the Xi terms may be a quadratic or other power
4624
transformation of the measured variables. Residuals were tested for serial dependence and, if appropriate,
4625
residual variability was modeled as a first-order autoregressive process. The small-sample or corrected
4626
Akaike Information Criterion (AICc, Hurvich and Tsai, 1989) was used to identify the most parsimonious
4627
model for predicting recruitment. The estimated level of uncertainty in the identified relationships was
4628
incorporated in the projections as described below to characterize uncertainty in future population
4629
trajectories.
4630
Modeling recruitment
4631
As an alternative to modeling S-R residuals estimated within the stock assessment model as a function of
4632
environmental variables, recruitment estimates from the assessment were used in a “post-assessment”
4633
analysis as response variable in a generalized Ricker model (Quinn and Deriso, 1999). The linear form of
4634
this model describes log-transformed recruitment as a function of spawning biomass and environmental
4635
variables as follows:
log(𝑅𝑡 ) = 𝛼 + 𝛽𝑆𝑆𝐵𝑡−1 + ∑𝑖 𝛾𝑖 𝑋𝑖,𝑡−1 + log(𝑆𝑆𝐵𝑡−1 ) + 𝜀𝑡
4636
(Eq. 3)
4637
where α and β are the productivity and carrying capacity parameters of the Ricker model, respectively, the
4638
Xi are one or more covariates acting on the egg and larval stages of walleye pollock prior to year t, the γi
4639
are the corresponding regression coefficients, and εt are either independent, normally distributed residuals
4640
or follow a first-order autoregressive process. The second SSB term was included in the model as an
4641
offset without a coefficient. In the exploratory stage, we also fit semi-parametric models with a liner term
4642
for SSB and smooth terms for the covariates Xi.
4643
Data sources
4644
Long-term indices of environmental variability that are potentially important to walleye pollock were
4645
constructed from various sources. Temperature indices were based on monthly extended reconstructed
4646
sea-surface temperatures (ERSSTv3, Smith et al., 2008), which are interpolated values on a 2° latitude x
4647
2°longitude grid, and were averaged over the southeast Bering Sea shelf inshore of the shelf break and
4648
extending to 61˚N for this analysis. Monthly ERSST data were used to develop two indices of
4649
temperature conditions: a spring transition index related to the timing of the non-ice-associated spring
207
4650
bloom and a late summer SST index. The spring transition index was constructed by interpolating
4651
between monthly mean shelf temperatures (assumed to reflect temperatures on the 15th of each month)
4652
using a cubic spline and estimating the day of the year when smoothed temperatures first exceeded 4°C.
4653
As an index of late summer upper layer temperature conditions, the monthly ERSSTs were averaged over
4654
the southeast Bering Sea shelf from July 1 through September 30.
4655
An index of the timing of sea ice retreat (ICE) for the period 1972-2003 was based on Palmer (2003) as
4656
described in Mueter et al. (2006) and was extended through 2008 using a regression-based proxy. The
4657
original index was defined as the International Organization for Standardization (ISO) week during which
4658
average ice concentration in the NMFS survey area first drops below 20%, where ice concentrations were
4659
obtained from digital ice charts provided by the Arctic Climatology Project, National Ice Center, NOAA
4660
(http://www.natice.noaa.gov). Predicted values of the index for 2004-2008 were obtained from a multiple
4661
linear regression of the original index on mean April air temperatures at St. Paul airport (57.15˚N,
4662
170.22˚W), mean April-May SST (ERSSTv3 averaged over southeast Bering Sea shelf as described
4663
above), and the mean February through April North-South wind speed component at 10m height near
4664
56.2˚N, 168.75˚W from the NCEP reanalysis (Kalnay et al., 1996). The best-fit linear regression
4665
explained 90% of the variability in the observed ice retreat and predicted values were obtained as follows:
4666
ICE = 19.196 – 1.672*SST – 7.020*v – 2.346*v2 – 3.329*airT – 2.504*airT2 – 2.187*airT3
4667
where SST, airT, and v are the sea-surface temperature index, air temperature index, and N-S winds,
4668
respectively, as described above. The best regression model was determined by first finding the additive
4669
model with smooth terms for each variable that resulted in the best leave-one out predictions (Wood,
4670
2006) and then substituting quadratic and cubic polynomial terms for smooth terms that had
4671
approximately 2 and 3 degrees of freedom, respectively. The parametric fit was almost identical to the
4672
additive smooth fit and was chosen for the analysis to simplify the computation of predicted values and to
4673
enhance transparency.
4674
To capture summer stratification, two alternative approaches were used. Data from a mooring at station
4675
M2 on the middle shelf for 1996-2007 (Stabeno et al., 2001; Stabeno et al., 2007) were used to compute
4676
an index of water column stability during late summer. The stability index was computed as the negative
4677
depth-integrated potential energy (J m-2), or the energy required to mix the water column, following
4678
Simpson et al. (1977). Daily average temperature and salinity profiles at M2 were estimated from discrete
4679
depth measurements by linearly interpolating between consecutive depths prior to computing the stability
4680
index. Daily indices of stability were averaged from July 1 through September 30. Because the measured
4681
time series of stratification was much shorter than the recruitment series, a proxy for water column
208
4682
stability based on a 1-D model of mixed-layer depth was used to extend the time series back to 1963
4683
(Carol Ladd, NOAA-PMEL, Seattle, pers. comm.). To quantify predation pressure, the only biotic
4684
variable in our analyses, diet compositions and consumption estimates from the early 1990s (Aydin et al.,
4685
2007) were used to construct an index of “potential predation pressure". The major sources of mortality
4686
for juvenile walleye pollock in the early 1990s, accounting for well over 50% of overall mortality, were
4687
adult walleye pollock, arrowtooth flounder, and flathead sole (Aydin et al., 2007). Therefore, an index of
4688
predation in year t (Predt) was computed as follows:
3
𝑃𝑟𝑒𝑑𝑡 = ∑ (𝑄⁄𝐵) × 𝐵𝑖,𝑡 × 𝑝𝑖
4689
𝑖
𝑖=1
4690
𝑄
where ( ⁄𝐵) is the consumption rate (consumption Q per unit biomass B) of predator i, Bi,t is the
4691
estimated total biomass of predator i in year t (obtained from NPFMC, 2009), and pi is the average
4692
proportion of juvenile pollock in the diet of predator i. This assumes that both the consumption rate and
4693
the proportion of pollock in a predator diet remain the same as those estimated for the early 1990s.
4694
Clearly, both assumptions may be violated as the consumption rate is a function of the age structure of the
4695
population and diet composition varies with the relative abundance of different prey, the spatial overlap of
4696
predators and prey, and other factors that affect prey availability. Therefore, potential effects of changes
4697
in spatial distribution of predators, as quantified by the center of gravity, on pollock recruitment were
4698
examined but were not found to be significant and are not further considered here. The index of predation
4699
was computed across the three major predators from 1977-2008, when biomass estimates for all three
4700
species were available, and was strongly correlated with the biomass of walleye pollock over this period
4701
(r = 0.91) because adult walleye pollock are the main predators on juvenile pollock due to their large
4702
biomass. Therefore, we also correlated pollock S-R residuals with the biomasses of each individual
4703
predator to confirm the negative relationship between the biomass of each predator and survival of
4704
walleye pollock from spawning to recruitment. Only arrowtooth flounder and flathead sole were included
4705
in the generalized Ricker model (eq. 3) because cannibalism by walleye pollock is implicitly captured by
4706
the density-dependent term in this formulation (effect of SSB on log-recruitment).
4707
Regional temperature forecasts
4708
To predict future recruitment of walleye pollock from global climate scenarios, regional forecasts of key
4709
environmental variables that drive recruitment variability of walleye pollock in the eastern Bering Sea are
4710
needed (Hollowed et al., 2009). A statistical downscaling approach was employed to forecast these
4711
variables, in particular summer SST over the shelf, from IPCC model projections. Because there are large
4712
uncertainties about any climate projection, a total of 82 climate scenarios were considered to reflect the
𝑖
209
4713
range of uncertainty in outcomes. Plausible future temperature scenarios to characterize the range of
4714
uncertainty were selected based on a subset of 9 IPCC models that performed best in capturing historical
4715
climate variability of the North Pacific in their 20th century hindcast simulations (Overland and Wang,
4716
2007). It is assumed that the models that are able to replicate the observed spatial and temporal
4717
characteristics of the Pacific Decadal Oscillation (PDO) are apt to be those that better handle the
4718
atmospheric forcing, air-sea interactions and upper-ocean circulations of the North Pacific including the
4719
Bering Sea, and that these models are therefore more reliable in terms of their simulations for the 21st
4720
century. On the other hand, the evaluation of model skill is hampered by the existence of only one
4721
realization of the past climate, and so it seems sensible to retain a relatively large number of individual
4722
models for the purpose of producing a meaningful ensemble mean and quantifying future uncertainties.
4723
For the 9 models chosen, 21st century projections with the low (B1), intermediate (A1B) and high (A2)
4724
CO2 emissions scenarios were considered. All three scenarios were used because they span a wide range
4725
in the simulated future climate forcing of the Bering Sea. In our use of IPCC climate model output we
4726
have essentially followed the “best practices” outlined by Overland et al. (in press).
4727
Stock projections
4728
Future recruitment series for walleye pollock were simulated under different climate scenarios and were
4729
used to drive a population dynamics model to explore the effects of climate variability on future
4730
population trajectories (Figure 10.2). Walleye pollock population numbers, biomass, and catches through
4731
2050 were projected starting with the 2009 numbers-at-age as estimated in the most recent stock
4732
assessment (Ianelli et al., 2009). To convert numbers to biomass, weights-at-age for all future years were
4733
assumed to be equal to the 1999-2008 observed average. Other parameter values for the projections were
4734
set equal to the values estimated by or used in Ianelli et al. (2009), including maturity-at-age, selectivity-
4735
at-age, and natural mortality. Projections used standard population dynamics equations as described in
4736
Ianelli et al. (2009) and a harvest control rule similar to the Tier 3 control rule used for many groundfish
4737
species off Alaska (NPFMC, 2002), but with an annual cap on total catches of 1.5 x 106 mt. This cap is
4738
close to recent maximum catches that have been achieved under the regulatory cap of 2 x 106 mt on total
4739
groundfish removals from the eastern Bering Sea as specified in the Fishery Management Plan (NPFMC,
4740
2002).
4741
Because our focus was on exploring the effects of future climate changes on recruitment, all population
4742
parameters except recruitment were fixed in the simulations. This included reference points used in the
4743
harvest control rule, which are the unfished level of spawning stock biomass (B100%, the projected
4744
spawning biomass under no fishing and assuming average recruitment based on the 1977-2008 period)
4745
and the corresponding fishing mortality rate (F40%) that would reduce spawning biomass to 40% of the
210
4746
unfished biomass (B40%). Under the assumed harvest control rule, the stock is fished at F40% if current
4747
biomass (B) is larger than B40% and fishing is reduced linearly if the biomass declines below B40%.
4748
Moreover, F is set to zero if the biomass declines below 20% of the unfished biomass as a precautionary
4749
measure that was implemented to protect the prey base for endangered Steller sea lions (Eumetopias
4750
jubatus).
4751
Recruitment at age-1 was simulated for each year of the projections using one of three general
4752
approaches:
4753
1) Random stock-recruitment residuals: As a control and for comparisons with scenarios that include
4754
environmental effects on recruitment, we generated future recruitment series by randomly drawing
4755
values from the observed S-R residuals for brood years 1977-2007 (εt in Eq. 1) and calculating
4756
recruitment in year t according to Eq. 1.
4757
2) Simulated stock-recruitment residuals (Type 1 model): Future S-R residuals under a given climate
4758
scenario were simulated from the AICc-best model(s) for describing the climate-recruitment
4759
relationship (Eq. 2). Values for εt in Eq. 2 (𝜀𝑡∗ ) were simulated by accounting for the full prediction
4760
uncertainty (i.e. parameter uncertainty plus residual uncertainty):
𝜀𝑡∗ = 𝛽0 + 𝛽1 𝑋1,𝑡−1 + 𝛽2 𝑋2,𝑡−1 + ⋯ + 𝑡𝑑𝑓 √𝜎𝛽2 + 𝜎𝜈2
4761
4762
where tdf is drawn from a Student t distribution with degrees of freedom equal to residual degrees of
4763
freedom of the best environmental model, 𝜎𝛽2 is the variance of the predicted value at the given level
4764
of the environmental variables, and 𝜎𝜈2 is the residual variance.
4765
3) Simulated recruitment from generalized Ricker model (Type 2): Future recruitments were simulated
4766
from a log-normal distribution with predicted means and variances estimated from the AICc-best
4767
model(s) of the form described in equation 3.
4768
The AICc-best models in both cases (Types 1 and 2) included predation terms. For Type 1 models, the
4769
predation index was annually updated in the projections to account for changes in walleye pollock
4770
biomass, while arrowtooth flounder biomass and flathead sole biomass were held constant at their 2009
4771
values. Under Type 2 models we explored two future scenarios for arrowtooth flounder, assuming either a
4772
constant biomass at the 2009 value for all future years (Model 2a) or a continuing linear increase in
4773
arrowtooth flounder biomass at the rate observed between 1991-2009 (Model 2b).
4774
Results
4775
Walleye pollock recruitment has undergone large fluctuations over recent decades (Figure 10.3). Large
4776
year classes tend to occur every 4-6 years and can cause considerable fluctuations in biomass. Similar to
211
4777
other fish species in the Bering Sea (Mueter et al., 2007), strong recruitment after the 1976/77 climate
4778
regime shift led to a strong increase in biomass. The average estimated recruitment at age-1 was similar
4779
prior to (18.3 billion) and following the regime shift (22.3 billion, t-test on log-transformed abundances:
4780
t=-0.56, p = 0.58). However, several strong year classes originated after the regime shift starting in 1977,
4781
followed by a significant decrease in average log-transformed recruitment (linear trend over time: t=-2.24,
4782
p = 0.033). In particular, there has been a considerable decrease in the strength of the largest year classes.
4783
For example, a quantile regression shows a significant decrease in the 80th percentile of log-transformed
4784
recruitment over time (Figure 10.3, top panel).
4785
Environmental variables were characterized by high interannual variability over most of the time series
4786
with a few multi-year periods of consistent cold (1971-1976) or warm (2001-2005) conditions (Figure
4787
10.3). As expected, temperature, ice, and spring transition indices were confounded with each other and,
4788
to a lesser extent, with the stratification index (Table 10.1). Importantly, the predation index was not
4789
correlated with any of the environmental indices.
4790
Correlations among environmental variables were used to reduce the number of variables to four
4791
significant principal components, which accounted for 97% of the overall variability (Table 10.1). The
4792
first PC contrasts warm years characterized by little ice, an early spring transition, and warm summer SST
4793
with cold years. Warm years tend to also be strongly stratified (positive loading for stratification index).
4794
However, variability in stratification and wind mixing are primarily captured by PC 2, which contrasts
4795
years characterized by weak wind mixing and strong stratification (positive values of PC 2) with years
4796
that have strong wind mixing and weak stratification (negative PC 2). The third PC primarily reflects
4797
predation with weak loadings on all of the environmental variables, while PC 4 in its positive phase was
4798
associated with unusual years characterized by both a late ice retreat and warm summers with strong wind
4799
mixing.
4800
Recruitment modeling
4801
For models of type 1, the best overall model based on AICc explained much of the variability in S-R
4802
residuals (adjusted R2 = 0.55, Figure 10.4) as a function of PC 1 (smooth fit with approximately 2 d.f., F
4803
= 5.36, p = 0.0088) and PC 3 (linear term, F = 25.5, p < 0.001). Neither PC 2 nor PC 4 were significantly
4804
related to variability in S-R residuals and there was no significant interaction between PC 1 and any of the
4805
other PCs. Thus variability in survival appears to be most strongly related to overall temperature
4806
variability (PC 1) and predation (PC 3), and survival anomalies were highest when temperatures were low
4807
to intermediate and when the predation index was low (Figure 10.4). Because of strong correlations
4808
among the variables (Table 10.1), the relative effects of spring and summer conditions on survival cannot
212
4809
statistically be separated. However, GAM models of S-R residuals as a function of individual variables
4810
suggest that survival was significantly related to late summer SST (3 d.f., adj. R2 = 0.24, F = 4.46, p =
4811
0.0082), but not to the timing of ice retreat (p = 0.775) or to the spring transition index (p = 0.382).
4812
Therefore, late summer SST appears to be more important in determining survival to recruitment than
4813
spring ice and temperature conditions.
4814
To predict future survival anomalies (S-R residuals) from available climate forecasts we further
4815
simplified the model as follows: First, we replaced PC 1 in the model with average Jul-Sep SSTs because
4816
ice conditions and the spring transition index were not significant individually and because ice retreat is
4817
difficult to forecast from current climate models. Second, we replaced the smooth terms with a quadratic
4818
term for SST and a linear term for predation to fit a linear regression model. The resulting model provided
4819
a reasonable fit (Figure 10.5, adj. R2 = 0.46, F = 9.51, p < 0.001) and will be referred to as model 1 in the
4820
projections. Further model comparisons confirmed that none of the other variables improved the model
4821
significantly when added individually (AICc always increased) and that there was no significant
4822
interaction between SST and predation (Difference in AICc values: ΔAICc = 4.5). Residuals from the best
4823
model were not significantly auto-correlated (Durbin-Watson test statistic = 1.82, p=0.243) and were
4824
close to normally distributed (Shapiro-Wilk test for normality, W = 0.968, p = 0.471). Although walleye
4825
pollock dominated the predation term in the model, the biomass of each individual predator had a
4826
negative effect on S-R residuals at the 90% significance level (walleye pollock: p < 0.001; arrowtooth
4827
flounder: p = 0.014; flathead sole: p = 0.065).
4828
An alternative model of type 2, using recruitment and spawner biomass estimates from the stock
4829
assessment to fit a generalized Ricker model, suggested similar significant effects of late summer
4830
temperatures on log-transformed recruitment. An exploratory model with a smooth term for summer SST
4831
suggested that (log-) recruitment was relatively high at lower SSTs, dropped steeply between
4832
approximately 9.2 and 9.8°C, and was low at higher temperatures. We therefore replaced the smooth term
4833
for SST with a threshold at 9.4°C, where the threshold value was estimated within the model (Figure
4834
10.5). The biomass of arrowtooth flounder was significantly and negatively related to pollock recruitment
4835
(simple linear regression, t = -2.115, p = 0.043), but the effect was not significant when arrowtooth
4836
flounder was included in the generalized Ricker model (p = 0.106). Nevertheless, a negative coefficient
4837
is consistent with a predation effect and with the model described above, hence arrowtooth flounder
4838
biomass was included in the model to examine the possible effects of different future arrowtooth
4839
trajectories on pollock recruitment (Figure 10.5, adj. R2 = 0.56, F = 13.9, p < 0.001). This model will be
4840
referred to as model 2 in the projections.
4841
Projections
213
4842
Ensemble predictions of future trajectories for average July – September SST under three emissions
4843
scenarios show very high variability in individual trajectories and a gradual increase in the ensemble
4844
mean through 2050 (Figure 10.6). On average, late summer SSTs in the eastern Bering Sea are expected
4845
to increase by about 1°C under IPCC emissions scenarios A2 and B1, and a somewhat larger increase
4846
under scenario A1B. The larger increase in the A1B scenario can be attributed to a slightly faster growth
4847
in global temperatures relative to the higher emissions A2 scenario until about 2050 (at which point
4848
temperatures under the A2 scenario begin to increase at a greater rate) and perhaps to a regional effect.
4849
Simulated population trajectories of walleye pollock were highly variable due to large variability in
4850
recruitment resulting from both high variability in future SST trajectories and large uncertainties in the
4851
estimated effects of SST and predation. Under model 1, average recruitment across the 3 emissions
4852
scenarios is expected to decline over the next 40 years by approximately 44% relative to the random
4853
recruitment scenario (based on 1977-2007 mean, see Figure 10.8). However, a 90% simulation envelope
4854
includes the 1977-2007 mean, implying approximately a 13% probability that average recruitment will be
4855
higher than the 1977-2007 mean in 2050. Similarly, spawning biomass and catches are expected to
4856
decline substantially with less than a 7.5% probability that spawning biomass will exceed B40% in 2050.
4857
Many of the catch trajectories resulted in zero catches during later years under model 1 as a result of
4858
spawning biomass frequently falling below the B20% threshold (Figure 10.7).
4859
The expected distribution of recruitment, spawner biomass, and catches are expected to decrease relative
4860
to the random recruitment scenario under both models (Type 1, 2), and with (2b) or without (2a) a
4861
continuing increase in arrowtooth flounder biomass (Figure 10.8). The simulated values had a similar
4862
distribution under all three emissions scenarios with slightly larger declines under scenario A1B. Declines
4863
under model 2a were moderate because of a strong compensatory response in recruitment implied by the
4864
generalized Ricker model (i.e. recruitment increases considerably, on average, at lower levels of SSB). If
4865
arrowtooth flounder biomass continues to increase, as assumed under model 2b, recruitment, spawner
4866
biomass and catches are expected to decrease substantially relative to the random recruitment scenario
4867
and relative to a model assuming arrowtooth flounder biomass remains at the current level (Figure 10.8).
4868
Discussion
4869
Empirical relationships between SSTs on the southeastern Bering Sea shelf during late summer and
4870
recruitment success of walleye pollock suggest that recruitment is reduced if regional average surface
4871
temperatures exceed approximately 9.4°C. The implications of these findings are that increasing
4872
temperatures in the eastern Bering Sea, as predicted by IPCC climate models under a range of scenarios
4873
(Figure 10.6), will likely reduce future recruitment, biomass, and harvests of walleye pollock (Figure
214
4874
10.8). However, large uncertainties in SST projections and in the estimated relationships imply large
4875
uncertainties in future population trajectories. Under the assumed harvest scenario, there is large overlap
4876
in simulated trajectories between the random recruitment scenario and all of the temperature-dependent
4877
scenarios, but under all models there is a very high probability that future biomass and catches will be
4878
lower than in the past. Simulations under a variety of plausible future management scenarios lead to
4879
similar conclusions (Ianelli et al., this volume).
4880
Environmental effects on recruitment
4881
Our results are consistent with recent findings that unusual warm conditions during the period from 2002
4882
to 2005 resulted in poor feeding conditions and low energy content of age-0 pollock in the fall months
4883
(Coyle et al., in press; Hunt et al., this volume). In spite of large abundances of age-0 pollock in surface
4884
waters during those years (Moss et al., 2009), the survival of the 2002-2005 year classes was low and
4885
resulted in very poor recruitment (Figure 10.3), presumably as a consequence of reduced overwinter
4886
survival from age-0 to age-1 (Hunt et al., this volume). Poor feeding conditions during the warm period
4887
resulted from a lack of the large copepd C. marshallae, euphausiids (in particular Thysanoessa raschii),
4888
and other large zooplankton species that are important prey for larval and early juvenile pollock. The
4889
mechanisms that caused low abundances of C. marshallae and other large zooplankton are poorly
4890
understood and may relate to a mismatch between the timing of the spring bloom and the prey needs of
4891
copepod nauplii (Baier and Napp, 2003; Hunt et al., this volume), or to a reduction in post-bloom
4892
production resulting from intense stratification and reduced nutrient supply into the surface layer (Coyle
4893
et al., in press; Coyle et al., 2008; Hunt et al., 2008). The latter hypothesis implies that reduced summer
4894
productivity limits food availability, growth, and subsequent survival of large zooplankton and may cause
4895
them to descend into deeper waters earlier in the season compared to cold years. Our results support the
4896
importance of late summer and fall conditions, but strong confounding between spring and summer
4897
temperature conditions (Table 10.1) do not allow us to statistically separate these non-exclusive
4898
hypotheses.
4899
Several studies provide evidence for the importance of upper layer temperatures and water column
4900
stratification, particularly during late summer and fall, to the survival of young-of-year walleye pollock.
4901
There is a very strong and negative correlation between a measure of water column stratification at the
4902
M2 mooring site during July-September and pollock survival (1996-2007, r = -0.86, p < 0.001, Coyle et
4903
al., in press), although this relationship became non-significant when the extended, model-based index of
4904
water column stratification (Figure 10.3) was used as a proxy for conditions at M2 (Mueter, unpublished
4905
data). This may indicate that the index is a poor measure of stratification in earlier years, that the
4906
importance of stratification has increased in recent years, or that the relationship is due to other factors
215
4907
associated with stratification. Strong stratification during warm years generally implies a shallower
4908
thermocline and a thinner upper layer, which may result in food limitations due to the reduction in
4909
available habitat and increased energetic demands due to warmer temperatures (Ciannelli et al., 1998).
4910
Similarly, during years with warm temperatures in the fall, walleye pollock in the Gulf of Alaska may
4911
suffer from higher predation and limited food availability (Ciannelli et al., 2004). Temperatures and
4912
stratification on the eastern Bering Sea shelf are weakly correlated (Table 10.1) and warm years (e.g.
4913
2000, 2001) can have low stratification, while cold years such as 2007 can have very high stratification
4914
(Carol Ladd, pers. comm.), providing some contrast between temperature and stratification. Results from
4915
the principal components analysis suggest that, to the extent that effects can be disentangled, temperature
4916
conditions (PC 1, Table 10.1) are more important than stratification (PC 2) to pollock survival.
4917
The finding that warm conditions in the eastern Bering Sea are associated with poor survival contradicts
4918
earlier findings that warmer years tend to produce strong year classes of walleye pollock (Hollowed et al.,
4919
2001; Mueter et al., 2006; Quinn and Niebauer, 1995). This apparent contradiction can be resolved by
4920
observing that warmer springs with an early ice retreat appear to favor survival of early larvae (Hunt et
4921
al., this volume), but excessive warm temperatures in the fall result in poor overwinter survival, resulting
4922
in a dome-shaped relationship between pollock survival and SST. The current analysis was limited to the
4923
post-regime shift period when SSTs were near the “optimum” SST range or on the descending limb of the
4924
hypothesized relationship (Figure 10.5). However, a dome-shaped relationship is clearly evident if the full
4925
time series of pollock recruitment (1963-2007) is plotted against late summer SST (Figure 9 in Coyle et
4926
al., in press). Similarly, Pacific cod (Gadus macrocephalus), whose recruitment is strongly correlated
4927
with that of walleye pollock in the eastern Bering Sea, switched from a positive relationship between SST
4928
and recruitment prior to the 1976-77 regime shift to a negative relationship after the regime shift (Mueter
4929
et al., 2009), when average temperatures were higher.
4930
The projections presented here assume that the mechanisms that caused low recruitment during the recent
4931
warm period, which were followed by a dramatic decline in pollock biomass, will continue to operate into
4932
the future. Although such empirical relationships frequently break down over time (Myers, 1998), recent
4933
field observations, spanning a period of contrasting warm and cold conditions, offer strong support for a
4934
decrease in recruitment under very warm conditions (Hunt et al., this volume). This support is based on a
4935
mechanistic understanding of variability in prey conditions affecting pollock survival; hence we believe
4936
that the estimated SST effect offers a reasonable basis for simulating future recruitment variability as long
4937
as the associated uncertainties are taken into account.
4938
Uncertainties in stock projections
216
4939
In this study we only considered uncertainty in future SST trajectories and in the relationship between
4940
SST and recruitment as estimated outside the assessment. Clearly, the population dynamics of walleye
4941
pollock are highly uncertain and future improvements should consider full uncertainty in the assessment
4942
and in the estimated management parameters. Moreover, the estimation of SST effects could be integrated
4943
within the assessment model for consistency between the retrospective estimation and future projections.
4944
Additional uncertainties about the effect of SST on future recruitment arise from extrapolating the
4945
estimated SST relationship beyond the range of observed temperatures. Projections of late summer SST
4946
used in this study exceed the maximum observed value (10.4°C) approximately 29% of the time, with
4947
temperatures under some scenarios exceeding 13.5°C in individual years. It is assumed here that
4948
recruitment is not further reduced when SST exceeds the observed maximum (Figure 10.5), an
4949
assumption that is likely to underestimate the true effect of climate warming on recruitment. However,
4950
prey and/or pollock dynamics could fundamentally change under continued warming, for example
4951
through northward shifts in the distribution of spawning and nursery areas. Such adaptive responses
4952
cannot be predicted at present, but populations may be particularly vulnerable during periods of
4953
adaptation, or the rate of change may overwhelm the ability of species to adapt (Brander, 2010), implying
4954
a need for additional precaution under rapid climate change.
4955
Additional uncertainties in the projections arise from the incorporation of density-dependent effects
4956
(including cannibalism) and predation in models of recruitment. The stock-recruitment relationship
4957
estimated within the stock assessment model implies a moderate level of density-dependence (Ianelli et
4958
al., 2009). The effect of density-dependence can be evaluated by comparing changes in recruitment under
4959
models with and without density dependence and by comparing models with different levels of density
4960
dependence. The projections of Ianelli et al. (this volume) predicted future (log-transformed) recruitment
4961
directly from SST without taking account of density dependence or predation. This is comparable to
4962
recruitment predictions from model 1, which included density-dependence, if predation is fixed at the
4963
observed mean. Results (not shown) suggest relatively minor differences in recruitment projections under
4964
this level of density dependence. In contrast, a generalized Ricker model fit to the assessment output
4965
implies a higher level of density-dependence, resulting in strong compensatory increases in recruitment at
4966
low levels of SSB, thereby moderating the influence of declining SST on recruitment (model 2a in Figure
4967
10.8).
4968
Strong density-dependence in the generalized Ricker model may be a consequence of within-cohort
4969
competition, within-cohort cannibalism, and cannibalism by adult pollock. Cannibalism is a major source
4970
of pollock mortality in the Bering Sea (Dwyer et al., 1987; Livingston and Lang, 1996) and was evident
4971
in the effect of the predation term in model 1 and was a likely reason for the strong density-dependence in
217
4972
model 2. Predation by arrowtooth flounder, a major predator on juvenile pollock (Aydin et al., 2007), also
4973
had a strong impact on future pollock dynamics. Arrowtooth flounder predation is believed to play a
4974
major role in regulating recruitment of walleye pollock in the Gulf of Alaska (Bailey, 2000; Hollowed et
4975
al., 2000) and our results suggest that a continued increase in arrowtooth flounder biomass in the eastern
4976
Bering Sea could have a strong impact on pollock recruitment (Figure 10.8).
4977
Conclusions
4978
Forecasts of the recruitment response of walleye pollock to future climate variability can be used within a
4979
management strategy evaluation framework to assess alternative harvest scenarios and to provide long-
4980
term strategic advice to managers who are confronted with a rapidly changing environment (Ianelli et al.,
4981
this volume). This study only considered the effects of temperature and major groundfish predators on
4982
recruitment as a first step in assessing the effects of climate change on a major commercial species in the
4983
Bering Sea. Efforts are currently underway as part of the BEST/BSIERP research program
4984
(www.bsierp.nprb.org) to develop an end-to-end model of the eastern Bering Sea for predicting the
4985
responses of a multi-species fish community to future climate variability, but the model is not yet
4986
operational (K. Aydin, NOAA-AFSC, Seattle, pers. comm.). In the near term, studies such as this and
4987
Ianelli et al. (this volume) in the Bering Sea, and a similar study in the Baltic Sea (ICES 2009, Anna
4988
Gårdmark, Swedish Board of Fisheries, pers. comm.) are likely to provide the best assessment of climate
4989
effects on single species as a basis for providing relevant management advice. These studies could be
4990
extended to include other effects of climate change, such as effects on growth and distribution, but are
4991
inadequate for capturing multi-species interactions or adaptive responses to warming.
4992
Acknowledgements
4993
NOAA ERSST v3 data, air temperatures at St. Paul airport, and NCEP Reanalysis data were provided by
4994
the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA, from their website at
4995
http://www.esrl.noaa.gov/psd/. We thank our colleagues in the BEST-BSIERP Project, which is
4996
supported by NSF and NPRB, for many valuable discussions and access to data and manuscripts. In
4997
particular, we thank Phyllis Stabeno for providing the M2 data, Carol Ladd for providing the stratification
4998
index, Robert Lauth for providing bottom trawl survey data for Figure 10.1 and Ed Farley for providing
4999
BASIS survey data for age-0 distributions in Figure 10.1. This is BEST-BSIERP contribution number
5000
XXX.
5001
References
218
5002
Aydin, K., Gaichas, S., Ortiz, I., Kinzey, D., and Friday, N. 2007. A comparison of the Bering Sea, Gulf
5003
of Alaska, and Aleutian Islands large marine ecosystems through food web modeling NOAA
5004
Technical Memorandum, NMFS-AFSC-178: 298.
5005
Bacheler, N. M., Ciannelli, L., Bailey, K. M., and Duffy-Anderson, J. T. 2010. Spatial and temporal
5006
patterns of walleye pollock (Theragra chalcogramma) spawning in the eastern Bering Sea
5007
inferred from egg and larval distributions. Fisheries Oceanography, 19: 107-120.
5008
5009
Baier, C. T., and Napp, J. M. 2003. Climate-induced variability in Calanus marshallae populations.
Journal of Plankton Research, 25: 771-782.
5010
Bailey, K. M. 2000. Shifting control of recruitment of walleye pollock Theragra chalcogramma after a
5011
major climatic and ecosystem change. Marine Ecology Progress Series, 198: 215-224.
5012
Brander, K. 2010. Impacts of climate change on fisheries. Journal of Marine Systems, 79: 389-402.
5013
Ciannelli, L., Brodeur, R. D., and Buckley, T. W. 1998. Development and application of a bioenergetics
5014
5015
5016
model for juvenile walleye pollock. Journal of Fish Biology, 52: 879-898.
Ciannelli, L., Chan, K.-S., Bailey, K. M., and Stenseth, N. C. 2004. Nonadditive effects of the
environment on the survival of a large marine fish population. Ecology, 85: 3418–3427.
5017
Coyle, K. O., Eisner, B., Mueter, F. J., Pinchuk, A. I., Janout, M. A., Cieciel, K. D., Farley, E. V., et al. in
5018
press. Climate change in the southeastern Bering Sea: impacts on pollock stocks and implications
5019
for the Oscillating Control Hypothesis. Fisheries Oceanography.
5020
Coyle, K. O., Pinchuk, A. I., Eisner, L. B., and Napp, J. M. 2008. Zooplankton species composition,
5021
abundance and biomass on the eastern Bering Sea shelf during summer: The potential role of
5022
water-column stability and nutrients in structuring the zooplankton community. Deep Sea
5023
Research Part II: Topical Studies in Oceanography, 55: 1775-1791.
5024
Dwyer, D. A., Bailey, K. M., and Livingston, P. A. 1987. Feeding habits and daily ration of walleye
5025
pollock (Theragra chalcogramma) in the eastern Bering Sea, with special reference to
5026
cannibalism. Canadian Journal of Fisheries and Aquatic Sciences, 44: 1972-1984.
5027
Hollowed, A. B., Bond, N. A., Wilderbuer, T. K., Stockhausen, W. T., Amar, Z. T., Beamish, R. J.,
5028
Overland, J. E., et al. 2009. A framework for modelling fish and shellfish responses to future
5029
climate change. ICES Journal of Marine Science, 66: 1584-1594.
5030
5031
Hollowed, A. B., Hare, S. R., and Wooster, W. S. 2001. Pacific basin climate variability and patterns of
Northeast Pacific marine fish production. Progress in Oceanography, 49: 257-282.
219
5032
Hollowed, A. B., Ianelli, J. N., and Livingston, P. A. 2000. Including predation mortality in stock
5033
assessments: a case study for Gulf of Alaska walleye pollock. ICES Journal of Marine Science,
5034
57: 279-293.
5035
Hunt, G. L., Jr. , Coyle, K. O., Eisner, L., Farley, E. V., Heintz, R., Mueter, F., Napp, J. M., et al. this
5036
volume. Climate impacts on eastern Bering Sea food webs: A synthesis of new data and an
5037
assessment of the Oscillating Control Hypothesis. ICES Journal of Marine Science.
5038
Hunt, G. L., Jr., Stabeno, P., Walters, G., Sinclair, E., Brodeur, R. D., Napp, J. M., and Bond, N. A. 2002.
5039
Climate change and control of the southeastern Bering Sea pelagic ecosystem. Deep Sea Research
5040
Part II: Topical Studies in Oceanography, 49: 5821-5853.
5041
Hunt, G. L., Jr., Stabeno, P. J., Strom, S., and Napp, J. M. 2008. Patterns of spatial and temporal variation
5042
in the marine ecosystem of the southeastern Bering Sea, with special reference to the Pribilof
5043
Domain. Deep Sea Research Part II: Topical Studies in Oceanography, 55: 1919-1944.
5044
5045
5046
Hurvich, C. M., and Tsai, C.-L. 1989. Regression and time series model selection in small samples.
Biometrika, 76: 297-307.
Ianelli, J., Hollowed, A., Haynie, A., Mueter, F., and Bond, N. this volume. Evaluating management
5047
strategies for Eastern Bering Sea walleye pollock (Theragra chalcogramma) in a changing
5048
environment. ICES Journal of Marine Science.
5049
Ianelli, J. N., Barbeaux, S., Honkalehto, T., Kotwicki, S., Aydin, K., and Williamson, N. 2009.
5050
Assessment of the Walleye Pollock stock in the Eastern Bering Sea. In Stock assessment and
5051
fishery evaluation report for the groundfish resources of the Bering Sea/Aleutian Islands regions.
5052
North Pacific Fishery Management Council, 605 W. 4th Ave., Suite 306, Anchorage, AK 99501.
5053
ICES. 2009. Report of the ICES/HELCOM Working Group on Integrated Assessments of the Baltic Sea
5054
(WGIAB), 16–20 March 2009, Rostock, Germany. ICES CM 2009/BCC:02. 80pp.
5055
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., et al. 1996. The
5056
NCEP/NCAR 40-Year Reanalysis Project. Bulletin of the American Meteorological Society, 77:
5057
437-472.
5058
Kotwicki, S., Buckley, T. W., Honkalehto, T., and Walters, G. 2005. Variation in the distribution of
5059
walleye pollock (Theragra chalcogramma) with temperature and implications for seasonal
5060
migration. Fishery Bulletin, 103: 574-587.
220
5061
Livingston, P. A., and Lang, G. M. 1996. Interdecadal comparisons of walleye pollock, Theragra
5062
chalcogramma, cannibalism in the eastern Bering Sea. NOAA Technical Report NMFS, 126:
5063
115.
5064
Moss, J. H., Farley Jr., E. V., Feldman, A. M., and Ianelli, J. N. 2009. Spatial distribution, energetic
5065
status, and food habits of Eastern Bering Sea age-0 walleye pollock. Transactions of the
5066
American Fisheries Society, 138: 497-505.
5067
Mueter, F. J., Boldt, J., Megrey, B. A., and Peterman, R. M. 2007. Recruitment and survival of Northeast
5068
Pacific Ocean fish stocks: temporal trends, covariation, and regime shifts. Canadian Journal of
5069
Fisheries and Aquatic Sciences, 64: 911-927.
5070
Mueter, F. J., Broms, C., Drinkwater, K. F., Friedland, K. D., Hare, J. A., Hunt, G. L., Jr., Melle, W., et
5071
al. 2009. Ecosystem responses to recent oceanographic variability in high-latitude Northern
5072
Hemisphere ecosystems. Progress in Oceanography, 81: 93-110.
5073
Mueter, F. J., Ladd, C., Palmer, M. C., and Norcross, B. L. 2006. Bottom-up and top-down controls of
5074
walleye pollock (Theragra chalcogramma) on the eastern Bering Sea shelf. Progress in
5075
Oceanography, 68: 152-183.
5076
5077
Myers, R. A. 1998. When do environment-recruitment correlations work? Reviews in Fish Biology and
Fisheries, 8: 285-305.
5078
NPFMC. 2002. Fishery Management Plan for the Bering Sea/Aleutian Islands Groundfish. 383 pp.
5079
NPFMC. 2009. Stock assessment and fishery evaluation report for the groundfish resources of the Bering
5080
5081
5082
5083
Sea/Aleutian Islands regions.
Overland, J. E., and Wang, M. 2007. Future climate of the North Pacific Ocean. EOS Transactions
American Geophysical Union, 88: 178-182.
Overland, J. E., Wang, M., Bond, N. A., Walsh, J. E., Kattsov, V. M., and Chapman, W. L. in press.
5084
Considerations in the selection of global climate models for regional climate projections: The
5085
Arctic as a case study. Journal of Climate.
5086
5087
5088
Palmer, M. C. 2003. Environmental controls of fish growth in the southeast Bering Sea. In Institute of
Marine Science, p. 60. University of Alaska, Fairbanks.
Quinn, T. J., II, and Deriso, R. B. 1999. Quantitative fish dynamics, Oxford University Press, New York.
221
5089
Quinn, T. J., II, and Niebauer, H. J. 1995. Relation of eastern Bering Sea walleye pollock recruitment to
5090
environmental and oceanographic variables. In Climate change and northern fish populations, pp.
5091
497-507. Ed. by R. J. BEAMISH. National Research Council of Canada, Ottawa.
5092
5093
5094
5095
5096
Rijnsdorp, A. D., Peck, M. A., Engelhard, G. H., Möllmann, C., and Pinnegar, J. K. 2009. Resolving the
effect of climate change on fish populations. ICES Journal of Marine Science, 66: 1570-1583.
Simpson, J. H., Hughes, D. G., and Morris, N. C. G. 1977. The relation of seasonal stratification to tidal
mixing on the contrinental shelf. Deep Sea Research II, 24: 327-339.
Smith, T. M., Reynolds, R. W., Peterson, T. C., and Lawrimore, J. 2008. Improvements to NOAA's
5097
historical merged land-ocean surface temperature analysis (1880-2006). Journal of Climate, 21:
5098
2283-2296.
5099
Stabeno, P. J., Bond, N. A., Kachel, N. B., Salo, S. A., and Schumacher, J. D. 2001. On the temporal
5100
variability of the physical environment over the south-eastern Bering Sea. Fisheries
5101
Oceanography, 10: 81-98.
5102
5103
5104
Stabeno, P. J., Bond, N. A., and Salo, S. A. 2007. On the recent warming of the southeastern Bering Sea
shelf. Deep Sea Research II, 54: 2599–2618.
Wespestad, V. G., Fritz, L. W., Ingraham, W. J., and Megrey, B. A. 2000. On relationships between
5105
cannibalism, climate variability, physical transport, and recruitment success of Bering Sea
5106
walleye pollock (Theragra chalcogramma). ICES Journal of Marine Science, 57: 272-278.
5107
5108
5109
Wood, S. N. 2006. Generalized Additive Models: An introduction with R, Chapman & Hall/CRC, Boca
Raton, FL, USA.
Wyllie-Echeverria, T. 1996. The relationship between the distribution of one-year-old walleye pollock,
5110
Theragra chalcogramma, and sea-ice characteristics. NOAA Technical Report NMFS, 126: 47-
5111
56.
5112
5113
5114
222
5115
Table 10.1: Pairwise Pearson’s correlation coefficients among independent variables used in analysis of
5116
walleye pollock recruitment with significance levels indicated by * (p < 0.1) and ** (p<0.05), and
5117
loadings of variables on first four principal components. Loadings larger than 0.4 are highlighted.
ICE
ST
SST
Wind
Strat
Ice retreat (ICE)
Spring transition
(ST)
Summer SST (SST)
Summer wind
(Wind)
Stratification (Str)
Predation (P)
0.77**
- 0.37**
0.61**
- 0.19
- 0.20
- 0.03
- 0.33*
0.04
- 0.06
0.24
0.59*
-
*
0.32*
0.07
0.12
5118
5119
223
0.04
PC1
PC2
PC3
PC4
(41%
(24%
(18%
(14%
)
)
)
)
-0.48
0.27
0.35
0.44
-0.59
0.09
0.11
0.24
0.54
0.15
0.21
0.40
0.17
-0.65
0.12
0.61
0.33
0.66
0.19
0.17
0.04
-0.21
0.87
-0.43
5120
5121
Figure 10.1: Distribution of pollock at various life stages including main concentrations of eggs based on
5122
Bacheler et al. (2010), smoothed distribution of age-0 walleye pollock in the fall from Bering-
5123
Aleutian Salmon International Survey (BASIS) research program, smoothed distribution of age-1
5124
(80-199mm) and age3+ walleye pollock (>=300mm) during summer averaged over 1982-2009
5125
National Marine Fisheries Survey (NMFS) bottom trawl surveys.
5126
224
5127
5128
Figure 10.2: Flow chart of modeling and projection approach. Rectangles denote main state variables of
5129
projection model (2009-2050), beginning with 2009 numbers-at-age (Na); long-dashed ovals denote
5130
fixed input parameters (Ft is the annual fishing mortality based on the harvest control rule described
5131
in the text). Rounded rectangles denote inputs for the statistical regression model to generate future
5132
recruitment; short-dashed lines denote main stochastic elements generated by random draws from
5133
an ensemble of climate projections (Ei) or from a specified distribution (εt). Shaded boxes denote
5134
quantities tracked and summarized in results.
5135
225
5136
5137
Figure 10.3: Time series of walleye pollock recruitment (bars) and biomass (solid line) in the eastern
5138
Bering Sea, 1963-2009, based on most recent stock assessment estimates (Ianelli et al., 2009) and
5139
anomaly time series of environmental variables used in the analysis. Vertical line in top panel
5140
denotes 1976/77 climate regime shift. Thin line and thin dashed lines in top panel indicate
5141
estimated 80th percentile of recruitment from 1977-2007 with 95% coverage interval based on a
5142
quantile regression of log-transformed recruitment over time (t = 2.57, p = 0.015).
5143
226
5144
5145
Figure 10.4: Predicted stock-recruitment residuals for walleye pollock in the eastern Bering Sea as a
5146
function of the first and third principal components from a PCA of 6 environmental indicators. PC 1
5147
reflects average temperature conditions, while PC 3 reflects predation pressure. Other PCs were not
5148
significant. Note that temperature and ice conditions in the spring are confounded with summer
5149
temperature conditions and their apparent effects on recruitment cannot be separated statistically.
227
5150
5151
Figure 10.5: Estimated effects of summer (Jul-Sep) SST and combined predation by walleye pollock,
5152
arrowtooth flounder, and flathead sole on survival anomalies (Stock-recruitment residuals) of
5153
walleye pollock (top panels), and estimated effects of summer SST and arrowtooth flounder
5154
biomass on log-transformed recruitment (density-dependent effect of spawning biomass not
5155
shown).
228
5156
5157
5158
Figure 10.6: Ensemble predictions for late-summer (July-September) SST through 2050 based on three
IPCC climate scenarios.
5159
229
5160
5161
Figure 10.7: Distribution of simulated recruitment, spawning biomass, and annual catches for 2011-2050
5162
under a model with no temperature effect (i.e. random recruitment) and under a model
5163
incorporating sea-surface temperature and predation effects on recruitment (Model 1). Thick lines
5164
indicate means over 1000 simulated trajectories under random recruitment (solid line) and under
5165
Model 1 (dashed line). Polygons denote simulation envelopes such that 90% of simulated values
5166
fall within the envelope in any given year. Future SST trajectories for these simulations were drawn
5167
at random from all of the trajectories in Fig. 6.
5168
230
5169
5170
Figure 10.8: Boxplots of recruitment, spawner biomass, and catches as observed in the past (1979:2008,
5171
Obs), and as projected into the future (2041-2050 averages) using different models and climate
5172
scenarios. Boxes labeled ‘Random’ assume no climate effect on recruitment (random draws from
5173
observed values). Other boxes show distribution of projected values under three different models to
5174
simulate temperature and predation effects on recruitment (see text for model details) and under
5175
three different IPCC emissions scenarios. Dashed horizontal reference lines denote mean values
5176
from ‘Random’ scenario, black bars denote median, boxes include central 50%, and whiskers
5177
include central 90% of simulated values. Note that the implemented harvest control rule limits
5178
catches to 1.5 x 106 mt, hence median, upper quartile, and upper whisker all coincide in some cases.
231
5179
Chapter 11: Evaluating management strategies for eastern Bering Sea walleye pollock (Theragra
5180
chalcogramma) in a changing environment
5181
(Abstract only)
5182
5183
James N. Ianelli1, Anne B. Hollowed1, Alan C. Haynie1, Franz J. Mueter2, and Nicholas A. Bond3
5184
5185
1
5186
Oceanic and Atmospheric Administration, National Marine Fisheries Service, 7600 Sand Point Way NE,
5187
Seattle, WA 98115, USA
5188
2
5189
USA
5190
3
5191
WA 98195, USA
Resource Ecology and Fisheries Management Division, Alaska Fisheries Science Center, National
School of Fisheries and Ocean Sciences, 315 Lena Point, 17101 Pt. Lena Loop Rd, Juneau, AK 99801,
Joint Institute for the Study of Atmosphere and Ocean, University of Washington, Box 354925, Seattle,
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
Citation: Ianelli, J.N., Hollowed, A.B., Haynie, A.C., Mueter, F.J., and Bond, N.A. 2011. Evaluating
5204
management strategies for eastern Bering Sea walleye pollock (Theragra chalcogramma) in a changing
5205
environment. ICES Journal of Marine Science: Journal du Conseil 68(6): 1297-1304.
232
5206
Abstract
5207
The impacts of climate change on fish and fisheries is expected to increase the demand for more accurate
5208
stock projections and harvest strategies that are robust to shifting production regimes. To address these
5209
concerns, we evaluate the performance of fishery management control rules for eastern Bering Sea
5210
walleye pollock stock under climate change. We compared the status quo policy with six alternative
5211
management strategies under two types of recruitment pattern simulations: one that follows temperature-
5212
induced trends and the other that follows a stationary recruitment pattern similar to historical
5213
observations. A subset of 82 Intergovernmental Panel on Climate Change climate models provided
5214
temperature inputs from which an additional 100 stochastic simulated recruitments were generated to
5215
obtain the same overall recruitment variability as observed for the stationary recruitment simulations.
5216
Results indicate that status quo management with static reference points and current ecosystem
5217
considerations will result in much lower average catches and an increased likelihood of fishery closures,
5218
should reduced recruitment because of warming conditions hold. Alternative reference point calculations
5219
and control rules have similar performance under stationary recruitment relative to status quo, but may
5220
offer significant gains under the changing environmental conditions.
5221
233
5222
Conclusions
5223
Through retrospective analyses of physical and biological time series for the eastern Bering Sea
5224
ecosystem, as well as the broader Northeast Pacific, we documented covariation between large-scale
5225
climate drivers and biology, synchrony and asynchrony among different biological components, and
5226
linkages between important climate drivers and the productivity of individual populations. At an
5227
ecosystem level, our results highlight the regime-shift like behavior of Northeast Pacific physical and
5228
biological systems and provide evidence for a directional trend in the biological system that cannot be
5229
accounted for by natural climate variability alone, but is consistent with a climate change effect (Chapter
5230
3). At the level of fish, seabird and marine mammal communities on the eastern Bering Sea shelf we
5231
found moderate to strong covariation among the productivity of some components that reflects
5232
similarities and differences in the mechanisms driving productivity (Chapter 2). For individual stocks,
5233
such as walleye pollock and snow crab, we identified empirical relationships linking temperature
5234
variability to recruitment. In the case of walleye pollock, this relationship is supported by a mechanistic
5235
understanding of the importance of temperature in determining prey conditions for juvenile pollock and
5236
their effect on subsequent survival (Chapter 5) and therefore allows plausible forecasts of recruitment
5237
under different climate scenarios (Chapter 10, 11).
5238
Based on a broad suite of physical and biological time series that span the Northeast Pacific we confirmed
5239
previous findings (Hare and Mantua 2000) that variability in Northeast Pacific ecosystems, as quantified
5240
by major modes (principal components) of variability, is best described by abrupt shifts (Chapter 3). In
5241
addition to the well-known regime shifts of 1976/77 and 1988/89, we identified a possible shift in climate
5242
conditions in 2007/2008 (Fig. 4f.1); however, the ecological importance of this shift remains to be
5243
determined, but coincides with a shift from a period of warm, low-ice conditions in the Bering Sea to the
5244
the more recent cold period which has persisted through at least the winter of 2013/2014. The leading
5245
modes of climate variability, including the PDO (Mantua et al. 1997) and the NPGO (Di Lorenzo et al.
5246
2008) are strongly related to regional climate variability in the eastern Bering Sea and Gulf of Alaska, but
5247
superimposed on these natural modes of variability is a strong residual climate trend presumably linked to
5248
global warming. Variability in the biological system cannot be explained by the natural modes of climate
5249
variability alone, and the climate trend explains a significant portion of the overall variability similar in
5250
magnitude to that explained by natural modes of climate variability (Chapter 3). While the observed
5251
relationship between an aggregate biological measure and the climate trend largely reflects long-term
5252
trends in both biology and climate, it does suggest that marine populations in Alaska's ecosystems have
5253
been and will likely continue to be affected by warming, in spite of the recent cold period.
234
5254
Our analysis of covariation among key fish, seabird, and marine mammal populations clearly highlighted
5255
that marine populations do not vary independently from each other but are affected by environmental
5256
variability in either similar or opposite ways. We found strong coherence in recruitment between two
5257
closely related gadid species, walleye pollock and Pacific cod, a moderate level of synchrony among
5258
several flatfish species and an inverse correlation between recruitment variability of these two groups.
5259
Variability in seabird productivity was uncorrelated with that of fish and fur seals, which may reflect their
5260
reliance on different prey and on a different and more restricted feeding area in the vicinity of the Pribilof
5261
Islands (including parts of the adjacent slope and basin). In contrast, variability in fur seal productivity,
5262
after removing a long-term declining trend, appeared to correlate with the observed variability in fish
5263
populations.
5264
The strong and remarkable coherence between year-class strengths of Pacific cod and walleye pollock,
5265
and the inverse relationships between some flatfishes and gadids have been noted previously (Mueter et
5266
al. 2007) but remain largely unexplained. Although pollock and cod share a similar environment during
5267
larval and early juvenile stages, they are associated with different assemblages (Siddon et al. 2011) and
5268
their diets differ substantially (Strasburger et al. in press). Nevertheless, both may benefit from the
5269
presence of larger, lipid-rich zooplankton species in the summer. This appears to be critical for late larval
5270
and early juvenile (age-0) pollock, which need to acquire sufficient energy reserves to survive their first
5271
winter (Heintz et al. 2013, Hunt et al. 2011, Siddon et al. 2013a). While it is plausible that Pacific cod
5272
similarly rely on large zooplankton prey, this hypothesis has not been explored for Pacific cod and should
5273
be the focus of future efforts to understand Pacific cod dynamics.
5274
The inverse relationship between gadid and flatfish recruitment likely reflects different mechanisms
5275
affecting recruitment that may be linked to the same or strongly correlated drivers. Recruitment of several
5276
flatfish species on the eastern Bering Sea shelf has been linked to enhanced eastward advection onto the
5277
shelf (Wilderbuer et al. 2013, Wilderbuer et al. 2002), while pollock recruitment was higher during cooler
5278
temperature conditions on the shelf in recent years (Chapter 5, 10). This could induce a negative
5279
correlation between pollock and flatfish recruitment because onshelf advection tends to be enhanced in
5280
warm years and reduced in cold years or, more accurately, advection contributes to differences in
5281
temperature on the shelf because stronger onshelf and northward advection will tend to bring warmer
5282
water onto the shelf (Danielson et al. 2011, Stabeno et al. 2012). However, cross-shelf flows onto the
5283
shelf are enhanced primarily during years with stronger winds from the southeast (Danielson et al. 2012),
5284
which do not always correspond to warmer years (Danielson et al. 2011).
5285
Alternatively, the inverse relationship between gadid and flatfish recruitment may reflect alternative
5286
pathways for the flow of energy that may either favor the more pelagic late larvae and juveniles of
235
5287
pollock and cod or the more benthic larvae and juveniles of flatfish following settlement in the summer or
5288
early fall. This was one of the BSIERP hypotheses which we were unable to test due to the lack of
5289
information on when and where flatfish in the eastern Bering Sea settle to the bottom and limited
5290
knowledge of their prey preferences before or after settlement. The hypothesis states that warmer
5291
conditions during spring are associated with more of the production staying in the water column,
5292
benefitting pelagic feeders, while more of the production sinks to the bottom during cold conditions to
5293
become prey for benthic feeders. However, contrary to initial expectations, recent warm years from 2001
5294
to 2005 were associated with poor walleye pollock and Pacific cod recruitment, while recruitment of both
5295
species was much higher on average during the following cold period (Ianelli et al. 2012). Whether gadids
5296
or flatfishes in the eastern Bering Sea are likely to be winners or losers in a warming climate is of key
5297
interest to researchers and managers alike.
5298
Poor recruitment of walleye pollock in the early 2000s resulted in a substantial drop in walleye pollock
5299
biomass estimates and associated fishing quotas (Ianelli et al. 2008). Because of the ecological and
5300
economic importance of walleye pollock, there was considerable concern about stock declines and a need
5301
to better understand the causes of recruitment fluctuations in this important stock. It was hoped that the
5302
BEST/BSIERP field work, combined with retrospective analyses and modeling could shed some light on
5303
the mechanisms driving the large variability in recruitment, which has important consequences for the
5304
fishery and its management. Sampling all life stages of pollock, and that of their prey, competitors and
5305
predators over three field seasons provided an unprecedented opportunity to resolve the seasonal and
5306
spatial dynamics of how larval and juvenile pollock grow and survive during their first year. The
5307
BEST/BSIERP field years were all characterized by anomalously cold conditions, providing less contrast
5308
in environmental conditions than hoped for. However, data on zooplankton species composition and their
5309
spatial distribution, and data on the energetic condition of walleye pollock was available from other
5310
studies conducted during some of the earlier, warm years. These data provided an excellent opportunity to
5311
examine conditions during two contrasting periods: a prolonged warm period from 2001 to 2005 followed
5312
by a prolonged cold period since 2007 and continuing through the present.
5313
This comparison of a warm and cold period formed the basis for a synthesis of new and earlier field
5314
observations that were coupled with empirical analyses of historical data series (Chapter 5, 10), leading to
5315
a revision of the Oscillating Control Hypothesis (Hunt et al. 2011, Hunt and Stabeno 2002). The revised
5316
OCH holds that an early ice retreat or a lack of ice on the southeastern Bering Sea shelf during
5317
exceptionally warm years leads to high abundances of smaller zooplankton species and a relative lack of
5318
large zooplankton. In contrast, the abundance of large, lipid-rich zooplankton (such as Calanus
5319
marshallae and Euphauusiids) increased during subsequent cold years, providing better feeding
236
5320
conditions for larval and juvenile walleye pollock and resulting in a much higher energy density of age-0
5321
walleye pollock during late summer (Heintz et al. 2013), as well as reduced cannibalism / predation on
5322
age-0 pollock (Coyle et al. 2011, Hunt et al. 2011). As a result, the overwinter survival from age-0 to age-
5323
1 and hence recruitment to age-1 was enhanced following cold years with lipid-rich age-0 juveniles
5324
(Heintz et al. 2013).
5325
These findings contradicted earlier studies which had suggested that walleye pollock survival in the
5326
eastern Bering Sea is enhanced during warm years (Hunt and Stabeno 2002, Mueter et al. 2006, Quinn
5327
and Niebauer 1995, Wyllie-Echeverria and Wooster 1998). Our retrospective analyses reconciled these
5328
contradictory findings by identifying a dome-shaped relationship between the survival of walleye pollock
5329
from spawning to recruitment and sea-surface temperature conditions over the shelf (Chapter 5, 10).
5330
Earlier studies were conducted when the range of SST was largely limited to the ascending limb of this
5331
dome-shaped relationship, while recent warm years apparently exceed the optimum temperature range for
5332
juvenile pollock survival in this area. Our analyses showed the strongest relationship occurred with SST
5333
during late summer (July-September), consistent with field observations that identified the importance of
5334
late summer feeding conditions for successful juvenile growth and survival.
5335
The empirical relationship between SST and survival, along with its associated uncertainty, was used to
5336
generate projections of walleye pollock recruitment and abundance over long time scales (2010-2050)
5337
under different climate scenarios. These scenarios were based on climate projections from global
5338
circulation models, downscaled to provide possible trajectories of future SSTs over the eastern Bering Sea
5339
shelf (Chapter 10). Projections of the pollock stock suggest that the anticipated warming of waters on the
5340
Bering Sea shelf may result in substantial declines in walleye pollock recruitment, abundance and
5341
allowable catches in the long term. These findings initially led to considerable concern in the pollock fleet
5342
(Paul McGregor, pers. comm.), but these concerns soon receded after the return of cold conditions in the
5343
Bering Sea resulted in several strong year classes of walleye pollock.
5344
While the retrospective analyses focused to a large extent on walleye pollock, other species on the eastern
5345
Bering Sea shelf will be affected by a changing climate. Because of the strong synchrony between Pacific
5346
cod and walleye pollock, Pacific cod may also be expected to decrease under future warming. Similarly,
5347
snow crab recruitment is reduced following warm years hence the snow crab population can be expected
5348
to decline in a warming climate (Chapter 6). Other species such as flatfish may benefit from a decline in
5349
walleye pollock recruitment if the inverse relationship between gadid and flatfish recruitment continues to
5350
hold. Projections of northern rock sole recruitment based on downscaled wind patterns through 2050
5351
suggest a moderate future increase in recruitment because stronger on-shelf winds are expected to lead to
5352
favorable conditions for recruitment (Wilderbuer et al. 2013).
237
5353
The retrospective analyses contributed not only to a better understanding of the importance of seasonal
5354
availability of suitable prey to the success of walleye pollock, but also suggest that a spatial mismatch
5355
between larval pollock and lipid-rich prey may have contributed to poor recruitment in warm years
5356
(Chapter 9). Temperature changes have previously been shown to affect the distribution of many fish
5357
species on the Bering Sea shelf (Chapter 8, Kotwicki and Lauth 2013, Mueter and Litzow 2008, Mueter et
5358
al. 2011b, Spencer 2008), which has important consequences for their relative spatial distribution and for
5359
species interactions (Ciannelli et al. 2008, Hunsicker et al. 2013). Our results (Chapter 9) highlight the
5360
importance of the spatial dynamics of fish and their zooplankton prey in a changing climate, which we
5361
consider to be a fruitful area for further research to better understand the consequences of future warming
5362
for fish populations.
5363
5364
BSIERP and Bering Sea Project connections
5365
Our project benefitted from being part of the larger BEST/BSIERP project primarily through the frequent
5366
exchanges with other PIs during regular phone conferences with both the modeling group and all BSIERP
5367
PIs, during annual PI meetings, and at various scientific meetings with sessions focused on BEST and
5368
BSIERP research. For us, these benefits primarily arose from two aspects of the program. First, the
5369
frequent exchange of new findings and ideas was important in enhancing interdisciplinary and holistic
5370
(system-level) thinking that may not have happened, or certainly would have happened to a lesser extent,
5371
in the absence of an integrated program. Second, frequent communication about (1) key research
5372
questions, (2) approaches to answering these questions, and (3) newly emerging scientific ideas resulted
5373
in the spread of a few unifying ideas and principles that sharpened the focus of individual researchers and
5374
research groups. Examples for each of these include (1) questions about the role of microzooplankton as a
5375
trophic link or the role of prey patches for upper trophic levels, (2) the approach of comparing conditions
5376
during contrasting warm and cold periods, which encompassed physical conditions as well as all trophic
5377
levels, and (3) the importance of late summer prey conditions to allow juvenile pollock to aquire energy
5378
reserves for winter. Each of these and a number of other ideas resulted in investigators from multiple
5379
components to focus on the same problem or using the same approach to study different aspects of the
5380
system, resulting in a more holistic understanding of the system.
5381
For our project, frequent communication and integration among groups also changed the direction and the
5382
outputs of the retrospective analyses in important ways. There was an initial perception that the
5383
retrospective project would provide predictions of future variability in fish populations. While the project
5384
always intended to support such predictions through a better understanding of past variability, we had not
238
5385
intended to produce forecasts as part of the retrospective work. However, there was an increasing interest
5386
in and emphasis on such predictions, in part motivated by two International Symposia on the effects of
5387
climate change on fish, and we were well positioned to attempt to make such predictions for at least the
5388
walleye pollock stock with a relatively simple, single-species model. Therefore, as a result of these
5389
discussions and interactions among multiple PIs, the work plan for this project component was
5390
considerably modified to be able to make long-term forecasts for the walleye pollock population in the
5391
eastern Bering Sea. This effort was informed by a number of the other components and resulted in
5392
contributions by this project to four publications involving at least 14 BEST and BSIERP investigators
5393
that were not part of the plan (Coyle et al. 2011, Hunt et al. 2011, Ianelli et al. 2011, Mueter et al. 2011).
5394
Integration across components naturally occurred because both Mueter and Kruse were also Co-PIs on the
5395
Correlative Biomass Dynamics Model component (B75) and Mueter was a Co-PI on the Ichthyoplankton
5396
component (B53). This overlap facilitated at least two important linkages across these components. First,
5397
results from analysis of covariation (Chapter 2) was used as a basis for choosing which species and
5398
groups to include in the multi-species biomass dynamics model that was developed by Tadayasu
5399
Uchiyama under B75. Second, as Co-PI on B53, Mueter worked closely with and advised Ph.D. student
5400
Elizabeth Siddon and contributed both results and data series from the retrospective component to her
5401
dissertation, including Siddon et al (2013, Chapter 10), which built on and extended her work conducted
5402
as part of the Ichthyoplankton component (B53 report).
5403
Other important collaborations occurred as a result of discussions among PIs at the annual PI meetings.
5404
For example, discussions between PIs on the seabird components, the patch-dynamics group, and the
5405
current project that occurred in preparation for the covariation analysis (Chapter 2) directly led to
5406
collaborative work with Heather Renner on a retrospective analysis of murre and kittiwake diets at the
5407
Pribilof Islands (Chapter 7). This collaboration would almost certainly not have occurred without the
5408
integration that was facilitated by BSIERP. Similarly, retrospective analyses contributed to several
5409
publications (Coyle et al 2011, Hunt et al 2011, see Chapter 5) that directly resulted from discussions at
5410
annual PI meetings which may not have occurred without these meetings.
5411
In addition to regular BSIERP conference calls, both PIs also participated in many of the modeling calls
5412
and meetings. While this participation was primarily intended to better integrate the Correlative Biomass
5413
Dynamics component with other modeling work (see B68 report), the retrospective component was
5414
invited to participate in these meetings to discuss how empirically estimated parameters from
5415
retrospective analyses could be used to inform and contribute to the development of the FEAST model. In
5416
the end, because the desire to develop the FEAST model based on first principals rather than empirical
5417
relationships, and because most of the required empirical-based parameters were estimated by scientists at
239
5418
the Alaska Fishery Science Center in Seattle based on data housed at the Center, the retrospective
5419
analyses and results did not directly contribute to the development of the model.
5420
Instead, discussions with PIs from both the modeling group and the fish group ultimately led to a major
5421
shift in focus of the retrospective component. This shift consisted of using empirical relationships
5422
between climate variability and fish populations to produce alternative projections of the possible
5423
responses of these populations to future climate variability using simpler single-species models. After
5424
discussions with NPRB and in collaboration with several Co-PIs from other BEST and BSIERP
5425
components (Bond, Hollowed, Ianelli), we projected future walleye pollock recruitment and abundance
5426
based on what we learned about the survival of juvenile pollock from field observations, laboratory
5427
analyses, and retrospective analyses. While this was clearly outside the scope of the originally proposed
5428
retrospective analyses, and came at the expense of some of the other objectives, this was ultimately one of
5429
the most rewarding aspects of the project and the resulting publication has received a lot of positive
5430
attention as an example of how to produce reasonable scenarios of future variability in a fish population
5431
based on global climate model projections. While these projections were based on retrospective modeling
5432
and analyses alone, the underlying empirical relationship is supported by mechanisms that were
5433
elucidated by the field and lab components of the BEST/BSIERP program and related work.
5434
Another benefit of the integrated nature of the project is the opportunity for ecosystem-level syntheses
5435
across multiple components, but this is also one of the more challenging aspects of a program of this
5436
magnitude. The opportunity for synthesizing findings was somewhat limited as the annual meetings were
5437
largely needed just to communicate some of the main findings to all of the investigators. Moreover, a true
5438
synthesis is hardly possible and any attempt will necessarily leave out many important findings. We feel
5439
that the main scientific break-throughs are only slowly taking shape and it will be many years before the
5440
enduring scientific legacy of the program will more clearly come into focus.
5441
5442
Management or policy implications
5443
Our findings have a number of implications for the management of fish stocks in the Bering Sea. First,
5444
the recognition of trade-offs between, for example, the productivity of flatfish and the productivity of
5445
gadids (Chapter 2) highlight the need for multi-species considerations and ecosystem-level constraints on
5446
removals. Such constraints are currently being explored through multi-species models, such as the multi-
5447
species biomass dynamics model (B75) and the Management Strategy Evaluations (B73). While an
5448
overall cap on removals is already being implemented for federally managed fisheries in the eastern
5449
Bering Sea, a better understanding of the trade-offs among different species or groups of species allows a
240
5450
more explicit consideration of these trade-offs. Second, our work has contributed to a better
5451
understanding of walleye pollock dynamics in the eastern Bering Sea and has alerted the North Pacific
5452
Fishery Management Council (NPFMC) and the industry to potential declines in walleye pollock given
5453
projected warming trends. Moreover, the presumed dome-shaped relationship between walleye pollock
5454
recruitment and late summer ocean temperatures based on our retrospective analyses (Coyle et al 2011,
5455
Hunt et al 2011, Mueter et al 2011, Chapters 5 and 10) has been incorporated in a population dynamics
5456
model to project future pollock trajectories and has been used in management strategy evaluation to test
5457
different harvest control rules (Ianelli et al, 2011, see Chapter 11). Results suggest that alternative harvest
5458
control rules may perform better under these projected changes than the current harvest control rule used
5459
by the NPFMC. While no specific management actions have yet been taken in response to these findings,
5460
there has been an increased recognition of the need for incorporating both climate variability and long-
5461
term changes into the management process. Our projections have provided realistic scenarios upon which
5462
to base future management actions such as choosing policies and management strategies that are robust to
5463
the predicted changes. Third, we have provided several ecosystem indicators for inclusion in the
5464
Ecosystem Considerations chapter of the Annual Stock Assessment and Fishery Evaluation reports
5465
(Chapter 4). These are annually communicated to the NPFMC plan teams and the Statistical and
5466
Scientific Committee to help inform the NPFMC's ecosystem-based approach to fisheries management.
5467
5468
Publications
5469
Coyle, K.O., Eisner, L.B., Mueter, F.J., Pinchuk, A., Janout, M.A., Cieciel, K., Farley, E.V., and
5470
Andrews, A.G. 2011. Climate change in the southeastern Bering Sea: impacts on pollock stocks
5471
and implications for the Oscillating Control Hypothesis. Fisheries Oceanography 20(2): 139-156.
5472
Hunt, G.L., Coyle, K.O., Eisner, L.B., Farley, E.V., Heintz, R.A., Mueter, F., Napp, J.M., Overland, J.E.,
5473
Ressler, P.H., Salo, S., and Stabeno, P.J. 2011. Climate impacts on eastern Bering Sea foodwebs:
5474
a synthesis of new data and an assessment of the Oscillating Control Hypothesis. ICES Journal of
5475
Marine Science 68(6): 1230-1243.
5476
Ianelli, J.N., Hollowed, A.B., Haynie, A.C., Mueter, F.J., and Bond, N.A. 2011. Evaluating management
5477
strategies for eastern Bering Sea walleye pollock (Theragra chalcogramma) in a changing
5478
environment. ICES Journal of Marine Science 68(6): 1297-1304.
5479
5480
5481
5482
Litzow, M.A., and Mueter, F.J. 2013. Assessing the ecological importance of climate regime shifts: an
approach from the North Pacific Ocean. Progress in Oceanography.
Litzow, M.A., Mueter, F.J., and Hobday, A.J. 2012. Four decades of climate-biology covariation in
Alaskan and North Pacific ecosystems, North Pacific Research Board Final Report 1024.
241
5483
Litzow, M.A., Mueter, F.J., and Hobday, A.J. 2013. Reassessing regime shifts in the North Pacific:
5484
incremental climate change and commercial fishing are necessary for explaining decadal-scale
5485
biological variability. Global Change Biology. DOI: 10.1111/gcb.12373
5486
Marcello, L.A., Mueter, F.J., Dawe, E.G., and Moriyasu, M. 2012. Effects of temperature and gadid
5487
predation on snow crab recruitment: comparisons between the Bering Sea and Atlantic Canada.
5488
Marine Ecology Progress Series 469: 249-261.
5489
Mueter, F.J., Bond, N.A., Ianelli, J.N., and Hollowed, A.B. 2011a. Expected declines in recruitment of
5490
walleye pollock (Theragra chalcogramma) in the eastern Bering Sea under future climate change.
5491
ICES Journal of Marine Science 68(6): 1284-1296.
5492
5493
5494
Mueter, F.J., Dawe, E.G., and Pálsson, Ó. 2012. Effects of climate and predation on subarctic crustacean
populations. Marine Ecology Progress Series 469: 191-193.
Mueter, F.J., Renner, H.M., and Kruse, G.H. In Prep. Patterns of covariation among fish, seabirds, and
5495
marine mammals in the eastern Bering Sea reflect spatial scales of distribution and tradeoffs
5496
among species. Target journal: Deep Sea Research II.
5497
Mueter, F.J., Siddon, E.C., and Hunt Jr., G.L. 2011b. Climate change brings uncertain future for subarctic
5498
marine ecosystems and fisheries. In North by 2020: Perspectives on Alaska’s Changing Social-
5499
Ecological Systems. Edited by A.L. Lovecraft and H. Eicken. University of Alaska Press,
5500
Fairbanks, Alaska. pp. 329-357.
5501
Renner, H.M., Mueter, F., Drummond, B.A., Warzybok, J.A., and Sinclair, E.H. 2012. Patterns of change
5502
in diets of two piscivorous seabird species during 35 years in the Pribilof Islands. Deep Sea
5503
Research II 65-70: 273-291.
5504
Siddon, E.C., Heintz, R.A., and Mueter, F.J. 2013a. Conceptual model of energy allocation in walleye
5505
pollock (Theragra chalcogramma) from larvae to age-1 in the southeastern Bering Sea. Deep Sea
5506
Research II 94: 140-149.
5507
Siddon, E.C., Kristiansen, T., Mueter, F.J., Holsman, K.K., Heintz, R.A., and Farley, E.V. 2013b. Spatial
5508
Match-Mismatch between Juvenile Fish and Prey Provides a Mechanism for Recruitment
5509
Variability across Contrasting Climate Conditions in the Eastern Bering Sea. PLoS ONE 8(12):
5510
e84526.
5511
Sigler, M.F., Stabeno, P.J., Eisner, L.B., Napp, J.M., and Mueter, F.J. 2014. Spring and fall phytoplankton
5512
blooms in a productive subarctic ecosystem, the eastern Bering Sea, during 1995–2011. Deep Sea
5513
Research II. DOI:10.1016/j.dsr2.2013.12.007.
5514
Strasburger, W.W., Hillgruber, N., Pinchuk, A.I., and Mueter, F.J. in press. Feeding ecology of age-0
5515
walleye pollock (Theragra chalcogramma) and Pacific cod (Gadus macrocephalus) in the
5516
southeastern Bering Sea. Deep Sea Research II.
242
5517
5518
5519
5520
Zador, S. 2013. Ecosystem considerations 2013. North Pacific Fishery Management 644 Council, 605 W
4th Ave, Suite 306, Anchorage, AK 99501.
Zador, S., and Gaichas, S. 2012. Ecosystem considerations 2012. North Pacific Fishery Management 644
Council, 605 W 4th Ave, Suite 306, Anchorage, AK 99501.
5521
5522
Poster and oral presentations (chronological order)
5523
Kruse, G.H. (Invited Presentation). Potential impacts of climate change on red king crabs in the eastern
5524
Bering Sea at the "Effects of Climate Change on the World's Oceans" Symposium in Gijon, Spain.
5525
5526
5527
Mueter, F.J., K. Coyle. (Oral presentation) "From physics to humans: Climate effects on Bering Sea food
webs and fisheries. PICES 27th Annual Meeting, Dalian, China, October 30, 2008.
Kruse, G.H., J. Zheng and J.E. Overland (Oral presentation) “A scenario approach to forecast potential
5528
impacts of climate change on red king crabs in the eastern Bering Sea”, PICES 27th Annual Meeting,
5529
Dalian, China, October 24, 2008.
5530
Zheng, J., G.H. Kruse and M.S.M. Siddeek (Oral Presentation) “Could the collapse of the Bristol Bay red
5531
king crab stock in the early 1980s have been avoided?”, PICES 27th Annual Meeting, Dalian, China,
5532
October 29, 2008.
5533
Kruse, G.H., J. Zheng and J.E. Overland (Seminar) “A scenario approach to forecast potential impacts of
5534
climate change on red king crabs in the eastern Bering Sea", School of Fisheries and Ocean Sciences,
5535
University of Alaska Fairbanks, Departmental Seminar, November 14, 2008.
5536
Coyle, K.O. A.I. Pinchuk, L.B. Eisner, J.H. Moss, E. Farley, J.M. Napp, F.J. Mueter (Oral Presentation).
5537
The potential role of water-column temperature, stability and nutrients in structuring the zooplankton
5538
community and influencing pollock survival of the southeastern Bering Sea shelf. BASIS Annual
5539
meeting, Seattle, Washington, November 17-19, 2008.
5540
Coyle, K.O. A.I. Pinchuk, L.B. Eisner, J.M. Napp, F.J. Mueter, J.H. Moss, E. Farley (Poster).
5541
Zooplankton species composition on the southeastern Bering Sea shelf during summer: the potential
5542
role of temperature and water column stability in structuring the zooplankton community and
5543
influencing the survival of pollock. Alaska Marine Science Symposium, Anchorage, Alaska, January
5544
19-22, 2009.
5545
Mueter, F.J. (Invited Seminar). Climate effects on Bering Sea food webs and fisheries. Juneau Center,
5546
School of Fisheries and Ocean Sciences, University of Alaska Fairbanks. Juneau, Alaska, January 23,
5547
2009.
5548
5549
Mueter, F.J. (Invited seminar). Environmental and Ecological Indicators for the Eastern Bering Sea.
NOAA, NMFS, Ted Stevens Marine Research Institute. May 4, 2009.
243
5550
Mueter, F.J., Siddeek Shareef, Jie Zheng (Oral presentation). Gadid-crustacean interactions in the Eastern
5551
Bering Sea and Gulf of Alaska. Ecosystem Studies of Subarctic Seas (ESSAS) Annual Meeting. June
5552
18, 2009.
5553
Mueter, F.J. (Invited Seminar). Climate effects on Bering Sea food webs and fisheries. University of
5554
Alaska Fairbanks, Fairbanks, AK. September 30, 2009. A copy of the presentation was submitted
5555
with the semi-annual report on October 1, 2009.
5556
Mueter, F.J. and M.A. Litzow (Invited Presentation). The spatial footprint of biological re-organization in
5557
a demersal community. PICES 28th Annual Meeting Science Board Symposium, Cheju Island, Korea,
5558
October 26, 2009.
5559
Mueter, F.J., Gordon H. Kruse, Vernon Byrd, and Heather Renner (Poster). Covariation among major
5560
fish, seabird, and mammal populations in the Eastern Bering Sea. Alaska Marine Science
5561
Symposium, January 20, 2010, Anchorage, AK.
5562
M.A. Litzow and F.J. Mueter (Poster). Four decades of climate-biology covariation in the northeast
5563
Pacific: 98 updated ecosystem indicators, 1965-2006. Alaska Marine Science Symposium, January
5564
20, 2010, Anchorage, AK.
5565
Mueter, F.J., Carol Ladd, Phyllis Stabeno, Ron Heintz, Ken Coyle, Gordon H. Kruse (Oral presentation).
5566
Environmental controls of gadid year class strength in the eastern Bering Sea. Alaska Marine Science
5567
Symposium, January 21, 2010, Anchorage, AK.
5568
Mueter, F.J. (Invited presentation). Climate variability on the eastern Bering Sea shelf: Effects on the
5569
distribution and productivity of fish populations. Western Alaska Interdisciplinary Science
5570
Conference (WAISC). 24 March 2010, Unalaska, AK.
5571
5572
5573
Mueter, F.J. (Invited presentation). Climate variability in the eastern Bering Sea. Western Alaska
Interdisciplinary Science Conference (WAISC). 24 March 2010, Unalaska, AK
Mueter, F.J., Bond, N.A., Ianelli, J.N, and Hollowed, A.B. (Oral presentation). Future recruitment of
5574
Bering Sea walleye pollock: (1) retrospective patterns & uncertainty. International symposium on
5575
"Climate change effects on fish and fisheries: Forecasting impacts, assessing ecosystem responses,
5576
and evaluating management strategies", Sendai, Japan, April 25-29, 2010.
5577
Mueter, F.J., Bond, N.A., and Ianelli, J.N. (Invited Presentation). Long-term forecasts of walleye pollock
5578
dynamics in the eastern Bering Sea based on estimated responses of recruitment and growth to
5579
climate variability. PICES 2010 Annual Meeting, Tuesday, October 26, 2010.
5580
5581
5582
5583
Mueter, F.J. (Oral Presentation). Spatial dynamics of Bering Sea groundfish: Effects of temperature and
density. Alaska Chapter, American Fisheries Society, Annual Meeting. November 4, 2010.
Mueter, F.J. Effects of temperature and density on spatial dynamics of Bering Sea groundfishes. Alaska
Marine Science Symposium. January 20, 2011.
244
5584
5585
5586
5587
5588
Mueter, F.J. (Invited seminar) The Bering Sea ecosystem: From climate to plankton to fish. University of
Alaska Southeast. Biological Sciences seminar, Juneau, AK, April 13, 2011.
Mueter, F.J. (Keynote) Groundfish in Hot Water: Challenges facing Fish and Fisheries in Alaska. Alaska
Department of Fish & Game. Statewide groundfish meeting in Anchorage, AK April 27, 2011.
Mueter, F.J., Stepanenko, M.A., Smirnov, A.V., and Yamamura, O. (Invited). Comparing walleye pollock
5589
dynamics across the Bering Sea and adjacent areas. International Symposium on “Comparative
5590
studies of climate effects on polar and sub-polar ocean ecosystems: progress in observation and
5591
prediction” (ESSAS Open Science Meeting). Seattle, WA, May 23, 2011.
5592
Litzow, M.A., Mueter, F.J., Urban, D. (Invited). Can rising variance predict sudden shifts in populations
5593
and ecosystems? A test using Alaskan crustacean data. American Fisheries Society 141st Annual
5594
Meeting, Seattle, WA, September 4-8, 2011.
5595
Litzow, M.A., Mueter, F.J., Urban, D. Rising variance as an early indicator of fishery collapses in
5596
Alaskan crab populations. Alaska Marine Science Symposium. Anchorage, Alaska, January 18, 2012.
5597
Mueter, F.J. (Invited seminar). The (F)Utility of Forecasting Climate Change Impacts on Marine Fish and
5598
Shellfish: Approaches and Examples. Global Change Seminar Series. UAF, Fairbanks, Alaska,
5599
October 19, 2011.
5600
Mueter, F.J., Bohaboy, E.C., Bundy, A., Fu C., Hjermann, D.O., Link, J.S. Common patterns, common
5601
drivers: comparative analysis of aggregate surplus production across ecosystems. American Fisheries
5602
Society 141st Annual Meeting, Seattle, WA, September 4-8, 2011.
5603
Mueter, F.J. Spatial dynamics of fish communities in subarctic and arctic seas under a changing climate.
5604
PICES-ICES workshop on a "Global assessment of the implications of climate change on the spatial
5605
distribution of fish and fisheries", St. Petersburg, Russia, May 22, 2013.
5606
Mueter, F.J., Litzow, M.A., Lauth, R.L., Danielson, S.L., and Spencer, P.D. Spatial dynamics of
5607
groundfish: the roles of temperature, abundance and advection. Ecosystem Studies of the Subarctic
5608
Seas (ESSAS) Annual Science Meeting, Hakodate, Japan, January 9, 2013.
5609
Mueter, F.J. (Keynote). Ecosystems, complexity, and sustainability from global to regional to local scales.
5610
American Fisheries Society, Alaska Chapter, Annual Meeting. Kodiak, Alaska, October 24, 2012.
5611
Mueter, F.J., Litzow, M.A., Lauth, R.L., Danielson, S.L., and Spencer, P.D. The roles of temperature,
5612
abundance and advection in modifying the spatial dynamics of groundfish at the Subarctic-Arctic
5613
boundary in the eastern Bering Sea. PICES 2012 Annual Meeting, Hiroshima, Japan, October 12,
5614
2012.
5615
5616
Scientific Conferences
245
5617

5618
5619
September 2008: Kruse participated in the Annual Science Conference of the International Council
for the Exploration of the Sea (ICES).

October 2008: Kruse and Mueter participated in the Annual PICES meeting in Dalian. Kruse co-
5620
convened the Science Board Symposium on “Beyond observations to achieve understanding and
5621
forecasting in a changing North Pacific Ocean” and convened the Fisheries Contributed Paper
5622
Session.
5623

Dr. Kruse chaired the Steering Committee for the 25th Lowell Wakefield Symposium on “Biology
5624
and Management of Exploited Crab Populations under Climate Change” held in Anchorage, AK,
5625
during March 10-13, 2009. The meeting was attended by approximately 80 participants from 6
5626
countries. Dr. Kruse was also lead editor for the symposium proceedings.
5627

From March 24-27, 2010, Mueter participated in the Western Alaska Interdisciplinary Science
5628
Conference (WAISC) in Unalaska. In addition to giving two presentations related to the BSIERP
5629
project, he participated in discussions with the community on global warming and ocean acidification
5630
issues.
5631

Mueter, with Laura Richards (DFO, Canada), John Field (NOAA, USA), and Sanae Chiba (Japan)
5632
organized a session at the 2010 PICES Annual Meeting in Portland, Oregon, entitled "Observations of
5633
ecosystem mixing under climate change" that featured an invited presentation by BEST/BSIERP PI
5634
Lorenzo Cianelli.
5635

5636
5637
Kruse organized both a session and an international symposium that both highlighted BEST/BSIERP
research at the 2010 PICES Annual Meeting in Portland, Oregon.

In November, 2010, Kruse chaired a scientific session titled, “Dynamics of Marine Ecosystems”, at
5638
the Annual Meeting of the Alaska Chapter, American Fisheries Society, in Juneau Alaska. This
5639
session included BSIERP-related talks, including a BEST/BSIERP overview talk by Mike Sigler.
5640

In November 2010, Gordon Kruse organized and chaired the Steering Committee for the 26th Lowell
5641
Wakefield Symposium titled, “Ecosystems 2010: Global Progress on Ecosystem-based Fisheries
5642
Management.” The symposium attracted 108 participants from 19 countries. Talks included multiple
5643
BEST/BSIERP presentations, including those delivered by Mike Sigler, Ivonne Ortiz, Kerim Aydin
5644
and others.
5645

In May 2011 both Mueter and Kruse participated in the ESSAS Open Science Meeting in Seattle,
5646
where Franz Mueter gave an invited BSIERP presentation on walleye pollock dynamics during the
5647
Bering Sea session and Gordon Kruse delivered a presentation on red king crab dynamics in the
5648
session on gadid-crustacean interactions.
246
5649

Mueter attended the Annual Meeting of the American Fisheries Society in Seattle in September 2011
5650
and presented in a session on increased variability in fish populations and in a session on surplus-
5651
production models. Both drew on some retrospective data compiled as part of BSIERP.
5652

Mueter participated in the 2nd International Symposium on the effects of climate change on the
5653
World's Oceans in Yeosu, South Korea, participated in a workshop and co-authored three papers
5654
partially supported by this project, presented by George Hunt, Paul Spencer, and Ting-Chun Kuo.
5655

Both Mueter and Kruse participated in the PICES Annual Meeting in Hiroshima, Japan, in October
5656
2012, participated in and helped organize workshops and scientific sessions, and co-authored several
5657
papers presented at the meeting that were partially supported by this project.
5658

5659
Both Mueter and Kruse participated in the PICES Annual Meeting in Nanaimo, BC, in October 2013,
participating in workshops, committee meetings, and scientific sessions.
5660
5661
Community Meetings
5662

August 12/13, 2008: Franz Mueter provided testimony on the impacts of climate change on fish and
5663
fisheries to a panel convened by the Aspen Institute in Fairbanks. A copy of the presentation was
5664
provided to Nora Deans, NPRB.
5665

In August 2011, Gordon Kruse testified in Dutch Harbor at a hearing on the Arctic by the Alaska
5666
State Legislature’s Joint Alaska Northern Waters Task Force. He summarized fisheries research being
5667
conducted by UAF in the Bering, Chukchi and Beaufort Seas. The presentation included an overview
5668
of the BSIERP program in general, with more detail on those BSIERP projects involving UAF
5669
scientists, including the current project.
5670

Gordon Kruse participated in the Alaska Young Fishermen’s Summit, sponsored by the University of
5671
Alaska Marine Advisory Program, and held in Juneau, Alaska on February 13, 2012. Gordon spoke to
5672
51 young Alaskan fishers about the role of science in state and federal fisheries management of
5673
groundfish and other species groups.
5674
5675
Workshops
5676

September 13-16, 2008: Mueter participated in the ESSAS (Ecosystem Studies of the Sub Arctic
5677
Seas) Annual Meeting in Halifax, Nova Scotia, including a workshop on climate variability in
5678
subarctic seas.
5679
5680

August 2008: Several earlier workshops sponsored by PICES and NPRB resulted in PICES Scientific
Report #34 “Forecasting Climate Impacts on Future Production of Commercially Exploited Fish and
247
5681
Shellfish”. Kruse prepared a section for the report on status of knowledge and proposed mechanisms
5682
linking climate change to the production of red king crab, Tanner crab and snow crab.
5683

On June 18, 2009, Franz Mueter co-chaired (with Earl Dawe, DFO, St. Johns, Newfoundland) a
5684
workshop on gadid-crustacean interactions in subarctic ecosystems at the ESSAS Annual Meeting in
5685
Seattle. He presented an overview (with Siddeek Shareef and Jie Zheng) of gadid and crustacean
5686
fisheries and dynamics in the Gulf of Alaska and eastern Bering Sea.
5687

November 16-18, 2009: Mueter gave an invited presentation by videoconference to the Alaska
5688
Community-Based Climate Change Adaptation Outreach Program Development Workshop on
5689
"Climate change impacts on fisheries".
5690

April 24, 2010. Mueter participated in a workshop on "Networking across global marine "hotspots",
5691
held in conjunction with the international symposium on "Climate change effects on fish and
5692
fisheries: Forecasting impacts, assessing ecosystem responses, and evaluating management
5693
strategies." in Sendai, Japan. As part of the workshop he gave an invited presentation on "Biological
5694
responses to recent climate variability on the eastern Bering Sea shelf".
5695

May 10-14, 2010: Mueter participated in an International Stock Production Modeling Workshop at
5696
Woods Hole, MA, contributing biological and physical time series of variability in the Bering Sea and
5697
Gulf of Alaska for analysis at the workshop.
5698

August 30 – September 1, 2010. Mueter and M.S. student Laurinda Marcello participated in the
5699
Ecosystem Studies of the Subarctic Seas (ESSAS) Annual Meeting in Reykjavik, Iceland, which
5700
featured a worshop on effects of climate variability in subarctic ecosystems and a workshop on gadid-
5701
crustacean interactions in subarctic seas. Mueter gave a invited presentation (co-authored with Mike
5702
Litzow) on linking climate and fish in the Northeast Pacific as part of the first workshop and Marcello
5703
presented results from her retrospective work on snow crab recruitment in the Eastern Bering Sea.
5704

In February 2011, Mueter and Kruse participated in a workshop on identifying approaches for
5705
achieving priority research for eastern Bering Sea snow crab biology and fishery management in
5706
Seattle. Kruse was one of the conveners of the workshop.
5707

On April 7-8, 2011, Franz Mueter participated in a workshop on stock-specific indicators at the
5708
Alaska Fisheries Science Center in Seattle and gave a presentation on forecasting pollock recruitment
5709
and growth.
5710

Mueter helped organize a half-day workshop and a scientific session on gadid-crustacean interactions
5711
at the ESSAS Open Science Meeting in May 2011. MS student Laurinda Marcello presented her work
5712
on snow crab (partially supported by BSIERP) at the meeting and submitted an associated
5713
manuscript.
248
5714

5715
On October 27-28, 2011, Franz Mueter participated via WebEx in a workshop on Management
Strategy Evaluations held at the Alaska Fisheries Science Center in Seattle.
5716

On February 6-9, 2012, Franz Mueter participated in a BEST Synthesis workshop in Bermuda.
5717

On April 4-5, 2012, Franz Mueter participated in a workshop on Workshop on Assessment and
5718
Management Issues Related to Recruitment at the Alaska Fisheries Science Center in Seattle and gave
5719
a presentation on environmental forcing of recruitment in the Bering Sea and Gulf of Alaska and its
5720
use in stock assessments and stock projections.
5721

In May 2012, Mueter participated in a full-day workshop on "Climate change and range shifts in the
5722
ocean" at the the 2nd International Symposium on the effects of climate change on the World's Oceans
5723
in Yeosu, South Korea.
5724

Mueter participated in a full-day workshop on interactions between the subarctic and Arctic at the
5725
PICES Annual Meeting in Hiroshima, Japan, in October 2012 and presented a paper on the spatial
5726
dynamics of fish stocks in the eastern Bering Sea.
5727

5728
On February 26-28, 2013, Franz Mueter participated in a BEST Synthesis workshop in Friday
Harbor, WA.
5729
5730
Outreach
5731

On September 16, 2008, Mueter participated in a meeting of the Scientific Steering Committee of
5732
ESSAS (Ecosystem Studies of the Sub Arctic Seas) in Halifax, Nova Scotia to provide information on
5733
BSIERP. The Steering Committee considered and approved an application from BSIERP to be
5734
formally affiliated with ESSAS. Since then, either Mike Sigler or Franz Mueter , who has served as
5735
co-chair of ESSAS since 2011, have reported to the ESSAS Scientific Steering Committee on the
5736
BSIERP program.
5737

On May 29, 2012, Mueter presented a public outreach seminar to the community of Nome as part of
5738
the " Strait Science Series" on "Limits to the northward movement of fishes in the Eastern Bering
5739
Sea", Nome, AK.
5740

On September 9, 2009, Gordon Kruse was invited to give an invited presentation on Climate Change
5741
and Marine Protected Areas: A Fisheries Perspective from Alaska to the Marine Protected Areas
5742
Federal Advisory Committee (FAC). Kruse also served on a panel of experts to field questions from
5743
the MPA FAC for two hours. This was reported on the evening news on KTUU (Anchorage) on
5744
September 9, 2009. Kruse reported on climate change effects on groundfish, crabs, herring, and other
5745
marine species and their implications to the design of marine protected areas.
249
5746

Gordon Kruse participated as a member of a panel to address the question: What will our fisheries
5747
and oceans look like in 20 years? The panel was convened during the Alaska Young Fishermen’s
5748
Summit hosted by the Alaska Marine Advisory Program in Anchorage on December 8, 2009.
5749

Franz Mueter had numerous e-mail exchanges with Wendee Holtcamp, a freelance reporter who
5750
participated in the 2010 BSIERP cruise and produced an article for BioScience (February 2011 / Vol.
5751
61 No. 2) on the BSIERP program. The exchanges focused on our revised understanding of the
5752
Oscillating Control Hypothesis.
5753

Gordon Kruse published two articles summarizing the 26th Lowell Wakefield Symposium titled,
5754
“Ecosystems 2010: Global Progress on Ecosystem-based Fisheries Management”; one was published
5755
in PICES Press and the other was published in the ICES 2010 Symposium Report. Copies of both
5756
articles are available for download from the Ecosystems 2010 Symposium website:
5757
http://seagrant.uaf.edu/conferences/2010/wakefield-ecosystems/index.php.
5758

On October 19, 2011, Gordon Kruse presented an overview of integrated ecosystem research in the
5759
BEST and BSIERP Programs to the North Pacific Marine Research Organization (PICES) Fishery
5760
Science Committee on October 19, 2011 in Khabarovsk, Russia.
5761

Franz Mueter presented a public seminar and other outreach activities in Nome on May 29, 2012. The
5762
seminar included an overview of some of the work conducted by this and the related retrospective
5763
project (B68) with a focus on the spatial distribution of subarctic species in the southeastern Bering
5764
Sea and their potential to expand northward into the northern Bering Sea.
5765

5766
Franz Mueter gave a public seminar in Juneau on November 30, 2012, as part of the Fisheries
Division seminar series, featuring results from BSIERP in addition to recent Arctic work.
5767
5768
Press Articles
5769

May 2008: An article on climate change and Alaska’s fisheries, written by Kruse, appeared in the
5770
Periodical "Alaska Seas and Coasts", published by Alaska Sea Grant and the Marine Advisory
5771
Program of the University of Alaska Fairbanks.
5772

5773
5774
July 13, 2008: Kruse was interviewed and quoted in a newspaper article in the Fairbanks Daily News
Miner on the effects of climate change on marine ecosystems of Alaska.

October 2008: Mueter was interviewed about responses of fish populations in the Bering Sea to
5775
climate variability by Ken Weiss from the LA Times for an article on climate warming and marine
5776
fishes Los Angeles Times, October 19, 2008: "Migrating Alaskan pollock are creating the potential
5777
for a new dispute with Russia"
250
5778

February 4, 2009. Dr. Kruse was interviewed by reporter Tom Kazzia for an article on the effects of
5779
climate change on Tanner crabs and other species in Alaska. The article “Crab comeback in
5780
Kachemak Bay” appeared in Anchorage Daily News on 2/21/2009. Kruse was quoted twice in the
5781
article.
5782

November 23, 2009: Franz Mueter was interviewed by Lu Snyder for an article on the effects of
5783
climate change on fishes that was published in the December 2009 edition of FISHRAP, the
5784
newsletter of the Northern Southeast Regional Aquaculture Association (Vol. 27, no. 2, p. 1).
5785

5786
5787
September 2, 2011: Mueter gave a phone interview on changes in arrowtooth flounder abundance to
Craig Medred of the Alaska Dispatch.

In May 2011 Gordon Kruse was interviewed for a UAF Aurora Magazine article that was published
5788
in fall 2011. He was heavily quoted concerning the collapse of many of Alaska’s king crab stocks,
5789
their failure to recover, and broader ecosystem changes in the GOA and EBS (see
5790
http://www.uaf.edu/aurora/).
5791
5792
Radio/Broadcast Interviews
5793

On August 13, 2009, both Kruse and Mueter were interviewed by Marc Kagan, video director and
5794
producer for NOAA, for a film on climate change and its effects on fish, crabs and other species in
5795
the marine ecosystems of Alaska. Gordon spoke about work related to this project, as well as other
5796
NPFB-funded research projects, such as Pacific herring. Franz spoke about changes in distribution of
5797
fish and shellfish in the eastern Bering Sea and other climate effects on the Bering Sea ecosystem,
5798
based in part on results from the retrospective analyses.
5799

On March 27, 2010, Mueter was interviewed by Anne Hillman from KUCB, Dutch Harbor, on the
5800
effects of climate variability on walleye pollock. The interview aired on the local public radio station
5801
and is available online.
5802

In August 2011 in a radio report titled, “New clues surface as to why Bering Sea red (king crabs)
5803
might be beating out blue ones”, Gordon Kruse was interviewed by a Dillingham reporter about
5804
changes in king crab populations in the Bering Sea and Gulf of Alaska, including potential
5805
competitive interactions among red and blue king crabs. The report was aired on August 10, 2011 by
5806
KDLG radio in Dillingham, Alaska. At the same time, Gordon was also interviewed by KDLG about
5807
his work on changes in size limit for EBS Tanner crab associated with reduced size of maturity for
5808
subsequent airing.
5809
5810

On September 9, 2012, Gordon Kruse was interviewed on camera for two hours by Luke GriswoldTergis for a future PBS documentary on fisheries management in Alaska. It is a story about the
251
5811
evolution of fisheries management and current issues in Alaska. The interview included discussion of
5812
some of the issues associated with the Bering Sea groundfish fishery and research resulting from the
5813
BSIERP project.
5814

On January 24, 2013, Franz Mueter was interviewed by Jaqueline Estes (APRN) on the potential
5815
effects of climate change on fish communities in the Arctic and Subarctic. The story aired on APRN
5816
on the following days.
5817

On January 24, 2013, Franz Mueter was also interviewed by Lauren Rosenthal from KUCB (Dutch
5818
Harbor), which led to an online article and a story on KUCB, largely focusing on the Arctic
5819
(http://www.alaskapublic.org/2013/01/24/chukchi-trawl-survey-sheds-light-on-unexplored-waters/) .
5820
5821
Acknowledgements
5822
We thank the numerous researchers who, over many decades, collected and processed fish and
5823
environmental data series that formed the heart of our analyses. We thank all of the BEST and BSIERP
5824
investigators for the open exchange of information and ideas, in particular George Hunt, Ken Coyle, Nick
5825
Bond, and Seth Danielson. We thank Anne Hollowed for her encouragement and for challenging us to
5826
attempt long-term projections. We thank the students at the UAF Juneau Center for many stimulating
5827
discussions that challenged our views and often led to new insights. We thank Mike Litzow for bringing
5828
his analytical skills and sharp thinking to a very fruitful collaboration. Finally, Franz Mueter thanks Earl
5829
Dawe and other members of ESSAS for encouraging us to apply lessons from other subarctic ecosystems
5830
to the Bering Sea.
5831
5832
Literature cited
5833
Ciannelli, L., Fauchald, P., Chan, K.S., Agostini, V.N., and Dingsør, G.E. 2008. Spatial fisheries ecology:
5834
Recent progress and future prospects. Journal of Marine Systems 71: 223–236.
5835
Coyle, K.O., Eisner, L.B., Mueter, F.J., Pinchuk, A., Janout, M.A., Cieciel, K., Farley, E.V., and
5836
Andrews, A.G. 2011. Climate change in the southeastern Bering Sea: impacts on pollock stocks
5837
and implications for the Oscillating Control Hypothesis. Fisheries Oceanography 20(2): 139-156.
5838
Danielson, S., Eisner, L., Weingartner, T., and Aagaard, K. 2011. Thermal and haline variability over the
5839
central Bering Sea shelf: Seasonal and interannual perspectives. Continental Shelf Research
5840
31(6): 539-554.
5841
Danielson, S., Hedstrom, K., Aagaard, K., Weingartner, T., and Curchitser, E. 2012. Wind-induced
5842
reorganization of the Bering shelf circulation. Geophysical Research Letters 39(8).
252
5843
Di Lorenzo, E., Schneider, N., Cobb, K.M., Franks, P.J.S., Chhak, K., Miller, A.J., McWilliams, J.C.,
5844
Bograd, S.J., Arango, H., Curchitser, E., Powell, T.M., and Rivière, P. 2008. North Pacific Gyre
5845
Oscillation links ocean climate and ecosystem change. Geophys. Res. Lett. 35(8): L08607.
5846
Drinkwater, K.F., Beaugrand, G., Kaeriyama, M., Kim, S., Ottersen, G., Perry, R.I., Pörtner, H.-O.,
5847
Polovina, J.J., and Takasuka, A. 2010. On the processes linking climate to ecosystem changes.
5848
Journal of Marine Systems 79(3-4): 374-388.
5849
5850
5851
5852
Essington, T.E., and Punt, A.E. 2011. Implementing Ecosystem-Based Fisheries Management: Advances,
Challenges and Emerging Tools. Fish and Fisheries 12(2): 123-124.
Francis, R.C., Hare, S.R., Hollowed, A.B., and Wooster, W.S. 1998. Effect of interdecadal climate
variability on the oceanic ecosystems of the NE Pacific. Fisheries Oceanography 7(1): 1-21.
5853
Hare, S.R., and Francis, R.C. 1995. Climate change and salmon production in the Northeast Pacific
5854
Ocean. In Climate change and northern fish populations. Edited by R.J. Beamish. National
5855
Research Council of Canada, Ottawa. pp. 357-372.
5856
5857
Hare, S.R., and Mantua, N.J. 2000. Empirical evidence for North Pacific regime shifts in 1977 and 1989.
Progress in Oceanography 47: 103-145.
5858
Heintz, R.A., Siddon, E.C., Farley, E.V., and Napp, J.M. 2013. Correlation between recruitment and fall
5859
condition of age-0 pollock (Theragra chalcogramma) from the eastern Bering Sea under varying
5860
climate conditions. Deep Sea Research Part II: Topical Studies in Oceanography 94: 150-156.
5861
5862
5863
Hollowed, A.B., and Wooster, W.S. 1992. Variability of winter ocean conditions and strong year classes
of northeast Pacific groundfish. ICES Marine Science Symposium 195: 433-444.
Hollowed, A.B., and Wooster, W.S. 1995. Decadal-scale variations in the eastern subarctic Pacific: II.
5864
Response of Northeast Pacific fish stocks. In Climate change and northern fish populations.
5865
Edited by R.J. Beamish. National Research Council of Canada, Ottawa. pp. 373-385.
5866
Hunsicker, M.E., Ciannelli, L., Bailey, K.M., Zador, S., and Stige, L.C. 2013. Climate and demography
5867
dictate the strength of predator-prey overlap in a subarctic marine ecosystem. PLoS ONE 8(6):
5868
e66025.
5869
Hunt, G.L., Coyle, K.O., Eisner, L.B., Farley, E.V., Heintz, R.A., Mueter, F., Napp, J.M., Overland, J.E.,
5870
Ressler, P.H., Salo, S., and Stabeno, P.J. 2011. Climate impacts on eastern Bering Sea foodwebs:
5871
a synthesis of new data and an assessment of the Oscillating Control Hypothesis. ICES Journal of
5872
Marine Science: Journal du Conseil 68(6): 1230-1243.
5873
Hunt, G.L., Jr, Allen, B.M., Angliss, R.P., Baker, T.N., Bond, N., Buck, G., Byrd, G.V., Coyle, K.O.,
5874
Devol, A., Eggers, D.M., Eisner, L., Feely, R., Fitzgerald, S., Frits, L.W., Gritsay, E.V., Ladd, C.,
5875
Lewis, W., Mathis, J., Mordy, C.W., Mueter, F.J., Napp, J., Sherr, E., Shull, D., Stabeno, P.,
5876
Stepanenko, M.A., Strom, S., and Whitledge, T.E. 2010. Status and trends of the Bering Sea
253
5877
region, 2003-2008. In Marine ecosystems of the North Pacific Ocean, 2003-2008. Edited by S.M.
5878
McKinnell and M. Dagg. pp. 196-267.
5879
5880
5881
Hunt, G.L., Jr., and Stabeno, P.J. 2002. Climate change and the control of energy flow in the southeastern
Bering Sea. Progress in Oceanography 55: 5-22.
Ianelli, J.N., Barbeaux, S., Honkalehto, T., Kotwicki, S., Aydin, K., and Williamson, N. 2008.
5882
Assessment of the Walleye Pollock stock in the Eastern Bering Sea. In Stock assessment and
5883
fishery evaluation report for the groundfish resources of the Bering Sea/Aleutian Islands regions.
5884
North Pacific Fishery Management Council, 605 W. 4th Ave., Suite 306, Anchorage, AK 99501.
5885
Ianelli, J.N., Hollowed, A.B., Haynie, A.C., Mueter, F.J., and Bond, N.A. 2011. Evaluating management
5886
strategies for eastern Bering Sea walleye pollock (Theragra chalcogramma) in a changing
5887
environment. ICES Journal of Marine Science: Journal du Conseil 68(6): 1297-1304.
5888
Ianelli, J.N., Honkalehto, T., Barbeaux, S., Kotwicki, S., Aydin, K., and Williamson, N. 2012.
5889
Assessment of the Walleye Pollock stock in the Eastern Bering Sea. In Stock assessment and
5890
fishery evaluation report for the groundfish resources of the Bering Sea/Aleutian Islands regions.
5891
North Pacific Fishery Management Council, 605 W. 4th Ave., Suite 306, Anchorage, AK 99501.
5892
pp. 51-156.
5893
Kotwicki, S., and Lauth, R.R. 2013. Detecting temporal trends and environmentally-driven changes in the
5894
spatial distribution of bottom fishes and crabs on the eastern Bering Sea shelf. Deep Sea Research
5895
Part II: Topical Studies in Oceanography 94: 231-243.
5896
5897
5898
5899
5900
Litzow, M.A., and Mueter, F.J. 2013. Assessing the ecological importance of climate regime shifts: an
approach from the North Pacific Ocean. Progress in Oceanography.
Litzow, M.A., Mueter, F.J., and Hobday, A.J. 2012. Four decades of climate-biology covariation in
Alaskan and North Pacific ecosystems, North Pacific Research Board Final Report 1024.
Litzow, M.A., Mueter, F.J., and Hobday, A.J. 2013. Reassessing regime shifts in the North Pacific:
5901
incremental climate change and commercial fishing are necessary for explaining decadal-scale
5902
biological variability. Global Change Biology.
5903
5904
5905
Lovecraft, A.L., and Eicken, H. 2011. North by 2020: Perspectives on Alaska’s Changing SocialEcological Systems. University of Alaska Press, Fairbanks, Alaska.
Mantua, N., Hare, S., Zhang, Y., Wallace, J., and Francis, R. 1997. A Pacific interdecadal climate
5906
oscillation with impacts on salmon production. Bulletin of the American Meteorological Society
5907
78(6): 1069-1080.
5908
Marcello, L.A., Mueter, F.J., Dawe, E.G., and Moriyasu, M. 2012. Effects of temperature and gadid
5909
predation on snow crab recruitment: comparisons between the Bering Sea and Atlantic Canada.
5910
Marine Ecology Progress Series 469: 249-261.
254
5911
Mueter, F.-J., Peterman, R.M., and Pyper, B.J. 2002. Opposite effects of ocean temperature on survival
5912
rates of 120 stocks of Pacific salmon (Oncorhynchus spp.) in northern and southern areas.
5913
Canadian Journal of Fisheries and Aquatic Sciences 59(3): 456-463.
5914
Mueter, F.J., Boldt, J., Megrey, B.A., and Peterman, R.M. 2007. Recruitment and survival of Northeast
5915
Pacific Ocean fish stocks: temporal trends, covariation, and regime shifts. Canadian Journal of
5916
Fisheries and Aquatic Sciences 64(6): 911-927.
5917
Mueter, F.J., Bond, N.A., Ianelli, J.N., and Hollowed, A.B. 2011a. Expected declines in recruitment of
5918
walleye pollock (Theragra chalcogramma) in the eastern Bering Sea under future climate change.
5919
ICES Journal of Marine Science: Journal du Conseil 68(6): 1284-1296.
5920
5921
5922
Mueter, F.J., Dawe, E.G., and Pálsson, Ó. 2012. Effects of climate and predation on subarctic crustacean
populations. Marine Ecology Progress Series 469: 191-193.
Mueter, F.J., Ladd, C., Palmer, M.C., and Norcross, B.L. 2006. Bottom-up and top-down controls of
5923
walleye pollock (Theragra chalcogramma) on the eastern Bering Sea shelf. Progress in
5924
Oceanography 68(2-4): 152-183.
5925
5926
5927
5928
5929
Mueter, F.J., and Litzow, M.A. 2008. Sea ice retreat alters the biogeography of the Bering Sea continental
shelf. Ecological Applications 18(2): 309-320.
Mueter, F.J., and 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. Fishery Bulletin 100(3): 559-581.
Mueter, F.J., Renner, H.M., and Kruse, G.H. In Prep. Patterns of covariation among fish, seabirds, and
5930
marine mammals in the eastern Bering Sea reflect spatial scales of distribution and tradeoffs
5931
among species. Deep Sea Research Part II: Topical Studies in Oceanography.
5932
Mueter, F.J., Siddon, E.C., and Hunt Jr., G.L. 2011b. Climate change brings uncertain future for subarctic
5933
marine ecosystems and fisheries. In North by 2020: Perspectives on Alaska’s Changing Social-
5934
Ecological Systems. Edited by A.L. Lovecraft and H. Eicken. University of Alaska Press,
5935
Fairbanks, Alaska. pp. 329-357.
5936
5937
5938
5939
Myers, R.A. 1998. When do environment-recruitment correlations work? Reviews in Fish Biology and
Fisheries 8: 285-305.
Myers, R.A., and Mertz, G. 1998. Reducing uncertainty in the biological basis of fisheries management
by meta-analysis of data from many populations: a synthesis. Fisheries Research 37: 51-60.
5940
NRC, N.R.C. 1996. The Bering Sea ecosystem. National Academy Press, Washington, DC, USA.
5941
Punt, A.E., A'mar, T., Bond, N.A., Butterworth, D.S., de Moor, C.L., De Oliveira, J.A.A., Haltuch, M.A.,
5942
Hollowed, A.B., and Szuwalski, C. 2013. Fisheries management under climate and environmental
5943
uncertainty: control rules and performance simulation. ICES Journal of Marine Science: Journal
5944
du Conseil.
255
5945
Quinn, T.J., II, and Niebauer, H.J. 1995. Relation of eastern Bering Sea walleye pollock recruitment to
5946
environmental and oceanographic variables. In Climate change and northern fish populations.
5947
Edited by R.J. Beamish. National Research Council of Canada, Ottawa. pp. 497-507.
5948
Renner, H.M., Mueter, F., Drummond, B.A., Warzybok, J.A., and Sinclair, E.H. 2012. Patterns of change
5949
in diets of two piscivorous seabird species during 35 years in the Pribilof Islands. Deep Sea
5950
Research Part II: Topical Studies in Oceanography 65-70: 273-291.
5951
Siddon, E.C., Duffy-Anderson, J.T., and Mueter, F.J. 2011. Community-level response of fish larvae to
5952
environmental variability in the southeastern Bering Sea. Marine Ecology Progress Series 426:
5953
225-239.
5954
Siddon, E.C., Heintz, R.A., and Mueter, F.J. 2013a. Conceptual model of energy allocation in walleye
5955
pollock (Theragra chalcogramma) from larvae to age-1 in the southeastern Bering Sea. Deep Sea
5956
Research Part II: Topical Studies in Oceanography 94: 140-149.
5957
Siddon, E.C., Kristiansen, T., Mueter, F.J., Holsman, K.K., Heintz, R.A., and Farley, E.V. 2013b. Spatial
5958
Match-Mismatch between Juvenile Fish and Prey Provides a Mechanism for Recruitment
5959
Variability across Contrasting Climate Conditions in the Eastern Bering Sea. PLoS ONE 8(12):
5960
e84526.
5961
Sigler, M.F., Stabeno, P.J., Eisner, L.B., Napp, J.M., and Mueter, F.J. 2014. Spring and fall phytoplankton
5962
blooms in a productive subarctic ecosystem, the eastern Bering Sea, during 1995–2011. Deep Sea
5963
Research Part II: Topical Studies in Oceanography(0).
5964
5965
5966
Spencer, P.D. 2008. Density-independent and density-dependent factors affecting temporal changes in
spatial distributions of eastern Bering Sea flatfish. Fisheries Oceanography 17(5): 396 - 410.
Stabeno, P.J., Kachel, N.B., Moore, S.E., Napp, J.M., Sigler, M., Yamaguchi, A., and Zerbini, A.N. 2012.
5967
Comparison of warm and cold years on the southeastern Bering Sea shelf and some implications
5968
for the ecosystem. Deep Sea Research Part II: Topical Studies in Oceanography 65–70(0): 31-45.
5969
Strasburger, W.W., Hillgruber, N., Pinchuk, A.I., and Mueter, F.J. in press. Feeding ecology of age-0
5970
walleye pollock (Theragra chalcogramma) and Pacific cod (Gadus macrocephalus) in the
5971
southeastern Bering Sea. Deep Sea Research II.
5972
Vert-pre, K.A., Amoroso, R.O., Jensen, O.P., and Hilborn, R. 2013. Frequency and intensity of
5973
productivity regime shifts in marine fish stocks. Proceedings of the National Academy of
5974
Sciences of the United States of America 110(5): 1779-1784.
5975
Wilderbuer, T., Stockhausen, W., and Bond, N. 2013. Updated analysis of flatfish recruitment response to
5976
climate variability and ocean conditions in the Eastern Bering Sea. Deep Sea Research Part II:
5977
Topical Studies in Oceanography 94: 157-164.
256
5978
Wilderbuer, T.K., Hollowed, A.B., Ingraham Jr., W.J., Spencer, P.D., Conners, M.E., Bond, N.A., and
5979
Walters, G.E. 2002. Flatfish recruitment response to decadal climatic variability and ocean
5980
conditions in the eastern Bering Sea. Progress in Oceanography 55: 235-247.
5981
5982
5983
5984
Wyllie-Echeverria, T., and Wooster, W.S. 1998. Year-to-year variations in Bering Sea ice cover and some
consequences for fish distributions. Fisheries Oceanography 7(2): 159-170.
Zador, S. 2013. Ecosystem considerations 2013. North Pacific Fishery Management 644 Council, 605 W
4th Ave, Suite 306, Anchorage, AK 99501.
5985
257