Abstract

A composite quantitative index for the percentage of primary production required to sustain fisheries and the average trophic level of catch (%PPR–TL c ) was employed to develop ecosystem-based reference functions suitable for fisheries management. Established ecosystem models, characterized by pairs of %PPR–TL c , were classified as either sustainably exploited or ecosystem overfished, on the basis of the results of factorial correspondence analysis applied to selected ecological indices, and on information from various sources. Canonical discriminant analysis of these pairs was applied to establish the discriminant function to separate the two exploitation classes. Next, reference functions related to different probabilities of ecosystem overfishing were developed to obtain an operational framework for ecosystem-based fisheries management. Values of ecosystem-based maximum sustainable catches associated with different probabilities of belonging to a sustainable situation were calculated. Overall, results show that most current fishing scenarios entail high risks of ecosystem overfishing.

Introduction

The primary production required to sustain fisheries (PPR; typically expressed as t km −2 y −1 ) has been conceived as an ecological footprint that highlights the role of fishing, in channelling marine trophic flows towards human use ( Pauly and Christensen, 1995 ), and is routinely deduced by Ecopath modelling ( Christensen and Walters, 2004 ). The trophic level (TL) identifies the position of organisms in the food chain ( Lindeman, 1942 ; Odum and Heald, 1975 ). By convention, primary producers and detritus have TL = 1; values for other groups are determined using mass-balance models, gut content analysis, or isotope data ( Briand, 2000 ). The average trophic level of the catch (TL c ) reflects the strategy of a fishery in terms of foodweb components selected, and is calculated as the weighted average of TL of harvested species ( Christensen and Walters, 2004 ).

Tudela (2003) proposed the use of %PPR (relative PPR, calculated by dividing by the total primary production available) in combination with TL c as a quantitative ecosystem index to capture the effect of fisheries. The reason given was that, for a given %PPR, a fishery with a higher TL c would be intrinsically less disruptive than a fishery with a lower one, and for a given TL c , a lower %PPR would be less disruptive than one with a higher %PPR. Therefore, the sensitivity of a system to the effects of fishing depends on both indicators, and it cannot be deduced from either metric independently. Accordingly, Tudela (2003) defined a theoretical framework to relate the %PPR–TL c parameter space to concepts of sustainable exploitation and ecosystem overfishing (EO; Figure 1 ). The challenge then is to establish the appropriate limit, as well as target reference functions ( Caddy and Mahon, 1995 ) based on available ecological information.

Figure 1

General theoretical framework based on the composite index %PPR–TL c and ecosystem overfishing status (after Tudela, 2003 ).

Figure 1

General theoretical framework based on the composite index %PPR–TL c and ecosystem overfishing status (after Tudela, 2003 ).

Our objective here is to define a quantitative boundary for EO in this framework, based upon available information on %PPR–TL c pairs and evidence of disruptive exploitation for different marine ecosystem types. In addition, the framework is used to evaluate ecosystems for which direct evidence of EO status was lacking, and to calculate the ecosystem-based maximum sustainable catches (EMSC) related to the chosen reference functions.

Material and methods

Pairs of %PPR and TL c were compiled on the basis of published Ecopath models ( Christensen and Walters, 2004 ) for different areas and periods ( www.seaaroundus.org ). Models for upwelling regions were excluded, because of difficulties in distinguishing between effects of fishing and those caused by environmental factors. Because %PPR values for primary producers only and for primary producers and detritus combined are highly correlated and yield conceptually similar results, we restrict ourselves here to analyses for primary producers only. Each Ecopath model was characterized by a %PPR–TL c pair, and classified if possible either as a sustainably fished, non-disrupted ecosystem, or as an overfished, structurally and functionally degraded ecosystem. Ecosystems were given an overexploited status when cumulative fishing impacts manifested at least one overexploitation symptom, based on the EO definition proposed by Murawski (2000) . The classification was based on specific information obtained from published and unpublished sources, on personal communication with experts on the ecosystem concerned, and on previous analysis of Ecopath-derived indices of ecosystem development, sensu Odum ( Odum, 1969 ; Christensen, 1995b ; Christensen and Pauly, 1998 ; Table 1 ). Factorial correspondence analysis (FCA), based on Pearson correlation coefficients and the WARD hierarchical method as aggregation algorithm (statistical package SPAD, version 5.2), was applied to a selection of Ecopath-derived indices ( Table 1 ) from a data set comprising 27 models for different periods and six ecosystems. Table 2 lists the 49 ecosystem models included in the analysis that could be classified in terms of sustainability/overexploitation criteria, and those that could not be classified.

Table 2

Ecosystem models (listed by main ecosystem types) that (a) could, and (b) could not be classified according to ecosystem overfishing criteria ( Murawski, 2000 ; EO status: +: overexploited, −: not overexploited).

Ecosystem model 
(a) Classified  TL c %PPR EO Reference 
Temperate shelves and seas 
1 Faroe Islands (1961) 3.9 14.7 −  Guénette et al . (2001)  
2 Icelandic fisheries (1950) 3.4 1.9 −  Guénette et al . (2001)  
3 North Sea (1880) 3.8 0.3 −  Guénette et al . (2001)  
4 North Sea (1963) 3.9 62.6 FC database 
5 North Sea (1974) 3.9 62.0 FC database 
6 North Sea (1981) 3.4 21.8 Christensen (1995a) 
7 Newfoundland (1900) 3.5 27.2 − Heymans (2003) 
8 Newfoundland (1985–1987) 3.9 6.3 Heymans (2003) 
9 Newfoundland (1995–2000) 3.1 15.1 Heymans (2003) 
10 Norwegian and Barents Sea (1950) 3.6 11.7 −  Guénette et al . (2001)  
11 Eastern Bering Sea (1950s) 3.4 30.9 −  Trites et al . (1999)  
12 Eastern Bering Sea (1980s) 3.3 15.2  Trites et al . (1999)  
13 Northern British Columbia (1750) 3.5 4.0 −  Ainsworth et al . (2002)  
14 Northern British Columbia (1900) 3.3 23.3 −  Ainsworth et al . (2002)  
15 Northern British Columbia (1950) 3.4 12.3  Ainsworth et al . (2002)  
16 Northern British Columbia (2000) 3.3 9.8  Ainsworth et al . (2002)  
17 Northern Gulf of St Lawrence (1985–1987) 3.6 22.9 − FC database 
18 Lancaster Sound Region (1980s) 4.1 3.3 −  Guénette et al . (2001)  
19 Georgia Strait (1950) 3.3 7.0 −  Pauly et al . (1998)  
20 West Greenland shelf (1997) 3.2 20.2  Guénette et al . (2001)  
21 Scotian shelf (1980–1985) 3.7 8.0 − Pitcher and Cochrane (2002) 
22 South Catalan Sea (1994–2000) 3.1 43.9  Coll et al . (2004)  
23 Azores archipelago (1997) 3.8 0.3 −  Guénette et al . (2001)  
24 Cantabrian Sea (1994) 3.8 82.3 Sánchez and Olaso (2004) 
Tropical shelves and seas 
25 Gulf of Thailand (1963) 3.0 1.8 − Christensen (1998) 
26 Gulf of Thailand (1980s) 3.1 16.1 Christensen (1998) 
27 Gulf of Thailand (1993) 3.0 6.7 FC database 
28 Southwest coast of India (1994) 2.6 13.2  Silvestre et al . (2003)  
29 Southwest coast of India (1995) 2.6 10.7  Silvestre et al . (2003)  
30 Southwest coast of India (1996) 2.6 11.7  Silvestre et al . (2003)  
31 Southern shelf of Brazil (1975–1979) 3.6 27.7 Vasconcellos and Gasalla (2001) 
32 Southern shelf of Brazil (1990–1994) 3.6 27.6 Vasconcellos and Gasalla (2001) 
33 Southeastern shelf of Brazil (1977–1980) 2.8 33.1 Vasconcellos and Gasalla (2001) 
34 Southeastern shelf of Brazil (1990–1995) 3.1 52.7 Vasconcellos and Gasalla (2001) 
35 Venezuelan northeastern shelf (1980s) 2.8 20.7 Christensen and Pauly (1993) 
36 Gulf of Mexico continental shelf (1990s) 2.6 31.6 Christensen and Pauly (1993) 
37 Brunei Darussalam (1980) 3.2 7.4 − Christensen and Pauly (1993) 
38 Vietnam–China shelf (1980) 3.3 9.7 − Pauly and Christensen (1993) 
39 South China deep sea (1980s) 3.5 10.0 − Pauly and Christensen (1993) 
40 Hong Kong (1990s) 3.0 27.2  Pitcher et al . (2002)  
41 Bay of Bengal (1984–1986) 2.7 8.3  Silvestre et al . (2003)  
42 San Pedro Bay (1994–1995) 3.3 3.1 −  Silvestre et al . (2003)  
Coastal areas and coral reefs 
43 Bay of Revellata, Corsica (1998) 3.8 11.9 − Pinnegar (2000) 
44 Prince William Sound, Alaska (1994–1996) 4.1 4.5 − Okey and Pauly (1998) 
45 Coast of Western Gulf of Mexico (1990s) 3.4 89.5 Christensen and Pauly (1993) 
46 Gulf of Lingayen (1990s) 3.3 51.6 Pauly and Christensen (1993) 
47 Maputo Bay (1980s) 2.5 20.3 Christensen and Pauly (1993) 
48 San Miguel Bay (1992–1994) 3.0 14.7 Bundy and Pauly (2001) 
49 Boliano reef flat (1991) 2.2 2.8 Christensen and Pauly (1993) 
Ecosystem model 
(a) Classified  TL c %PPR EO Reference 
Temperate shelves and seas 
1 Faroe Islands (1961) 3.9 14.7 −  Guénette et al . (2001)  
2 Icelandic fisheries (1950) 3.4 1.9 −  Guénette et al . (2001)  
3 North Sea (1880) 3.8 0.3 −  Guénette et al . (2001)  
4 North Sea (1963) 3.9 62.6 FC database 
5 North Sea (1974) 3.9 62.0 FC database 
6 North Sea (1981) 3.4 21.8 Christensen (1995a) 
7 Newfoundland (1900) 3.5 27.2 − Heymans (2003) 
8 Newfoundland (1985–1987) 3.9 6.3 Heymans (2003) 
9 Newfoundland (1995–2000) 3.1 15.1 Heymans (2003) 
10 Norwegian and Barents Sea (1950) 3.6 11.7 −  Guénette et al . (2001)  
11 Eastern Bering Sea (1950s) 3.4 30.9 −  Trites et al . (1999)  
12 Eastern Bering Sea (1980s) 3.3 15.2  Trites et al . (1999)  
13 Northern British Columbia (1750) 3.5 4.0 −  Ainsworth et al . (2002)  
14 Northern British Columbia (1900) 3.3 23.3 −  Ainsworth et al . (2002)  
15 Northern British Columbia (1950) 3.4 12.3  Ainsworth et al . (2002)  
16 Northern British Columbia (2000) 3.3 9.8  Ainsworth et al . (2002)  
17 Northern Gulf of St Lawrence (1985–1987) 3.6 22.9 − FC database 
18 Lancaster Sound Region (1980s) 4.1 3.3 −  Guénette et al . (2001)  
19 Georgia Strait (1950) 3.3 7.0 −  Pauly et al . (1998)  
20 West Greenland shelf (1997) 3.2 20.2  Guénette et al . (2001)  
21 Scotian shelf (1980–1985) 3.7 8.0 − Pitcher and Cochrane (2002) 
22 South Catalan Sea (1994–2000) 3.1 43.9  Coll et al . (2004)  
23 Azores archipelago (1997) 3.8 0.3 −  Guénette et al . (2001)  
24 Cantabrian Sea (1994) 3.8 82.3 Sánchez and Olaso (2004) 
Tropical shelves and seas 
25 Gulf of Thailand (1963) 3.0 1.8 − Christensen (1998) 
26 Gulf of Thailand (1980s) 3.1 16.1 Christensen (1998) 
27 Gulf of Thailand (1993) 3.0 6.7 FC database 
28 Southwest coast of India (1994) 2.6 13.2  Silvestre et al . (2003)  
29 Southwest coast of India (1995) 2.6 10.7  Silvestre et al . (2003)  
30 Southwest coast of India (1996) 2.6 11.7  Silvestre et al . (2003)  
31 Southern shelf of Brazil (1975–1979) 3.6 27.7 Vasconcellos and Gasalla (2001) 
32 Southern shelf of Brazil (1990–1994) 3.6 27.6 Vasconcellos and Gasalla (2001) 
33 Southeastern shelf of Brazil (1977–1980) 2.8 33.1 Vasconcellos and Gasalla (2001) 
34 Southeastern shelf of Brazil (1990–1995) 3.1 52.7 Vasconcellos and Gasalla (2001) 
35 Venezuelan northeastern shelf (1980s) 2.8 20.7 Christensen and Pauly (1993) 
36 Gulf of Mexico continental shelf (1990s) 2.6 31.6 Christensen and Pauly (1993) 
37 Brunei Darussalam (1980) 3.2 7.4 − Christensen and Pauly (1993) 
38 Vietnam–China shelf (1980) 3.3 9.7 − Pauly and Christensen (1993) 
39 South China deep sea (1980s) 3.5 10.0 − Pauly and Christensen (1993) 
40 Hong Kong (1990s) 3.0 27.2  Pitcher et al . (2002)  
41 Bay of Bengal (1984–1986) 2.7 8.3  Silvestre et al . (2003)  
42 San Pedro Bay (1994–1995) 3.3 3.1 −  Silvestre et al . (2003)  
Coastal areas and coral reefs 
43 Bay of Revellata, Corsica (1998) 3.8 11.9 − Pinnegar (2000) 
44 Prince William Sound, Alaska (1994–1996) 4.1 4.5 − Okey and Pauly (1998) 
45 Coast of Western Gulf of Mexico (1990s) 3.4 89.5 Christensen and Pauly (1993) 
46 Gulf of Lingayen (1990s) 3.3 51.6 Pauly and Christensen (1993) 
47 Maputo Bay (1980s) 2.5 20.3 Christensen and Pauly (1993) 
48 San Miguel Bay (1992–1994) 3.0 14.7 Bundy and Pauly (2001) 
49 Boliano reef flat (1991) 2.2 2.8 Christensen and Pauly (1993) 
Ecosystem 
(b) Not classified  TL c %PPR Partial information available Reference 
Temperate shelves and seas 
1 Faroe Islands (1997) 3.7 33.2 Recovering from heavy exploitation  Guénette et al . (2001)  
2 Icelandic fisheries (1997) 2.9 1.2 Decrease in exploited biomass; cod heavily exploited  Guénette et al . (2001)  
3 Norwegian and Barents Sea (1997) 3.5 17.9 Medium-heavily exploited  Guénette et al . (2001)  
4 Central North Pacific (1998) 3.8 20.4 Fishery on top predators FC database 
5 West Greenland shrimp trawling area (1994) 3.3 3.6 Regime shift; large cod bycatch  Guénette et al . (2001)  
6 Gulf of Maine, Georges Bank (1982) 3.5 6.5 Heavily exploited  Guénette et al . (2001)  
7 Atlantic coast of Morocco (1980s) 3.2 5.5 Heavily exploited  Guénette et al . (2001)  
Tropical shelves and seas 
8 Campeche Bank of Yucatan shelf (1990s) 4.1 50.4 Heavily exploited; some overexploited resources Christensen and Pauly (1993) 
9 Gulf of Thailand (1973) 2.7 3.3 Moderately exploited FC database 
10 Kuala Trengganu (1980s) 3.4 26.1 Moderately exploited Pauly and Christensen (1993) 
11 Bali Strait (1990s) 2.9 6.9 Heavily exploited, sardine possibly overexploited Pitcher and Cochrane (2002) 
Coastal areas and coral reefs 
12 Schlei fjord (1984) 3.1 43.4 Moderately exploited Christensen and Pauly (1992) 
13 Beach-seine fishery area of Gulf of Mexico (1990s) 3.1 52.0 Fishing pressure has declined Christensen and Pauly (1993) 
14 Shallow areas of Gulf of Thailand (1979) 2.6 2.6 Small-scale fisheries Pauly and Christensen (1993) 
15 North coast of central Java (1979–1980) 3.0 11.4 Heavily exploited  Silvestre et al . (2003)  
16 Great Barrier Reef, prawn ground in the far north (1994–1996) 2.4 7.9 Fully exploited; no evidence of overexploitation FC database 
Ecosystem 
(b) Not classified  TL c %PPR Partial information available Reference 
Temperate shelves and seas 
1 Faroe Islands (1997) 3.7 33.2 Recovering from heavy exploitation  Guénette et al . (2001)  
2 Icelandic fisheries (1997) 2.9 1.2 Decrease in exploited biomass; cod heavily exploited  Guénette et al . (2001)  
3 Norwegian and Barents Sea (1997) 3.5 17.9 Medium-heavily exploited  Guénette et al . (2001)  
4 Central North Pacific (1998) 3.8 20.4 Fishery on top predators FC database 
5 West Greenland shrimp trawling area (1994) 3.3 3.6 Regime shift; large cod bycatch  Guénette et al . (2001)  
6 Gulf of Maine, Georges Bank (1982) 3.5 6.5 Heavily exploited  Guénette et al . (2001)  
7 Atlantic coast of Morocco (1980s) 3.2 5.5 Heavily exploited  Guénette et al . (2001)  
Tropical shelves and seas 
8 Campeche Bank of Yucatan shelf (1990s) 4.1 50.4 Heavily exploited; some overexploited resources Christensen and Pauly (1993) 
9 Gulf of Thailand (1973) 2.7 3.3 Moderately exploited FC database 
10 Kuala Trengganu (1980s) 3.4 26.1 Moderately exploited Pauly and Christensen (1993) 
11 Bali Strait (1990s) 2.9 6.9 Heavily exploited, sardine possibly overexploited Pitcher and Cochrane (2002) 
Coastal areas and coral reefs 
12 Schlei fjord (1984) 3.1 43.4 Moderately exploited Christensen and Pauly (1992) 
13 Beach-seine fishery area of Gulf of Mexico (1990s) 3.1 52.0 Fishing pressure has declined Christensen and Pauly (1993) 
14 Shallow areas of Gulf of Thailand (1979) 2.6 2.6 Small-scale fisheries Pauly and Christensen (1993) 
15 North coast of central Java (1979–1980) 3.0 11.4 Heavily exploited  Silvestre et al . (2003)  
16 Great Barrier Reef, prawn ground in the far north (1994–1996) 2.4 7.9 Fully exploited; no evidence of overexploitation FC database 
Table 1

Ecopath-derived ecological indices applied to factorial correspondence analysis, the expectedly/supposedly increasing (≫) or decreasing (≪) trend of fishery development according to ecological theory ( Odum, 1969 ; Christensen, 1995b ; Christensen and Pauly, 1998 ), and the percentage of analysed models behaving as expected.

Index Description Change (%) 
B/T Biomass/throughput ratio ≪60 
P/R Production/respiration ratio ≫87 
PP/B Total primary production/biomass ≫73 
B/P Biomass/production ratio ≪67 
R/B Respiration/biomass ratio ≪60 
GE Gross efficiency of the fishery ≫87 
RO Relative overhead ≪67 
FCI Finn cycling index (proportion of system throughput recycled) ≪67 
PL Path length index (measure of the complexity of ecosystems) ≪73 
SOI System omnivory index (related to variance in the TL of prey groups) ≫67 
Index Description Change (%) 
B/T Biomass/throughput ratio ≪60 
P/R Production/respiration ratio ≫87 
PP/B Total primary production/biomass ≫73 
B/P Biomass/production ratio ≪67 
R/B Respiration/biomass ratio ≪60 
GE Gross efficiency of the fishery ≫87 
RO Relative overhead ≪67 
FCI Finn cycling index (proportion of system throughput recycled) ≪67 
PL Path length index (measure of the complexity of ecosystems) ≪73 
SOI System omnivory index (related to variance in the TL of prey groups) ≫67 

Because of heteroscedasticity in the data, %PPR–TL c pairs were standardized by log-transformation. The discriminant functions associated with the classified ecosystems were then obtained through canonical discriminant analysis (CDA; using SPSS, version 11) to check classification efficiency. Limit reference functions (following the nomenclature of Caddy and Mahon, 1995 ) were established on the basis of 50%, 70%, and 90% probabilities that %PPR–TL c pairs belonged to a sustainably fished situation. Graphically, these functions can be used to identify the parameter space corresponding to different probabilities of EO. The discriminant functions were then applied to categorize previously unclassified models ( Table 2b ), and the classification obtained was evaluated against available information on exploitation effects.

Following Pauly and Christensen (1995) , catches can be expressed in terms of %PPR by  

formula
Calculating %PPR from the limit reference functions defined above allows us to estimate the ecosystem-based maximum sustainable catch (EMSC) for each ecosystem type, knowing the associated TL c and primary production (PP) values.

Results

FCA results for historical series of exploited ecosystems confirm that, among different %PPR–TL c pairs from a given ecosystem, the suite of ecological indices chosen was useful in tracking structural and functional changes within the ecosystem driven by changes in exploitation scenarios ( Table 1 ), so were broadly consistent with the theory that resilience decreases if fishing pressure increases.

Figure 2 presents paired estimates of %PPR–TL c for classified cases ( Table 2a ). Sustainably fished ecosystems are limited by values of TL c ranging from 3.0 to 4.1, and of %PPR ranging from 0.3 to 14.7, with few exceptions (Newfoundland 1900; Eastern Bering Sea 1950s; northern British Columbia 1900). Overfished ecosystems have a wider TL c range of 2.2–3.9 and %PPR of 2.8–89.5, except in the case of Newfoundland (1985–1987), which has a high TL c and a low %PPR.

Figure 2

Ecosystem-based reference framework based on %PPR–TL c , with reference functions related to 50%, 70%, and 90% belonging to a sustainably fished situation.

Figure 2

Ecosystem-based reference framework based on %PPR–TL c , with reference functions related to 50%, 70%, and 90% belonging to a sustainably fished situation.

CDA applied to the %PPR–TL c pairs from previously classified models correctly reclassify 88% of the cases ( Figure 3 ). Figure 2 also shows limit reference functions associated with probabilities of the EO status, using the discriminant function from CDA. Sustainably fished ecosystems involve a TL c > 3.0 and a low to moderate %PPR, and the range of sustainable levels for %PPR increases with TL c . The likelihood of EO status clearly increases as TL c decreases, and models located within this area are characterized by fisheries exploiting small fish and invertebrates ( Table 2a ).

Figure 3

Discriminant scores from canonical discriminant analysis of pairs of %PPR–TL c (the sequence of models following that in Table 2a ).

Figure 3

Discriminant scores from canonical discriminant analysis of pairs of %PPR–TL c (the sequence of models following that in Table 2a ).

Pairs of %PPR–TL c for unclassified models ( Table 2b ) were plotted within the general framework defined ( Figure 4 ). Faroes (1997) and Norwegian and Barents Sea (1997) models were located above the (50% probability) threshold curve for EO. The rest of the temperate models fitted within the range for sustainable exploitation. All models from tropical shelves and seas and coastal areas and coral reefs were located above the threshold curve, indicating that present exploitation implies a high risk of EO status. In general, the results fit the fragmentary information available from these ecosystems ( Table 2b ).

Figure 4

Previously unclassified pairs of %PPR–TL c ( Table 2b ) plotted within the ecosystem-based operational reference framework.

Figure 4

Previously unclassified pairs of %PPR–TL c ( Table 2b ) plotted within the ecosystem-based operational reference framework.

Table 3 lists a compilation of estimates of ecosystem-based maximum sustainable catches (EMSC) for tropical and temperate shelves and seas and coastal areas and coral reefs, in comparison with current annual catches. Reported catches are far in excess of EMSC 50 and EMSC 70 estimates, by factors of 1.7 and 2.9, respectively, in temperate shelves and seas, 2 and 3.4 in tropical shelves and seas, and 2.7 and 4.6 in coastal areas and coral reefs.

Table 3

Ecosystem-based maximum sustainable catches (EMSC) at a 50% and 70% probability of being exploited sustainably for different ecosystem types, compared with current levels of catch and discards.

Ecosystem type  PP * (gC m −2 y −1 )   TL c Catch and discards (t km −2 y −1 ) *†  EMSC 50 (t km −2 y −1 )   EMSC 70 (t km −2 y −1 )  
Tropical shelves and seas 310 3.3 2.87 1.46 0.84 
Temperate shelves and seas 310 3.5 2.31 1.38 0.80 
Coastal areas and coral reefs 890 2.5 10.5 3.96 2.28 
Ecosystem type  PP * (gC m −2 y −1 )   TL c Catch and discards (t km −2 y −1 ) *†  EMSC 50 (t km −2 y −1 )   EMSC 70 (t km −2 y −1 )  
Tropical shelves and seas 310 3.3 2.87 1.46 0.84 
Temperate shelves and seas 310 3.5 2.31 1.38 0.80 
Coastal areas and coral reefs 890 2.5 10.5 3.96 2.28 

1988–1991.

Discussion

Some limitations of the data set need to be considered when it comes to interpreting the ecosystem-based framework developed for fisheries management, and to applying it to new pairs of %PPR–TL c characterizing exploited ecosystems. First, more information was available for temperate areas than for other marine ecosystem types, and modelled ecosystems included in the analysis referring to reconstructed past situations are mainly from temperate regions. The models for historical situations are fundamental to this approach ( Jackson et al ., 2001 ), even if their precision tends to be lower than for more recent models ( Pitcher, 2001 ). Furthermore, catches are subject to high uncertainty, owing to difficulties in estimating discards, bycatches, and illegal, unreported, or unregulated catches. However, the proposed ecosystem-based framework for fisheries management and the related reference functions can be modified easily whenever more accurate data become available.

The initial hypothesis on the behaviour of %PPR and TL c with respect to ecosystem overfishing ( Tudela, 2003 ) has been confirmed, and can now be described mathematically, allowing limit reference functions to be established. Any new pair of %PPR–TL c for any exploited ecosystem can now be directly assessed in terms of its deviation from a situation desirable to preclude EO status. Overall, the framework appears to represent a clear step forward in the difficult process of translating concepts of “ecosystem overfishing” into operational definitions that can be used in fisheries management ( Larkin, 1996 ), and it provides a holistic ecosystem-based evaluation of a harvesting strategy, taking into account marine foodweb structure and functionality.

Discriminant analysis correctly reclassified a large number of models. Exceptions refer to reconstructed Newfoundland (1900), eastern Bering Sea (1950s), and northern British Columbia (1900) systems, which were characterized by great fishing pressure on marine mammals and a high %PPR. Newfoundland (1985–1987) was classified as an overfished ecosystem ( Heymans, 2003 ), but it is located within the range of sustainable exploitation. This may be due to a transitional state of the ecosystem, caused by a supposed regime shift ( Steele, 1998 ). This stresses the importance of detailed and wide-ranging historical information when interpreting ecosystem state.

Results highlight the fact that ecosystem overfishing is reached sooner following an increase in %PPR operating on low-TL species. Clearly, disrupting energy flows lower in the trophic web has the farthest-reaching effects on the whole ecosystem. This is important in the context of evidence about the rapid worldwide depletion of predator fish, and the consequent decrease of the TL c ( Christensen et al ., 2003 ; Myers and Worm, 2003 ).

According to our results, %PPR values of ≥20% are only compatible with sustainable exploitation if fisheries are harvesting high in the foodweb (TL c > 3.5). The sustainable ranges for TL c and %PPR values identified contrast with estimates reported by Pauly and Christensen (1995) . This raises serious concern about the sustainability of marine ecosystems under current exploitation regimes, mainly characterized by fisheries focusing on low-TL species and a high harvesting intensity. If reconstructed situations from the past are eliminated, Figure 2 provides a good graphic overview of the overall impact of current fishing practices on marine ecosystems: clearly, overexploited systems now dominate.

Estimated EMSC values below current catch levels agree with previous assessments of exploited ecosystems dominated by EO situations, and point to current catches being well beyond sustainable levels. Moreover, EMSC estimates could be overestimated because they are theoretical maxima, considering a mean TL c . If the fishery is operating on a broad spectrum of TLs, sustainable maxima will be lower than those proposed, owing to the exponential relationship relating %PPR and TL c .

Catches exceeding EMSC might be maintained within chronically degraded ecosystems, although the long-term viability of such a situation is doubtful, and large-scale resource collapses may continue ( Jackson et al ., 2001 ). This conclusion reinforces the need to apply a more generous precautionary approach to fisheries management.

We thank Villy Christensen for providing the database and useful advice, various scientists for supplying essential information about, and personal experience with, the ecosystems modelled, and Maria Lourdes Palomares and Niels Daan for constructive comments on the draft manuscript.

References

Ainsworth
C.
Heymans
J. J.
Pitcher
T.
Vasconcellos
M.
Ecosystem models of northern British Columbia for the time periods 2000, 1950, 1990 and 1750
2002
 
Fisheries Centre Research Reports, 10(4). 41 pp
Briand
F.
Fishing down the Mediterranean food webs?
2000
 
CIESM Workshop Series, 12. Kerkyra, Greece, July 2000
Bundy
A.
Pauly
D.
Selective harvesting by small-scale fisheries: ecosystem analysis of San Miguel Bay, Philippines
Fisheries Research
 , 
2001
, vol. 
53
 (pg. 
263
-
281
)
Caddy
J. F.
Mahon
R.
Reference points for fisheries management
1995
 
FAO Fisheries Technical Paper, 347. 83 pp
Christensen
V.
A model of trophic interactions in the North Sea in 1981, the year of the stomach
Dana
 , 
1995
, vol. 
11
 (pg. 
1
-
28
)
Christensen
V.
Ecosystem maturity – towards quantification
Ecological Modelling
 , 
1995
, vol. 
77
 (pg. 
3
-
32
)
Christensen
V.
Fishery-induced changes in a marine ecosystem: insight from models of the Gulf of Thailand
Journal of Fish Biology
 , 
1998
, vol. 
53
 
Suppl A
(pg. 
128
-
142
)
Christensen
V.
Guenette
S.
Heymans
J.J.
Walters
C.
Zeller
D.
Pauly
D.
Hundred-year decline of North Atlantic predatory fishes
Fish and Fisheries
 , 
2003
, vol. 
4
 (pg. 
1
-
24
)
Christensen
V.
Pauly
D.
Ecopath II. A software for balancing steady-state models and calculating network characteristics
Ecological Modelling
 , 
1992
, vol. 
61
 (pg. 
169
-
185
)
Christensen
V.
Pauly
D.
Trophic Models of Aquatic Ecosystems
1993
, vol. 
26
 
ICLARM Conference Proceedings
 
390 pp
Christensen
V.
Pauly
D.
Changes in models of aquatic ecosystems approaching carrying capacity
Ecological Applications
 , 
1998
, vol. 
8
 
Suppl 1
(pg. 
104
-
109
)
Christensen
V.
Walters
C.
Ecopath with Ecosim: methods, capabilities and limitations
Ecological Modelling
 , 
2004
, vol. 
172
 (pg. 
109
-
139
)
Coll
M.
Palomera
I.
Tudela
S.
Sardà
F.
Mass-balance approach applied to the north western Mediterranean Sea: the case study of the Delta of Ebro river area
2004
Fourth World Fisheries Congress Abstracts. The American Fisheries Society
May 2004
Vancouver, Canada
 
p. 172
Guénette
S.
Christensen
V.
Pauly
D.
Fisheries impacts on North Atlantic ecosystems: models and analyses
Fisheries Centre Research Reports
 , 
2001
, vol. 
9
 
4
 
344 pp
Heymans
J. J.
Ecosystem models of Newfoundland and Southeastern Labrador: additional information and analyses for “Back to the Future”
Fisheries Centre Research Reports
 , 
2003
, vol. 
11
 
5
 
81 pp
Jackson
J.B.C.
Kirby
M.X.
Berger
W.H.
Bjorndal
K.A.
Botsford
L.W.
Bourque
B.J.
Bradbury
R.H.
Cooke
R.
Erlandson
J.
Estes
J.A.
Hughes
T.P.
Kidwell
S.
Lange
C.B.
Lenihan
H.S.
Pandolfi
J.M.
Peterson
C.H.
Steneck
R.S.
Tegner
M.J.
Warner
R.R.
Historical overfishing and the recent collapse of coastal ecosystems
Science
 , 
2001
, vol. 
293
 (pg. 
629
-
638
)
Larkin
P.A.
Concepts and issues in marine ecosystem management
Reviews in Fish Biology and Fisheries
 , 
1996
, vol. 
6
 (pg. 
139
-
164
)
Lindeman
R.L.
The trophic-dynamic aspect of ecology
Ecology
 , 
1942
, vol. 
23
 (pg. 
399
-
418
)
Murawski
S.A.
Definitions of overfishing from an ecosystem perspective
ICES Journal of Marine Science
 , 
2000
, vol. 
57
 (pg. 
649
-
658
)
Myers
R.A.
Worm
B.
Rapid worldwide depletion of predatory fish communities
Nature
 , 
2003
, vol. 
423
 (pg. 
280
-
283
)
Odum
E.P.
The strategy of ecosystem development
Science
 , 
1969
, vol. 
104
 (pg. 
262
-
270
)
Odum
W.E.
Heald
E.J.
Cronin
L.E.
The detritus-based food web of an estuarine mangrove community
Estuarine Research, 1
 , 
1975
New York
Academic Press
Okey
T.
Pauly
D.
A trophic mass-balance model of Alaska's Prince William Sound ecosystem, for the post-spill period 1994–1996, 2nd edn
Fisheries Centre Research Reports
 , 
1998
, vol. 
7
 
4
 
146 pp
Pauly
D.
Christensen
V.
Sherman
K.
Alexander
L.M.
Gold
B.D.
Stratified models of large marine ecosystems: a general approach and an application to the South China Sea
Stress, Mitigation and Sustainability of Large Marine Ecosystems
 , 
1993
Washington, DC
AAAS Press
(pg. 
148
-
174
)
Pauly
D.
Christensen
V.
Primary production required to sustain global fisheries
Nature
 , 
1995
, vol. 
374
 (pg. 
255
-
257
)
Pauly
D.
Pitcher
T.
Preikshot
D.
Back to the future: reconstructing the Strait of Georgia ecosystem
Fisheries Centre Research Reports
 , 
1998
, vol. 
6
 
5
 
99 pp
Pinnegar
J. K.
Planktivorous fishes: links between the Mediterranean littoral and pelagic
2000
 
PhD thesis, University of Newcastle-upon-Tyne, UK. 213 pp
Pitcher
T.J.
Fisheries managed to rebuild ecosystems: reconstructing the past to salvage the future
Ecological Applications
 , 
2001
, vol. 
11
 (pg. 
601
-
617
)
Pitcher
T.
Buchary
E.
Trujillo
P.
Spatial simulations of Hong Kong's marine ecosystem: ecological and economic forecasting of marine protected areas with human-made reefs
Fisheries Centre Research Reports
 , 
2002
, vol. 
10
 
3
 
75 pp
Pitcher
T.
Cochrane
K.
The use of ecosystem models to investigate multispecies management strategies for capture fisheries
Fisheries Centre Research Reports
 , 
2002
, vol. 
10
 
2
 
156 pp
Sánchez
F.
Olaso
I.
Effects on fisheries on the Cantabrian Sea shelf ecosystem
Ecological Modelling
 , 
2004
, vol. 
172
 (pg. 
151
-
174
)
Silvestre
G.
Garces
L.
Ahmed
M.
Stobutzki
I.
Santos
R. A. V.
Luna
C.
Lachica-Aliño
L.
Christensen
V.
Pauly
D.
Assessment and management of tropical coastal fisheries in developing Asian countries
2003
, vol. 
67
 
ICLARM Conference Proceedings
 
952 pp
Steele
J.
Regime shifts in marine ecosystems
Ecological Applications
 , 
1998
, vol. 
8
 
Suppl 1
(pg. 
S33
-
S36
)
Trites
A.W.
Livingston
P.A.
Mackinson
S.
Vasconcellos
M.C.
Spriner
A.M.
Pauly
D.
Ecosystem change and the decline of marine mammals in the eastern Bering Sea: testing the ecosystem shift and commercial whaling hypotheses
Fisheries Centre Research Reports
 , 
1999
, vol. 
7
 
1
 
106 pp
Tudela
S.
The pair %PPR and TL fishery : a possible composite indicator of ecosystem impact aligned with a holistic vision of the ecosystem-based fisheries management approach. Ecofish
2003
In Workshop on Sustainability Indicators in Fisheries
April 2003
Dakar
Vasconcellos
M.
Gasalla
M.A.
Fisheries catches and carrying capacity of marine ecosystems in southern Brazil
Fisheries Research
 , 
2001
, vol. 
59
 (pg. 
279
-
295
)