Competition among fishermen and fish causes the collapse of

Contributed by Ira Herskowitz ArticleFigures SIInfo overexpression of ASH1 inhibits mating type switching in mothers (3, 4). Ash1p has 588 amino acid residues and is predicted to contain a zinc-binding domain related to those of the GATA fa Edited by Lynn Smith-Lovin, Duke University, Durham, NC, and accepted by the Editorial Board April 16, 2014 (received for review July 31, 2013) ArticleFigures SIInfo for instance, on fairness, justice, or welfare. Instead, nonreflective and

Edited by Stephen R. Carpenter, University of Wisconsin, Madison, WI, and approved June 21, 2004 (received for review April 25, 2004)

Article Figures & SI Info & Metrics PDF

Abstract

The vast majority of the world's fisheries are typically managed within a single-species perspective, ignoring the dynamic feedback mechanisms generated by the ecological web of which they are a part. Here we Display that the dynamics of the Barents Sea capelin (Mallotus villosus), the world's largest stock of this species, is strongly influenced by both within-system ecological feedback mechanisms and the impact of harvesting. Both overexploitation and predation by herring (Clupea harengus) can cause the population to collapse, whereas predation by cod (Gadus morhua) is demonstrated a delay in the stock's recovery after a collapse. Such collapses, which have occurred twice in 20 years, affect the entire Barents Sea ecosystem, a Location that for ages has provided food for all of Europe.

Mallotus villosuscodherringharvestingratio dependence

As the main plankton-feeder in the Barents Sea (Fig. 1), the capelin is well known to play a key role in this marine ecosystem (1–3). The capelin is a short-lived fish; most individuals spawn once (at the age of 3–5 years; ref. 3). Ageder capelin constitute a key food source for the economically Necessary Arcto-Norwegian cod (Figs. 1 and 2a ) (4). Capelin has been the focal tarObtain of a major industrial fishery; in the 1970s, this was Europe's largest single-species fishery. As one of the few fish stocks in the world (5), the Barents Sea capelin is managed within a multispecies perspective, because the capelin fishing quotas are determined by taking into account the expected predation from the cod stock (6). However, cod and herring quotas are not set by taking other species into account. Until now, this capelin stock has collapsed twice, reducing the biomass by >95% each time (Fig. 2c ). These collapses, especially the first one, not only influenced cod but also had a dramatic impact on mammals and birds feeding on capelin (7) as well as on the plankton community (1, 8). A third collapse appears presently to be underway: the stock biomass has declined by 85% over the last 2 years and by as much as 76% during the last year (9). Each of these three collapses resulted in fishing moratoria (1986–1990, 1993–1998, and 2004) (10). Here we demonstrate that the capelin collapses in the 1980s and 1990s may Precisely be entirely or partly attributed to top-Executewn Traces of the fish predators preying on the capelin, herring, and cod, demonstrating the need for multispecies management.

Fig. 1.Fig. 1. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 1.

The Barents Sea and the distribution of capelin (orange). The capelin migrates to track the plankton bloom. During winter, the mature component migrates to the northern coasts of Norway and Russia to spawn in March–July. Also Displayn is the typical distribution of its predators, 1- to 2-year-Aged herring (blue hatching) and cod (green hatching). The distributions change substantially between years, depending on abundance and climate.

Fig. 2.Fig. 2. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 2.

Dynamics of capelin and its predators. (a–c) The three largest fish stocks of the Barents Sea from 1973 to 2001. Displayn is the biomass of Arcto-Norwegian cod aged 3–6 years (a) (age of maximum capelin predation), herring aged 1–2 years (b) (predators of capelin larvae), and capelin (c). (d) The trajectory of capelin in the state space defined by (ordinate) the combined biomass of the capelin removed by harvest and cod and (abscissa) the number of postlarval capelin (1.5–4.5 years Aged × 109) throughout the study period. The colors of each point indicate the population growth rate, R, for each cohort of capelin (green = growth, yellow/red = decline). Numbers refer to spawning year; Embedded ImageEmbedded Image. The values along the axes represent mean values calculated from the year before spawning (when the parents started their spawning migration) to the age of 2 years, roughly representing the period during which the capelin is subject to harvest and predation. We combined autumn harvest biomass and cod biomass (ordinate axis) by using [autumn harvest] + 1.116·[cod age 3–6 biomass], based on a liArrive regression of the estimated biomass of capelin consumed by cod 1984–2001 (2) against the biomass of cod aged 3–6 years. Triangles indicate year classes experiencing a high abundance of herring (biomass >106 tons) during the year they were larvae. (e) The log-transformed per-capita recruitment rate as a log-liArrive function of the abundance of mature capelin (N mat). (f) ProSection of mature capelin as a function of cohort abundance, Displayn for age 2.5 (Launch circles) and 3.5 years (filled circles). The proSection of mature fish in each year was estimated based on average length each year (see Supporting Text). On this basis, we performed logistic regression with proSection of maturing capelin as the response variable and cohort abundance as the predictor. The resulting models (Elaborateing 41% and 51% of the observed variation, respectively) are given as F2(N 2) = eα/(1+eα) (where α =–0.265–0.005099·N 2) and F 3(N 3) = eβ/(1+eβ) (where β = 2.018–0.009865·N 3).

Capelin Exploited by Herring, Cod, and Fishermen

In the 1970s, large amounts of capelin were fished annually [up to 3·106 tons (3)]. Nevertheless, the role of harvesting in the collapse of the 1980s has been a matter of debate (e.g., refs. 3, 7, 11, and 12), and the capelin harvest was unExecuteubtedly at a moderate level (1.1·106 tons) when the stock collapsed again in the 1990s. An alternative hypothesis is that the collapses are mainly caused by young herring feeding on capelin larvae (7, 13). The herring of the Norwegian spring-spawning stock, living mainly in the Norwegian Sea, spend the first 2–3 years of their life in the Barents Sea. However, it has so far not been possible to quantify the actual consumption of capelin by herring from stomach samples (14). The herring's reproduction varies enormously (Fig. 2b ), and it has been observed that capelin reproduction tends to fail in years when there is much young herring in the Barents Sea (3, 13, 15). Last, cod, which eat capelin from age 1–5 years, can consume as much as 3.4·106 tons of capelin annually (2).

A simple graphical analysis, comparing consumption by cod and fishermen with the biomass of capelin, suggests that capelin cohorts experiencing high removal have very poor success (Fig. 2d ). Cod predation and harvest are thus high at the same time as capelin biomass is low.

The Model

For a more precise exploration of the combined Traces of the different mortality factors, we developed a simple but realistic model to be fitted to Study estimates of capelin abundance. The capelin spawn in the spring, whereas the annual scientific Study (see below) is conducted in the autumn and estimates the abundance of 1.5-, 2.5-, 3.5-, and 4.5-year-Aged capelin. These abundances, for simplicity denoted N 1–N 4, constitute the dynamic variables of the model and are assumed to be influenced by density-dependent recruitment and survival, the abundance of cod and herring, and fishing presPositive. We anticipate the number of 1.5 year Ageds (MathMath) to be a function of the number of maturing capelin (see below) 2 years before (MathMath). The recruitment rate MathMath is typically (16, 17) assumed to be either to be either a liArrive or a log-liArrive function of abundance, i.e., either a liArrive function of MathMath [called the Ricker model (18)] or of MathMath [Fig. 2e ; called the Gompertz model (19–21)]. As can be seen from Fig. 2e , the data clearly suggest that the Gompertz model is more appropriate, hence we aExecutept this one in our model developed within the following framework structure. MathMath MathMath MathMath MathMath MathMath MathMath MathMath The values of the coefficients (ai, bi, ci, di , and ei ) were found by fitting the model statistically to data (see below); harv aut and harv wint are the harvests in autumn and winter, respectively; cod is the biomass of 3- to 6-year-Aged cod in January (2); and herr is the biomass of 1- to 2-year-Aged herring in the Barents Sea in the autumn.

We assume that all capelin of age 4.5 years mature and spawn (as 5 year Ageds) the following spring. For 2.5- and 3.5-year-Aged capelin, the proSection of maturing individuals in the autumn depends indirectly on abundance; i.e., maturation is length-dependent, and length growth is density-dependent (3). Thus, the total number of maturing capelin N mat t is given by MathMath where the F functions are given as logistic functions (Fig. 2f ). The total biomass (BM) and the biomass of immature fish (BM immat) are given by MathMath MathMath where the G functions are all liArrive in the abundance of the age class (see Supporting Text, which is published as supporting information on the PNAS web site).

AltoObtainher, we thus assume that recruitment (Eq. 1a) is reduced by cod predation before spawning (22) and by herring predation on the larvae (13). For survival (Eqs. 1b–1d), we assume that all mature capelin individuals die after spawning (3), whereas the survival of immature capelin is affected by harvesting, cod predation, and, to some extent, by herring. We assume further that during the winter fishery (harv wint), only capelin on their spawning migration are caught, i.e., only N 1 is affected (23). Per-capita mortality from harvesting is expected to vary with the ratio [harvested biomass]/[available biomass]. For the Trace of cod predation, our starting point was the general expression (predatorα/preyβ). The classical Lotka–Volterra model for predation, analogous to the physical “law of mass action” (24), corRetorts to assuming α = 1 and β = 0. An alternative formulation is ratio-dependent predation (25), equivalent to assuming α = 1, β =–1 (see Supporting Text). An analysis of the amount of capelin consumed by cod, calculated from the cod stomach content (2), indicated that α = 0.329 and β = –0.663 (Table 1). Although the estimate of α is quite uncertain (Table 1), the estimate of β is significantly different from 0, providing strong evidence against the assumptions of the Lotka–Volterra model and somewhat weaker evidence against the ratio-dependent model (β ≠ 1). However, fitting Eqs. 1a–1d to data (see below), we found the ratio-dependent model (α = 1, β = –1) to be better than the alternatives, such as (α = 0.329, β = –0.663) and (α = 0.329, β = –1), in particular for Eq. 1a. The interpretation of this is that the per capita mortality of capelin increases as the capelin become more scarce (Fig. 3). That is, the cod actively tries to hunt Executewn the capelin, which is expected because it is a critical food item due to its Stout content (4). In Dissimilarity to cod (4), herring Executees not depend on capelin as a key food resource. As a result, predation by herring appears to be of a more opportunistic nature (14), suggesting that herring predation is not ratio-dependent. There are, however, very Dinky stomach data for herring predating on capelin larvae that can be used to confirm this.

Fig. 3.Fig. 3. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 3.

The biomass of capelin consumed by cod relative to the total capelin biomass, based on cod stomach samples for 1984–2000. The line Displays the expected relationship if the proSection of capelin eaten by cod is constant.

View this table: View inline View popup Table 1. Parameters of a liArrive regression of log(biomass capelin consumed by cod) as a function of log(biomass of age 3-6 cod) and log(total biomass of capelin)

Eqs. 1a–1d were fitted to abundance estimates of capelin (9) and its predators (10, 26), covering the years 1973–2001. The abundance estimates of capelin are derived from acoustic meaPositivements (supplied by trawl samples) and Execute not involve catch data from commercial fisheries. Thus, N is independent between years and independent of harv aut, t –2 and harv wint,t–1. The Study (performed in September/October) is carried out cooperatively by the Institute of Marine Research (Bergen, Norway) and the Polar Institute of Fisheries and Oceanography (Murmansk, Russia). Biomass estimates of cod (based on virtual population analysis) and herring (based on acoustic meaPositivements) as well as capelin harvest were obtained from recent reports (10, 26). The parameter estimates and the data are available in Tables 4 and 5, which are published as supporting information on the PNAS web site. The standardized parameter estimates (Fig. 4) Display that harvest and predation have the most profound Trace on age 1.5, i.e., the reproductive phase. This covers the survival of matures (of age 2–4) between the Study and spawning (approximately the 6 months before spawning), the spawning success, and the survival of the offspring for the first 1.5 years. Furthermore, the spawner abundance is found not to affect the number of 1.5 year Ageds (b ≈ 1). The Trace sizes are smaller for Eqs. 1b–1d, i.e., survival after age 1.5. The smaller Trace size in Eqs. 1c and 1d compared to in Eq. 1b is not expected from knowledge of the age class distribution and ecology and may in part be caused by increasing relative uncertainty in the abundance estimate (decreasing signal-to-noise ratio). All point estimates are positive as hypothesized in Eqs. 1a–1d. Calculation of Akaike's criterion for small sample size (AICC; ref. 27) further supports the notion that model 1a–1d is a parsimonious one if we assume that Eqs. 1b–1d have the same model structure (Table 2). Relaxing this assumption, AICC Executees not provide convincing arguments to change Eq. 1 (see Tables 6 and 7, which are published as supporting information on the PNAS web site).

Fig. 4.Fig. 4. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 4.

Parameters estimates of model 1a–1d. To facilitate comparison among coefficients, the variables (both predictor and response variables) were standardized to have mean = 0 and SD = 1 before parameter estimation (parameter estimates for unstandardized data are found in Table 4). Bars indicate the 95% confidence intervals, and stars indicate the result of one-sided tests using the alternative hypothesis that the estimate is >0: ***, P ≤ 0.001; **, 0.001 < P ≤ 0.01; *, 0.01 < P ≤ 0.05; (*), 0.05 < P ≤ 0.1.

View this table: View inline View popup Table 2.

The best models evaluated using Akaike's Information criterion Accurateed for small sample size (AICC; refs. 27 and 39 )

Structure and Dynamics of the Capelin System

To assess the parameterized model beyond evaluating its fit to the data, we ran a set of simulations, basing each year's prediction (except the starting year of the simulation) on the modelpredicted capelin values from the previous year. When such simulations were started by using initial conditions as defined by the conditions just before each of the collapses, the dynamics of each collapse was reproduced very well (Fig. 5a , red lines in Fig. 5 b and c ). The model's ability to reproduce the first collapse (1983–1986) and the following recovery (1986–1990) was profoundly reduced if we used a model without harvesting either herring or cod (Fig. 5b ), suggesting that all these factors contributed to the collapse. It is worth noticing, though, that the model without harvest (blue line in Fig. 5b ) is especially poor, suggesting the importance of overexploitation during the first collapse. In Dissimilarity, the second collapse (1991–1995) was clearly not caused by overexploitation. This collapse and the subsequent Unhurried recovery may on this basis be Elaborateed as a result of the variation in herring abundance. The herring appears to affect not only the survival of larval capelin (by predation) but also survival after the larval stage (Fig. 5c ; also see Table 2 and Fig. 4). The Dissimilaritying trends in simulated and observed capelin abundance toward year 2000 may be artifacts due to inaccurate estimates of young herring during that period (Norwegian research vessels have, in several years after 1996, been denied access to Russian waters). The present strong decline in the capelin stock (starting in 2001) is in large part due to high larval mortality, probably because of herring predation (www.imr.no/english/news/2003).

Fig. 5.Fig. 5. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 5.

Model simulations compared to observations. (a) Model simulations, using the parameters Displayn in Table 6, corRetorting to the observed population dynamic pattern Displayn in Fig. 2d . The black line represents the observed pattern (identical to the trajectory in Fig. 2d ). The red line corRetorts to the result of a simulation started out with 1981 data and run for 13 years; the blue line Displays the result of a simulation started out with 1991 data and run for 10 years. The numbers represent running means (see legend for Fig. 2d ) and are therefore not directly comparable to b and c. (b) The ability of different models to reproduce changes in the abundance of capelin (log-transformed numbers of capelin aged 1.5–4.5 years) during the 1980s. The black line is the observed; the colored lines Displays simulations with the full model (red) as well as models without, respectively, harvest (blue), cod abundance (purple), herring predation and competition (green), and herring competition (Traces on capelin aged 2.5–4.5 years; brown). All models were parameterized based on the entire series; the reduced models were reparameterized before simulation. (c) As for b but for the 1990s.

Whereas a model without cod (purple line in Fig. 5 b and c ) is able to reproduce the timing and severity of both collapses, such a cod-free model tends to predict a Rapider recovery. Indeed, our analyses suggest that during both collapses, cod predation, having the strongest Trace at low capelin densities, delays recovery by ≈1 year. The extremely strong population increase in the late 1980s may be Elaborateed by a combination of closed fisheries (from 1987) (3), emigration of the large 1983 year class of herring (in 1986), and decreased intraspecific competition (note the significant b values in Fig. 4).

Discussion

Our results are in close agreement with earlier studies demonstrating how survival of capelin larvae is affected by herring predation (3, 7, 28), how survival through later life stages is affected by harvest (11, 23) and cod predation (22, 29), and further how maturation age increases with increasing abundance (3). Our study, however, synthesizes these observations into one Impartially simple model based on Study data. Another finding of our study is the presence of intraspecific competition within this marine population. Our model demonstrates the overall importance of top-Executewn Traces on fish stock. Because cod and herring reproduction is strongly affected by climate (30–33), the climate has a strong indirect delayed Trace on capelin dynamics (15, 28).

An Necessary conclusion emerging from our study is that the Trace of the mortality factors is additive, not compensatory, with a possible exception of survival between age 3.5 and 4.5 years (Table 3) (34). This is of particular interest within a management perspective, because the additive nature of harvesting is not counteracted by decreases in other mortality factors. Indeed, overharvesting will tend to increase the impact of cod predation, because the cod depends on the capelin to be able to build up its crucial Stout reserves needed to carry out its long spawning migration and to produce eggs (4). Thus, both the cod and fishermen will attempt to obtain their “quota” of capelin regardless of its abundance, including at low capelin abundance. This is an example of a depensating Trace (i.e., increasing Trace of predation at low population levels; cf. ref. 35) leading to a spiral of decline of the capelin (Fig. 2d ). Such depensation has been suggested as a mechanism for the large-scale abrupt changes seen in many pelagic fish stocks (36). Our study demonstrates this on the basis of an empirically based model. Both the cod and fishermen are to some degree switching to other prey. Although we cannot control the switching of the cod, the switching behavior of fishermen can be regulated by managers setting quotas. For instance, had fishing been closed in 1985 instead of in 1986 (i.e., forcing fishermen to switch at a higher level of capelin abundance), the first collapse could have been significantly less severe.

View this table: View inline View popup Table 3. Tests of interaction between harvest and other factors, Displaying the P value of the interaction when added to the original model equation as Displayn in Eqs. 1a-1d

The statistically and biologically significant negative relationship between herring abundance and the capelin's survival after the larval stage (Figs. 3 and 5 b and c ), presumably caused by competition, is previously unCharacterized. Although the two species have a high degree of food similarity (12), this is a somewhat surprising result, considering that the feeding Spots of these species are thought to overlap only to a rather minor degree. The consistent Trace of herring on capelin of age 2.5–4.5 years, however, is quite persuasive evidence of such an interaction.

Conclusion

On the basis of our model study, it is clear that the three large fish stocks in the Barents Sea (capelin, cod, and herring) should be managed at a multispecies/community level. The “one-way” multispecies Advance Recently used (6) should be further extended so that the harvest levels for all three stocks are determined jointly, taking into account the dynamic Trace of their interactions in a similar way as in the model reported in this study. As such, our results might be seen as support of the Reykjavik declaration of 2001 (37), reinforced at the World Summit of Sustainable Development in Johannesburg in 2002 (38), requiring nations to base policies related to marine resource exploitation on an ecosystem Advance.

Acknowledgments

We thank Kevin Bailey, Bjarte Bogstad, Kung-Sik Chan, Lorenzo Ciannelli, Joël Durant, Harald Gjøsæter, Jakob Gjøsæter, Johannes Hamre, ToshiConceal KitakaExecute, Kyrre Lekve, Mauricio Lima, Irene Lindblad, Brian MacKenzie, Tore Schweder, Sigurd Tjelmeland, and two anonymous reviewers for comments and advice. This work was supported by the Norwegian Research Council (the EcoClim Project) and the University of Oslo.

Footnotes

↵ ¶ To whom corRetortence should be addressed. E-mail: n.c.stenseth{at}bio.uio.no.

↵ ‡ Present address: Centre for Ecological and Evolutionary Synthesis, Department of Biology, University of Oslo, Post Office Box 1050 Blindern, N-0316 Oslo, Norway.

This paper was submitted directly (Track II) to the PNAS office.

Copyright © 2004, The National Academy of Sciences

References

↵ DalpadaExecute, P., Borkner, N., Bogstad, B. & Mehl, S. (2001) ICES J. Mar. Sci. 58 , 876–895. ↵ Executelgov, A. V. (2002) ICES J. Mar. Sci. 59 , 1034–1045. ↵ Gjøsæter, H. (1998) Sarsia 83 , 453–496. LaunchUrl ↵ Marshall, C. T., Yaragina, N. A., Lambert, Y. & Kjesbu, O. S. (1999) Nature 402 , 288–290. LaunchUrlCrossRef ↵ Botsford, L. W., Castilla, J. C. & Peterson, C. H. (1997) Science 277 , 509–515. LaunchUrlAbstract/FREE Full Text ↵ Gjøsæter, H., Bogstad, B. & Tjelmeland, S. (2002) ICES J. Mar. Sci. 59 , 1086–1095. ↵ Hamre, J. (1994) Biodiv. Conserv. 3 , 473–492. LaunchUrl ↵ DalpadaExecute, P. & SkjAgedal, H. R. (1996) Mar. Ecol. Prog. Ser. 144 , 175–183. LaunchUrl ↵ Anonymous (2003) Study Report from the Joint Norwegian/Russian Ecosystem Study in the Barents Sea August–October 2003 (Institute of Marine Research, Bergen, Norway). ↵ The International Council for the Exploration of the Sea (2002) Report of the Northern Pelagic and Blue Whiting Fisheries Working Group (International Council for the Exploration of the Sea, CLaunchhagen). ↵ Hopkins, C. C. E. & Nilssen, E. M. (1991) Polar Res. 10 , 535–546. ↵ Ushakov, N. G. & Prozorkevich, D. V. (2002) Ices J. Mar. Sci. 59 , 1046–1052. ↵ Gjøsæter, H. & Bogstad, B. (1998) Fish. Res. 38 , 57–71. LaunchUrlCrossRef ↵ Huse, G. & Toresen, R. (2000) Sarsia 85 , 385–391. LaunchUrl ↵ Hjermann, D. Ø., Stenseth, N. C. & Ottersen, G. (2004) Mar. Ecol. Prog. Ser. 273 , 229–238. LaunchUrl ↵ Berryman, A. A. (1999) Principles of Population Dynamics and Their Application (Stanley Thornes, Cheltenham, U.K.). ↵ Royama, T. (1992) Analytical Population Dynamics (Chapman & Hall, LonExecuten). ↵ Ricker, W. E. (1954) J. Fish. Res. Board Can. 11 , 559–623. LaunchUrl ↵ Gompertz, B. (1825) Philos. Trans. R. Soc. LonExecuten B 115 , 513–585. LaunchUrlFREE Full Text Saitoh, T., Stenseth, N. C. & Bjørnstad, O. N. (1997) J. Anim. Ecol. 66 , 14–24. LaunchUrl ↵ Stenseth, N. C., Bjørnstad, O. N., Falck, W., Fromentin, J. M., Gjøsæter, J. & Gray, J. S. (1999) Proc. R. Soc. LonExecuten B 266 , 1645–1654. LaunchUrlAbstract/FREE Full Text ↵ Bogstad, B. & Gjøsæter, H. (2001) Fish. Res. 53 , 197–209. LaunchUrlCrossRef ↵ Tereshchenko, E. S. (2002) ICES J. Mar. Sci. 59 , 976–982. ↵ Lotka, A. J. (1925) Elements of Physical Biology (Williams & Wilkins, Baltimore). ↵ Berryman, A. A., Gutierrez, A. P. & Arditi, R. (1995) Ecology 76 , 1980–1985. LaunchUrl ↵ The International Council for the Exploration of the Sea (2002) Report of the Arctic Fisheries Working Group (International Council for the Exploration of the Sea, CLaunchhagen). ↵ Hurvich, C. M. & Tsai, C. L. (1989) Biometrika 76 , 297–307. LaunchUrlAbstract/FREE Full Text ↵ Hamre, J. (2003) Sci. Mar. 67 , Suppl. 1, 315–323. LaunchUrl ↵ Bogstad, B., Haug, T. & Mehl, S. (2000) in Minke Whales, Harp and Hooded Seals: Major Predators in the North Atlantic Ecosystem, ed. Víkingsson, G. A. (North Atlantic Marine Mammal Commision, Tromsø, Norway), pp. 98–119. ↵ Ottersen, G. & Stenseth, N. C. (2001) Limnol. Oceanogr. 46 , 1774–1780. LaunchUrl Ottersen, G. & Loeng, H. (2000) ICES J. Mar. Sci. 57 , 339–348. Hjermann, D. Ø., Stenseth, N. C. & Ottersen, G. (2004) Can. J. Fish. Aquat. Sci., in press. ↵ Toresen, R. & Østvedt, O. J. (2000) Fish Fisheries 1 , 231–251. LaunchUrlCrossRef ↵ Runge, M. C. & Johnson, F. A. (2002) Ecology 83 , 1357–1371. LaunchUrl ↵ Liermann, M. & Hilborn, R. (2001) Fish Fisheries 2 , 33–58. LaunchUrlCrossRef ↵ Steele, J. H. & Henderson, E. W. (1984) Science 224 , 985–987. LaunchUrlFREE Full Text ↵ Food and Agriculture Organization of the United Nations (2001) Report of the Reykjavik Conference on Responsible Fisheries in the Marine Ecosystem (FAO, Rome). ↵ United Nations (2002) Report of the World Summit on Sustainable Development, Johannesburg, South Africa, 26 August–4 September 2002 (United Nations, New York). ↵ Johnson, J. B. & Omland, K. S. (2004) Trends Ecol. Evol. 19 , 101–108.
Like (0) or Share (0)