A. ZOOGEOGRAPHICAL CATEGORIES RELEVANT TO FISHERIES MANAGEMENT IN THE MEDITERRANEAN


Abstract

An overview is provided of the potential application of various geographical categories such as natural management areas, biocoenoses, unit stock areas and species refugia, as well as areas of influence of local ports, as they may apply to fisheries management in the Mediterranean. A brief overview of oceanographic features is provided. These are surprisingly stable from year to year and, together with high variation in physical characteristics, may imply a fairly high diversity of stocks for shelf and lagoonal species.

Species distributions from FAO Identification Sheets themselves are classified into faunal provinces, which generally coincide quite well with GFCM sub/divisions. The fragmentary information and criteria for deciding on unit stocks are reviewed. A range of scales is evident, from wide stock ranges for highly migratory species which extend from the Mediterranean into the Atlantic, through one or a few stock units for migratory and other large pelagic stocks. For small pelagics, the situation is more diverse, and stock units may reflect standing oceanographic features mentioned earlier. Strictly local stock areas may apply for demersal resources inhabiting long, narrow shelf areas with local mixing.

Although extensive shelf areas such as the Adriatic and the Gulfs of Lions and of Gabès are probably characterized by one or a few unit stocks, but long narrow shelf areas separated by "deep basins" are characteristic of much of the Mediterranean littoral and continental shelves. Under these conditions, larval diffusion from one side to the other of the Mediterranean seems unlikely, and for these narrow shelf areas either small stocks are likely to occur in sequence along the shelf or a continuous transition of genotypes is likely to be the dominant mode. From a management perspective, these latter resources could perhaps be best addressed as if local small stocks exist, by dividing the shelf into sectors each with its local ports and fleets and considering management first at the local scale, with integration as convenient to larger areas subsequently.

Introduction

In its suggestions for the work of the Scientific Advisory Committee, the Twenty-third session of GFCM proposed that this body place an early focus on the definition of fishery management units. This paper attempts to look at the possible geographical categorizations that might be appropriate or useful for various fishery management purposes in the Mediterranean Sea.

The main outline for the paper resembles that provided earlier (Caddy, 1989) for the WECAFC area, which was accepted as a basis for defining Natural Management Areas by that Commission. It may be noted, however, that a consultation of experts to attempt to define stock ranges and degree of sharing of resources had preceded the last-mentioned paper and facilitated the discussion. Such an analysis has not yet been carried out for the Mediterranean, as it has for the Black Sea (Ivanov and Beverton, 1985), and it is suggested that an analysis of the type implied by Figure 1 needs to be given priority in the Mediterranean region proper for the key migratory species. From general faunistic considerations, Fredj et al. (1992) pointed out that some 7-8 000 metazoan species have been registered from the Mediterranean in the MEDIFAUNE data bank but roughly 1/5 of these have been collected since 1950, meaning that we are not only dealing with a highly diverse system, but one where there are still considerable gaps in our knowledge.

Fig. 1. Distribution and regulations of Atlantic Bonito (Sarda sarda, Bloch) in the Black Sea from Ivanov and Beverton (1936)

Marine Populations: An Overview of Some Key Concepts

It would be useful to first examine some key concepts in population biology. From the perspective of fisheries assessment, it would be ideal if fisheries could be regarded as harvesting local populations of a single species, where a local population is defined by Sinclair (1987) as `a self-sustaining component of a particular species'. These effectively share a common gene pool and are implicit in the biological species concept of Mayr (1942), which is defined as "groups of actually or potentially interbreeding natural populations, which are reproductively isolated from other such groups" and we may note that this definition is valid for genetically isolated stocks also. Sinclair notes that fisheries scientists (see, for example, Heincke, 1898 and Hjort, 1914) were perhaps the first to recognize that such local groups of organisms were the ideal `working units' for fisheries management, even though, for reasons mentioned later, in relatively few cases are we able to say with certainty that this rigorous version of the `unit stock' concept applies to real-world fisheries. Sinclair also initiated the concept of `population richness' (Fig.2), which is the characteristic of some species as a result of their history and/or geographical discontinuities, to split into smaller self-sustaining local populations. Thus, there is a range of phenomena from single-population species such as the common eel where all Atlantic populations breed in the Sargasso Sea, through typical groundfish species such as cod or shelf pelagics such as herring, of which there are a number of local stocks, to species such as salmon and shad, which may have as many genetically distinct local populations as there are spawning streams. He also noted that reproductive isolation is important but can occur without geographical barriers (e.g. stocks A and B in Fig. 2), especially since, according to the scheme of Harden Jones (1968) in Figure 3, migratory fish may mingle on nursery or feeding grounds.

Fig. 2. Sinclair (1987) noted that a number of stocks may coexist in an area, and this represents its "population richness". If they are reproductively isolated, they may even co-occur on the same grounds

Fig. 3. Showing an identified typical scheme of life history migrations, with terminology after Harden-Jones (1968) from Sinclair (1987)

Categorizations of Species Ranges

An examination of species ranges of some characteristic organisms given by the maps in Fischer et al. (1987) shows the variety of phenomena the General Fisheries Commission for the Mediterranean needs to deal with (see Annex). Evidently, there is a wide range of different types of species distributions within the approximately 1 900 species of commercial and potentially commercial fish, invertebrates and plants listed by Fischer et al. (1987) for the Mediterranean. Despite a considerable variety of distribution patterns for individual species, a brief overview first allows one to discriminate a smaller number of characteristic distributions, a few examples of which are listed in the annex. It is of interest to note that for the Mediterranean fauna as a whole, Fredj et al. (1992) concluded that 69% of the fauna is of Atlantic origin, endemic species make up 28%, while some 5% are immigrants from the Red Sea via the Suez Canal.

The categorizations in stated in Annex may in some cases provide an indication of stock management units for migratory species, but in others it is not clear how many smaller stock units they include. Before going into further detail in the search for meaningful management units, it may be useful to look first at some general spatial or geographical categorizations commonly used in fisheries and marine ecology. Logically, it would seem important first to briefly review the main geographical features of the Mediterranean water masses to see what they may suggest in terms of standing structures and processes which will affect fish stock distributions.

Differentiation of Mediterranean Habitats Based on Oceanographic Criteria

According to Sinclair (1987), the completion of life histories of oceanic species is subject to spatial constraints more than trophodynamic ones, and such populations are regulated to a large extent in size and location by physical oceanographic processes, where he refers to oceanographic features which ensure closure of the life cycle as `geographic opportunities'. Adjacent stocks may mingle on the feeding grounds but are separated at spawning by each having different spawning area/season `windows' which are much more restricted than the stock range, at least for pelagics. Both Sinclair (1987) and Bakun (in press) suggest that these are defined by hydrographic conditions, thus Iles and Sinclair (1982) reported that "The number of herring stocks and the geographical location of their respective spawing sites are determined by the number, location, and extent of geographically stable larval retention areas". Small stocks, they note, are associated with small hydrographic features and vice versa. Given the structural complexity of Mediterranean shorelines and oceanographic features, such a categorization would seem to suggest, for shelf resources, that apart from extensive flat bottom areas such as the Adriatic, it is likely that a fairly large number of small stocks exist, rather than one or a few large population units. For migratory pelagic resources, the number of stocks may be much smaller, or a sin. In the Western Atlantic, for example, the Atlantic mackerel, Scomber scombrus, is divided into two populations, differentiated by large scale oceanic circulation features. One or a few stocks is probably the situation that applies for medium to large-sized pelagic migratory Mediterranean species, such as those in categories 4, 8 and 9 in the annex.

Available information from oceanography throws some light on the behaviour of water masses, which may help better understand some characteristics of the whole Mediterranean basin, faunistically speaking. A gross simplification is that surface water enters from the Atlantic through the Straits of Gibraltar, and proceeds to the eastern Mediterranean, en route progressively losing nutrients and increasing in salinity through evaporation. It eventually passes back out to the Atlantic at depth through the Straits of Gibraltar to form the high saline component of the Atlantic circulation. Among the local exceptions to this general scheme are a north-south nutrient gradient due to incoming nutrients from northern rivers (the Rhone in the West, the Po into the Adriatic and Black Sea inflows through the Dardanelles into the Aegean). The effect of this gradient of nutrient availability and other anthropogenic effects on trends in fisheries production has been described by Caddy et al. (1995) and also seems borne out by analysis of GFCM Statistics (Fiorentini et al., 1997). Of immediate relevance from the point of view of a possible differentiation of local stocks is the observation (Le Vourch et al., 1992) that there are many small scale oceanographic features in the Mediterranean and that these seem relatively stable. Apart from gyres and fronts, such features that lead to stock separation may also include estuarine circulations, coastal embayments, etc.

From the perspective of oceanographic circulation, Le Vourch et al. (1992) call attention to the presence of a system of cyclonic and anticyclonic gyres and fronts in the Mediterranean, and the following is largely extracted from their account. Many of these structures (see Tables 1 and 2) seem to have a standing or semi-permanent characteristic, being present from year to year though varying seasonally. One may suppose they play an important role, especially, but not exclusively, for pelagic fish and their differentiation by stocks or even populations. Although Sinclair inferred that larvae and juvenile fish show vertical migrations or even migrate actively to the nursery grounds against residual currents, the hydrographic regime is an important factor that helps ensure that a higher proportion arrive on the nursery grounds. The relevance of a stable hydrographic situation again comes from Sinclair (1987), who notes the importance of `geographically stable larval retention areas', and that "there has to be a water-column physical oceanographic (or geographic) system that exists in the same geographic location from year to year". We may consider the Mediterranean from this perspective using data from Le Vourch et al. (1992):

Referring to Figure 4a, we have, from west to east:

- Atlantic water entering the western Alboran Sea shows a strong anticyclonic gyre with upwelling along the northern edge. It is hard to imagine that one or more unit stocks of small pelagic fish are not associated with this structure.

- Like the last structure, there is evidence for the eastern Alboran Sea being partly separated by an Almeira-Oran front from the rest of the Western Mediterranean.

- The Algerian coast to Tunisia is an area of chaotic high energy and wide spatial variability in time, with transient gyres, and intermittent communication with Spanish coastal water. This may not be conducive to formation of stable stocks.

a)

b)

Fig. 4. General surface circulation in a) Western and b) Eastern Mediterranean of possible relevance to larval transport (from Le Vourch et al., 1992 )

Table 1. Distribution of thermal fronts in January (1980-87 inclusive), analysed from Le Vourch et al. (1992)

Date

Alboran

Sea

Gulf of

Valencia

Gulf of

Lions

Gulf of

Genoa

Sardinia to Tunisia

Gulf of

Gabès

NW to SW Adriatic

Off

Albania

SE of

Sicily

Off R'as

al Hilal

Off

Peloponnesus

N-Central

Aegean

NE of

Crete

Rhodos-

Kas

1/80

1

1

2

1

 

2

3

1

3

1

 

2

 

1

1/81

1

2

1

1

2

2

3

2

2

   

1

   

1/82

3

1

2

1

2

1

3

2

3

2

 

1

1

 

1/83

 

2

1

2

 

2

3

2

2

 

1

3

   

1/84

1

1

1

1

1

1

2

1

1

 

1

2

 

1

1/85

1

1

1

   

2

3

1

 

1

 

1

   

1/86

 

1

 

1

1

2

3

1

 

2

 

2

   

1/87

   

1

   

2

2

 

2

 

1

1

   

Table 2. Distribution of thermal fronts in July (1979-87 inclusive)

Date

Alboran

Sea to

Algeria

Gulf of

Lions to

Lies

d'Hyères

NE

Corsica

Corsica-Sardinia

Sardinia - Tunisia

Gulf of

Gabès

W+S

Sicily

S.

Messina

W.

Adriatic

Off

Albania

Peloponnesus

N.

Aegean

W . of Crete

Rhodos-

Kas

7/79

2

1

   

1

1

2

2

 

1

 

1

1

 

7/80

3

3

2

2

3

3

3

3

 

2

2

1

1

1

7/81

3

3

2

2

3

1

2

1

 

2

1

1

2

2

7/82

2

1

1

1

1

1

3

2

 

2

2

3

2

1

7/83

2

1

     

2

2

1

1

2

 

1

   

7/84

2

1

1

1

2

1

2

1

1

2

1

2

 

1

7/85

2

1

 

2

 

2

3

1

 

2

2

2

2

2

7/86

1

1

1

 

1

 

1

   

1

 

1

 

1

7/87

2

2

1

 

1

2

1

2

 

3

2

2

2

2

- Incoming Atlantic water masses often move north at the Straits of Sicily and take a reverse path back towards the Alboran Sea along the northern coast of the Western Mediterranean (Fig.3a). From this we might deduce that the most likely dispersion direction for drifting pelagic larvae and juveniles in the western Mediterranean, is from east to west in the north and vice versa in the south.

- Water masses entering the eastern Mediterranean move southeast along the African coast and up around the Levant to enter the Aegean, warming in the process, presumably with similar implications as for the last-described area above (Fig.4b).

For small pelagic stocks, however, we may look to evidence from oceanography to discriminate potential stock areas, either for feeding or reproduction. Tables 1 and 2 and Figure 5 from Le Vourch et al. (1992) show the distribution of thermal fronts in January and July respectively for a series of years. The remarkable thing is that, although there are some seasonal differences in the location of these structures, they tend to retain their identity from year to year. Although it is too soon to say that each of these is associated with its own pelagic stock, it appears a useful point of departure for investigations on stock structure to start with maps such as shown in Figures 4 and 5.

Fig. 4c shows clearly the large proportion of narrow shelf area referred to earlier and the relatively limited areas of extensive trawlable bottom in the Mediterranean (shaded), where unitary stocks may occur.

Some Categorizations of Relevance to Fisheries and Marine Ecology

In attempting to define Natural Management areas in an ecologically diverse area, the Carribbean and Gulf of Mexico (Caddy, 1989), four groups of geographical categorization were discussed which, with some modification, are as follows:

(1) the Faunal Province and the Natural Management Area;

(2) the area occupied by a homogeneous ecological community or biocoenosis (with its linkage to the related concept of the food web);

(3) the critical habitat and the refugium;

(4) the area occupied by a unit stock (the stock distribution area).

For completeness, we may add several other geographical categories of particular relevance for the Mediterranean, namely:

(5) the shelf area under the influence of a local fishing port.

The previously-cited WECAFC paper entered into a detailed discussion of categorizations of shared resources falling across maritime boundaries. Although these will not be the main thrust of this paper, they inevitably need to be borne in mind in the discussions that follows; especially for the few extensive shelf areas.

All of the above five concepts are widely (and loosely) used in fisheries and marine ecological literature, both in the Mediterranean and elsewhere. It is useful therefore to suggest some working definitions of these, especially since in the literature there may be other interpretations of these in use in different fields.

(1) Faunal Provinces and Natural Management Areas

The concept of the Faunal Province is not widely used in fisheries but implies a geographically extensive area where a set of interlinked fish communities or biocoenoses occur. The concept of a Natural Management Area (NMA) shares some of the characteristics of a faunal province but also implies that the whole of the area is managed, or can be managed, under a common framework. Thus, it goes beyond the Faunal Province in implying not only that the resources of this area are 'characteristic' but says something about the feasibility of joint management of the resources of the NMA from a socio-economic or political perspective. An example is the small island shelves of the Lesser Antilles island chain which are similar ecologically, and are considered to make up a NMA coordinated by the Committee for the Management and Development of the Lesser Antilles of WECAFC (Mahon, 1993).

Fig. 4c . Mediterranean shelf areas in relation to a line 12 miles distant from shore

Fig. 5. Distribution of thermal fronts (black and shading) from Le Vourch et al. (1992)

A similar analysis to Mahon's (op.cit.), similarly based on species distribution maps, in this case those in Fischer et al. (1987), has been attempted by Garibaldi and Caddy (in press), based on similarity/dissimilarity between points on a grid overlaid on the distribution maps in Fischer et al. (1987). In general, this analysis reveals two major features of the Mediterranean fish fauna:

(a) As would be expected from the dominant contribution of the Atlantic to the Mediterranean fauna (Fredj et al., 1992), the species richness declines from west to east (Fig.6a), with relatively fewer species in the eastern Mediterranean, and even fewer in the Black Sea, which resembles the North Atlantic in species diversity.

(b) The dissimilarity indices calculated from overlays of maps for 536 species occurring down to the 1 000-m isobath, suggested a series of discontinuities which might constitute faunal provinces (Fig.6a). These incidentally, tend to coincide spatially with the ten GFCM Mediterranean and Black Sea divisions used by GFCM (Fig.6b).

Essentially, the situation shown by the last-cited paper is the following;

- the most westerly faunal province coincides roughly with GFCM divisions 1.1 + 1.3 and includes the whole western Mediterranean, from the Straits of Gibraltar to the Straits of Sicily.

- GFCM Div. 1.2 (Gulf of Lions) is also distinct, but tends to continue eastward beyond the confines of the statistical area.

- The upper Adriatic forms a distinct area with lower diversity and coincides with the northern half of GFCM Division 2.1.

- The lower Adriatic and northern Ionian Sea (2.1 South and 2.2 North) are similar faunistically.

- GFCM Divisions 2.2 South, 3.2 West and 3.1 Aegean are all faunistically similar to each other.

- GFCM Divisions 3.2 East (Levant), 4.1 (Marmara), 4.2 (Black Sea) and 4.3 (Azov) each have their distinctive fauna from a perspective of biodiversity.

What other conclusions can we draw from this very preliminary analysis? In general, that existing GFCM Divisions seem to have a degree of validity, and that the high heterogeneity of the Mediterranean is confirmed from a biodiversity perspective. This does not mean of course that these classifications are ideal for all management purposes. These may be too coarse for local shelf stocks and too small for large pelagic fishes. They do, however, seem to provide some management units that can be used in `fall-back' mode in absence of better information on stock structure.

(2) Biocoenoses, Assemblages, Fish Communities and Food Webs

The `provisional definition' of a fish community from the above-cited WECAFC paper is as follows:

Areas where a statistically similar resident fauna/flora occurs, usually at a similar depth and bottom type, including a fairly well defined group of commercial species.

Commonly-used wordings such as `the biomass of fish in the area' often refer to a similar multi-species categorization. `Species Assemblage' as used by Tyler et al. (1982) implies a group of species that can be shown statistically to co-occur in the samples taken with a fishing gear in a given area. The commercial species in this fish community or biocoenosis, as a group, are roughly equivalent to the poorly-defined concept of `Multi-species stock'. A review of the situation for the Mediterranean is provided by Stergiou et al. (1997) for Hellenic Seas, who noted that a strong stratification by depth applies, with separate demersal assemblages recognizable from 0-50 m, 50-200 m and over 200 m.

Fig. 6a. Areas having similar numbers of species from Garibaldi and Caddy (in press)

Fig. 6b. Overlay of the GFCM divisions with the faunal provinces from Garibaldi and Caddy (in press)

Usually fish communities or biocoenoses are considered of scientific as opposed to managerial interest; however, heavy fishing can change species compositions, often resulting in new species replacing those formerly targeted. Such an effect seem to have been documented in the Gulf of Lions (Dremière, 1982).

Evidently, species that share the same habitat for most or all of the time may interact with each other in a number of ways, of which trophic or feeding interactions are currently given high prominence in the literature. The trophic categorization of Mediterranean fish communities would, in fact, be a first step towards fitting either an ECOPATH or ECOSIM model. There are, however, considerable practical difficulties in fitting either these or Multi-species VPA models as applied in the North Sea, to the geographically and taxonomically diverse Mediterranean system. There is little point in constructing such food web models if the species do not coexist for a significant proportion of the time over most of their species ranges, which again suggests focusing first on this aspect of the problem.

It may be imagined that given the likely spatial mosaic of fish communities, with the number of species concerned and the spatial heterogeneity of the Mediterranean, fitting a single trophic or multi-species model would be of dubious value. One would want first to determine homogenous faunal provinces as above and then see to what extent a single food web adequately explains interactions between the main species concerned. This theme is not explored further in this paper.

One conclusion tentatively drawn from this brief analysis is that although fish communities or assemblages exist in the Mediterranean, the strong local heterogeneity of many Mediterranean environments, and their stratification with depth, makes this type of classification not very useful for most fishery management purposes.

(3) The Critical Habitat and the Refugium

Reference is made by these terms to geographically restricted areas which have a particular importance for one or more species at one or more stages in their life history, which they are obliged to pass inside such an area. The particular utility of such definitions is in identifying and protecting areas which are believed to be especially susceptible to human influences. In MAP (Mediterranean Action Plan) parlance, these are also called `sensitive areas'. Several of them that may be mentioned are sea-grass beds which can be critical habitats for shrimp and some inshore fish and mollusc life histories, and deep water outcrops, which are often habitat for precious red coral, or spiny lobsters. We may also identify these areas with one or several fish communities or assemblages.

In general, spawning and nursery areas are considered `critical habitats' where they are restricted in extent and vulnerable to human damage of environmental change. For many demersal species (see, e.g., Fig.7), completion of the biological cycle implies seasonal vertical and/or horizontal migrations, such that the resource is rarely equally vulnerable in all areas and seasons. Though rarely used in this context, narrow transit areas (e.g. straits for migratory fish) may be considered also covered by the term `critical habitat'. Many `marine parks' aim to include some areas of critical habitats within them.

One conceptual framework that may be placed under the heading of critical habitat is the necessity for pelagic fish species of a combination of oceanic features being present in order to complete life histories. In his `triad' hypothesis, Bakun (in press) suggests that three classes of physical processes need to be present for the successful completion of life histories of many types of fish:

(i) An enrichment process (such as an upwelling or other source of nutrients);

(ii) A concentration process such as a convergence or frontal formation where food and larvae accumulate;

(iii) Retention processes such as those favouring the progeny staying within the `stock range' or drift towards an appropriate (e.g. nursery) habitat.

Fig. 7. Vertical displacements during life histories of 3 common demersal fish species(after Doumerge)

Evidence for the utility of this hypothesis comes from the north-central Adriatic, where sardine spawning has been linked to a frontal zone and an upwelling (Regner et al.,1988). If this is a tenable hypothesis, as noted earlier, a search for standing fronts and nutrient sources could help identify potential stock units of small pelagics in other areas.

The term refuge, "A shelter or protection from danger or distress" (Webster`s Dictionary) was used by Beverton and Holt (1957) to imply the effect of cover in controlling/reducing mortality. One other useful concept that has spatial implications is that of the `refugium' (Latin refugere, to escape), which is distinguished from the simple, mechanical protection offered by `cover'. Refugium is seen as referring to the ensemble of habitat and harvesting method which allows persistence of species. The particular short-term usage proposed for `refugium' in Anthony and Caddy (1980) is "A spatial or temporal enclave within the normal range of distribution of a species (elsewhere subject to catastrophic mortality prior to maturity), or a characteristic of the scheduling of harvesting mortality, which allows a proportion of the stock to survive to maturity and reproduce the population as a whole".

As we see elsewhere in this paper, the local heterogeneity of fishing pressure along the Mediterranean littoral can also contribute to the creation of a refugium. The two mechanisms implicit in the last definition (i.e. as an effective spatio-temporal closure and/or a characteristic of the harvesting process that allows escapement), have both been implicated by Abella et al. (1998) for Mediterranean hake. Persistence of this species despite highly intense fishing on juveniles is seen as the net result of the uneven distribution of trawl effort spatially, the distribution of mature fish outside the main trawling grounds and the relatively low vulnerability of large fish to fine mesh trawls. As must be evident, these effects were not planned in the past but should be taken into account in planning future management measures. The possibility that discontinuities of harvesting between successive ports along a long narrow shelf may be considered (e.g. Fig. 8) as providing conditions for a refugium to exist.

(4) The Area Occupied by a Unit Stock (the Stock Distribution Area)

The `stock-distribution area' is the classical fishery management unit, and the limited information available on this is examined at in the following section, preceded by a summary of current conceptual approaches to defining a 'stock' used in other world areas.

Definitions of Stocks

A good review of the pros and cons of different stock definitions is given in Carvalho and Hauser (1994), varying from the Darwinian population of 'genetic stock' to empirical and pragmatic definitions based on the need to separate species into discrete management units. Their conclusion that "there is no universally applicable definition of the term stock" deserves attention, since it make this term dependent on its use and context.

Two main usages of the term stock are found in the literature: one refers to separately manageable adult populations, the other is a more strict, `Darwinian', definition. As noted, this latter definition sees a stock as effectively identical to a self-reproducing, isolated and even genetically distinct population.

The pragmatic usage for a `Unit Stock', that was particularly promoted by J.A. Gulland, is to consider a stock as a useful working definition of the biomass and cohorts of adults of a species in a given area, which is subject to a well defined fishery, and is believed to be distinct and with limited interchange of adults, from biomasses or cohorts of the same species in adjacent areas (Gulland 1983).

As he noted, "Little more than passing attention is normally given in the early stages of an investigation to the question of what constitutes a unit stock. A common course is to take the data referring to the fishery of direct concern and proceed directly to the analysis of major interest (estimating growth and mortality, plotting CPUE against average effort, etc.). The implicit assumption is that the fish do effectively constitute a unit stock. If possible this assumption should be checked".

Implicit in this last sentence is the admission that a `working definition' of a unit stock can be adopted in lieu of later, more detailed analysis, if the fisheries of the area suggest a discontinuity of adult distribution of the `unit stock' from adjacent ones.

Later in the same book he suggests that the possible existence of separate stocks within the group of fish managed within in a region does not constitute a serious problem, as long as the distribution of fishing between them remains unchanged and as long as there is not an expansion of the range of the fishery which might progressively add new populations to the supposed `unit stock' with time. If this occurs, the rise in catches from the `new' stock may be balanced by the decline in catches from the `old' one, such that a misleadingly stable production appears to be achieved. Based on experience, inter alia, with groups of tropical shrimp species, many would regard the above stock definition as not fully satisfactory or even dangerous. In many cases, including the Mediterranean, this assumption de facto has usually been made for lack of data.

Some further comments attributed to Gulland (from Gauldie, 1991) make it clear that, in his view, a genetically isolated population is the ideal stock definition but is rarely achieved, and that empiricism is often needed:

"Ideally, a unit stock is a self-contained and self-perpetuating group, with no mixture from the outside and within which the biological characteristics and impact of fishing is uniform. Such a stock would also be a genetic unit. Few groups of fish form such neat units, and the choice of what should be considered a unit stock has usually to be made empirically.....In practice, it is usually convenient to treat as large a group as possible as a unit stock.... Even when the group of fish being treated as a unit in fact contains several independent unit stocks, serious errors will not be incurred if the values of growth, mortality, etc., do not vary between the stocks".

Fig. 8. Distribution of fishing effort/mortality across a long, narrow shelf and the possible creation of refugia

Several criteria were offered by Gulland for distinguishing stocks. These are given in the following table, and it is suggested that they be applied to management units of Mediterranean resources:

Suggested actions to discriminate

adjacent stocks

Data and research needs

1. Examine the distribution of fishing in space and time (since seasonal migrations may occur). Is there a gap in effort distribution between this resource and adjacent `stocks'?

Need seasonal data on distribution of all vessels fishing the local and adjacent units of the species in question.

2 Spawning areas and seasons should be local and distinct from other spawning areas, and show some correspondence spatially to the units being fished locally. Spawning areas should also be identified for adjacent units believed to be distinct.

Need identification of spawning areas and seasons, perhaps by locating ripe and mature fish or by egg and larval survey.

3. Values of population parameters are different between adjacent fisheries. Here he makes special reference to differences in growth/mortality rates between areas believed to correspond to distinct stocks.

Recent GFCM work has focused on parameter measurement. [Comment: Differences in growth rate are suspect, given that growth curve fitting notoriously depends on outliers. Differences in mortality or age/size composition may depend on geographical segregation by age in the same stock, and hence also be suspect].

4. Morphological or physiological characters may be distinct as may be parasitic populations.

Careful measurement/enumeration of characteristics from all possible populations in a region, preferably when separated on the spawning grounds. This may be feasible if there is perfect stock mixing, but the same `population' may show a progressive `drift' in a characteristic with distance (see text).

5. Tagging and distribution of returns differs geographically between adjacent fisheries.

The cost of tagging tends to confine this approach to large/valuable species. For deep water fish and other species there is a problem of low survival/returns. To be used selectively!

6. Genetic/physiological discrimination.

This method was formerly prohibitively costly, but it is progressively more economically feasible to use methods such as DNA, blood-protein typing, etc., but still on a selective basis. Vertebral fin ray counts have also been used, but are rarely unambiguous and may vary due to individual environments in development.

In conclusion, criteria 1 and 3 seem promising, but will need good data on spatial distribution of fleets (which will also be needed for effort control), and would be automatically available if `black box' telemetry systems were established on a wide scale. Tagging and genetic discrimination are more definitive, however (see next section), and are beginning to be more widely used.

Discriminating Stocks from Oceanographic Data

An earlier section has touched on this question, but commenting further on the use of oceanographic data to differentiate stocks, we may note the following:

(a) There is clearly a problem with definition of stocks that inhabit narrow shelf bathometric zones. In the Mediterranean these may extend for a thousand or more kilometres but may be only several kilometres or less wide. For example, a narrow shelf spans 16 degrees East-West from the Straits of Gibraltar to the Sicily Channel, and extensive strips of narrow shelf also occur, especially in the southern and eastern Mediterranean. Are the shelf resources here to be treated as `long narrow stock units'?

At least potentially, there seems the likelihood of a sequence of small stocks along the shelf, or for allopatric species, differentiation has occurred to result in a `cline' along the narrow shelf area. Riggio and Chemello (1992) have in fact suggested that coastal lagoons and other structural features may have contributed to local `endemisms'. These in turn, could have recolonized the Mediterranean proper, leading presumably to a wide diversity of genotypes and small local population units. Adults of many non-migratory demersal fish will almost certainly not move from one extreme point on such narrow shelf areas to the other, and even larval propagation will not result in perfect mixing. Based on experience elsewhere, a `cline' of characteristics may result (e.g. Waples, 1987), perhaps interrupted at one or more points by a genetic discontinuity. It would require extensive investigation to establish this type of phenomenon for the Mediterranean, but evidence so far does not provide a clear case for separate populations occurring with precise boundaries along narrow coastal strips.

The definition of self-reproducing units, especially for semi-sedentary or territorial species, may be replaced for the moment in practical management terms by `local stocks', defined as: "those resources within the (exclusive) range of action of boats from a given port or ports", hence, the reason for including category 5 in our list of geographical concepts.

(b) In theory, for a semi-enclosed sea where "deep basins" separate opposite shelves and their adult populations, the possibility exists of larval dispersion from one shelf to another. If it regularly occurs, it could have one other result, namely that larval loss rates (or `vagrants' in Sinclair's terms) due to flushing and loss of larvae from shelf waters may be much lower for shelf resources, than along open oceanic coastlines. Evidence from Mallorca (Oliver, 1994), seems to preclude this happening regularly and, in fact, from movements of water masses Oliver estimated it would take 1 month for passive drift to transport a larva between the Catalan coast and the Baleares platform. This can be contrasted with two months for a hake to grow from egg to an alevin of 3-4 cm length. If this argument is followed, with the current system in the Western and Eastern Mediterranean (see figs. 4a and b), larvae are unlikely to cross "deep basin" waters from one shelf to the other. Some indications such as the accentuated fluctuations in hake landings in Mallorca, compared with those for the Spanish mainland, may suggest for example that the hake stock on the Mallorcan shelf is a largely separate population from those on mainland shelves.

Larvae from the western Mediterranean are also likely to stay there, and, if larval dispersion plays a role in recruitment of Mediterranean shelf resources, it is likely that dispersion along-shelf is going to be the main modality.

It would seem unwise therefore, to assume that planktonic larvae have not evolved specific mechanisms to return to their stock of origin, and in a similar contex - small island reefs - Bakun (1986) demonstrated that hydrographic structure can result in maintenance and self-regeneration of local stocks. Sinclair's `member-vagrant' hypothesis implies that a significant proportion of offspring do not return to the nursery area of the stock but that their future contribution is lost to the stock as `vagrants'. For a semi-enclosed sea one must ask the question, however, if these larvae are inevitably lost or added to other populations with consequent stock mixing. Available information seems inadequate at present to fully resolve this question for the Mediterranean, and in fact there seems little evidence from the Mediterranean fishery literature that the issue of stock discrimination has been looked at very seriously.

Discriminating Stocks from Biological Data

With respect to using biological data to discriminate stocks, growth rate studies in Italy (Sicily) had concluded for red mullet that Sicilian Channel stocks were different from other stocks in the Mediterranean (Levi et al., 1990). Similar conclusions were drawn from growth data for sardine from Aegean and Ionian seas by Tserpes and Tsimenedes (1991).

Conclusions on stock separation were drawn from differing biological data of hake and red mullet stocks in Greek waters (Vassilopoulou and Papaconstantinou, 1988). Separate hake stocks were believed to be each associated with a spawning area in the Patraikos Gulf and another in deeper Ionian waters. Unit hake stocks may reasonably be associated with nursery or spawning areas such as the Jabuka pit (Alegría-Hernández and Jukic, 1988) in the Adriatic. For species of Mugil and Dicentrarchus that ascend into coastal lagoons or estuaries to spawn, we may also expect a proliferation of local stocks associated with local lagoons and rivers.

Evidence on Stock Structure Gained from Genetic Studies

So what is the evidence as to the genetic population units of Mediterranean fish species given that Roberti et al. (1993) have little faith in morphometric information to distinguish stocks: "the characterizations of fish stocks... carried out from phenotypic data sets.... are known to be accompanied by a considerable uncertainty". A good review of relevant considerations in using genetic information for stock discrimination is given in Carvalho and Hauser (1994).

For large pelagics, Alvarado-Bremer et al. (1995) found swordfish samples taken globally contained two clades, one ubiquitous, the other which diverged some 550 000 years ago and likely originated in the Mediterranean during the Pleistocene. Although this study did not look at stock differentiation of swordfish within the Mediterranean, it provides some support to at least looking at Mediterranean swordfish stocks separately, without considering Atlantic stocks as for bluefin tuna. In fact, de la Serna et al. (1992) suggested at least two stock units in the Mediterranean: one, in the western Mediterranean, was indistinguishable from Atlantic populations, other(s) occurring further east, presumably proprietary to the Mediterranean proper. Further work on this issue for large pelagic fish is underway at ICCAT.

Small pelagics: Several sardine stocks are believed to exist in the Western Mediterranean, with semi-independent although not completely isolated populations, and a north-south cline has been detected genetically. The stock of sardine in the Alboran Sea were confirmed to be separate from the rest of Mediterranean populations. For anchovy, Bembo et al. (1996) found that Aegean fish differed significantly

from western Mediterranean fish sampled but that, even within the Adriatic, some heterogeneity of anchovy populations apparently exists.

Demersals: There is some evidence for sea bass that coastal fish may be differentiated genetically. Some preliminary information now emerging from genetic analysis of selected Mediterranean marine species used as aquaculture broodstock, provides some conflicting results. For the nearshore and estuarine Mediterranean sea bass, Castilho and McAndrew (1998) found high levels of polymorphism at microsatellite loci which suggests populations are structured at a small geographic scale. For example population units are perhaps separate between, inter alia, Tyrrhenian Sea, the South and the North Adriatic, Crete and East Sicily, Gulf of Valencia and the Gulf of Lions.

For gilthead sea bream, Magoulas (1998) did not find conclusive evidence of population differentiation from a limited number of samples of this species from several Mediterranean locations, which may suggest from very preliminary results that for fully marine species there is a fair degree of mixing of populations.

For other areas outside the Mediterranean, a study of inshore fishes along an extensive shoreline down the Californian coast (Waples, 1987) showed that significant levels of genetic differentiation occurs with distance in these fish and that this is determined primarily by a balance between gene flow and genetic drift and is not related to dispersive capability of adults or larvae. Relevant here is the conclusion on an American flat fish, where Brown et al. (1987) showed that adjacent stocks which had an estimated exchange rate of 10%, still reacted independently to exploitation, and Carvalho and Hauser )1994) suggest this percentage as a limit, above which treating stocks separately at higher levels of intermingling is impractical, and could lead to incorrect management results.

Demersal stocks tend to be considered as bounded by semi-enclosed seas such as the Adriatic, and Aegean, where existing GFCM sub-areas probably form quite reasonable stock boundaries. However, genetic analysis (Spain) suggest that for western Mediterranean hake at least two stocks (genetic strains) coexist: one native to this sea, the other similar to Atlantic hake stocks and found mainly on the southern shelf. The division of hake into eastern and western races seems supported by vertebral counts (Orso-Relini et al., 1994). Other shelf resources such as the common sole (Kotoulas et al., 1995) show a significant correlation between genetic and geographic distances, within its range with an estimate that panmictic populations lie within a radius of the order of 100 km.

Some particular problems must be faced in the case of the Mediterranean and other semi-enclosed seas in relation to stock definition, which are not evident from experience with extensive Atlantic fishing grounds, and stem from an understanding of bottom topography and associated oceanography.

The Shelf Area Under the Influence of a Local Fishing Port

In a recent paper, Caddy and Carocci (MS) proposed a flexible approach to modelling the distribution of fishing effort with distance from port, which would especially apply for local ports operating in sequence along narrow shelves (Fig.9a). This approach proposes that inshore fishing effort applied from a port is bounded by geographical features of adjacent shelf areas. For day trips, effort and fishing intensity would then usually be functions, amongst other factors, of the distance from port. They also suggested (Figs.9b and c) that areas of interactions of adjacent ports be identified and, where possible, used as sites for closed areas or reserves. Since these `boundary zones' would potentially be areas of contention between vessels from adjacent ports, such a siting of e.g. marine parks, may be more easily accepted by the fishermen from adjacent ports. Identifying shelf areas where fleets from one or several ports act alone or in combination, to fish local resources, would ideally lead to a sequence of small management units fishing the unit resources of long narrow coastal areas and continental shelves. These fishery-based units are equivalent to the 'harvest stocks' of Carvalho and Hauser (1994).

Some support for this approach is provided by the Atlas of the fisheries of the Western and Central Mediterranean by Garcia and Charbonnier (1985), which divided the fishery into largely arbitrary `sectors', which may not always correspond to the national administrative units but are presumably intended to correspond to ecologically similar areas within which one or a few main ports and their coastal fishing grounds are situated. An example of a map of such sectors is referred to as Fig. 2 in the other paper in this issue of Studies and Reviews. It is suggested that such sectors, or the national subdivisions that best correspond to them, be erected as working units for collection and analysis of local data, to be combined, if this seems appropriate for the fishery or resource distribution in question.

Fig. 9a. Showing the predicted effect of distance from port on fishing intensity for four ports

Figs 9b and 9c. Probabilities of conflict between vessels from 4 adjacent ports given the ports in Fig. 9a. Areas where a roughly equal share (40-60%) of effort is predicted to come from adjacent ports are delineated in dark grey.

Summary

1. Shelf Resources

In summary, although there are significant differences in species composition from sub-region to sub-region, we can say that many of the commercially important shelf species are ubiquitous, but since they are confined to shelf areas, stock distributions are likely to occur in sequence along narrow shelves or form a cline of gradually changing characteristics.

Some local stocks, e.g. demersals lying on narrow shelves within territorial waters, are purely national and exclusively fished by the local port or ports. Here, as for lagoon fisheries, in consistency with the empirical approach of Gulland to defining stocks as `fishable units of adults', the smallest fishery unit would consist of one or several ports and their local fishing grounds. One would, of course, need to look more broadly in defining population reproductive units; however, it seems unlikely for shelf resources of a semi-enclosed sea with "deep basins", that larvae from stocks on one side of the sea recruit to shelf areas on the other.

The author's conclusion, therefore, is that, in absence of other data, it would be practical to use GFCM sub-areas as NMAs and within these larger units, especially for narrow shelves, define arbitrary port-ground units as the smallest management units. For larger extensive shelf areas such as the Adriatic and Gulf of Lions/Gabès, the conventional approach of assuming, unless otherwise evident, that unit stocks of the key species coincide with these areas, seems the logical interim course of action.

2. High Sea Resources

Large pelagics have generally wide stock distribution units. For bluefish, bonito, and some of the large scombrids, following Gulland's criteria, one can reasonably assume a Mediterranean-wide distribution (and even beyond) for some stocks until further information is available. Dolphin fish is also believed to enter Mediterranean from the Atlantic seasonally, and in this sense perhaps has a stock status similar to bluefin tuna, although confined to the warmer parts and seasons of the Mediterranean. For bonito and chub mackerel the stock range formerly included the Black Sea, but they are relatively rare there now though still abundant in the Mediterranean, hence their range may now be smaller than formerly. For swordfish, the possibility that there may be at least two stocks in the Mediterranean seems a reasonable working hypothesis. Like Bluefin tuna the Atlantic swordfish stock also enters the Straits of Gibraltar and the Western Mediterranean, but it is presumed that at least one other stock occurs further to the east.

With respect to bluefin tuna and dolphin fish, there seems good evidence that these form unit stocks that extend beyond the Mediterranean into the Atlantic; for the latter, this seems necessary in the winter months when Mediterranean temperatures fall below the species minimum temperature tolerance. This might also be the case for a number of large- and medium-sized pelagics such as bluefish, bonito and mackerels, but these are more likely to form unit Mediterranean stocks, judging from their species ranges (see Annex). For the last two mentioned species, their range formerly also included the Black Sea but, due to environmental degradation, it is not clear to what extent this is still the case. In general, there is a decline in species diversity going from the Western to the Eastern Mediterranean, via the Black Sea to Azov (where some freshwater species were formerly abundant).

A review of possible geographical categorizations of importance to fisheries management confirms that, from a biogeographical perspective, the different sub-area of GFCM can be regarded as Natural Management Units, even though the situation with respect to stock distributions does not uniformly support any one geographical categorization as applying to all species.

For several groups of species distributions, where these almost exclusively coincide with sub-areas such as the Black Sea and Marmara, and the Levant region for Red Sea immigrants, we can probably assume they form isolated unit stocks.

A division between east and west Mediterranean at the Straits of Sicily seems likely to apply for many species at the stock level and also poses a barrier and limit to distribution and eastward movement of others

Small pelagic species may be distributed in stock units that are probably linked to areas of upwelling or other nutrient sources and zones of convergence.

For demersals and even small pelagic resources, semi-enclosed basins such as the Gulf of Lions, Adriatic and Gulf of Gabès, the concept of a unit resource probably applies. For the Aegean, this is probably also the case for the small pelagics, but the demersal resources of the narrow and indented coast and island archipelagos, as well as the demersal resources of larger islands such as the Balearic shelf, Sardinia, Corsica, Sicily, Malta, Cyprus and Crete, may reasonably be regarded as unit populations.

Limited genetic information currently becoming available supports the idea that, for inshore species of narrow continental margins, apart from the extensive areas of shelf just mentioned, these long narrow coastal and shelf areas are inhabited by a series of local or sub-regional populations which may mix with their adjacent local populations. As such, the genetic characteristics of the species change gradually with distance along the shelf. Such evidence as exists does not seem to support extensive larval mixing between the same species on opposite sides of the Mediterranean, and this may also be unlikely between the larger islands and the mainload.

Given the above considerations, it would seem that the best working solution for shelf resources would be to decide on a relatively fine scale of unit areas such as the `sectors' erected by Garcia and Charbonnier (1982) for data collection. Data from these could be combined as appropriate, if it is necessary to consider stock or natural management areas with wider ranges of distribution.

Given the geographical characteristics of the Mediterranean, with local occurrence of untrawlable bottom along narrow shelves and along the continental slope, some areas of shelf may be lightly fished. These may act as refugia allowing recruits to survive to reproductive size and move to adult distribution areas, often further offshore. The distribution of fishing effort on demersal resources is therefore probably concentrated on trawlable areas and in the vicinity of ports. Relatively few boats from Mediterranean ports exert distant water effort offshore from the local fishing grounds of other countries, and fishing effort on the deeper shelf and slope is constrained by bottom topography, except for long line and trammel nets. Such a situation, even if at present largely inferential, supports the utility of creation of `corridors' or closed areas or marine parks between adjacent ports. In addition to improving escapement, these would reduce conflict between vessels from adjacent ports and might provide the basis for separating individual fishing zones along the shelf within which limited effort quotas could be allocated without risk of overlap.