The control of fishing effort is discussed in the Mediterranean context, including categorization of effort units, a discussion of effort control in multi-species fisheries, simple approaches comparing effort distribution and survey biomasses and some options for fitting of various models that use effort data. Some general questions of strategy in designing an effort control system are discussed. An annex provides some relevant definitions and technical information.
Over the last 20 years of GFCM activities, the major emphasis within sub-regional GFCM Expert Consultations on stock assessment has been on various aspects of the fishery-related to the marine-renewable resources, with relatively minor attention given to the characteristics of the vessels harvesting the stock. Attention has been focused in the seventies on applying standard age-based and production models such as those developed and used in northern Atlantic and European Seas. This has not been particularly effective, largely because the age composition data has not been widely available, which, in turn, has been due to problems of sampling high-value, very diverse landings, especially by technologically varied demersal fleets using a variety of gear and fishing techniques. Some of the issues directly relevant to fishery management, together with existing information on Mediterranean stocks, are given in Caddy and Oliver (1996). For the Black Sea, a recent review (Prodanov et al., 1997) has provided more detailed information on the state of resources of the Black Sea.
The situation as summarized by Caddy and Oliver (1996) indicates that coordinated work leading to formal assessments of the state of the main resources, such as we have seen in the North Sea for example, has been relatively ineffective here. The geographical configuration of the Mediterranean, as already noted in the companion paper in this volume, plays a major role, together with the geographical separation of most national fishing grounds and resources. This reduces shared stock and mixed fleet problems compared with the North Atlantic. Except for wide shelf areas such as the Gulf of Lions and the Adriatic, vessels tend to operate locally from local ports along a narrow shelf and littoral, often each with their specific fishing grounds. One other consequence of this that is of relevance to management, is scepticism that data can always be pooled to calculate a mean state of exploitation for the larger statistical areas of the GFCM, given local variations in fishing intensity and uncertain stock areas.
One biological approach that has been applied with varying degree of success in the Mediterranean stems from techniques developed in tropical seas, which derive fish stock parameters from size frequency analysis (e.g. Pauly and Morgan, 1982). The problem with this approach is that, although it tends to confirm a high degree of exploitation for most demersals, it does not readily lead to specific management recommendations other than supporting the general recommendation of all assessment work so far to reduce fishing intensity on demersal stocks.
Another biologically-inspired initiative coming out of the difficulty of commercial sampling of catches has been the initiative to carry out routine surveys and direct biomass estimation for demersals, for example in the MEDITS programme. This approach has also been applied to acoustic estimation of some pelagic stocks (Tsimenides, 1989). To date, however, formal methods of using this biomass data set in stock assessments are still at an early stage.
From a management perspective, apparently there has also been a reluctance (and perhaps lack of infrastructure) to face the necessary complications involved in applying a quota management system to the very diverse Mediterranean demersal resources. Largely for this reason GFCM has placed emphasis on direct control of capacity and fishing effort, and a priority of the GFCM Scientific Advisory Committee will be to encourage provision of the information base and methodologies for application of such an approach.
Such an approach seems consistent with experience in other areas. Hillis and Arnason (1995) are not alone in suggesting that a combination of technical measures and TACs will not alone place an effective cap of overall fishing mortality without an overall control of fishing effort.
In general, limited licence schemes have been widely adopted as a first approach to control of fishing capacity, both in the Mediterranean and elsewhere. Despite this, national control of fishing capacity faces immediate difficulties due to very incomplete reporting of national fleet size and composition.
There is fragmentary evidence that a number of countries have increased capacity markedly over the last ten years. Despite questions as to the actual fleet size, estimates of fleet size since the early nineties (Caddy and Oliver, 1996) suggested that an overall increase in fleet size of 6% or so per annum over the last few decades would not be in excess of reality. Given also technological improvements, it seems unlikely that the increase in rate of exploitation has been less than this figure. One first measure that could lead to a stabilization of fishing capacity would therefore be for member countries to enumerate, and control over time, the number and characteristics of fishing units operating in their waters. The importance of also linking real-time vessel enumeration to the local ports from which such vessels operate will hopefully emerge from the following account.
There are obvious problems in enumerating vessels due to the wide size range of boats operating from Mediterranean ports. Many small boats fall below the 15-m limit agreed to by GFCM as the minimum size to which effort control in international waters will first be applied. It is important, however, to make parallel estimates of the number of vessels smaller than this, and a sampling frame approach based on ports has been recommended for this purpose.
In the following sections, a number of approaches, from the general to the specific, are suggested for scientific research in support of effort control in the Mediterranean. As a basis for discussion on this topic, some elementary concepts relating to fishing effort definition are given in an Annex.
Various objectives are possible for an effort control programme, and each of these has its different definition of fishing effort and its measurement. The most important distinction is between the use of fishing effort in determining the economic performance of fishing units and its use as a measure of fishing mortality. In this paper we will be principally concerned with the second use, although this does not exclude that economic measures can be used as a means of controlling fishing effort. One example here is provided by Brendel (1990), who found that fuel consumption is a reliable measure of total effort consumed and one that integrates the problem of calibrating relative fishing power of different classes of vessels.
Cunningham and Whitmarsh (1980), and see Annex, make another distinction between:
Nominal fishing effort, `the volume the of resources devoted to fishing, quantified either in monetary or physical units, and:
Effective fishing effort, which is used as a measure of fishing mortality and is proportional to the biomass extracted by fishing as a proportion of mean population size over the interval."
The problem they note is that catch rate or CPUE = Density * Catchability, and that CPUE is a measure of fish density if catchability is constant (and has been used with various degrees of success for this; see, for example, Gulland 1974, and Ulltang 1977). It is also used for calibration of relative fishing power of two or more vessels, assuming biomass or fish density remains constant (see Annex).
Most methods of using CPUE data for fishing power calibration and for converting nominal to effective effort, take into account that both catchability and biomass may vary simultaneously, and general linear models explicitly take this into account.
As noted by Gulland (1956) in his classic study, any serious approach to effort control will require the creation of homogeneous vessel categories. The classic approach has then been that one of these categories is selected as a standard. Smit (1996) points out that the standard group must ideally be defined as one where "nothing much has happened" in the period considered, i.e. no technical improvements occurred nor was there a `drain' of the better skippers to other categories such as might occur during a period where a strong trend in number of vessels in any given category was observed.
Mediterranean fisheries are generally categorized as multi-engine, multi-species activities, although this emphasis on complexity is sometimes an exaggeration in that some 5-6 species make up 50% of the landings, as opposed to 2-3 for the North Sea (Fig.1). However, there is also a wide multiplicity of gear types to make a full effort calibration involving all gear/vessel combinations problematic, even if a single-species approach were followed.
It may be more practical to aim at one of several approaches falling short of a full calibration of separate single-species effort units in terms of multi-engine catch rates. Trying to arrive at a full matrix of gear/species catchabilities, such as the hypothetical approach illustrated in the table below, is probably going to be impossible with the type of data currently available. Using a principal-component approach to discriminate the fleet into separate `metiers' (e.g. Lewy and Vinther, 1994) is one common approach, but Biseau (1998) observes that this gives undue weight to separating metiers on the basis of commercially insignificant species caught incidental to the main target species in the catch.
Referring to the table below, ideally, it would be desirable to have estimates for all the relative catchabilities of species/gear+vessel combinations. However, as noted in the table, there are some species which are `target species' for a particular vessel/gear category, and these are distinguished by capital Q in table 1, where the species in question makes up (say) more than 10-15% of the catch. Perhaps the analysis should first be concentrated on these.
Table 1. Hypothetical example of the relative catchability problem in a mixed species fishery for target (Q) and incidental (q) species. Finding the actual values q, Q would be difficult, but relative fishing power would be roughly in the ratio of respective CPUE's for the different gear/vessel classes in relation to a standard for one of the Q species.
Species |
Small trawler (A) |
Medium trawler (B) |
Large trawler (C) |
Rapido (D) |
Gillnet (E) |
Longline (F) |
Other (G) |
SPEC 1 |
q (1,A) |
Q (1,B) |
q (1,C) |
q (1,D) |
Q (1,E) |
q (1,F) |
q (1,G) |
SPEC 2 |
Q (2,A) = 1.0 |
Q (2,B) |
q (2,C) |
q (2,E) |
q (2,G) | ||
SPEC 3 |
Q (3,A) |
Q (3,B) |
q (3,D) |
Q (3,E) |
Q (3,F) |
||
SPEC 4 |
q (4,A) |
q (4,C) |
Q (4,D) |
q (4,G) | |||
SPEC 5 |
q (5,A) |
Q (5,B) |
Q (5,C) |
q (5,D) |
q (5,F) |
||
SPEC 6 |
Q (6,B) |
Q (6,C) |
q (6,G) | ||||
SPEC 7 |
q (7,C) |
q (7,D) |
q (7,E) |
q (7,G) | |||
SPEC 8 |
q (8,A) |
Q (8,B) |
q (8,C) |
q (8,D) |
|||
SPEC 9 |
Q (9,B) |
q (9,D) |
Fig.1. Ranked cumulative landings by species in the East and West Mediterranean and the North Sea
As noted in the Annex, a conventional approach to fishing power calibration uses one gear/vessel type as standard, preferably that which is most ubiquitous and least changed by technological improvements in time. In the above hypothetical table, small trawlers have been chosen for this role, but the question of characterizing vessel/gear categories and choosing a suitable standard gear/vessel type is an urgent question, that needs to be addressed early on by SAC.
Our first priority, in one sense, is to arrive at a reduced table consisting of estimates for the Q values of 5-6 key species, and to convert these Qs into relative fishing powers, given as the relative CPUE's of (say) `species 2 caught by small trawlers', which can be considered as unity.
This may be considered a `keystone' approach, and the problem of trawl fishery assessment is then reduced to one of estimating optimal fishing conditions for one or several of these `keystone' species/gear+vessel combinations. The assumption is that the situation with respect to exploitation of the other species will be brought closer to an optimal condition if that for the keystone species is close to ideal. In this connection, it has already been suggested that seeking to optimize conditions for harvesting of larger, slower-growing species such as hake will ensure that most smaller species are not overexploited, since these have faster growth rates, higher natural mortality rates and lower ages at maturity than hake. As a result, they will tend to be optimally- or even under-exploited, if the hake stock is properly exploited.
One approach that may be adaptable to the Mediterranean is described by Biseau (1998), who observed that selecting which CPUE data to use as an index of relative abundance, rather than calculating an index by the classical ratio 'total landings/total effort', may be the way to go.
His definition of a target species whose contribution to the overall catch is used to determine the metier, might be useful in some circumstances, namely that it must meet most of the following general criteria:
A target species:
(i) must be representative of a single metier and targeted by only a small number of vessels in a given time period;
(ii) to lead to meaningful metiers, should have either large total landings or high prices;
(iii) could not be defined as such if it did not make up more than (say) 50% of the landings.
Evidently, points (i) and (iii) leads one to question whether this is a useful approach in high-diversity Mediterranean trawl fisheries, although it seems to merit further investigation and modification to fit local conditions. Nonetheless, he suggests `thresholds' for percent contribution to the overall catch in directed trips which would allow these trips to be considered a `directed' one and suggests only these trips be used to calculate the CPUE index for the species. Some further details on this approach are given in the annex.
We start first with suggesting relatively simple and attainable objectives and move towards more sophisticated approaches but place emphasis on the need to take into account the geographical dimension implied by the other paper in this volume.
Earlier efforts to map the distribution of fishing capacity in the Western Mediterranean (Charbonnier and Garcia, 1985) gave a quick synoptic view of the distribution of fishing power of trawlers in the early eighties all along the littoral of the Western Mediterranean. Figure 2 gives the arbitrary geographic sub-units they used, and Figures 3 and 4 are extracted from the above publication by scanning the charts in this document and entering them into a GIS system for demersal and small pelagis respectively. They summarize local variations in fleet capacity along stretches of coastline and shelf and the accompanying fishing intensity that is assumed to apply after this fishing capacity is expressed per area of shelf (Fig. 3).
The first data set is calculated from the landings of demersal and benthic resources in Charbonnier and Garcia (1985), divided by the area of shelf down to 200 m (Fig. 3) and similarly for the pelagics (Fig. 4).
Fig. 2. Fisheries sub sectors used in "Atlas des pêcheries de la Méditerranée occidentale et centrale" by Charbonnier and Garcia (1985)
GIS maps prepared by F. Carocci FAO-FIRM
Date: July 1998
Projection Mercator
Fig. 3. Catch and effort per unit of surface estimates for demersal species in Atlas des pêcheries de la Méditerranée occidentale et centrale by Charbonnier and Garcia (1985)
FAO CGPM-GFCM CEE-EEC
GIS maps prepared by F. Carocci FAO-FIRM
Date: July 1998
Projection Mercator
Fig. 4. Catch and effort per unit of surface estimates for pelagic species in Atlas des pêcheries de la Méditerranée occidentale et centrale by Charbonnier and Garcia (1985)
FAO CGPM-GFCM CEE-EEC
GIS maps prepared by F. Carocci FAO-FIRM
Date: July 1998
Projection Mercator
Evidently, such an approach is valid to the extent to which effort remains local to home ports, which we believe is generally the case, at least for the demersal and benthic resources.
This GIS approach seems well adapted to the Mediterranean configuration of continental shelves with their series of local fisheries which operate largely independently in sequence along the coast. It illustrates the wide spatial variations in catches and fishing intensity exerted throughout the Western Mediterranean, as calculated using this data set from the eighties. An updating of this approach with modern information is highly desirable and very feasible once member countries provide details on their local fishing fleets. Although we cannot say what levels of fishing mortality the different local intensities corresponded to, we believe such an approach illustrates that solving the problem of effort control is a local one and has to be mainly addressed locally along the generally narrow Mediterranean shelf.
From this data set, for demersal resources, the problem of high fishing intensity and the need for effort control, is most urgent for these species in the sectors off Castellón de la Plana, along the narrow shelf from Nice to Genoa, in the eastern Tyrrhenian Sea, and in the Northwest Adriatic (Fig.3). Interestingly enough, the effort expressed per unit of surface appears well correlated with catches per unit area of surface (Fig.3, lower), perhaps providing evidence that there are local centres of abundance and fishery productivity and that the effort is directed onto them.
For small pelagics (Fig.4), the situation revealed by this preliminary analysis in most respects seems relatively similar to the picture for demersals.
One interesting concept that seems to emerge incidentally from this analysis and is touched on in the companion paper in this volume is that in some areas, both demersals and small pelagics are more lightly fished than in others. These areas, for example, Sardinia and Corsica and southern Sicily, the waters of northern Tunisia, and the Eastern Adriatic, may have constituted local refugia for the resources concerned. Certainly, ensuring that some lightly fished areas stay this way may be a rather effective approach to ensuring reproductive continuity for the whole GFCM Division concerned.
One suitable management approach that emerges from this analysis is that fishing intensity estimates by sector can be compared with biological estimates derived from survey data. Local adjustments can be made to fleet capacity until local fishing intensity has been reduced to levels corresponding to some Target Reference Point which has been agreed to by fisheries experts.
A number of production models have been fitted to data series of single and multi-species catches and fishing effort in the Mediterranean which are worth mentioning in this review. Single species production models have been fitted to several Tunisian fisheries (e.g. Ben Mariem et al., 1996), and a multi-species equilibrium production model was fitted by the GFCM ad-hoc Working group on Management of Stocks in the Western Mediterranean for the Gulf of Lions Fisheries (GFCM, 1988). This latter was perhaps the first cooperative assessment produced by a GFCM working group. These models have not attempted to correct for departures from equilibrium and could usefully be repeated using the latest FAO software (Punt and Hilborn, 1996) once the appropriate data are available.
A simple approach to modelling the impact of fishing intensity on overall multispecies catch rates using the Schaefer model (Munro, 1979; Caddy and Garcia, 1982) suggests that it is possible to relax basic assumptions of this type of model and assume comparability of production from ecologically similar sub-areas. In this `Composite Production Modelling' approach (see Annex) it is assumed first that the ecosystems of a series of sub-areas have similar basic fisheries productivities as judged by a mean (multi-species) catch rate. It is assumed, also, that catch rates in similar areas react in a similar way to comparable changes in fishing intensity. It also assumes that rapid changes in overall multispecies abundance are not occurring from year to year (which is perhaps a more valid assumption for the Mediterranean demersal fishery than elsewhere). If this assumption is correct, then mean catch rates in different sub-areas experiencing different fishing intensities (expressed as total HP or tonnage of trawlers per depth range and local area fished) can be combined in a common assessment. This can be done for the same year in different sub-areas or for a limited number of years in a series of sub-areas before fitting a common production model.
Garcia (1984) fitted this type of model for the fisheries of the Spanish littoral. His analysis suggested, for example, that the fisheries of South and Southwest Spain, Castellón and Alicante were significantly overfished with respect to MSY, but that Valencia was somewhat underfished. Such an analysis can be readily repeated once suitable sub-areas are identified which receive local fishing pressure from adjacent fleets/ports. It would provide the first step towards a visualisation of the actual situation of exploitation of key areas as a second step to the simple GIS approach described earlier. Obviously, a precondition for such an approach is to identify which areas of fishing grounds are fished by boats of which local ports.
In an extension of the Composite Production Model approach, also incorporating concepts from Csirke and Caddy (1983), Abella et al. (in press) estimated the overall mortality rate Z by size-frequency analysis of the survey catches from the MEDITS joint EC survey. These mortality rates were compared with the mean catch per unit effort for the same area. This analysis was repeated separately for three key species (red mullet, hake and Norway lobster) within each of 12 distinct sub-areas in the Ligurian and Tyrhennian Seas. These separate fishing unit areas were believed to be subject to different exploitation rates by different local fleets.
The estimated overall mortality rate for each species was expressed as a logistic curve fitting the standard survey catch rate (CPUEX) against the overall mortality rate (ZX) for the xth area studied. The peak of the curve of total production (fishing plus natural mortality) corresponds to the Maximum Biological Production (MBP) which was shown by Caddy and Csirke (1983) to be to the left of MSY conditions. Die and Caddy (1998) showed the mortality rate corresponding to MBP to be a reasonable and fairly conservative Target Reference Point.
The conclusion of the study by Abella et al. (in press) was that red mullet and hake are harvested above MBP levels in some areas. In one area where a complete fishing ban was imposed in 1990, catch rates of red mullet were very high (this shows, incidentally, that stocks can recover if lightly fished for a while). In general, the modified composite production modelling approach did not work well for Norway lobsters, which were believed to be either lightly exploited in most areas or not well sampled by the survey gear.
A number of vessel-gear categories are interchangeable in Mediterranean fisheries where a vessel can change from one gear/target species to another seasonally. Thus, the effective effort (f) exerted by vessels of category j which fish a given species at some time in the year is
F j << [No. vessels]* j [relative fishing power] *j [annual number days at sea] j
In the circumstance that the vessel is currently targetting species i, it would be misleading to calculate that the latent effort not used by this vessel for species i is automatically available to fish species i + 1..
Clearly, problems of this kind will need to be carefully resolved in discussing effort control for multi-species fisheries, but the advantages of concentrating on 1 or 2 target species assessments seem evident.
In order to further specify the level of effort exerted on each fishery, one must have some idea of the legislative and licensing requirements in place, which goes somewhat beyond the scope of this document, but this question cannot be avoided completely. In fact, three questions must be resolved at the policy level before further technical details can be addressed by scientific advisors. These are as follows:
1. To what extent should the fishing licence specify fishing method, species or gear type?
2. Is it feasible to go beyond issuing a multi-species licence and specify that a vessel be allowed to participate in more than one fishery, for example a demersal, small pelagic, and/or large pelagic fishery?
3. To what extent should the licence specify the area within which fishing is allowed?
Question 1: A tentative answer to the first question would note that specifying more than one allowable fishing method would reflect the reality of sequential fisheries for seasonally available resources and in fact would allow the fleet to divert effort to seasonally abundant resources. This might provide a motive for fishers to avoid exploiting those currently low in abundance.
At the same time, allowing too much flexibility in transfer could mean that all vessels might focus on a depleted resource with high unit value, causing even greater problems to this resource. This suggests that a legislative approach which allows selective closure of harvesting of specified fisheries where such resources are believed to be in danger, could be a useful adjunct to an effort control system as long as this does not increase discarding.
Question 2: Probably the issuing of licenses separately by species will be more feasible for species groups which co-occur in catches by the same gear. Apart from bluefin tuna, where ICCAT regulations, endorsed by GFCM, specify levels of harvest, four types of fishery can be categorised:
(a) Large pelagic resources, including bluefin tuna, swordfish, albacore and others of less importance;
(b) Small pelagic resources, including sardine, anchovy and possibly other species such as horse mackerel, bogue, etc.;
(c) Mixed demersal resources;
(d) The possibility exists, and seems worth considering, of adding specific licences for resources such as vongole, mussels, lobsters, shrimp, etc.
Question 3: Specifying fisheries access by area seems to allow closer 'tuning' of management response, especially where small management areas are feasible and would promote local responsibility and the comparative management approach. In this case, it is desirable to compare levels of productivity between areas subject to different fishing intensities (see Figs 3 and 4).
Seasonal openings and closures have been successfully applied where demersal resources are vulnerable to harvest shortly after recruitment to the bottom.
One approach to effort control in a multi-species trawl fishery used in Queensland, Australia may be of interest. There, each vessel participating in multi-species fisheries is assigned a hull mark for each fishery it is allowed to participate in. These marks together constitute a `package` of licenses that not only include gear/target species allowed, but areas/seasons where these may be fished. For example, hypothetically, a vessel may have the license to exploit:
- target species: A, D, E (with discarding for other species specified)
- with gear: trawl (specifications) for A,D + longline (specifications) for species A, E
- in zones and during seasons:
zone (1b): 1Jan to 31 Dec - Trawl
zone (2a): 15 Apr to 30 Oct - Trawl and longline
zone (2b): 1 Feb to 15 Apr - Trawl
The local authorities must be notified when the vessel switches between zones. In the Mediterranean, this change would have to be recorded by port captains, unless a transponder system is obligatory. In this manner, a nominal value of effort units would be accumulated for each area, gear type and zone fished and checked against that allocated to the vessel. In such a way, the current flexibility to exploit in sequence a number of fisheries and areas is not sacrificed. It is suggested that SAC commission a comparative study of the application of this and other systems of effort control as applied to multi-species fisheries elsewhere in the world.
