Content uploaded by Boris Worm
Author content
All content in this area was uploaded by Boris Worm on Dec 17, 2015
Content may be subject to copyright.
Content uploaded by Boris Worm
Author content
All content in this area was uploaded by Boris Worm on Dec 17, 2015
Content may be subject to copyright.
..............................................................
Rapid worldwide depletion of
predatory fish communities
Ransom A. Myers & Boris Worm
Biology Department, Dalhousie University, Halifax, Nova Scotia, Canada
B3H 4J1
.............................................................................................................................................................................
Serious concerns have been raised about the ecological effects of
industrialized fishing
1–3
, spurring a United Nations resolution on
restoring fisheries and marine ecosystems to healthy levels
4
.
However, a prerequisite for restoration is a general understand-
ing of the composition and abundance of unexploited fish
communities, relative to contemporary ones. We constructed
trajectories of community biomass and composition of large
predatory fishes in four continental shelf and nine oceanic
systems, using all available data from the beginning of exploita-
tion. Industrialized fisheries typically reduced community bio-
mass by 80% within 15 years of exploitation. Compensatory
increases in fast-growing species were observed, but often
reversed within a decade. Using a meta-analytic approach, we
estimate that large predatory fish biomass today is only about
10% of pre-industrial levels. We conclude that declines of large
predators in coastal regions
5
have extended throughout the
global ocean, with potentially serious consequences for eco-
systems
5–7
. Our analysis suggests that management based on
recent data alone may be misleading, and provides minimum
estimates for unexploited communities, which could serve as the
‘missing baseline’
8
needed for future restoration efforts.
Ecological communities on continental shelves and in the open
ocean contribute almost half of the planet’s primary production
9
,
and sustain three-quarters of global fishery yields
1
. The widespread
decline and collapse of major fish stocks has sparked concerns about
the effects of overfishing on these communities. Historical data
from coastal ecosystems suggest that losses of large predatory fishes,
as well as mammals and reptiles, were especially pronounced, and
precipitated marked changes in coastal ecosystem structure and
function
5
. Such baseline information is scarce for shelf and oceanic
ecosystems. Although there is an understanding of the magnitude of
the decline in single stocks
10
, it is an open question how entire
communities have responded to large-scale exploitation. In this
paper, we examine the trajectories of entire communities, and
estimate global rates of decline for large predatory fishes in shelf
and oceanic ecosystems.
We attempted to compile all data from which relative biomass at
the beginning of industrialized exploitation could be reliably
estimated. For shelf ecosystems, we used standardized research
trawl surveys in the northwest Atlantic Ocean, the Gulf of Thailand
and the Antarctic Ocean off South Georgia, which were designed to
estimate the biomass of large demersal fish such as codfishes
(Gadidae), flatfishes (Pleuronectidae), skates and rays (Rajiidae),
among others (see Supplementary Information for detailed species
information). In all other shelf areas for which we could obtain data,
industrialized trawl fisheries occurred before research surveys took
place. For oceanic ecosystems, we used Japanese pelagic longlining
data, which represent the complete catch-rate data for tuna (Thun-
nini), billfishes (Istiophoridae) and swordfish (Xiphiidae) aggre-
gated in monthly intervals, from 1952 to 1999, across a global
58£58grid. Pelagic longlines are the most widespread fishing gear,
and the Japanese fleet the most widespread longline operation,
covering all oceans except the circumpolar seas. Longlines, which
resemble long, baited transects, catch a wide range of species in a
consistent way and over vast spatial scales. We had to restrict our
analysis of longlining data to the equatorial and southern oceans,
because industrialized exploitation was already underway in much
of the Northern Hemisphere before these data were recorded
11,12
.
Longlining data were separated into temperate, subtropical and
tropical communities (see Methods).
For each shelf and oceanic community, i, we estimated
NiðtÞ¼Nið0Þ½ð12diÞe2ritþdið1Þ
where N
i
(t) is the biomass at time t,N
i
(0) is the initial biomass
Figure 1 Time trends of community biomass in oceanic (a–i) and shelf ( j–m)
ecosystems. Relative biomass estimates from the beginning of industrialized fishing (solid
points) are shown with superimposed fitted curves from individual maximum-likelihood
fits (solid lines) and empirical Bayes predictions from a mixed-model fit (dashed lines).
letters to nature
NATURE| VOL 423 | 15 MAY 2003 | www.nature.com/nature280 © 2003 Nature Publishing Group
before industrialized exploitation, and r
i
is the initial rate of decline
to d
i
, the fraction of the community that remains at equilibrium.
The initial rate of decline in total biomass
—
that is, the fraction lost
in the first year
—
is ð12diÞð12e2riÞ:Then we combined all data
using nonlinear mixed-effects models
13
, where ri,Nðmr;j2
rÞand
log di,Nðmd;j2
dÞ;to estimate a global mean and variance of r
i
and d
i
.
In the open ocean communities, we observed surprisingly con-
sistent and rapid declines, with catch rates falling from 6–12 down
to 0.5–2 individuals per 100 hooks during the first 10 years of
exploitation (Fig. 1a–i). Rates of decline were similar in tropical and
subtropical regions, but consistently highest in temperate regions in
all three oceans (Fig. 1c, f, i and Table 1). Temperate regions also
showed the lowest equilibrium catch rates (Table 1). Spatial pattern
of expansion and decline of pelagic fisheries are shown in Fig. 2.
During the global expansion of longline fisheries in the 1950s to
1960s, high abundances of tuna and billfish were always found at the
periphery of the fished area (Fig. 2a–c). Most newly fished areas
showed very high catch rates, but declined to low levels after a few
years. As a result, all areas now sustain low catch rates, and some
formerly productive areas have been abandoned (Fig. 2d). In shelf
communities, we observed declines of similar magnitude as in the
open ocean. The Gulf of Thailand, for example, lost 60% of large
finfish, sharks and skates during the first 5 years of industrialized
trawl fishing (Fig. 1j). The highest initial rate of decline was seen in
South Georgia (Fig. 1k), which has a narrow shelf area that was
effectively fished down during the first 2 years of exploitation
14
.
Less-than-average declines were seen on the Southern Grand Banks
Table 1 Meta-analysis of time trends in predatory fish biomass
Region
r
i
(£100) d
i
(£100)
Individual fit CL Mixed model Individual fit CL Mixed model
...................................................................................................................................................................................................................................................................................................................................................................
Tropical Atlantic 16.6 13.5–19.7 16.7 12.1 10.0–14.5 11.9
Subtropical Atlantic 12.9 10.1–15.7 13.0 8.1 6.4–10.2 8.3
Temperate Atlantic 21.4 15.8–26.9 20.3 4.7 3.2–6.9 5.3
Tropical Indian 9.2 7.1–11.4 9.5 17.6 14.9–20.6 16.8
Subtropical Indian 6.5 5.1–7.8 6.8 8.2 5.5–12.3 9.2
Temperature Indian 30.7 23.7–37.8 27.7 5.5 3.9–7.7 6.3
Tropical Pacific 12.1 9.4–14.8 12.4 15.5 13.0–18.6 14.9
Subtropical Pacific 12.8 8.5–17.1 13.5 23.5 18.9–29.3 21.5
Temperate Pacific 20.8 14.3–27.3 20.4 8.2 5.6–12.1 8.5
Gulf of Thailand 25.6 18.2–33.0 22.2 9.3 6.8–12.6 9.8
South Georgia 166.6 2.2–331.1 30.8 20.9 17.5–25.0 16.0
Southern Grand Banks 4.0 2.9–5.1 5.7 0.0 – 10.0
Saint Pierre Banks 5.1 0.1–10.1 6.3 2.7 0.0–36600 7.9
Mixed model mean 16.0 10.3
Mixed model CL 10.7–21.3 7.7–13.9
Distribution 4.5–31.6 4.6–23.6
...................................................................................................................................................................................................................................................................................................................................................................
Two parameters were estimated: r
i
is the initial rate of decline (in per cent per year), and d
i
the residual biomass proportion at equilibrium (in per cent). Point estimates and 95%confidence limits (CL)
are presented for the individual maximum likelihood fits, and for the mixed-effects model that combined all data (see Methods for details). The random-effects distribution (95% limits) provides a
measure of the estimated parameter variability across communities.
Figure 2 Spatial patterns of relative predator biomass in 1952 (a), 1958 (b), 1964 (c) and
1980 (d). Colour codes depict the number of fish caught per 100 hooks on pelagic
longlines set by the Japanese fleet. Data are binned in a global 58£58grid. For complete
year-by-year maps, refer to the Supplementary Information.
letters to nature
NATURE| VOL 423 | 15 MAY 2003 | www.nature.com/nature 281
© 2003 Nature Publishing Group
(Fig. 1l) and Saint Pierre Bank (Fig. 1m); these communities may
already have been affected by intense pre-industrial fisheries
15
.
By combining all data using a mixed-effects model, we estimated
that the mean initial rate of decline, r
i
, is 16% per year, and the mean
residual equilibrium biomass, d
i
, is 10% of pre-exploitation levels
(Table 1). So, an 80% decline typically occurred within 15 years of
industrialized exploitation, which is usually before scientific moni-
toring has taken place. The proportion of residual biomass, d
i
,
showed remarkably little variation between communities (Table 1):
the mixed-effects model estimates imply that 95% of communities
would have a residual biomass proportion between 5% and 24%.
We believe that these still represent conservative estimates of total
predator declines for the following reasons: (1) pre-industrial
removals from some of the shelf communities
15
; (2) gear saturation
at high catch rates in the early longlining data, as well as higher
initial levels of shark damage leading to an underestimation of
initial biomass
16
(see Supplementary Information); (3) increasing
fishing power of longline vessels over time owing to improved
navigation and targeting of oceanographic features
17
; and (4)
targeting of some migratory species, such as southern bluefin
tuna (Thunnus maccoyii), at their tropical spawning grounds before
widespread exploitation in temperate areas occurred
18
. Further-
more, declines in other large predators such as sharks are not fully
captured by our data, but may be of similar or greater magnitude
than those of bony fishes
19,20
.
One mechanism that could compensate for the effects of over-
fishing is the increase in non-target species due to release from
predation or competition
21
. In our analyses, we see evidence for
species compensation in both oceanic billfish and shelf groundfish
communities (Fig. 3). According to the longlining data and to early
surveys
11,12
, blue marlin was initially the dominant billfish species,
but declined rapidly in the 1950s (Fig. 3a). Simultaneous increases
in faster-growing species such as sailfish were observed, followed by
a decrease, possibly due to increased ‘bycatch’ mortality (Fig. 3a;
neither species was targeted by the Japanese fleet). Coincidentally,
swordfish catch rates increased until these fish became prime
targets of other fleets in the late 1980s. Surprisingly consistent
patterns of compensatory increase and decline were seen in most
pelagic communities (see Supplementary Information). Similarly,
in the North Atlantic demersal communities, we observed rapid
initial declines, particularly in large codfishes, but also in skates
and rockfish. Although the dominant codfishes declined sixfold
between 1952 and 1970, sixfold increases were seen in the flatfishes,
which were not initially targeted by the trawl fishery (Fig. 3b).
Some increase in the gadoids occurred when implementation of the
200-mile limit in 1977 curtailed foreign overfishing in Canadian
waters. However, as in the billfish data, we observed an ultimate
decline in all species groups (Fig. 3b) as fishing pressure from
Canadian and other fleets intensified in the late 1980s, leading to
the collapse of all major groundfish stocks
10
. We conclude that
some species compensation was evident, but often reversed within
a decade or less, probably because of changes in targeting or
bycatch.
Our analysis suggests that the global ocean has lost more than
90% of large predatory fishes. Although it is now widely accepted
that single populations can be fished to low levels, this is the first
analysis to show general, pronounced declines of entire commu-
nities across widely varying ecosystems. Although the overall
magnitude of change is evident, there remains uncertainty about
trajectories of individual tuna and billfish species. Assessments of
these species are continually improved by the international manage-
ment agencies. However, most scientists and managers may not be
aware of the true magnitude of change in marine ecosystems,
because the majority of declines occurred during the first years of
exploitation, typically before surveys were undertaken. Manage-
ment schemes are usually implemented well after industrialized
fishing has begun, and only serve to stabilize fish biomass at low
levels. Supporting evidence for these conclusions comes from the
United Nations Food and Agriculture Organization (FAO) data set,
which indicates declining global catches
22
and a consistent decline
in the mean trophic level of the catch
23
, which is a result of removing
predatory fishes. Furthermore, on seamounts and on continental
slopes, where virgin communities are fished, similar dynamics of
extremely high catch rates are observed, which decline rapidly over
the first 3–5 years of exploitation
24
. We suggest that this pattern is
not unique to these communities, but simply a universal feature of
the early exploitation of ecosystems.
Our results have several important management implications.
First, we need to consider potential ecosystem effects of removing
90% of large predators. Fishery-induced top–down effects are
evident in coastal
5
and shelf
25
ecosystems, but little empirical
information is available from the open oceans. This is worrisome,
as any ecosystem-wide effect is bound to be widespread, and
possibly difficult to reverse, because of the global scale of the
declines (Fig. 2). Another serious problem in heavily depleted
communities is the extinction of populations, particularly those
with high ages of maturity
26
. Local extinctions can go unnoticed
even in closely monitored systems such as the northwest Atlantic
27
,
let alone in the open ocean. Finally, the reduction of fish biomass to
low levels may compromise the sustainability of fishing, and
support only relatively low economic yields
3
. Such concerns have
motivated a recent UN resolution to restore fish stocks to healthy
Figure 3 Compensation in exploited fish communities. a, Oceanic billfish community in
the tropical Atlantic, showing the catch per 100 hooks (c.p.h.h.) of blue marlin (Makaira
nigricans; solid circles, solid line), sailfish (Istiophorus platypterus; open triangles, dashed
line) and swordfish (Xiphias gladius; open circles, dotted line). b, Demersal fish
community on the Southern Grand Banks, showing the biomass of codfishes (Gadidae;
solid circles, solid line) and flatfishes (Pleuronectidae; open circles, dotted line). Lines
represent best fits using a local regression smoother.
letters to nature
NATURE| VOL 423 | 15 MAY 2003 | www.nature.com/nature282 © 2003 Nature Publishing Group
levels
4
. Our analysis shows that it is appropriate and necessary to
attempt restoration on a global scale, and provides a benchmark
against which community recovery could be assessed. A
Methods
Data selection
For shelf communities, we compiled data from research trawl surveys from the Southern
Grand Banks (43–468N, 49–538W) and Saint Pierre Banks (45–478N, 55–588W) (ref. 28),
the Gulf of Thailand (9–148N, 100–1058W) (ref. 29) and South Georgia (53–568S,
35–408W) (ref. 14). All other trawl data sets that we considered (for example, North Sea,
Georges Bank and Alaska) did not capture the beginning of industrialized exploitation. We
included only demersal predators; pelagic species, which were not well sampled by the
trawl gear, were excluded. Longlining data obtained from the Japanese Fishery Agency
were divided into temperate (Atlantic, 40–458S; Indian, 35–458S; Pacific, 30–458S),
subtropical (Atlantic, 10–408S; Indian, 10–358S; Pacific, 15–308S) and tropical
communities (Atlantic, 208N–108S; Indian, 158N–158S; Pacific, 10–158S). These
divisions were based on their dominant species: yellowfin (T. albacares), albacore
(T. alalunga) or southern bluefin tuna (T. maccoyii), respectively, and excluded areas
previously fished by the Japanese, Spanish and US fleets. Running the models with
alternative divisions (^58) did not change the results significantly. The catch rates in each
community were determined as the sum of the catches divided by the sum of the effort in
each region in each year. Years with very low effort (,20,000 hooks for the entire region)
were excluded. Alternative treatment of the data, including removing seasonal effects and
taking the average catch rates over 58£58squares, had little effect on the results. For
longlines, we assume that the catch rate is an approximate measure of relative biomass,
which is probably conservative because the average individual weights of fish in exploited
populations tend to decline over time. Our data capture the abundance of larger fishes that
are vulnerable to baited hooks and bottom trawls, respectively.Many smaller species have
low catchabilities and are not recorded reliably by these methods. Changes in the longline
fishery occurred around 1980 when the fishery began to expand into deeper regions;
however, this was only after the declines in biomass were observed. For more details on
species composition, data treatment and interpretation of trends, refer to the
Supplementary Information.
Data analysis
Our model (equation (1)) assumes that for each community, i, the rate of decline to
equilibrium is exponential with rate r
i
from a pre-exploitation biomass N
i
(0), where t¼0
is the first year of industrialized fishing. Exploitation continues until equilibrium is
approached, where a residual proportion, d
i
, of the biomass remains. We fit this model
separately to each community under the assumption of a lognormal error distribution
using nonlinear regression (Procedure NLIN in SAS, version 8). We also used nonlinear
mixed-effects models
13
to determine whether the patterns were similar across
communities. Mixed-effect models were fitted by maximizing the likelihood integrated
over the random effects using adaptive gaussian quadrature (Procedure NLMIXED in
SAS). Toaccount for the fact that biomass was recorded in different units (kilotonnes (kt),
catch rates), the initial biomass, N
i
(0), was assumed to be a fixed effect for each
community with appropriate units. For South Georgia, N
i
(0) was fixed at the first biomass
estimate to capture the high initial rate of decline. This first estimate (750 kt; ref. 14) was
considered to be realistic because it was very close to the sum of total removals (514 kt;
ref. 30) plus the residual biomass estimate (160 kt; ref. 14) after the first 2 years of fishing.
Autocorrelation in the residuals of some time series may cause the standard errors to be
underestimated. The results were robust to alternative error assumptions (separate error
variances for the time series and alternative error distributions); for example, under the
assumption of normal errors, the rate of decline was 13.9% and residual biomass was
10.9%, respectively.
Received 25 November 2002; accepted 25 March 2003; doi:10.1038/nature01610.
1. Pauly, D. & Christensen, V. Primary production required to sustain global fisheries. Nature 374,
255–257 (1995).
2. Tegner, M. J. & Dayton, P. K. Ecosystem effects of fishing. Trends Ecol. Evol. 14, 261–262 (1999).
3. Pauly, D. et al. Towards sustainability in world fisheries. Nature 418, 689–695 (2002).
4. United Nations. World Summit on Sustainable Development: Plan of Implementation khttp://
www.johannanesburgsummit.org/html/documents/summit_docs/2309_planfinal.html(United
Nations, New York, 2002).
5. Jackson, J. B. C. et al. Historical overfishing and the recent collapse of coastal ecosystems. Science 293,
629–638 (2001).
6. Steele, J. H. & Schumacher, M. Ecosystem structure before fishing. Fish. Res. 44, 201–205 (2000).
7. Worm, B., Lotze, H. K., Hillebrand, H. & Sommer, U. Consumer versus resource control of species
diversity and ecosystem functioning. Nature 417, 848–851 (2002).
8. Pauly, D. Anecdotes and the shifting baseline syndrome of fisheries. Trends Ecol. Evol. 10, 430
(1995).
9. Field, C. B., Behrenfeld, M. J., Randerson, J. T. & Falkowski, P. Primary production of the biosphere:
integrating terrestrial and oceanic components. Science 281, 237–240 (1998).
10. Myers, R. A., Hutchings, J. A. & Barrowman, N. J. Why do fish stocks collapse? The example of cod in
Atlantic Canada. Ecol. Appl. 7, 91–106 (1997).
11. Bullis, H. R. Preliminary report on exploratory longline fishing for tuna in the Gulf of Mexico and the
Caribbean Sea. Comm. Fish. Rev. 17, 1–15 (1955).
12. Shomura, R. S. & Murphy,G. I. Long line Fishery for Deep-Swimming Tunas in the Central Pacific, 1953.
US Fish and Wildlife Service Special Scientific Report: Fisheries No.157 (USFWS, Washington, 1955).
13. Davidian, M. & Giltinan, D. M. Nonlinear Models for Repeated Measurement Data (Chapman & Hall,
New York, 1995).
14. Kock, K.-H. & Shimadzu, Y. in Southern Ocean Ecology: The BIOMASS Perspective (ed. El-Sayed, S. Z.)
287–312 (Cambridge Univ. Press, Cambridge, 1994).
15. Hutchings, J. A. & Myers, R. A. in The North Atlantic Fisheries: Successes, Failures and Challenges
(eds Arnason, R. & Felt, L.) 38–93 (The Institute of Island Studies, Charlottetown, Prince Edward
Island, Canada, 1995).
16. Rothschild, B. J. Competition for gear in a multiple-species fishery. J. Cons. Int. Explor. Mer. 31,
102–110 (1967).
17. Lyne, V., Parslow, J., Young, J., Pearce,A. & Lynch, M. Development, Application and Evaluation of the
Use of Remote Sensing Data by Australian Fisheries (CSIRO Marine Research, Hobart, Australia, 2000).
18. Caton, A. E. A review of aspects of southern bluefin tuna biology, population and fisheries. Inter-Am.
Trop. Tuna Comm. Spec. Rep. 7, 181–350 (1991).
19. Stevens, J. D., Bonfil, R., Dulvy, N. K. & Walker, P. A. The effects of fishing on sharks, rays, and
chimaeras (chondrichthyans), and the implications for marine ecosystems. ICES J. Mar. Sci. 57,
476–494 (2000).
20. Baum, J. K. et al. Collapse and conservation of shark populations in the Northwest Atlantic. Science
299, 389–392 (2003).
21. Fogarty, M. J. & Murawski, S. A. Large-scale disturbance and the structure of marine systems: Fishery
impacts on Georges Bank. Ecol. Appl. 8(suppl.), 6–22 (1998).
22. Watson, R. & Pauly, D. Systematic distortions in world fisheries catch trends. Nature 414, 534–536
(2001).
23. Pauly, D., Christensen, V., Dalsgaard, J., Froese, R. & Torres, F. Jr Fishing down marine food webs.
Science 279, 860–863 (1998).
24. Roberts, C. M. Deep impact: the rising toll of fishing in the deep sea. Trends Ecol. Evol. 242, 242–245
(2002).
25. Worm, B. & Myers, R. A. Meta-analysis of cod–shrimp interactions reveals top–down control in
oceanic food webs. Ecology 84, 162–173 (2003).
26. Myers, R. A. & Mertz, G. The limits of exploitation: a precautionary approach. Ecol. Appl. 8(suppl.),
165–169 (1998).
27. Casey, J. M. & Myers, R. A. Near extinction of a large, widely distributed fish. Science 281, 690–692
(1998).
28. Casey, J. M. Fish Assemblages on the Grand Banks of Newfoundland. Thesis, Memorial Univ.
Newfoundland, Canada (2000).
29. Pauly, D. in Fish Population Dynamics (ed. Gulland, J. A.) 329–348 (Wiley, New York, 1988).
30. Commission for the Conservation of Antarctic Marine Living Resources Statistical Bulletin Vo l . 1 4
(CCAMLR, Hobart, Australia, 2002).
Supplementary Information accompanies the paper on www.nature.com/nature.
Acknowledgements We thank J. Casey, A. Fonteneau, S. Hall, J. Hampton, S. Harley, J. Ianelli,
I. Jonsen, J. Kitchell, K.-H. Kock,H. Lotze, M. Maunder, T.Nishida, M. Prager, T. Quinn, G. Scott
and P. Ward for data, comments and suggestions, N. Barrowman and W.Blanchard for statistical
advice, and D. Swan for technical assistance. This research is part of a larger project on pelagic
longlining initiated and supported by the Pew Charitable Trusts.Further support was provided by
the Deutsche Forschungsgemeinschaft and the National Sciences and Engineering Research
Council of Canada.
Competing interests statement The authors declare that they have no competing financial
interests.
Correspondence and requests for materials should be addressed to R.A.M.
(Ransom.Myers@dal.ca).
..............................................................
Attractor dynamics of network
UP states in the neocortex
Rosa Cossart, Dmitriy Aronov & Rafael Yuste
Department of Biological Sciences, Columbia University, New York,
New York 10027, USA
.............................................................................................................................................................................
The cerebral cortex receives input from lower brain regions, and
its function is traditionally considered to be processing that input
through successive stages to reach an appropriate output
1,2
.
However, the cortical circuit contains many interconnections,
including those feeding back from higher centres
3–6
,andis
continuously active even in the absence of sensory inputs
7–9
.
Such spontaneous firing has a structure that reflects the coordi-
nated activity of specific groups of neurons
10–12
. Moreover, the
membrane potential of cortical neurons fluctuates spontaneously
between a resting (DOWN) and a depolarized (UP) state
11,13–16
,
which may also be coordinated. The elevated firing rate in the UP
state follows sensory stimulation
16
and provides a substrate for
letters to nature
NATURE| VOL 423 | 15 MAY 2003 | www.nature.com/nature 283
© 2003 Nature Publishing Group
