Computer modelling studies

to determine the target sizes of marine reserve systems

By Dr J Floor Anthoni (2003)

Lacking empirical (experimental) data, scientists have resorted to computer modelling studies in order to prove that marine reserves can help fisheries management, and to what extent. Their aim was to use marine reserves as a direct fishery management tool. But fishermen disagree. They have a better plan.
Introduction to this chapter and to the methods of computer modelling, their limitations and expected accuracies. Discussion of problems and what the computer models suggest.
the fisherman's
Fishermen are a proud and independent folk. They have an intimate knowledge of the sea against which everything scientists say is tested. Not surprisingly they are against marine reserves as a blunt fishery management tool. They propose a better way, deserving of support from scientists.
A quick overview showing the range of claims and target sizes for various aims.
details of studies
Summaries of each study, highlighting their methods, aims and recommendations.
For comments, suggestions and improvements, please e-mail the author, Floor Anthoni.
-- Seafriends home -- conservation index -- marine conservation -- Rev 20030715,

Marine reserve proponents have always claimed fishery benefits from marine reserves, but only few studies have shown actual benefits, although these too are questionable. In general, the fishery benefits of marine reserves came from fishery related measures such as banishing destructive practices, buying fishermen out or providing subsidies to fish further out with bigger boats. Lacking experimental evidence for their claims, scientists have used computer models to estimate perceived benefits in all areas of fishing, from stock enlargement and protection of genetic diversity to higher catches and designing marine reserve networks.

In order to understand the issues, let's begin with the ill effects of fishing:

Fishermen would tackle the above problems by tackling them at their root causes but this is not always possible, for instance: It is thought that fishing refuges or marine reserves can help. However, these too suffer from the above compliance problems.

Cynics would say that computer models come to the rescue where all other measures failed. They allow people to sit at their office desks, designing virtual realities according to which the natural world should behave. But computer models can provide new insight, either because they can mimic complexities beyond our ability to comprehend, or they can simplify a situation such that basic insight forms. A computer model consists of:

Some models are able to introduce uncertainty by doing various runs for randomly varying assumptions. As one can see, every model is based on assumptions, particularly where the environment is still poorly understood. In this respect they are no better than fisheries management models, which have the enormous benefit of feedback from actual catches, size distributions and effort. Note in this respect that the marine reserve as a fishery management tool was born from people's concern about fishery collapses, even where these were well controlled by fishery scientists.
"The view of fishery scientists is that it was partly a failure of science that caused the collapse of the Newfoundland cod. Scientific advice to managers was not always correct or timely, and management failed to act in time on either the erroneous advice or the corrected advice." [1, p30]. It fails to say that the marine environment with its many actors and environmental fluctuations is still very poorly understood.
It is an important principle of models that they become rapidly less reliable as their number of assumptions increases. Indeed, in cases where little is known, intuition can be better than even the simplest of models. In the case of the marine environment, complicated models must be viewed with great suspicion.

Here are some of our concerns regarding the assumptions:

Apart from the above technical or ecological difficulties, scientists make other questionable assumptions about marine reserves, covered in our Frequently Asked Questions: Taking all the above in mind, what do the model studies tell us?

The fisherman's perspective
The fisherman's perspective is quite different. It is true that fishing had its cowboy years, moving from one collapsed stock to another. But now that the limits of harvest are reached, they have to fish in a more sustainable way. 
Here in New Zealand the Quota Management System (QMS) was introduced with Individual Transferable Quotas (ITQ) based on freemarket principle of best allocation by price signals, while being controlled by the Minister of Fisheries. For all its shortcomings, it is starting to work by providing flexibility within a framework for all. Every year an enormous amount of discussion goes into the allocation of Total Allowable Catches (TAC), often hotly debated by recreational fishermen who wish to fish with more precaution. The mood has changed, and fishermen are really concerned about the future for their children. Yet they are against marine reserves created for political or ideological reasons in places where these won't work. There is a better way. In the face of the above sentiments, the work of marine reserve scientists is bound to be no more than a cry in the night. It would be more productive for them to see how science can support the fishermen's objectives.

[1] Source: NRC 2002: Marine protected areas: tools for sustaining ocean ecosystems. National Academy Press, Washington DC.

This table gives an overview of typical recommended target sizes of closed areas as percentage of total fished area, for the various reasons indicated. Note that some authors are not listed in the next chapter.

Targets for networks of marine reserves or non-fishing refugia
Of the studies marked with an asterisk * more detail can be found in the next chapter
target or area 
reason reference
10% 1993 protect spawning stock of fast growing species DeMartini 1993 *
1-20% 2000 protect genetic diversity Trexler & Travis 2000 *
10-40% 1990 spawning stock & yield of Atlantic cod Polachek 1990 *
10-15% 1994 to provide and detect significant fishery benefits NMFS 1994
10-20% 1998 representative habitats Ballantine 1991*, 
Dayton et al 1995
15-29% 1996 to recover collapsed fishery, red snapper Gulf of Mexico Holland & Brazee 1996 *
20% 1997 marine conservation Lubchenko et al
20-30% 1990 protect spawning potential of reef fisheries Plan Development Team 1990
20-50% 1998 alternative strategy for sustainable fisheries Yoklavich 1998
25-29% 1999  biodiversity, fishing impacts (Georges Bank/ Gulf of Maine) Jegalien 1999
>30% 1993 protect spawning stock of slow growing species DeMartini 1993 *
40-50% 1992 abalone (paua) protecting egg production Nash 1992 (in Shepherd & Brown 1993)
50+% 1998 fisheries Lauck et al 1998 *,
Guenette et al 1998 *
60-80% 1997 fisheries exploitation Pitcher eds 1997
90% 1998 fisheries long-lived rock fish Walters 1998, cited in Perry et al 1999

Source of the above table: Forest & Bird, New Zealand in their submission to the Fiordland MPA proposal.

Details of the marine reserve modelling studies
Reader, pleas note that percentages could relate either to the closed area or to the stock biomass. When relating to stock biomass, they mean the percentage of the unexploited (by humans) fish stock biomass.

Allison et al., in review: ?%
Was interested in the benefit of reserves for natural and human catastrophes. He calculated that the more frequent a disturbance, and the longer the recovery time, the larger the area in protection should be.

Attwood & Bennett, 1995: 25-65%
Looked at three species of surf zone fish, concluding that marine reserves would benefit catches of two and reduce the risk of overfishing for one. Catches peak for one species at 65% area protection, whereas 25-30% would be optimal for the other species. (South Africa)

Ballantine, 1997: 10-20%
Wants to protect a target of 10% of all of the marine habitats in New Zealand. The key principle is that we should not fish everywhere. Some areas should be set aside as refuges for ethical reasons. Ten percent "has a long traditional use as a figure that signifies importance without serious hurt", contrasting favourably with the 90% left open to exploitation and is less than the national parks area in New Zealand. It is a call for arms, only the beginning of more to come. Note that the author does not support his claims with modelling studies.

Botsford et al., 1999: 8-33%
Studied the exploitation of Californian red sea urchins (Strongylocentrotus franciscanus) by computer model. He showed that if the species is resilient to fishing, reserves would reduce benefit, whereas if the species were sensitive to overfishing, it would provide a positive benefit. In the case of uncertainty, a closed area of 8-33% would be needed. A closed area of 17% would increase catches by 18% over the long term.

Bustamante et al. 1999: 36%
In order to protect all areas of high biological importance around the Galapagos Islands, in 5 biogeographic zones, would require 36% in no-take marine reserves.

Daan, 1993: 10-25%
Simulated the benefits of closed areas on Atlantic cod (Gadus morhua). He found that 10% area protection would reduce mortality by only 5% if leakage from reserves were low. Likewise, protecting 25% would reduce mortality by 10-14%. In his model, the fish were assumed to be distributed evenly, as was fishing effort.

DeMartini, 1993: 20-50%
Used a yield per recruit model to examine the effects of closed areas on fishing the Pacific coral reefs. His conclusions were similar to those of Polachek, that biomass increases inside reserves, and could serve as an insurance for overfishing outside. He concluded that 20-50% would be needed, at a cost to the fishery. Benefits from spawn were ignored.

Goodyear, 1993: 20%
Used fishery models to estimate that stock levels above 20% would avoid recruitment overfishing. From this others concluded that 20% in closed areas would be the minimum to avoid stock collapses. This is where the magical 20% comes from!

Guénette et al., 2000: 20%-80%
Studied the migratory cod (Gadus morhua) fishery collapse in eastern Canada. Using a computer model he found that without any fisheries measures, a no-fishing area of 80% should be set aside, but together with seasonal fishery closure a closed area of 20% should be enough to prevent fishery collapse.

Halfpenny & Roberts, in review: 10%
Were interested in the design of a representative system of marine reserves in all biogeographic regions, including duplication. Two systems of 10% were sufficient in achieving replication for most, but not all of the biogeographic regions and habitats.

Hanneson, 1998: 70-80%
Used a bioeconomic model to examine the effects of reserves on spawning stock, size, catches and costs of fishing for mobile species like the Georges Bank Atlantic cod (Gadus morhua). When assuming uncontrolled catching outside these areas, he concluded that reserves had to be 70-80% in order to produce fish catches equal to a controlled fishery at 60% stock level. At 50-80% of the area, reserves increased spawning stocks by 1.4-2.3 times. The required closed area reduces when fisheries controls are applied. But the closed areas increased the cost of fishing and tended to promote overcapacity.

Holland & Brazee, 1996: 15-29%
Used models to simulate red snapper (Lutjanus campechanus) fishery in the Gulf of Mexico. They found that reserves would not benefit catches until the stock was overfished. Reserves needed to be 15-29% as fishing pressure increased. But for optimal short-term economic benefit, they needed to be somewhat smaller.

Lauck et al., 1998: 30-70%
This group studied fisheries management errors in estimating stock productivity, mortality and population size. In a simple model they showed that reserves should be between 31 and 70% of the fishing grounds to maintain populations above 60% of their unexploited size, which he considered optimal. The less certain stock estimates are, the larger the protected area should be, up to 70%.

Mace & Sissenwine, 1993: 35%
Studied 91 fish stocks of 27 species and concluded that for some fast growing species 20% would suffice, but for some slow growing species 70%.

Mace, 1994: 40%
If the relationship between population size and recruitment is unknown (true for most stocks), a precautionary approach would aim for a 40% stock level. (not a closed area)

Man et al., 1995: 20-40%
Used a model that followed several populations distributed over various habitat patches. In such a situation, a patch can be fished out as fishing pressure increases. His findings showed that a closed area of 20-40% would be beneficial as a resource of offspring at the highest levels of fishing.

Mangel, 2000: 20-30%
In order to maintain fish populations above target levels (35% of unfished stock), closed areas of 20-30% could sustain fisheries. However, if more variability and less certainty prevailed, these percentages need to be higher. He reasoned that such reserves increased cumulative yields when fish populations were initially heavily exploited.

Pezzey et al (in press): 0-50%
Studied the small tropical coral reserves of Belize, St Lucia and Jamaica, concluding that larger marine reserves were needed as fishing pressure increased, in order to prevent stock collapses.

Polachek, 1990: 10-40%
Examined the reserve effects on spawning stock biomass and yield, of the Atlantic cod Gadus morhua using a yield per recruit model, assuming that stocks moved freely from reserves to the fished area. Although reserves increased biomass, they reduced fish catches unless stock moved easily out of the reserves. Assuming leakage of 50% per year, reserves should be between 10-40%, depending on fishing intensity. Polachek ignored benefits from spawn.

Quinn et al., 1993: 50%
Used a population model to study area protection for the Californian red sea urchin (Strongylocentrotus franciscanus). The propagation of this species is suspected to depend on densities (Allee effect) for both successful fertilisation and juvenile recruitment. Population sizes and catches were greatest at 50% area closure, for all except the slightest of exploitation. The model also considered larval dispersal from reserves.

Roberts, in review: 5-30%
Was interested in the density of reserves within an area for optimal connectivity. Connectivity increased rapidly as the number of reserves increased. By protecting 5-30%, reserves were getting closer to one another by 76%. Assuming that the degrees of target horizon to nearby reserves is decisive in recruitment spill from one reserve to another, he concluded that 30% in closed areas was 4 times more effective than 5%.

Roughgarden, 1998: 75%
To avoid recruitment overfishing he thinks that 75% should be set aside to avoid overfishing in the absence of any other controls.

Sladek Nowliss & Roberts, 1997, 1999: 40-80%
Using a single-species model, applied to four different species, found that reserves were effective in increasing catches only when stocks were overfished. In most of the extensively fished Caribbean, reserves should cover 75-80%, but 40% would already give higher yields.

Sladek Nowliss & Yoklavich, 1998: 20-27%
Studying the Pacific rockfish Boccacio (Sebastes paucispinis) with a population model, they found that reserves could enhance catch depending on how overfished the stock was. For optimal long term catches, reserves should range from 20-27%.

Soh et al., 1998: protect hotspots
Studied and modelled the effect of closing hotspots (areas of aggregation) while allowing catches outside, for two species of rockfish (a kind of bass). This fish is caught as by-catch while fishermen high-grade their catches by discarding 15-60%. The reserves allowed all rockfish to be landed, reducing discards, while increasing biomass.

Sumailia, 1998: 30-50%
Studied the Barents Sea cod fishery using a bioeconomic model, showing that closed areas reduced overall yield but increased stocks. His closed area assumed a leakage of 40-60% of fish to the fished areas. His model showed that closed areas between 30-50% provided protection without greatly reducing current economic benefit, even in the presence of several years of recruitment failure.

Turpie et al. in press: 29%
Explored designs for marine reserves by dividing the South African coast into 52 sections of 50km. He was interested in a design that saved most of the endemic species, and found that 10% protection would save 97.5% of species but not 15 endemic ones, whereas 29% area protection would. To represent all species in their ranges would require 36% in reserves.

Trexler & Travis, 2000: 1-20%
Were interested in the ill effects of fishing on genetic diversity. A closed area of 1% already provides remarkable benefits; a 10% area decreased directional selection by 60%, while 20% protection would eliminate selective effects from the population entirely.

Source: NRC 2002: Marine protected areas: tools for sustaining ocean ecosystems. National Academy Press, Washington DC.