Northeast Fisheries Science Center Reference Document 06-18
Environmental
Preferences of Atlantic Herring
under Changing Harvest Regimes
by Kevin
D. Friedland1, John
E. O'Reilly1, Jonathan A. Hare1, Grayson
B. Wood1, William J. Overholtz2,
and Matthew D. Cieri3
1National Marine Fisheries Service, Office of Marine
Ecosystem Studies, 28 Tarzwell Dr., Narragansett RI 02882
2National Marine Fisheries Service, Population Dynamics Branch, 166 Water
St.,
Woods Hole MA 02543
3Maine Department of Marine Resources, West Boothbay Harbor ME 04575
Print
publication date August 2006;
web version posted September 6, 2006
Citation: Friedland
KD, O’Reilly JE, Hare JA, Wood GB, Overholtz WJ, Cieri MD. 2006.
Environmental preferences of Atlantic herring under changing harvest
regimes. U.S. Dep. Commer., Northeast Fish. Sci. Cent. Ref. Doc.
06-18; 34 p.
Download complete PDF/print version
Abstract
The meso-scale distribution of filter-feeding fishes, such as Atlantic
herring, is usually associated with oceanographic features. These
species are often concentrated along fronts, which demarcate boundaries
between water masses and are frequently areas of increased primary
and secondary production. Fishing operations can use these
oceanographic features as a predictive tool to find fish more efficiently. These
habitat features are also partly responsible for the local availability
of fish, which can be an important factor if a fishery is allocated
regionally or by fleets. The distribution of the Atlantic herring
purse seine fleet was studied to evaluate the response of the fish
to oceanographic features. Catch rates were compared to remotely-sensed
sea surface temperature, chlorophyll concentration, primary production
rate, and frontal occurrence probability. Temperature, chlorophyll,
and primary production were poor predictors of catch location, whereas
frontal probability was associated with fishing. This association
changed dramatically in 1995 when purse seine fishers seemed to target
and find fish in distinctly different oceanographic conditions. This
transition in the distribution of purse seine effort was coincident
with the increased activity of the mid-water trawl fleet in the Gulf
of Maine, suggesting that gear interaction may have influenced the
selection of target fishing areas.
INTRODUCTION
The distribution and local availability of fish can
be influenced by environmental effects, thus masking changes in stock
condition and harvest interaction (Gillis and Frank 2001). This is of particular
concern with migratory species such as Atlantic herring (Clupea harengus
harengus), which are a highly mobile species capable of significant
annual migrations and within-season movements. Thus, movement of
stock components may confound assessments by varying their availability
to various fishing operations.
There is growing concern that changes in local availability of Gulf
of Maine herring on traditional fishing grounds has resulted in significant
relocation of fishing operations. Neal (2003) suggests that spawning
by coastal herring has progressively decreased over the last two decades
and that herring are not as abundant in Maine waters. This is posited
as a result of the fish spawning later in the season, while still geographically
centered on the traditionally identified core spawning areas of eastern
Maine and New Brunswick. These data are in contrast to a general
picture of high stock abundance coming from stock assessment data and
fishery independent surveys (Overholtz et al. 2005). Thus, we are
faced with a complex set of issues relating to the local availability
of the species and the potential influence of a range of factors capable
of causing stock components to migrate at variance to the stock complex.
Herring, like other pelagic species, appear to use gradient search strategies
to find plankton food concentration. However, our interpretations
of herring distribution must also take into account the behaviors associated
with spawning, thus drawing on the information on the areal extent and
timing of spawning, as well as the magnitude of spawning (Smith and Morse
1993). Zooplankton distribution is likely to be a predictor of
herring distribution, considering the importance of zooplankton in the
diet of both larvae and adults (Sherman and Honey 1971, Sherman and Perkins
1971). Larval herring feed on smaller copepods (Psuedocalanus sp., Paracalanus
parvus, and Centropages typicus), while adult herring feed
predominantly on euphausiids, copepods, and chaetognaths (Reid et al.
1999).
Changes in the distribution and abundance of prey can affect the feeding
ecology of herring (Foy and Norcross 1999), suggesting a potential linkage
to the observed decrease in size-at-age for herring on the Northeast
Shelf (Overholtz et al. 2005). Relationships to zooplankton data
would provide the opportunity to evaluate whether these decreases in
size-at-age are related to changes in prey abundance, distribution, and
community composition. Smith and Morse (1993) used ichthyoplankton
data to document the fall and rise of herring on Georges Bank. However,
despite the utility of plankton studies in describing decadal patterns
in abundance, local scale distribution cannot be explained with the historically
collected zooplankton data owing to its lack of spatial and temporal
resolution (Jossi et al. 2003).
Remotely-sensed oceanographic data offer proxy variables that can be
related to the distribution of herring food resources, and thus allow
tests of hypotheses relating environmental condition and local availability. From
a variety of data sources it is clear the annual phytoplankton production
cycle in the Gulf of Maine is out of phase with that observed for Georges
Bank and the Middle Atlantic Bight (O’Reilly and Zetlin 1998, Yoder
et al. 2002). In the Gulf of Maine, the fall phytoplankton
bloom is superior to the spring bloom, whereas on Georges Bank and in
the Middle Atlantic Bight, the annual peak in phytoplankton biomass is
observed during the spring bloom in March. Phytoplankton chlorophyll
levels throughout the July-September period are also significantly higher
in the Gulf of Maine than in Georges Bank and Middle Atlantic Bight. These
high levels of phytoplankton standing stocks and primary production in
the Gulf during summer and early fall coincide with the period of intense
feeding, growth, and spawning of herring in the region. Moreover,
there is considerable interannual variability in the meso-scale distributional
patterns of sea surface temperature and primary production within the
Gulf during the summer, and particularly so during the October fall bloom.
Oceanographic fronts, which are regions of sharp gradients in temperature
or density, are well known to have a concentrating effect on a range
of marine organisms and have been examined on both micro- and meso-scale
spatial dimensions. Fronts provide structure within the water column
that often results in the upwelling of nutrients and subsequent stimulation
of primary production. Fronts can also form density barriers that
can serve to concentrate primary consumers, which forage for species
like herring. Fronts are more easily synoptically measured than
are zooplankton patterns, thus they are features that serve as a proxy
variable for zooplankton abundance, which may not be measured as effectively.
Zinkevitch (1967) analyzed the data from the foreign fleets fishing
on the Northeast Shelf and showed that set frequency for herring was
concentrated in frontal regions along Georges Bank and the Middle Atlantic
Bight. Fronts have also been associated with pre-spawning aggregations
of herring (Maravelias 1997), which may be of significance to the historical
fishing patterns in the Gulf of Maine. The Gulf has a complex set
of frontal patterns associated with bathymetry, vigorous tidal mixing,
and the overarching pattern of circulation. Frontal associations
for migratory tuna in the Gulf suggest that the frontal associations
that may involve herring could form a complex trophic cascade (Schick
et al. 2004).
The Northeast U.S. Shelf Large Marine Ecosystem is complex and highly
dynamic, and known to have variable inter-annual and spatial patterns
of hydrography and primary and secondary production. In light of
the documented rebuilding and recent large population estimates of herring
from stock assessments (Overholtz et al. 2005), the goal of this investigation
is to examine a range of physical and biological oceanographic datasets
in an exploratory fashion to develop hypotheses about potential factors
affecting the local availability of herring. The specific objectives
of this study are to examine the correspondence between spatial patterns
in oceanographic parameters and the distributional patterns of herring
catch in the Gulf of Maine. We will also seek to analyze these
patterns in the context of changing fisheries regimes where the manner
of harvest may influence the local abundance patterns of the stock.
METHODS
Atlantic herring vessel trip report data
Currently the vessel trip report database has approximately 80,000 records
or trips reporting herring. Some gears reported in the database
have consistently produced large catches of herring over wide geographic
areas and over relevant time periods; these gears will provide the basis
for our analyses. The most active gear in the database is purse
seines, which have targeted herring over the full temporal extent of
the database, 1960-2004 (Table 1). Purse seine catches and trips
provide a depiction of the distribution of herring schools along the
Maine coast. Fishing areas for this gear are reported by ten minute
squares; thus fine scale oceanographic features may not be useful in
analyzing the behavior of this gear. Stop seines and weirs were
a major gear in the fishery until the availability of herring juveniles
changed and the fishery was first reduced in the 1980s and ended in the
1990s. A number of oceanographic and biologic datasets could be
re-examined to study changes in the historical fishery.
The contemporary fishery includes trawl gears that both target herring
and take herring as by-catch. The principal targeting gear is mid-water
trawls. These trip reports have specific latitude and longitude
locations and lend themselves to analyses of fine scale oceanographic
features such as fronts. The non-target gears, such as groundfish
and shrimp otter trawls, do not have concentrated catches of herring,
but since they are not targeting herring, their catch rate data may be
useful as a relative index of abundance. These non target gears,
which sample over broad areas, may be useful in testing assumptions related
to the data from targeted gears, which sample over smaller areas. Of
the gears targeting herring, we focused on the catches of the purse seine
fleet that operates along the Maine coast. This fleet provides
a long time series of data that exceeds the time periods of coverage
associated with the oceanographic datasets.
Sea surface temperature
Daily high resolution (4 km/pixel) maps of the distribution of sea
surface temperature (SST) throughout the Northeast U.S. Continental
Shelf Ecosystem have been generated by combining nighttime SST data
from three polar orbiting satellite sensors: NOAA’s satellites
equipped with AVHRR sensors, and NASA’s MODIS terra and Aqua
SST sensors. All
data were obtained from NASA JPL (http://podaac.jpl.nasa.gov/). AVHRR
data were processed using the Pathfinder method and covered the period
from 1985 through July 2005. MODIS Terra SSTs were available from
Feb. 2000 to the present and MODIS Aqua SSTs from July 2002 through the
present. By combining SST data from two or more sensors after 2000,
we were able to increase the number of good (cloud-free) SST estimates
and ecosystem coverage. Nighttime SST data were used exclusively
to avoid unrepresentative SSTs in daytime scenes resulting from diurnal
heating. Our resulting 4 km SST time series, encompassing the period
from January 1985 to the present, was binned to 10’ squares to
match the spatial resolution of the vessel trip report data.
Chlorophyll and primary productivity estimates
Since September 1997, remotely-sensed satellite data from the SeaWiFS
ocean color sensor has provided us with daily synoptic views of surface
concentrations of chlorophyll throughout the Northeast continental shelf
ecosystem. The concentration of chlorophyll, the dominant green
pigment in phytoplankton, is considered an index of phytoplankton abundance
and biomass. Our daily average chlorophyll time series was developed
from 4,765 1 km-resolution scenes of the Northeast acquired by SeaWiFS
from September 1997 through July 2005. As with SST, the chlorophyll
data were binned to 10’ squares to match the resolution of the
vessel trip report data.
We were also interested in exploring relationships between patterns
in the level of phytoplankton primary production and herring distributions. Our
estimates of daily phytoplankton primary production are based on remotely-sensed
SST, chlorophyll, and PAR (photosynthetically active radiation) from
SeaWiFS, and the VGPM2a model, a variation of the vertically generalized
productivity model (VGPM) developed by Behrenfeld and Falkowski (1997). The
conventional method used for measuring phytoplankton primary production
is the 14C-uptake method. While the in situ 14C-uptake
method provides a precise estimate of primary productivity, this method
is expensive and labor-intensive, and consequently it is difficult to
obtain sufficient spatial and temporal coverage to assess annual variability
and long-term trends. At present, combining remotely-sensed data
from satellites with productivity algorithms (Campbell et al. 2002) represents
the only feasible method for resolving seasonal, annual and climate-related
variability of primary productivity throughout large marine ecosystems. Our
daily primary productivity time series spans the period from September
1997 to present, beginning with the first available chlorophyll maps
from SeaWiFS.
We consider our satellite-based estimates of primary production to
be reliable because the general spatial and seasonal patterns of primary
production developed from contemporary satellite data and the VGPM2a
model (O’Reilly and Ducas 2004) agree well with historical patterns
based on in situ measurements of 14C-uptake made during
MARMAP surveys of the Northeast from 1977 through 1987 (O’Reilly
et al. 1987).
Frontal
probability
Remote sensing not only provides data on sea surface temperature and
chlorophyll, but also provides information on sea surface temperature
gradients (i.e., fronts). Many pelagic species aggregate in the
vicinity of fronts, and we evaluate the hypothesis that herring catches
are associated with frontal locations and that changes in frontal locations
are related to changes in the distribution of the herring fishery in
the Gulf of Maine. We use a data set developed by satellite oceanographers
at the University of Rhode Island Graduate School of Oceanography (Ullman
and Cornillon 1999). The fronts dataset is limited to the years
1985 to 2000. It was necessary to retrieve all the original sea
surface temperature images used to identify the fronts, so an estimate
of good pixels could be made for each data bin. Good pixels are
cloud-free ocean surface pixel locations where a sea surface temperature
could be estimated by the satellite radiometer. Frontal probability
is simply the ratio of frontal pixels identified in the fronts database
to good pixels available from the original sea surface temperature image.
Data analyses
The catch data were linked with mean sea surface temperature, chlorophyll,
primary productivity, and frontal probability for each trip location
by standard 10’ square. With the data linked, we examined
the range of each parameter associated with the fishery and the correlation
between the catch and the respective parameters. The most obvious
limitation of these analyses is that they only sample the temperature,
chlorophyll, and fronts data where the fishery occurs; thus if the distribution
of the fishery is being affected by a coastal gradient, the data from
the trip locations may not characterize the gradient. Therefore,
we also examined the full meso-scale gradients of these parameters by
examining the parameters where the fishery occurred annually versus where
the fishery had occurred during the time series.
RESULTS
Atlantic herring purse seine catch
The purse seine fishery occurs primarily during the summer with most
of the catch landed during the months of July through October (Figure
1). The fishery targets adult herring for the bait markets, which
in turn supplies lobster fishers. The bait fishery can extend well
into the fall to satisfy the needs of the export lobster fishery. Only
in some years has the fishery duration been constrained by the quota. Over
the last decade, the number of purse seine trips has declined, reflecting
the lower number of vessels participating in the fishery (Table 2). The
contemporary purse seine fleet comprises only four boats, down
from over ten vessels just a decade ago. The average landings per
trip have increased slightly during the peak months of the fishery, possibly
reflecting the effect of decreased competition for fish (Table 3). Landings
from individual trips have ranged from 0 to 11,725 mt. Most trips
land less than 100 mt and the distribution of trip size appears to be
log normal (Figure 2). The presence of zero and low tonnage catches
in the database suggests that the dynamic range of the data may be sufficient
to evaluate and contrast associations between low versus high catch rate
areas and oceanographic conditions.
The purse seine fleet has utilized most of the coastal ocean from eastern
Maine to Massachusetts Bay. We unitized the reported location of
individual trips to study the distribution of the fishery and considered
ways of characterizing the central tendency of the fishery distribution
and indices that characterize the range as well. The distribution
of the fishery within the area encompassing all catch locations is a
function of the time period over which catches are summed and the factors
affecting the search decisions and success of the participants in the
fishery. Our
first characterization of distribution of the fishery was to compute
mean catch-weighted longitude and latitude by month and year. We
restricted our analysis to the used purse seine trip data from the Gulf
of Maine, eliminating records south of 42°N
from the analysis. These data show that the fishery has been centered
all along the western Gulf of Maine (Figure
3). The early season
fishery (catches made during May) is distributed more to the south and
west compared to the summer fishery catch of July and August. In
time series, these data suggest that during the collapse of the stock,
most of the fishing activity was further to the south and west. The
time series of catch-weighted mean longitude show that the contemporary
fishery is occurring around 69°W, whereas during the years of the
stock collapse, the fishery was closer to 70°W (Figure
4). The
catch-weighted mean latitude data show a similar trend over time with
the contemporary fishery occurring mostly around 44°N and the historic
fishery occurring around 43°N (Figure
5). We are limited to
the contemporary fishery (since 1985) for our comparisons to the principal
oceanographic datasets; it clear the fishery is now constrained to a
more narrow range of both longitude and latitude. Despite this
narrowing of the range of the mean locations, there has been a dramatic
increase in the number of 10’ square boxes visited by the fishery
beginning in 1996 (Figure
6).
Sea surface temperature
Sea surface temperature of trip locations generally follows the seasonal
pattern of temperatures in the Gulf of Maine. SST associated with
May trip locations averaged 7°C and increase to a mean of approximately
13°C during the August fishery (Figure 7). The inter-quartile
range for the August fishery captured temperatures in excess of 15°C. Trip
location temperatures began to decrease with fall cooling during September
into October.
We examined the relationship between herring catch and sea surface temperature
from two perspectives, within the area fishing occurred and between areas
where fishing occurred versus where it had occurred during the time series. For
the first part of the analysis, the relationship between the mean catch
of herring by ten minute square and the mean sea surface temperature
of the square was analyzed. There are no significant trends between
herring catch and sea surface temperature (Figure
8). There is
a tendency for correlation to be mostly positive in the late season months,
but few of these correlations are significant.
For the second part of the analysis, the sea surface temperature in
ten minute squares where fishing occurred was compared to the temperature
in squares where fishing had occurred during the time series. In
a given year, the fishery will occur typically in one-fourth of the squares
where trips have occurred. The sea surface temperature in the fishery
versus the unvisited squares area was similar for all years and months,
with no systematic differences between areas (Figure
9).
Chlorophyll
and primary production
Chlorophyll a at trip locations also shows a monthly pattern
mostly likely reflecting the fall bloom. Mean chlorophyll a is
approximately 2 mg/m3 from May to July and increases to 4
mg/m3 during August and September (Figure
10). Primary
production is more highly patterned showing a progressive seasonal increase
(Figure
11).
The distribution of the fishery versus chlorophyll a concentration
and primary production rate was analyzed in the same way as sea surface
temperature. Correlation between catch and chlorophyll a was
non-significant and without trend (Figure 12). Likewise, chlorophyll a was
not significantly different in the fished versus the un-fished areas
(Figure 13). There were some significant correlations between catch
and primary production rate, but there were both positive and negative
correlations, suggesting they were not meaningful (Figure 14). There
was also a slight tendency for primary production to be higher in un-fished
areas during early- to mid-summer (Figure 15).
Frontal Probability
The probability of SST fronts increased seasonally, with the most fronts
evident in September sea surface temperature scenes (Figure 16). Most
correlations between catch and frontal probability are positive and all
significant correlations are also positive (Figure 17). The most
striking trend occurs with the September data, where the period 1985
to 1992 yielded seven significant correlations. The correlation
between catch and frontal probability appears to have been stronger in
the early part of the time series; at some point in the mid-1990s the
correlations tended to be more neutral or negative.
Distribution related to fronts is also seen in the analysis of fronts
in areas fished versus un-fished. During the summer months, fishing
occurred in lower frontal probability areas during the early part of
the time series, but switched to higher frontal probability areas again
during the 1990s (Figure
18). This switch in frontal probability
associated with fishing is most dramatic in the August data, where nearly
all the years prior to 1995 show fishing in low frontal probability areas
and years after 1994 show fishing in high frontal areas.
This result is inconsistent with the idea that fishing is responding
solely to the distribution of fronts. In fact, when we consider
when mid-water trawlers entered the fishery, we see a relationship between
the intensity of mid-water trawler trips and the August frontal probabilities
associated with the purse seine fleet (Figure 19). It would appear
that before mid-water trawlers were fishing herring in the Gulf of Maine,
the purse seine fleet concentrated their effort in low frontal probability
regions but found greater success fishing in the higher probability frontal
areas within these regions. After the mid-water trawlers entered
the fishery, the purse seine fleet appears to have selected higher frontal
probability regions to search for fish, and its success is no longer
related to the distribution of fronts within these areas.
However, these observations must be tempered by the fact that there
has been change in the environment with respect to the distribution of
fronts. The distribution of fronts in the Gulf of Maine has changed
over time, in particular during the month of August, where the western
segment of the Gulf started to have more fronts develop in the early
1990s (Figure 20). The environmental change from low to high front
years does not match the 1994-1995 transition seen in the fisheries data,
nor does it address the spatial issue of where fishing occurred, but
it does maintain the possibility of an environmentally driven causality.
DISCUSSION
Most coastal pelagic fish species are migratory, moving
toward the equator in winter and toward the poles in summer (Fréon
and Misund 1999). Spawning and feeding activities are embedded
within these seasonal movements, and the specific patterns vary among
species and among local populations within species. The habitat
of these species is described by a combination of abiotic (temperature,
salinity, fronts) and biotic (food availability, predator distribution)
variables related to the oceanography and bathymetry of their ecosystems. Fishing
targets individuals at different points in the life cycle and seasonal
cycle; this intersection determines the distribution of the fishing relative
to the distribution of the species.
Within this context, Atlantic herring move seasonally within
the Gulf of Maine, and three general migratory patterns are recognized
and associated with general stock structure (Sindermann 1979). Herring
in Nova Scotian waters spawn in the late-summer and fall along the southwestern
coast and overwinter along the northeastern coast. Georges Bank/Nantucket
Shoals herring overwinter south of Cape Cod, spend the spring and summer
in the Gulf of Maine, and spawn in the fall. The movements of herring
that spawn along coastal Maine are less well known; overwintering likely
occurs in the vicinity of or south of Cape Cod, while spring and summer
are spent in the Gulf of Maine and spawning occurs in the fall. The
amount of mixing among stocks is unknown. Given the general importance
of environmental factors in defining pelagic fish habitat, the distribution
of herring likely changes interannually in response to environmental
variation in the Gulf of Maine, and potentially in areas to the south
(e.g., Mid-Atlantic) and north (e.g., Nova Scotia).
U.S. fishing in the Gulf of Maine targets the coastal Gulf
of Maine and Georges Bank/Nantucket Shoals herring and is primarily composed
of purse seiners and mid-water trawlers (Overholtz et al. 2005). Our
study focused on the purse seine fishery and our results suggest that
the extent of the ecosystem used by the purse seine fleet has increased
over the past decade. This change coincides with a change in reporting
for all fisheries, so it may be an artifact of reporting practices. However,
it also coincides with the large increase in fishing by mid-water trawlers,
thus it may reflect more dispersed schools of fish caused by the interaction
of the gears.
Another factor in the interpretation of the catch data
is the effect of spawning closures on the distribution of catches. Spawning
closures are made by area and at different times each year. The
closures are based on within-season samples, thus the timing of closures
is not the same each year. When there is the potential for a closure,
there is evidence that smaller fish are targeted to avoid triggering
a closure; this targeting of smaller fish could result in a change in
the distribution of the fishery. These factors have to be taken
into account before drawing any conclusion about the environmental analysis
presented here.
Our analyses indicate that sea surface temperature and
characteristics of primary production are not an important determinant
of success or location of the purse seine fishery. Although several studies
have indicated that temperature is important factor in the distribution
of herring (Maravelias and Reid 1997, Corten 2001), we found no effect. Temperature
in and of itself is more of a regional-scale feature and is probably
too coarse a parameter to be related to the fishery within the Gulf of
Maine and within the fishing season. In looking at the range of
temperature of the fishery, we see that it varies seasonally and reflects
the range of temperature tolerances of the fish itself. As with
sea surface temperature, neither plankton parameter appears to be correlated
to the location or intensity of fishing. Again this is in contrast
to other studies (Maravelias and Reid 1997, Corten 2001), but not unexpected
because we are examining data within the Gulf of Maine and within the
fishing season. Further, since herring do not directly consume
phytoplankton, a direct link is not likely. However, if we were
able to examine zooplankton abundances on the same temporal and spatial
scales as herring catches, we would expect to see an association (see
Maravelias and Reid 1997, Corten 2001, Kvamme et al. 2003).
Our results suggest that fronts play an important role
in the distribution of herring, and similar results have been found in
other studies. Off the coast of South Africa, anchovy (Engraulisi
capensis) and round herring (Etrumeus whiteheadi) tend to
concentrate around fronts, whereas sardine (Sardinops sagax) show
no aggregation near fronts (Agenbag et al. 2003). In the North
Sea, Atlantic herring (Clupea harengus) tended to avoid stratified
and frontal areas (Maravelias and Reid 1997). In the Gulf of Maine,
frontal probability appears to be positively related to the selection
of fishing area and the success or intensity of catch. However,
the dramatic changes in the fishery relative to fronts in the mid-1990s
suggests that other factors are influencing the distribution and catch
of the purse-seine fishery. In fact, the dominant change in the
purse-seine fishery catch data co-occurs with the initiation of the midwater
trawler fishery. Thus, the changes in the purse-seine fishery are
linked to the mid-water trawl fishery, but we cannot rule out the possibility
of an environmental influence related to the intensity and distribution
of fronts in the Gulf of Maine.
The analyses presented here indicate that fishery-dependent
data may be influenced by management changes in a fishery (e.g., closures),
interactions between fisheries (e.g., purse seiners and mid-water trawled),
and the environmental variables that define habitat (e.g., fronts). Additionally,
the fishery-dependent data examined here could be influenced by differences
in abundance and migration of local populations that use the Gulf of
Maine during the spring, summer, and early fall (see Sinclair 1988, McQuinn
1998). We show that the distribution of the purse-seine fishery
has changed dramatically relative to the distribution of fronts in the
Gulf of Maine. The time of this change coincided with the beginning
of the mid-water trawl fishery, suggesting an interaction between the
fisheries. However, there has also been an underlying change in
the distribution of fronts. Previous studies defining pelagic fish
habitat using fishery-dependent data have tried to limit the influence
of the fishery on the analyses of environmental variables (Agenbag et
al. 2003), but ultimately these analyses are still influenced by fishery
dynamics. Future approaches could define the environmental effects
using fishery-independent data (e.g., Kvamme et al. 2003) and then examine
the interaction between fisheries using fishery dependent data, taking
into account the previously defined environmental effects.
ACKNOWLEDGMENTS
We thank D. Ullman, University of Rode Island, for providing
access to the SST fronts data.
LITERATURE CITED
Agenbag JJ, Richardson AJ, Demarcq H, Fréon P, Weeks S, Shillington
FA. 2003. Estimating environmental preferences of South African
pelagic fish species using catch size- and remote sensing data. Prog
Oceanogr 59: 275-300.
Behrenfeld MJ, Falkowski PG. 1997. Photosynthetic rates
derived from satellite-based chlorophyll concentration. Limnol
Oceanogr 42(1):1-20.
Campbell J, Antoine D, Armstrong R, Arrigo K, Balch W, Barber R, Behrenfeld
M, Bidigare R, Bishop J, Carr ME, Esaias W, Falkowski P, Hoepffner N,
Iverson R, Kiefer D, Lohrenz S, Marra J, Morel A, Ryan J, Vedernikov
V, Waters K, Yentsch C, Yoder J. 2002. Comparison of
algorithms for estimating ocean primary production from surface chlorophyll,
temperature, and irradiance. Global Biogeochem Cyc Vol. 16(3):
Art. No. 1035.
Corten A. 2001. Northern distribution of North Sea herring
as a response to high water temperatures and/or low food abundances. Fish
Res 50:189-204.
Foy RJ, Norcross BL. 1999. Spatial and temporal variability
in the diet of juvenile Pacific herring (Clupea pallasi) in Prince
William Sound, Alaska. Can J Zool 77: 697-706.
Fréon P, Misund OA. 1999. Dynamics of pelagic fish
distribution and behaviour: effects of fisheries and stock assessment. Cambridge
UK: Cambridge University Press.
Gillis DM, Frank KT. 2001. Influence of environment and
fleet dynamics on catch rates of eastern Scotian Shelf cod through the
early 1980s. ICES J Mar Sci 58:61-69.
Jossi JW, John AWG, Sameoto D. 2003. Continuous Plankton
Recorded sampling of the east coast of North America: history and status. Prog
Oceanogr 58: 313-326.
Kvamme C, Nøttestad L, Fernö A, Misund AA, Dommasnes A,
Axelsen BE, Dalpadado P, Melle W. 2003. Migration patterns
in Norwegian spring-spawning herring: why young fish swim away from the
wintering area in late summer. Mar Ecol Prog Ser 247: 197-210.
Neal B. 2003. Eastern Gulf of Maine Atlantic herring spawning
area survey project, year 6 summary. 9 p.
Maravelias CD. 1997. Trends in abundance and geographic
distribution of North Sea herring in relation to environmental factors. Mar
Ecol Prog Ser159:151-164.
Maravelias CD. 2001. Habitat associations of Atlantic herring
in the Shetland area: influence of spatial scale and geographic segmentation. Fish
Oceanogr 10: 259-267/
Maravelias CD, Reid DG. 1997. Identifying the effects of
oceanographic features and zooplankton on prespawning herring abundance
using generalized additive models. Mar Ecol Pro Ser 147: 1-9.
McQuinn IH. 1997. Metapopulations and the Atlantic herring. Rev
Fish Biol Fish 7: 297-329.
O’Reilly JE, Ducas T. 2004. Seasonal and annual variability
in primary production in the Northeast U.S. Large Marine Ecosystem. Poster
presented at NASA Ocean Color Research Team Meeting, Washington DC, April
2004.
O’Reilly JE, Zetlin CA. 1998. Seasonal, horizontal,
and vertical distribution of phytoplankton chlorophyll a in the
Northeast US continental shelf ecosystem. NOAA Tech Rep NMFS 139,
120 p.
O’Reilly JE, Evans-Zetlin C, Busch DA. 1987. Primary
production. Chapter 21 in: Georges Bank. Backus RH, ed. Cambridge
MA: MIT Press, p. 220-233.
Overholtz W, Jacobson L, Melvin G, Cieri M, Power M, Libby D, Clark
K. 2005. Stock assessment of the Gulf of Maine - Georges
Bank Atlantic herring complex, 2003. Northeast Fisheries Science
Center Ref Doc 04-06; 290 p.
Reid RN, Cargnelli LM, Griesbach SJ, Packer DB, Johnson DL, Zetlin CA,
Morse WW, Berrien PL. 1999. Essential fish habitat source
document: Atlantic herring, Clupea harengus, life history and
habitat characteristics. NOAA Tech Memo NMFS-NE-126.
Schick RS, Goldstein J, Lutcavage ME. 2004. Bluefin tuna
(Thunnus thynnus) distribution in relation to sea surface temperature
fronts in the Gulf of Maine (1994-96). Fish Oceanogr 13: 225-238.
Sherman K, Honey KA. 1971. Seasonal variations in the food
of larval herring in coastal waters of Maine. P V Réun Cons
Int Explor Mer 160: 121-124.
Sherman K, Perkins HC. 1971. Seasonal variations in the
food of juvenile herring in coastal waters of Maine. Trans Am Fish
Soc 100: 121-124.
Sinclair M. 1988. Marine populations: an essay on population
regulation and speciation. Seattle WA: Washington Sea Grant Program;
252 pp.
Sindermann CJ. 1979. Status of northwest Atlantic herring
stocks of concern to the United States. Sandy Hook Lab Tech Ser
Rep 23, 449 p.
Smith WG, Morse WW. 1993. Larval distribution patterns:
early signals for the collapse/recovery of Atlantic herring Clupea
harengus in the Georges Bank area. Fish Bull 91:338-347.
Ullman DS, Cornillon PC. 1999. Satellite-derived sea surface
temperature fronts on the continental shelf off the northeast U.S. coast. J
Geophys Res 104:23,459-23,478.
Yoder JA, Schollaert SE, O'Reilly JE. 2002. Climatological
phytoplankton chlorophyll and sea surface temperature patterns in continental
shelf and slope waters off the northeast U.S. coast. Limnol Oceanogr
47(3), 672–682.
Zinckevich VN. 1967. Observations on the distribution of
herring, Clupea harengus L., on Georges Bank and in adjacent waters
in 1962-1995. ICNAF Res Bull 4: 101-115.
