© 2004 by ICES/CIEM International Council for the Exploration of the Sea/Conseil International pour l'Exploration de la Mer
Fishing strategy development under changing conditions: examples from the French offshore fleet fishing in the North Atlantic
IFREMER, 150 quai Gambetta BP 699, 62 321 Boulogne sur Mer, France
*Correspondence to P. Marchal: tel: +33 321 995600; fax: +33 321 995601. e-mail: paul.marchal{at}ifremer.fr.
A typology is presented for the French offshore fleet fishing in the North Atlantic. The investigation was carried out separately for all years between 1985 and 2002. Two methods were combined: PCA and cluster analysis. In 1985, most vessels targeted saithe off western Scotland (vessels from Lorient) and in the northern North Sea (vessels from Boulogne and Fécamp). Then, probably because of the decline in biomass of saithe, some vessels started to target deepwater species from the early 1990s. In 2002, some vessels fished exclusively for saithe (>80%), while others targeted mainly deepwater species. In all, 12 fisheries are identified for the period 1985–2002. Results are given of a preliminary investigation trying to identify the external factors (stock biomass, catch limits, price) that may influence the shifts in fishing strategies.
Keywords: deepwater species, fisheries dynamics, fishing strategy, mixed fisheries, multivariate analyses, saithe
Received 2 January 2004; accepted 29 July 2004.
 |
Introduction |
|---|
 Top Introduction Material and methodsResultsDiscussionReferences |
|---|
In waters of the European Community, fisheries are regulated primarily through single-stock total allowable catches (TACs) and quotas. However, a large number of fisheries are of a mixed nature. In mixed fisheries, several species are caught simultaneously during a haul, and one species may be fished by different gears (trawls, gillnets, etc.). Ignoring the mixed nature of fisheries may result in inappropriate management. For instance, fishing for one species could lead to discards of another, whose quota had already been exceeded. Both the scientific community and decision-makers have acknowledged the need to account explicitly for the mixed nature of fisheries in advice and management (Laloë et al., 1995), and actions have been taken in that respect (ICES, 1991, 1992, 2003; EC, 2002). In order to shift from single-stock to mixed-fisheries advice, scientists first need to define operational fisheries, as a cornerstone on which future advice to management bodies can build (ICES, 2003).
Different approaches have been proposed to identify fisheries based on catch and/or effort data. One set of approaches is based on multivariate analyses. Biseau and Gondeaux (1988) proposed a method based on principal components analysis (PCA) to define French fisheries (referred to as métiers) in the Celtic Sea, carrying out two analyses of two types of variable. The first was based on the relative landing proportion of each of the main species per vessel, and the second on the time spent by each vessel in different spatial units. Those authors then identified fisheries by contrasting the results of the two PCAs. Lewy and Vinther (1994) identified Danish North Sea trawl fisheries through a hierarchical agglomerative cluster (HAC) analysis, based on the fraction of the value of different species to the total landings value for each fishing trip. More recently, Pelletier and Ferraris (1999) used a two-step multivariate approach to identify fishing tactics in Senegalese fisheries. First, catch profiles were obtained from trip-by-trip landings composition, with a non-normalized PCA and an HAC. Then, a multiple correspondence analysis (MCA) and an HAC were applied to catch profile, fishing location, gear, and month. One major drawback of the above approaches is that fisheries were identified over a limited period of time. Given the volatility of fleet dynamics, such fisheries definitions cannot reasonably be extrapolated to other time periods.
The alternative approach found in the fisheries literature consists of classifying fishing trips on the basis of arbitrary criteria. Such criteria could be either a minimal catch proportion of target species in the total landings (Biseau, 1998), or a combination of fisheries inputs (gear, mesh size, fishing area) defined on the basis of expert knowledge (EC, 2002; ICES, 2003). The major drawback of this approach is the arbitrary character of the criteria used to identify the fisheries.
This paper proposes an approach based on multivariate analyses (PCA, HAC) to identify fisheries, to describe the dynamics of these fisheries over a period of time, and to gain insights into possible mechanisms that may have induced the dynamics. The concept of fisheries is here assumed to be equivalent to that of fishing strategies, which in turn are defined as the sum of fishing tactics over a calendar year (Laloë and Samba, 1991; Ferraris, 1995). Consistent with Laloë and Samba (1991) and Laurec et al. (1991), fishing tactics, or métiers, are here characterized by a combination of target species, fishing area, gear, and time of year. Identification of fishing strategies consisted of constructing groups (or clusters) of vessels that fished for the same species with the same gear in the same area during a year. Additional information was used to characterize these clusters. In contrast with earlier studies, the analyses were carried out separately for each year of a time period in which different fisheries possibly emerged, shifted, or disappeared. Such an approach should allow insights into the strategic evolution of fleets and vessels. The shifts in fishing strategies were then contrasted with variables that characterize, to some extent, the dynamics of the external environment of fishing vessels (catch limits, stock abundance, price of landed species). The analysis was applied to French offshore otter trawlers fishing in the North Atlantic over the period 1985–2002.
|
Material and methods |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Commercial catch and effort data
The fleet selected for the analysis was French otter trawlers at least 40 m long. Catch and effort data by fishing trip and fishing area were made available over the period 1985–2002. During this period, from 1994 to 1998, data from St Malo vessels were not registered in the database. In 1994 and 1995, data from freezer vessels (four vessels) were not available.
The locations of the home harbours of the fleet analysed are shown in Figure 1. The main fishing harbours were Lorient and Boulogne. From 1985 to 2002, the number of French offshore vessels decreased drastically, but two categories can be distinguished. The first included vessels of 350–800 grt and power >750 kW. These vessels fished exclusively in the Northeast Atlantic. The second category included vessels of grt >1000 t, which fished in the Northeast Atlantic and Canadian waters. Trawlers fishing in Canadian waters were mainly registered in St Malo and Bordeaux (Figure 1). In the 1980s, vessels fished essentially demersal species, generally saithe (Pollachius virens), cod (Gadus morhua), and a few deepwater species, such as blue ling (Molva dypterygia) and ling (Molva molva) (Figure 2). From the early 1990s, trawlers started exploiting new deepwater species (Figure 3), specifically roundnose grenadier (Coryphaenoides rupestris), black scabbardfish (Aphanopus carbo), orange roughy (Hoplostethus atlanticus), and deepwater sharks (mainly Centroscymnus coelolepis). The spatial distribution of the catches of the main species in 2002 is given in Figure 4. This study was based on the main species harvested (Table 1). The stock definitions considered in this study are those used by ICES (2002)1, those harvested outside ICES waters (i.e. in Canadian waters) being referred to as "Canadian stock", and those harvested outside any assessment area being referred to as "other stock".
|
|
|
|
|
The spatial distribution of catch and effort by ICES rectangle in 1985 is poorly documented. Missing data amounted to 39% (fishing effort), 40% (saithe catches), and 70% (blue ling catches). By 1992, missing data amounted to 13% (fishing effort), 6% (saithe catches), 23% (blue ling catches), and 25% (roundnose grenadier catches). The spatial distribution of effort and catch by ICES Division was fully documented over the whole period of investigation, so in the analyses that follow, the basic spatial unit chosen to aggregate catch and effort data is the ICES Division (Figure 1).
One aim of this work was to obtain an annual typology of French offshore trawlers >40 m long fishing in the North Atlantic. Vessel adherence to a type/strategy was based on the weight of their landings (Biseau and Gondeaux, 1988). The proportion of landings of a stock relative to the total landings was
|
|
|
Factorial analyses and classification
A principal components analysis (PCA) was first applied annually to Xy(V,S). Consistent with Biseau and Gondeaux (1988), the only transformation applied to the data was centring. Rescaling the data to unit variance would imply that the same weighting was given to target species and bycatches, an inappropriate procedure in the context of this study, in which fishing strategies are expected to depend primarily on target species rather than bycatch. Consequently, data were not rescaled. A PCA may be used to show principal variability sources and to identify optimal factorial plans (Blanc et al., 1976; Lebart et al., 1995).
A hierarchical agglomerative clustering (HAC) was then applied annually to Xy(V,S). This technique aims to group the V vessels into clusters, which are both well separated and as homogeneous as possible with respect to the catches observed. Clusters are built by single link, and distance between individuals is a measure of dissimilarity of the closest pair. HAC agglomerates vessels by adding the shortest possible link (Jardine and Sibson, 1971). Several measures of distance can be used, but given that input data are percentages, Euclidean distance appeared to be the reasonable choice (Gordon, 1981). Results are given in the form of dendrograms, which allow identification of fishing strategies. Stock landings were grouped into four categories: <20%, 20–40%, 40–75%, and >75% of total landings. PCA and HAC were run with the statistical software S-Plus 6.1 (2001).
The only theoretical limit to the number of clusters considered is the total number of vessels in the fleet. Increasing the number of clusters leads to a gain in precision in their definition, but this gain in precision decreases as the number of clusters increases. If the fishing strategies derived from the clusters are to be used as operational units for modelling purposes, there is no point inflating excessively the number of clusters up to a level at which the gain in precision is negligible. To our knowledge, there is no objective method for identifying the most appropriate number of clusters (and of fishing strategies). ICES (2003) provide some guidelines to help identify the number of fishing strategies that ensures reasonable balance between precision and operational requirements. That approach, which combines the outcome of multivariate analyses with expert knowledge, has been applied in this paper. First, clusters are described at their most accurate level. Second, the number of clusters is reduced on the basis of expert knowledge. Harbour origin seemed to be an element of distinction between fishing strategies, and it appeared to make little sense to group vessels from Lorient with those from northern France within the same fishing strategy. Knowledge of the ecology and distribution of deepwater species was also taken into account in reducing the number of clusters.
Roundnose grenadier are distributed along the slopes of the temperate North Atlantic at depths of 180–2200 m (Fontaine, 1979; Haedrich and Merrett, 1988; Quéro, 1997), but north and west of the United Kingdom (ICES Subareas VI and VII), they are most abundant at about 800–1000 m (Ehrich, 1983; Magnússon and Magnússon, 1996; Quéro, 1997). Black scabbardfish and deepwater sharks are usually distributed on the same fishing ground (ICES, 2002). Further, all these species are generally scattered over large areas (ICES, 2002), so trawlers targeting roundnose grenadier may catch black scabbardfish and deepwater sharks as bycatch, and total catches of these three species can be considered to constitute a single fishing strategy. Blue ling and orange roughy have a different distribution from the other three species/groups. Blue ling aggregate and are most abundant in shallower water (350–500 m; Quéro, 1997). Orange roughy also live within dense aggregations (Lorance and Dupouy, 2001), but they are most abundant in deeper water (900–1200 m; Quéro, 1997). Therefore, the fisheries targeting blue ling and orange roughy are not grouped with those targeting the three other deepwater species. These two criteria, harbour of origin and deepwater species ecology, were used to affirm identification of the main fishing strategies.
Finally, a preliminary approach was followed to generate insights into what may stimulate shifts in fishing strategy. Information on the status of deepwater stocks was not available for the study period. However, it was available for saithe, the main historical target species of the French fleet, so the approach employed was based on the fishing strategies identified for Boulogne's trawlers targeting saithe in ICES Division IVa. First, we examined the Pearson correlation between the uptake of (i.e. proportion of vessels participating in) these strategies, and the market price, spawning-stock biomass (SSB), and TAC of saithe. Then, the influence of all these parameters in determining fishing strategies was discussed qualitatively.
|
Results |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
A PCA was run for every year from 1985 to 2002. There was no correlation between the landings proportion of the different stocks. Although the two first factorial axes explained 80–95% of the variability, the PCA did not facilitate definition of a typology of strategies. For example, in 1985 (Figure 5), the two first factorial axes explained 89% of the variability. This first factorial plan highlighted the main features of the fleet. It allowed vessels targeting cod in Canadian waters to be distinguished from a single vessel targeting blue ling in Faroese, western Scottish, and western Irish waters. Identification of the fishing strategies of the other vessels appears to be more complex, requiring consideration of additional factorial axes, despite the low variability they explained. Moreover, marginal vessels may over-influence the estimation of factorial axes, so for the case study investigated here, clusters were not identified only by running a PCA.
|
An HAC was then run separately for all years between 1985 and 2002. In 1985, the first group of vessels (Figure 6a, strategy A) targeted almost exclusively cod in Canadian waters (>80% for six vessels). The vessel from Fécamp did not follow exactly the same strategy. It fished mainly cod in Canadian waters (60% of total landings), but it also fished herring (Clupea harengus) and saithe in the North Sea. In the late 1980s, the number of vessels fishing cod in Canadian waters decreased, and the fishing strategy of the remaining vessels changed. Some vessels started to exploit Northeast Arctic cod (Gadus morhua). In 1989 (Figure 6b), only three vessels from St Malo targeted cod, mainly in Canadian waters. In 1990, the number of vessels fishing for cod in Canadian waters decreased dramatically, and from 1991, no vessels fished in Canadian waters. From 1994 to 1998, the strategy of those vessels could not be determined because data were not recorded. Only one vessel targeted Northeast Arctic cod from 1999 onwards.
|
Most of the smaller trawlers (350–800 t) targeted North Sea saithe in 1985 (Figure 6a). One group (strategy D) fished almost exclusively for saithe in the North Sea (70–75%), but none of this group were operating in 1986. The main group (strategy E) fished saithe (40–63%), with a variable part of their activity targeted at blue ling (3–22%). Bycatch composition gives additional information on subgroup structure. Vessels from Boulogne caught herring, whiting (Merlangius merlangus), and haddock (Melanogrammus aeglefinus) in the North Sea and eastern English Channel. Those from Lorient caught haddock, ling, and cod off western Scotland and western Ireland. Consequently, vessels from Lorient targeted saithe more specifically in ICES Subareas VIa (West Scotland), and trawlers from Boulogne fished more specifically in ICES Division IV (North Sea). The difference in fishing strategy resulted from difference in harbour of origin.
From 1985 to 1989, vessels from Lorient (and some from Boulogne) followed consistently the same strategy. However, most vessels from Boulogne adopted a more specialist activity. In 1989, one group (strategy I) caught almost exclusively North Sea saithe. All trawlers from group I were registered in Boulogne, except one registered at Lorient. However, that vessel belonged until 1988 to an owner from Boulogne, so it is not unreasonable to assume that it followed a strategy similar to that of other Boulogne vessels in group I. Another part of the Boulogne fleet targeted blue ling (strategies G1, G2, H; Figure 6b). In 1989, the Lorient fleet and some Boulogne vessels began to land new deepwater species (mainly roundnose grenadier), caught during experimental fishing (FROMNord, 1990).
In the early 1990s, the deepwater fisheries burgeoned. In 1992 (Figure 6d), trawlers from Lorient increased their catches of roundnose grenadier (20–30%), and one (strategy O) started to fish for orange roughy off western Ireland. Six vessels from Boulogne caught almost exclusively deepwater species (strategy M7) in 1991, mainly roundnose grenadier, followed by orange roughy, and blue ling. The number of vessels following a similar strategy increased in 1992. Three vessels still targeted almost exclusively saithe (strategy I) in 1991 (Figure 6c), but in 1992, these three vessels were no longer part of the fleet, and strategy I disappeared (FROMNord, 1992, 1993). Some vessels always followed the "traditional" strategy (E6; seven vessels in 1991, six in 1992), targeting North Sea saithe.
Deepwater sharks were landed from 1994 (Figure 6e), and most vessels landed a substantial quantity of deepwater species. Several strategies can be identified, based on the proportion of saithe, orange roughy, and black scabbardfish in the bycatch.
The percentage of saithe landed by vessels from Lorient continued to decrease, and the proportion of deepwater species landings increased. Vessels from Boulogne developed two strategies. Some increased their percentage of saithe landings, others targeted almost exclusively deepwater species. Orange roughy landings decreased regularly, and in 2002 (Figure 6g), no vessel appeared to target this species. The proportion of blue ling landings dropped during the early 1990s, increased in the second part of the 1990s (Figure 6f), but decreased again from 2000 onwards. In 2002, one vessel group fished almost exclusively saithe. A second vessel group caught principally deepwater species, mainly roundnose grenadier. Depending on the relative importance of saithe, roundnose grenadier, and blue ling, eight subgroups could be identified (Figure 6g).
This data exploration suggests that the strategies of the French offshore fleet changed over the period 1985–2002. In order to define operational fishing unit from the outcomes of this study, it was necessary to simplify the comprehensive set of strategies identified here. Fisheries knowledge and biological aspects have therefore been used as criteria to group similar strategies into categories.
In terms of the Lorient fleet (Figures 7a, 8a), all vessels historically caught principally saithe (40–75%; Table 3, strategy L1), but with a substantial proportion of blue ling. Different bycatches (anglerfish, Lophius spp., haddock, hake, cod) were caught mainly in ICES Divisions VIa and VII (except VIId–e). In the early 1990s, this strategy was replaced by an intermediate strategy (L2). Vessels caught saithe (>20%) and deepwater species (20–40%). From 1995, vessels targeted principally deepwater species (40–75%), and alternatively blue ling or saithe (strategy L3). An alternative strategy (L4) was followed in 1993 and 1994, and in 2000 and 2001 by another vessel. That vessel targeted mainly roundnose grenadier, and alternatively orange roughy (20–40%).
|
|
|
The fishing strategy of the Boulogne fleet differed (Figures 7b, 8b). In 1985, the vessels followed a strategy (Table 3, B1) similar to that of the Lorient fleet. The main difference was the fishing area. In contrast with the Lorient vessels, which fished principally off western Ireland, trawlers from Boulogne fished in ICES Divisions IV and VIId. From 1986 to 1989, some vessels stopped fishing for blue ling (strategy B2), but others specialized in fishing for blue ling, with small bycatches of saithe (B3). Specialization continued, with some vessels targeting almost exclusively saithe (strategy B4), from 1988 onwards. Then, from the early 1990s, strategy B4 disappeared. Saithe fishing was replaced by new strategies based on different species (deepwater species, orange roughy). Some vessels (strategy B6) caught almost exclusively deepwater species (roundnose grenadier, black scabbardfish, deepwater sharks), a few followed an intermediate strategy (B5), catching both saithe (25–50%) and deepwater species (20–40%), and another strategy (B8) combined catches of deepwater species (40–75%) and orange roughy (>20%). Strategy B3, based on blue ling, was replaced by different strategies. Vessels targeted blue ling with different bycatch proportions of saithe, orange roughy, and deepwater species. In the late 1990s, the number of fisheries targeting almost exclusively saithe (strategy B4) increased. Real specialization took place, with vessels landing more than 90% of saithe. By contrast, other vessels specialized in deepwater species, and the intermediate strategy (B5) disappeared. The proportion of orange roughy in one vessel's landings increased until 1999, but then dropped, and the strategy disappeared in 2002. From 2001, a new strategy (B7) appeared. Some vessels from Boulogne targeted deepwater species in Faroese waters and saithe in the North Sea.
In summary, 12 strategies could be identified, four for Lorient, and eight for Boulogne–Dieppe–Fécamp (Table 3).
To explain the reasons for the changes in strategy, the correlation of different parameters with the proportion of vessels targeting saithe mainly in the North Sea (saithe strategy uptake) was tested (Table 4). The saithe strategy uptake appeared to be correlated with the saithe TAC (r = 0.82), but not with the saithe SSB. One interpretation is that catch limits, rather than stock abundance, had an effect in determining fishing strategies. However, the skippers' decision-making mechanisms are certainly more complex. For instance, since 1992, the saithe strategy uptake has increased more quickly than the saithe TAC (Figure 9). Moreover, when both SSB and TAC of saithe increased substantially from 2000 to 2002, the number of vessels targeting the species did not increase as expected. The development of this strategy could have been limited by market forces, influenced most by the decrease of the saithe market price between 2000 and 2002.
|
|
|
Discussion |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Interactions between fleets and species play a major role in fisheries dynamics. Therefore, management of fisheries requires an analysis of fishing strategies to understand the adaptability of fishing fleets. Consequently, the identification of fishing tactics and strategies is still an important ongoing issue for fisheries managers. Classical methods were based on PCA and cluster analysis (Biseau and Gondeaux, 1988; Lewy and Vinther, 1994; Taquet et al., 1997; Pelletier and Ferraris, 2000; Ulrich et al., 2001). Generally, those methods were based on the landings of the most important species, and the method developed in this paper is similar in that respect. However, to the best of our knowledge, the methods developed in the fisheries literature were of a static nature, because they did not incorporate a time dimension. The idea of this study was therefore to use such methods annually over a time period, and to analyse which fisheries appeared, and which disappeared.
To consider the French fishery for cod in Canadian waters, of eight vessels registered in 1985, none fished in Canadian waters from 1991 onwards. The history of the Canadian cod stock probably explains this evolution. In 1980, the TACs of groundfish species were increased by 25%, the increase mainly applying to cod in NAFO areas 2J3KL. Consequently, the foreign allocation of the cod quota also increased (Mitchell, 1997), and the French fleet's TAC made cod fishing there an economically viable activity. Then, from 1989 onwards, Canada announced a reduction in TAC, to the point where cod fishing could not be economically viable for the French fleet. As a consequence, the vessels, mainly from St Malo, transferred their activity towards cod in the Arctic.
Historically, vessels from Lorient, Boulogne, Fécamp, and Dieppe followed a similar strategy. They landed principally saithe, with a component of blue ling and different traditional species (cod, haddock, whiting). The main difference was fishing areas. Vessels from northern France fished principally in the North Sea, while those registered in Lorient fished off northern and western Ireland. This difference was traditionally attributable to the vessels fishing on stock aggregations located as close as possible to their home harbour. From 1994, the French national quota for several species, including saithe, was shared among producer organizations (PO). Three quotas are set for saithe in European Union (EU) waters, and these are distributed in three different fishing areas: the North Sea [ICES Divisions IIa (EU waters), III (EU waters), IV], the Northern Shelf [ICES Divisions Vb (EU waters), VI, XII, XIV], and the Southern Shelf [ICES Divisions VII, VIII, IX, X, and COPACE34.1.1 (EU waters)]. Most trawlers registered in northern France belong to a PO that has most of the North Sea saithe quota. Most of those registered in Lorient belong to POs that have most of the Northern Shelf and Southern Shelf saithe quotas. Differences in the share of the quota between POs clearly represent another explanation of the different fishing strategies pursued by trawlers registered in northern France and Lorient.
From 1985 to 2002, there were clear strategy shifts, and we have examined some mechanisms that may have induced such changes. First, changes in the saithe fisheries appeared to some extent to be linked with the saithe TAC (Table 4). In 1985, the biomass of the main saithe stock (North Sea, Skagerrak, West Scotland) was below the precautionary level Bpa, and it continued to decrease until 1992. As a result, the saithe TAC decreased from 200 000 to 127 000 t between 1989 and 1992. Vessels targeting saithe had to adapt their strategy to maintain a viable fishing activity. An option was to find alternative species to balance smaller saithe catches. First, blue ling fisheries developed. Then, in the early 1990s, experimental trips allowed fishers to identify new fishing areas for roundnose grenadier, black scabbardfish, and orange roughy. Blue ling and deepwater species were mainly caught off western Scotland and western Ireland, close to the traditional fishing areas of the Lorient fleet. Subsequently, vessels slowly increased their catches of the newly discovered deepwater species. The reasons why this increase was slow could be that vessels and gears had to be adapted to fish on deeper grounds, and also that a market had to be developed to facilitate the sale of these new products. Trawlers from northern France quickly adapted: some continued to target almost exclusively saithe, others targeted deepwater species. Further, within the groups of trawlers targeting deepwater species, strategies changed. Initially, fishing trips were mostly experimental, but catches increased as fishers learnt where and how these species could be fished, and probably also because the market developed to accept them.
From 1995 onwards, the biomass of saithe recovered gradually, exceeding Bpa from 1999, and the TAC increased. Following this recovery, some vessels from Boulogne directed more of their effort towards saithe in the North Sea, while others continued to target deepwater species in the more remote areas off western Scotland and western Ireland. However, despite the increase in saithe biomass, vessels from Lorient continued to target principally roundnose grenadier, which was commercially more valuable. In addition, and unlike vessels from Boulogne, those from Lorient could harvest deepwater species on fishing grounds that were relatively close to those where they traditionally fished for saithe. Although economic information was not available for this study, one would expect the difference in operating costs for these vessels to have been offset by the gain in gross revenue brought about by the more attractive landing value of deepwater species.
In the later 1990s, interest in orange roughy waned. Historically, the species was caught in ICES Division VI, but catches there dropped as catches in Division VII increased. Despite its high value, catches continued to decrease, and there has been no targeted fishing since 2002. The sharp decrease in orange roughy landings is thought to result from either biomass depletion (most likely), or from market competition with other suppliers fishing elsewhere in the world. The species lives in aggregations and is slow-growing, and fleets likely exploit local aggregations sequentially, fishing out first one aggregation, then moving on to the next (ICES, 2002).
TACs may play an important role in fisheries dynamics (Table 4). Adoption of saithe-fishing strategies decreased concomitantly with the decline in the saithe TAC during the early 1990s, suggesting that the TAC may have influenced fisher behaviour. However, although the saithe TAC increased again over the period 1999–2002, a saithe-fishing strategy did not prove attractive over the same period, suggesting that management is not the only determinant of strategy shifts, and that other factors could come into play. One such factor is the possibility of targeting other species. In the mid-1980s, although saithe biomass was low, no alternative species was readily available to fishers. At that time, neither fishing technology nor market conditions were suitable for catching and commercializing deepwater species. A second factor could be species availability. Thus, the decline in abundance of orange roughy could have had an effect on the decline of the strategy based on that species. A third factor could be market conditions. From 1999, although saithe biomass and TAC increased, the proportion of vessels targeting the species decreased. Thus, saithe landings then could have been restricted by the decrease in market price for the species (Figure 9).
The method outlined here was based on the application of PCA, followed by HAC, to annual landings. There are several reasons for classifying according to a factorial analysis (Lebart et al., 1995; Pelletier and Ferraris, 1999). The factorial axes and the projections of vessels on them derived from PCA may be difficult to interpret. For instance, although the first two axes usually explain most of the variance, it may in some cases be necessary to consider the higher ranked axes (Biseau and Gondeaux, 1988). Moreover, some marginal vessels may be very influential in the construction of the axes. Consequently, defining clusters through a PCA may be unrealistic when information is diluted on many axes, and when marginal vessels excessively influence the definition of factorial axes. Classification methods address these reservations by summarizing information in a manner that is easier to interpret than individual projections. In addition, these methods are to some extent more robust to marginal individuals than factorial analysis, because of the iterative algorithms used in HAC. Nevertheless, there are also several reasons for running a factorial analysis before classification (Pelletier and Ferraris, 1999). Notably, factorial analyses provide a geometric description of the vessels, the variables, and the relationships between them, which is helpful in exploring the structure of the data set.
Different linkages (simple, average, or complete) were available to run the HAC. In our case, we chose a simple linkage, which is the shortest distance between two groups, i.e. that the distance between two groups is equal to the shortest distance between one element of the first group and one element of the second. This method has been selected mainly for its simplicity (one vessel is always aggregated with the most similar vessels), and Jardine and Sibson (1971) argue that only this method has all the desirable properties of a clustering method. Nevertheless, one limitation of the method may be the "chain effect", and some authors argue that average or complete-linkage could sometimes be better metrics. In any case, we tested these different linkage methods and obtained similar results. The results were therefore robust to the metric used.
One difficulty of the current method is in defining the most appropriate number of clusters. It is in theory possible to define 1 to n clusters, with n specifying the number of individuals (vessels). Selecting too few clusters belittles the quantity of information available. By contrast, exaggerating the number of clusters would adversely affect the analytical possibilities based on the clusters. In the absence of objective benchmarks, the compromise number of clusters was found by combining visual exploration of the clusters' properties with expert knowledge of the French offshore fisheries. This approach is consistent with current practice (Lebart et al., 1995), and also with ICES recommendations for identifying fisheries groups (ICES, 2003).
Neither PCA nor cluster analysis is a new method. However, the scope of the paper was not to develop a new methodology to identify fisheries, but rather to apply existing methods to characterize fleet dynamics in terms of shifts in fishing strategies. Further, classification methods are routinely used by some ICES stock assessment working groups (e.g. the Working Group on the Assessment of Hake, Monk, and Megrim, and the Working Group on the Assessment of Southern Shelf Demersal Stocks) to define tuning fleets. In addition, cluster analysis has been advocated by ICES (2003) as one of the methods to be used to identify fishing units in carrying out mixed-fisheries forecasts.
Alternative multi-table analytical methods, including multiple factorial analysis (MFA) (Escofier and Pagès, 1984), and the STATIS method (Lavit et al., 1994), could be considered. Such methods allow characterization of the underlying common structure present in several tables as well as the variability of each compared with the common structure. However, MFA requires a constant number of observations (vessels in this study) over the years, so could not be applied here because the number of vessels belonging to the fleet examined changed from one year to the next. The STATIS method is more flexible. It has been used in the past to analyse sensory profiling data, but it has also recently been applied to characterize the spatial and temporal distribution of marine biological data (Licandro et al., 2001; Gaertner et al., 2002). Although, to our knowledge, the STATIS method has not been applied yet in classifying fisheries, it could prove promising as an alternative approach for future investigations.
The present approach was based on analysis of catch weights. When differences in market price are great, fishing strategies may be better reflected by the value of landings than by their weight. However, the value of a landing may also be a misleading variable in identifying fishing strategies, because it depends on normal price fluctuations, and does not account for operating costs, which may vary dramatically from one trip to another.
The fishing strategies identified in this study are defined, in certain cases, on the basis of the fishing behaviour of a single vessel. In any statistical analysis, marginal points are considered as outliers or points of specific interest. In our study, however, a marginal point is a fishing strategy, which can be pursued year after year by just one vessel, but which may subsequently be adopted by other vessels. For example, in 1985, only one vessel from Fécamp targeted blue ling, but a few years later, other vessels started in the fishery. In this case, the first trawler should be considered as a precursor, rather than as an outlier.
The results of this study seem to support the plasticity of fisher behaviour, and bear out the findings of other studies that suggest that fishers react quickly to modification of their external environment (Ferraris, 1995). In the case study investigated here, a number of traditional fishing opportunities were restricted by biological and management constraints (low biomass, low TAC), so some fishers explored new opportunistic fishing strategies, such as strategy L2 in 1992 (Figure 6d). As this strategy proved economically attractive, more fishers adopted it. Notwithstanding, the mechanisms underlying fishing strategies and fisher behaviour are certainly more complex than those investigated here.
Although some historical dynamics of the fishing strategies of the French offshore trawler fleet have been identified, it is not possible to forecast future fisheries development on the basis of this study. From 2003, TACs for orange roughy, roundnose grenadier, black scabbardfish, and blue ling have been established; the French fleet owns most of the quotas. A critical issue could therefore be to evaluate the influence of the quotas for these species on future fishing strategies. There is growing interest in analysing and modelling the processes underlying fisher behaviour and decision-making. Various aspects of fleet dynamics have already been investigated, including spatial allocation of fishing effort (Gillis et al., 1993; Holland and Sutinen, 1999), gear development (Pech and Laloë, 1997), and discarding (Stratoudakis et al., 1998). There is scope in further developing such approaches into a comprehensive operational model, which could be applied by fisheries managers to forecast fleet responses and adaptation to management measures. This issue will be addressed in a companion study.
|
Acknowledgements |
|---|
This work was funded by the European Union (DG XIV, study no. QLRT-2001-01291). This support is gratefully acknowledged, as are the thoughtful and helpful comments of two anonymous referees, and ICES for providing stock estimates.
|
Footnotes |
|---|
1 http://www.ices.dk/committee/acfm/comwork/report/asp/acfmrep.asp
|
References |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
-
Biseau A. (1998) Definition of a directed fishing effort in a mixed-species trawl fisheries, and its impact on stock assessment. Aquatic Living Resources 11:119–136.[CrossRef][Web of Science]
Biseau A. and Gondeaux E. (1988) Apport des méthodes d'ordination en typologie des flottilles. Journal du Conseil International pour l'Exploration de la Mer 44:286–296.
Blanc F., Chardy P., Laurec A., Reys J.P. (1976) Choix des métriques qualitatives en analyse d'inertie. Implications en écologie marine benthique. Marine Biology 35:49–67.[CrossRef]
EC. (2002) Report of the sub-group on resource status (SGRST) of the Scientific, Technical and Economic Committee for Fisheries (STECF). SEC(2002), 1373. 127 pp.
Ehrich S. (1983) On the occurrence of some fish species at the slopes of the Rockall trough. Archiv für Fischereiwissenschaft 33:105–150.[Web of Science]
Escofier B. and Pagès J. (1984) L'Analyse factorielle multiple: une méthode de comparaison de groupes de variables [Multiple factorial analysis: a method to compare groups of variables]. Data Analysis and Informatics 3:41–55.
Ferraris J. (1995) Démarche méthodologique pour l'analyse des comportements tactiques et stratégiques des pêcheurs artisans sénégalais. Questions sur la Dynamique de l'Exploitation Halieutique. Colloques et séminaires(Institut francais de Recherche scientifique pour le Developpement en Cooperation (ORSTOM), Paris) 263–295.
Fontaine B. (1979) Le grenadier. Sciences de la Pêche, Bulletin de l'Institut des Pêches Maritimes 294:15–21.
FROMNord. (1990) Rapport d'activité 1989. Fonds Régional d'Organisation du Marché du Poisson. 42 pp.
FROMNord. (1992) Rapport d'activité 1991. Fonds Régional d'Organisation du Marché du Poisson. 44 pp.
FROMNord. (1993) Rapport d'activité 1992. Fonds Régional d'Organisation du Marché du Poisson. 45 pp.
Gaertner J-C., Bertrand J.A., Souplet A. (2002) STATIS-CoA: a methodological solution to assess the spatio-temporal organization of species assemblages of the French Mediterranean Sea. Scientia Marina 66:Suppl. 2, 221–232.[Web of Science]
Gillis D., Peterman R., Tyler A. (1993) Movement dynamics in a fishery: application of the ideal free distribution to spatial allocation of effort. Canadian Journal of Fisheries and Aquatic Sciences 50:323–333.
Gordon A.D. (1981) Classification: Methods for the Exploratory Analysis of Multivariate Data(Chapman & Hall, London).
Haedrich R.L. and Merrett N.R. (1988) Summary atlas of deep-living demersal fishes in the North Atlantic Basin. Journal of Natural History 22:1325–1362.[CrossRef][Web of Science]
Holland D. and Sutinen J. (1999) An empirical model of fleet dynamics in New England trawl fisheries. Canadian Journal of Fisheries and Aquatic Sciences 56:253–264.
ICES. (1991) Report of the Working Group on fishery units in sub-areas VII and VIII. ICES Document, CM 1991/Assess: 24. 215 pp.
ICES. (1992) Report of the Study Group on fishery units in sub-areas VII and VIII. ICES Document, CM 1992/G: 15. 70 pp.
ICES. (2002) Report of the Working Group on the biology and assessment of deep-sea fisheries resources. ICES Document, CM 2002, ACFM: 16 Ref. G. 253 pp.
ICES. (2003) Report of the Study Group for the development of fishery-based forecasts. IVCES Document, CM 2003, ACFM: 08 Ref. D. 37 pp.
Jardine N. and Sibson R. (1971) Mathematical Taxonomy(John Wiley, London).
Laloë F., Rey H., Durand J.L. (1995) Avant-propos: questions sur la dynamique de l'exploitation halieutique. Questions sur la dynamique de l'exploitation halieutique. Colloques et séminaires(Institut francais de Recherché scientifique pour le Developpement en Cooperation (ORSTOM), Paris) 15–24.
Laloë F. and Samba A. (1991) A simulation model of artisanal fisheries of Senegal. ICES Marine Science Symposia 193:281–286.
Laurec A., Biseau A., Charueau A. (1991) Modeling technical interactions. ICES Marine Science Symposia 193:225–236.
Lavit C., Escoufier Y., Sabatier R., Traissac P. (1994) The ACT (Statis method). Computational Statistics and Data Analysis 18:97–119.[CrossRef]
Lebart L., Morineau A., Piron N. (1995) Statistique Exploratoire Multidimensionnelle(Dunod, Paris) 439 pp.
Lewy P. and Vinther M. (1994) Identification of Danish North Sea trawl fisheries. ICES Journal of Marine Science 51:263–272.
Licandro P., Ibanez F., Zunini T.S. (2001) Long term variations of zooplankton in the Ligurian Sea (western Mediterranean). Archivio di Oceanografia e Limnologia 22:159–166.
Lorance P. and Dupouy H. (2001) CPUE abundance indices of the main target species of the French deep-water fishery in ICES sub-areas V–VII. Fisheries Research 51:137–149.[CrossRef][Web of Science]
Magnússon J.V. and Magnússon J. (2000) The distribution, relative abundance and biology of the deep-sea fishes of the Icelandic slope and Reykjanes ridge. In Hopper A.G. (Ed.). Deep-water Fisheries of the North Atlantic Oceanic Slope(Kluwer Academic, Amsterdam) pp. 161–199.
Mitchell C.L. (1997) Fisheries management in the Grand Banks, 1980–1992 and the straddling stock issue. Marine Policy 21:97–109.[CrossRef][Web of Science]
Pech N. and Laloë F. (1997) Use of principal component analysis with instrumental variables (PCAIV) to analyse fisheries catch data. ICES Journal of Marine Science 54:32–47.
Pelletier D. and Ferraris J. (2000) A multivariate approach for defining fishing tactics from commercial catch and effort data. Canadian Journal of Fisheries and Aquatic Sciences 57:51–65.
Quéro J-C. (1997) Les Poissons de Mer des Pêches Françaises(Delachaux et Niestlé, Lausanne) 304 pp.
S-Plus 6.1 J-C. (2001) S-Plus 6 for Windows User's Guide(Insightful Corporation, Seattle).
Stratoudakis Y., Fryer R.J., Cook R.M. (1998) Discarding practices for commercial gadoids in the North Sea. Canadian Journal of Fisheries and Aquatic Sciences 55:1632–1644.
Taquet M., Gaertner J.C., Bertrand J. (1997) Typologie de la flottille chalutière du port de Sète par une méthode de segmentation. Aquatic Living Resources 10:137–148.[CrossRef][Web of Science]
Ulrich C., Gascuel D., Dunn M.R., Le Gallic B., Dintheer C. (2001) Estimation of technical interactions due to the competition for resource in a mixed-species fishery, and the typology of fleets and métiers [trades] in the English Channel. Aquatic Living Resources 15:267–281.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Forcada, C. Valle, J. L. Sanchez-Lizaso, J. T. Bayle-Sempere, and F. Corsi Structure and spatio-temporal dynamics of artisanal fisheries around a Mediterranean marine protected area ICES J. Mar. Sci., October 26, 2009; (2009) fsp234v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Marchal A comparative analysis of metiers and catch profiles for some French demersal and pelagic fleets ICES J. Mar. Sci., May 1, 2008; 65(4): 674 - 686. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||