A comparative analysis of métiers and catch profiles for some French demersal and pelagic fleets
IFREMER, 150 Quai Gambetta, BP 699, 62321 Boulogne sur mer, France
tel: +33 321995600; fax: +33 321995601; e-mail: paul.marchal{at}ifremer.fr
Marchal, P. 2008. A comparative analysis of métiers and catch profiles for some French demersal and pelagic fleets. – ICES Journal of Marine Science, 65: 674–686.A quantitative comparison between métiers and resulting catch profiles was carried out for seven French demersal and pelagic fleets operating in the North Sea, the eastern Channel, and the Bay of Biscay. Typologies for four métiers have been attempted, based on different factors (gear, mesh size, fishing ground, and/or a priori target species), data sources (logbooks or harbour enquiries), and aggregation levels. Catch profiles were selected through cluster analysis. The linkage between métiers and catch profiles was quantified using uncertainty coefficients, which depended on the métier typology being used and the fleet under consideration, but were not subject to substantial inter- or intra-annual fluctuations. Future catch profiles and métiers were forecast in 2005 based on the 2001–2004 métier/catch profile correlations. When contrasted with the 2005 observations, the forecasting score was greatest (80–85%) for pelagic trawlers and gillnetters, and lowest for demersal trawlers (5–60%).
Keywords: catch profiles, fishing intention, fleets, métiers, uncertainty coefficients
Received 18 June 2006; accepted 19 February 2008; advance access publication 25 March 2008.
 |
Introduction |
|---|
 Top Introduction Material and methodsResultsDiscussionReferences |
|---|
Fisheries data collection, advice, and management have traditionally been based on a single-stock basis. However, this approach has long been recognized as inadequate, particularly when applied to mixed fisheries, which are subject to technical interaction between fishing units and across species. Ignoring such interactions could lead to an undesirable situation where fishing for one species may lead to discarding of another, the quota of which has already been exceeded.
Several steps have been undertaken in the past decade to incorporate explicitly technical interactions between fishing units within the cycle of observing, assessing, forecasting, and managing fisheries. First, in waters of the European Union, a sampling programme for fisheries biology and economics data that will come into force in 2008 will explicitly include fishing units as sampling strata (EC, 2005, 2006). Second, modelling approaches have been developed to account for technical interactions between fishing units in forecasting the status of fish stocks (Murawski and Finn, 1986; Laurec et al., 1991; Marchal and Horwood, 1996; Vinther et al., 2004). Third, direct effort limitations, applied to specific fishing units, are increasingly being used to complement single-species total allowable catches (TACs), to improve the efficiency of managing fisheries all around the world, as well as in EU waters.
The cornerstone of all these approaches is the fishing unit. ICES (the International Council for the Exploration of the Sea) considers three types of fishing unit: the fleet, the fishery, and the métier (ICES, 2003). A fleet is a physical group of vessels sharing similar characteristics in terms of technical features and/or major activity. A fishery is a group of vessel voyages targeting the same (assemblage of) species and/or stocks, using similar gear, during the same period of the year and within the same area. A métier is a homogeneous subdivision, either of a fishery by vessel type, or of a fleet by voyage type. Two of the three fishing units are well enough defined to categorize fishing vessels and fishing trips, so I focus on fleets and métiers.
Several methods may be contemplated to identify fleets and métiers. The fleet classification depends on: (i) the technical feature considered to categorize the physical characteristics of fishing vessels (e.g. vessel length, horsepower, tonnage, or economic structure); (ii) the variable used to characterize a fishing activity (e.g. fishing effort by gear, area, target species, or métier); (iii) the value of the activity threshold above which a vessel falls into a given fleet category; (iv) the period chosen to calculate fishing activity in relation to the threshold identified in (iii) (e.g. calendar year, quarter).
Métiers should reflect the fishing intention, e.g. the species targeted, the area visited, and the gear used, at the start of a fishing trip. However, there are many situations where fishing intention cannot be observed directly, and can only be estimated retrospectively by examining the catch profiles resulting from fishing trips. The approaches used in the past to identify métiers may then be classified into input-based, output-based, and combined methods. Input-based methods either make use of existing records of the technical features of fishing trips, which are typically available in fishers' logbooks, e.g. gear and mesh size used, fishing grounds visited, season (Ulrich et al., 2001; Marchal et al., 2006), or build on direct interviews with stakeholders (Neis et al., 1999; Christensen and Raakjer, 2006).
Output-based methods assume that catch profiles perfectly reflect fishing intention. The simplest approach consists of selecting the fishing trips where a certain catch proportion (in weight or value) of selected key species is exceeded. Each set of fishing trips discriminated by this approach may then be drawn into a métier category (Biseau, 1998). Another approach consists of conducting multivariate analyses of catch profiles, then grouping fishing trips of similar catch profiles into métiers. Métiers can be identified by direct visual inspection using principal component analysis (Biseau and Gondeaux, 1988; Laurec et al., 1991) or automatically through a hierarchical cluster analysis algorithm (Léwy and Vinther, 1994; Holley and Marchal, 2004). Combined methods have categorized métiers by clustering catch profiles (outputs), then relating these clusters to fishing trip characteristics (inputs) using multivariate analysis (Pelletier and Ferraris, 2000; Ulrich and Andersen, 2004).
Both methods, input- and output-based, have benefits and limitations. Collecting information on fishing intentions through direct face-to-face interviews has at least two merits. First, there are some fleet segments for which reporting fisheries data in logbooks is not mandatory. This derogation applies in particular to EU vessels <10 m, although these may represent up to 70% of the entire EU fleet (EC, 2005, 2006). For such vessels, direct interviews are the only way to collect quantitative information on fleets and fisheries. Second, interviews are the only way to collect information on the species initially targeted when this field is not recorded in logbooks. In countries such as New Zealand, the target species is reported in logbooks (Anon, 2004), and it is frequently used as a criterion to classify fishing trips and to derive the tuning series used in stock assessments. The target species is, however, not documented in EU logbooks, although it is considered a key factor in grouping fishing intentions into the categories, which ideally would be used as the sampling strata for collecting stock assessment data (EC, 2005, 2006).
The main drawback of the face-to-face interview approach is that data collection may be time-consuming. A second best proxy to estimate fishing intention is based on the gear, mesh size, and fishing grounds recorded in fishers' logbooks. There are two main drawbacks to this approach. First, the quality of information recorded in logbooks may be variable. In particular, anecdotal evidence suggests some misreporting of mesh size used and of fishing grounds visited, especially when management is restrictive. Second, fishing intention may sometimes be expressed not only by combinations of gear, mesh size, and area, but also by fine-scale tactical features—referred to as the skipper effect—which are generally not reported in logbooks.
Catch profiles should be the method of last resort for defining métiers, because there are at least two reasons why fishing intentions may not be reflected by species composition. First, catch profiles are often estimated from landings, so the discard fraction is ignored. Many mixed fisheries are subject to discarding, and ignoring that component of the original catch may seriously bias the catch composition estimated from landings. Second, because all species caught in the same fishery do not necessarily have the same temporal and spatial dynamics, the relative species catch composition may differ from that expected. Despite these problems, catch profiles often remain the only source of information available to classify métiers.
The main aim here is to evaluate in a quantitative manner the extent to which (i) fishing intentions can be used to forecast catch profiles, and (ii) catch profiles may be reasonable proxies of fishing intentions. Different typologies of fishing métier are drawn up, based on both input and output methods. The linkage between input- and output-based métiers is then quantified using uncertainty coefficients, and the variations of these are analysed. Finally, catch profiles and fishing intentions are forecast using different predictors and different annual lags, and the study is applied to the main French demersal and pelagic fleets over the period 2001–2005.
|
Material and methods |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Data
I used logbook data and national fisheries registers and activity calendars registered by the French Fishery Ministry (DPMA) and extracted from Harmonie, the database of the French Fisheries Information System managed by IFREMER, the French Research Institute for Exploitation of the Sea. The catch-and-effort data used for the analysis are derived from fishers' logbooks and national fisheries registers. Data are disaggregated by vessel, fishing trip, statistical rectangle [surface 1° longitude x 0.5° latitude, or
30 nautical miles (hereafter referred to as miles) x 30 miles (1 mile = 1.853 km)], and gear used. The recorded vessel characteristics are length, tonnage, and horsepower. The type of gear (otter trawl, pair trawls, beam trawl, gillnet, etc.) and, for most fleets, the mesh size used were also made available. These data were made available over the period 2001–2005. Since 2000, IFREMER has implemented an annual census in all harbours located on the Channel and the Atlantic coast (i.e. from Dunkerque to Bayonne), to collect the activity calendar of each vessel belonging to the official national fleet register. First, fishers record on a dedicated form, the fishing slip, all métiers (combination of gear, target species, and fishing areas) practiced by a vessel during a month and for each month of a given year. The information collected in the fishing slips is then checked and validated by a network of observers, who have a regional portfolio of vessels to survey. Part of the validation process consists in meeting face-to-face with a sample of fishers, following a sampling plan specified by IFEMER. The target is for the skippers of at least 30% of the French fishing fleet to be interviewed each year. However, because these interviews are held on a voluntary basis, achievement of the targeted interview rate may vary between years. Finally, the information validated by the observer network is entered into the Harmony database and recorded as activity calendars.
Fleet and métier typologies
Fishing vessels registered in fishers' logbooks and activity calendars were grouped into fishing fleets. Consistent with ICES (2003), a fleet is defined as a group of vessels sharing similar characteristics in terms of technical features and (or) major activity. In practical terms, fleets are defined as a combination of vessel harbours, vessel length range, and main gear used during a calendar year. Subsequently, each vessel can belong to just one fleet during a calendar year, but it can move to another fleet segment for another year. The study focused on a selection of seven major pelagic and demersal French fleets, mainly fishing in the Bay of Biscay, the eastern Channel, and the North Sea, over the period 2001–2005. The characteristics of and the codes given to these fleets are listed in Table 1. Note that the logbooks of a number of vessels belonging to fleet FL1 were not recorded in 2004.
|
Input- and output-based methods were applied to group fishing trips into métiers. Four input-based approaches were attempted to estimate fishing intention, two building on métiers as recorded in activity calendars, and two building on different combinations of gear, mesh size, and fishing grounds recorded in fishers' logbooks.
Input-based approaches: estimating fishing intention
Monthly métiers, as reflected in activity calendars, are listed as a combination of gear, a priori target species, and fishing area. This is the finest level at which information can be disaggregated, and it is here referred to as "level 2". To harmonize métier definitions across EU countries, a coarser level of aggregation has been suggested by EC (2006), and this is here referred to as "level 1". The difference between levels 1 and 2 is that the target species identified at level 2, e.g. cod (Gadus morhua), whiting (Merlangius merlangus), hake (Merluccius merluccius), or Norway lobster (Nephrops norvegicus), are grouped into broader categories at level 1 (demersal fish, or mixed crustaceans and demersal fish). The grouping was primarily based on available knowledge of the biology and ecology of the target species (crustaceans, molluscs, cephalopods, demersal fish, small/large pelagic fish, and deep-water species). Further, some target species were aggregated in mixed categories to reflect clear technical interactions. For instance, French fishers targeting Norway lobster or squid with otter trawls also catch substantial quantities of demersal fish. Therefore, the level 2 métiers "Norway lobster" and "squid", as reported in activity calendars, respectively, became "mixed crustaceans and demersal fish" and "mixed cephalopods and demersal fish" at level 1. Examples of levels 1 and 2 métiers are shown in Tables 2 and 3.
|
|
To group the fishing trips defined in logbook data into métiers at levels 1 and 2, as defined in activity calendars, a link needs be generated between the two datasets. The difficulty in generating that link is that logbook information is aggregated at the level of a fishing trip, whereas activity calendars are reported by month. To circumvent that difficulty, a selection of fishing trips was made from the logbook dataset. When several métiers were operated by the same vessel during the same calendar month, all the corresponding fishing trips were eliminated. That selection procedure was applied to métiers defined at levels 1 or 2. In the restricted logbook dataset, information may then be aggregated by month, because only one métier is operated per month. The aggregation level of logbooks is then consistent with that of activity calendars, and the two datasets can be linked directly.
Métiers derived from national logbooks were defined as a combination of gear, mesh size, and fishing area. Gear was defined as for levels 1 and 2 (Table 2). Mesh sizes were grouped into 13 categories defined based on current management regulations: 1–69, 70–79, 80–89, 90–99, 100–109, 110–119, 120–129, 130–139, 140–159, 160–179, 180–199, 200–249, and >250 mm. Two levels of aggregation were considered for fishing areas. The coarse aggregation level includes just two fishing areas: the North Sea and the Northeast Atlantic. Fine-scale aggregation is to that of the ICES rectangle, which represents the best spatial resolution achievable with logbook data. Table 4 shows the number of métier categories, depending on the typology being applied.
|
Output-based approach: estimating catch profiles
The output-based approach applied here is a cluster analysis (Léwy and Vinther, 1994; Holley and Marchal, 2004), which has been implemented using the SAS/STAT (1999) procedures CLUSTER and TREE. Clusters were defined using the Ward method. The variables analysed were the catch proportions (by value) of each species relative to the total catch. The analysis was conducted for the whole dataset rather than by fleet, so each clustered catch profile derived from the analysis may potentially be allotted to fishing trips operated by any of the seven fleets under consideration. A decision on the most appropriate number of clusters was made by inspecting dendrograms and a number of statistics (cubic clustering criterion, r2 and partial t2), which have been made available for each cluster configuration.
Linking fishing intention with catch profiles
Uncertainty coefficients (Goodman and Kruskal, 1979) were used to quantify the association between clustered catch profiles, on the one hand, and the four estimates of fishing intention.
The first series of uncertainty coefficients (IOi) measures the proportion of uncertainty in the catch profiles that is explained, a priori, by fishing intention. The second series of uncertainty coefficients (OIi) measures the proportion of uncertainty in the fishing intention that is explained, a posteriori, by catch profiles. Index i characterizes the method used to estimate fishing intention: i = 1, métiers recorded in activity calendars aggregated at level 1; i = 2, métiers recorded in activity calendars aggregated at level 2; i = 3, combined gear, mesh size, and area aggregated in two levels; i = 4, combined gear, mesh size, and area disaggregated by ICES rectangle. IOi and OIi may be derived from:
|
| (1) |
|
| (2) |
The different uncertainty coefficients for each fleet were subject to two analyses. First, the variations of each uncertainty coefficient were evaluated by GLM (explanatory variables, year and quarter as class variables, fishing effort as continuous variable; distribution function, binomial; link function, logit). Second, pairwise comparisons of the different uncertainty coefficients (IOi vs. IOj
i and OIi vs. OIji) were carried out.
Forecasting fishing intentions and catch profiles
For each fleet, the extent to which the knowledge of past and present knowledge of the correlation between fishing intentions and catch profiles could be used to forecast was tested with (i) future catch profiles given fishing intention known, and (ii) future fishing intentions given catch profile known. For the purpose, the probability of catch profile c given fishing intention m and the probability of fishing intention m given catch profile c were estimated based on the frequency of fishing trips and vessels for each category, for each fleet, and for years 2001–2004. For each m, the 2001–2004 catch profile category with the greatest probability was used to forecast the 2005 catch profile given m. Conversely, for each catch profile category c, the 2001–2004 fishing intention with the greatest probability was used to forecast the 2005 fishing intention given c. The 2005 fishing intention and catch profile forecasts were then contrasted with the 2005 observations.
To evaluate further the benefits of using the past correlations between fishing intentions and catch profiles to make forecasts, the extent to which the past seasonal distribution of fishing intentions and catch profiles could be used to forecast those two factors was investigated. For each quarter q, the 2001–2004 catch profile (respectively fishing intention) category with the greatest probability was used to forecast the 2005 catch profile (respectively fishing intention), given q. The 2005 fishing intention and catch profile forecasts were then contrasted with the 2005 observations.
|
Results |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Fleet and métier typologies
Fleet size did not change much between 2001 and 2005 (Figure 1a and b). In 2005, the number of vessels ranged between
35 (FL2) and 150 (FL5) units.
|
Table 5 lists the occurrences where 1, 2, 3, 4, or 5 métiers (defined at level 1 or 2) were operated by the same vessel during the same calendar month. Most trips followed the same métier during a month (70–80%, and 60–73% when métiers are defined at levels 1 and 2, respectively), and this number increased over the period 2001–2005. Métiers at both levels 1 and 2 are needed in the analyses. If only one métier defined at level 2 operates in a given month, then only one métier defined at level 1 operates during that month. However, when one métier defined at level 1 operates in a given month, several métiers defined at level 2 can operate during the same month. To ensure that a single métier defined at level 1 or 2 operates each month, the fishing trips were selected based on métiers defined at level 2. The number of trips excluded based on the level 2 métier definition (23–40% of fishing trips) is only slightly higher than if the selection had been based on the level 1 métier definition (20–30% of fishing trips excluded).
|
Figure 1c and d suggests that the relative proportion of fishing effort allotted to each fleet and each métier (as defined at level 1) changed between 2001 and 2005. However, most métiers operating in 2001 were still doing so in 2005, although in different proportion.
The results of the cluster analysis of catch profiles are shown in Table 6 and Figures 2 and 3. There is a peak in the pseudo-t2 statistic at 16 clusters (Figure 2d), and significant increases in the cubic clustering criterion are observed up to 16 clusters, but not with a larger number of clusters (Figure 2b). With 16 clusters, the r2 is 85%, and there is little increase with a higher level of clustering (r2 < 90% with 24 clusters). Based on these criteria, 16 clusters were chosen to classify catch profiles in subsequent analyses. Table 6 lists the species composition of each cluster. For 11 of the 16 clusters, only one species contributes to >50% of the landings. For the remaining five clusters, the bulk of the landings is from two or three species. Figure 3 suggests that the catch profile composition of fleets changed over time, more particularly so for fleets FL1, FL2, and FL6. The quasi-absence of catch profile c8 (Atlantic tuna) for fleet FL1 in 2004 is attributable to missing information (see Data above).
|
|
|
) relative to total landings.Linking fishing intention to catch profiles
The results of the GLM analyses of the different uncertainty coefficients did not reveal any annual or seasonal effects. The detailed results of that analysis are not presented here. Pairwise comparisons of the uncertainty coefficients IO1–IO4 and OI1–OI4 are shown for each fleet in Figures 4 and 5.
|
|
As could be anticipated, the more disaggregated the métiers, the better explained were the catch profiles, so consistently, IO1 < IO2 and IOi
4 < IO4. The linkages between IO1 and IO3, on the one hand, and IO2 and IO3 on the other are clearly fleet-dependent. Defining métiers based on gear and a priori target species, as aggregated at level 2, better explains catch profiles than defining métiers based on gear and mesh size (IO2 > IO3), for all bottom trawling fleets (FL2, FL4, FL5, FL6) and also for Bay of Biscay gillnetters (FL7). The same conclusion is reached when métiers are defined based on gear and a priori target species, as aggregated at level 1 (IO1 > IO3) for all bottom trawling fleets except FL6. For pelagic trawlers (FL1), métiers based on gear and mesh size contribute more to subsequent catch profiles than métiers based on gear and target species, irrespective of whether they are aggregated at level 1 or 2 (IO1 < IO2 < IO3). For IO2, the ability to explain catch profiles ranged from a low 20% (FL2) to a high 80% (FL7). FL2 is the fleet for which catch profiles are least explained by métiers, irrespective of the typology method applied.
As a general rule, OIi4 > OI4 and OI1 
OI2 for all fleets. The linkages between OI2 (or OI1) and OI3 are fleet-dependent. Catch profiles better contribute to explaining métiers defined based on gear and a priori target species (aggregated at level 1 or 2) than métiers defined based on gear and mesh size (OI1 > OI3, and OI2 > OI3), for all fleets except FL1, for which OI1, OI2, and OI3 have comparable low values. To consider OI2, the ability to explain métier categories based on catch profiles ranged from a low 25% (FL2) to 70% (FL7).
Overall, grouping métiers as combinations of gear and a priori target species, aggregated at level 2, generally appears to be the most appropriate typology, in terms of explaining both catch profiles given métiers (as reflected by IO2), and métiers given catch profiles (as reflected by OI2). This typology was therefore retained to generate forecasts.
Forecasting fishing intentions and catch profiles
Figure 6a–d indicates that the precision of forecasts (catch profiles or métiers) does not depend much on the year where information on past correlations between catch profiles, metiers, and seasons was last made available (2001, 2002, 2003, or 2004), except for fleets FL1 (Figure 6a–d) and FL6 (Figure 6b and d). For fleet FL1, the drop in 2004 (Figure 6a) is an artefact attributable to the misrepresentation of vessels having a major contribution to catch profile c8 (see Data above, and Figure 3). As for fleet FL6, the linkage between catch profiles and métiers over the most recent period (2003–2004) was a better predictor of métiers than the 2001–2002 correlation (Figure 6b).
|
The proportion of fishing trips and vessels for which catch profiles were correctly forecast in 2005, given availability of past correspondence between catch profiles and métiers, was greatest (90%) for the gillnet fleets and lowest for pelagic and demersal trawlers (20–75%), with the minimum value attained by FL2 (Figure 6a). These results are consistent with the values taken by IO2, except for FL3, for which IO2 was second lowest (Figure 5). Comparison of Figure 6a and c suggests that the past correlation between catch profile and métiers is a better predictor of future catch profiles than the past seasonal distribution of catch profiles, except for fleets FL2 and FL3.
The proportion of fishing trips and vessels for which métiers were correctly forecast in 2005, given availability of past correspondence between catch profiles and métiers, is greatest (80–85%) for pelagic trawlers and the gillnet fleets, and lowest for demersal trawlers (5–60%), with the minimum value attained by FL2 (Figure 6b). These results are consistent with the values taken by OI2, except for FL3, for which OI2 was second lowest (Figure 5). Comparison of Figures 6b and 6d suggests that the past correlation between catch profile and métiers is a better predictor of future métiers than the past seasonal distribution of métiers, except for fleets FL2 and FL3.
|
Discussion |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
The framework suggested here allows one to assess the extent to which métiers and catch profiles are interchangeable concepts. Although earlier studies have characterized catch profiles using input-based métier definition, and vice versa (Laurec et al., 1991; Pelletier and Ferraris, 2000; Ulrich and Andersen, 2004), to my knowledge, this is the first time that the linkage between these two concepts has been (i) quantified for different fleets using specific indicators, and (ii) used to forecast catch profiles and métiers in the short term.
Four typologies have been proposed to classify métiers. Overall, grouping métiers as combinations of gear and a priori target species, at the most disaggregated level, generally appears to be the most appropriate typology, in terms of explaining catch profiles given métiers and métiers given catch profiles. This result suggests that initiatives to collect information on fishing intention before a fishing trip (e.g. target species), which cannot currently be accessed directly from logbooks, should be encouraged.
The linkage between métiers and catch profiles, and the ability to forecast them, were clearly fleet-dependent. It is striking that the best scores, in terms of forecasting either catch profiles (Figure 6a) or métiers (Figure 6b) was achieved by the two gillnet fleets (FL3 and FL7) and the pelagic fleet (FL1). For the latter, the drop observed in 2004 (Figure 6a) should be interpreted as an artefact, as explained above. Gillnetters and pelagic trawlers are generally more species-selective than bottom trawlers. Trammelnets are used to target sole (Solea solea) at night (catch profile 11 in Table 6), and fixed nets to target either cod in the North Sea (catch profile 13) or hake in the Bay of Biscay with small mesh size (catch profile 4), or anglerfish (Lophius sp.) in the Bay of Biscay with a large mesh size (catch profile 15). Depending on the mesh size used, the pelagic fleet may target anchovy (Engraulis encrasicolus, catch profile c7), mackerel (Scomber scombrus, catch profile c6), sea bass (Dicentrarchus labrax, catch profile c5), or Atlantic tuna (Thunnus alalunga, catch profile c8). Therefore, relatively fewer catch profiles, and consequently a lower probability of mispredicting those catch profiles, may be expected by métier category, and vice versa. One would also have anticipated the uncertainty coefficients IO2 and OI2 to be greatest for both gillnet fleets (Figure 5). However, this was the case for FL7 (IO2 and OI2) and FL1 (OI2), but not for FL3 (IO2 and OI2) and FL1 (IO2), which deserves some comment. Of all seven fleets, FL3 and FL1 are the ones for which IO2 and OI2 vary most (across fishing vessels and trips). Despite that variability, when averaging over vessels and trips by métier (respectively catch profile), the catch profile (respectively métier) with the greatest probability is generally the one observed in future. The score reached is 90% (respectively 75%) for forecasting the catch profiles of FL3 (respectively FL1), and 85% (respectively 80%) for forecasting the métiers operated by FL3 (respectively FL1). This suggests that the relatively small value taken by the median value of uncertainty coefficients only reflects large variability between vessels and trips, rather than a bias.
In comparison with the gillnetters and pelagic trawlers, the bottom trawlers (FL1, FL4, FL5, and FL6) are less species-selective, so reach a lower score (in terms of forecasting both catch profiles and métiers). These results are consistent with the relative values taken by the uncertainty coefficients. The forecasting scores achieved by the three Bay of Biscay bottom trawl fleets (FL4, FL5, and FL6) are of the same order of magnitude (at least for the forecasts building on the 2003–2004 catch profile/métier correlations), suggesting no effect of vessel length.
The scores and the uncertainty coefficients values obtained with the North Sea bottom trawl fleet (FL2) are the lowest of the seven fleets under investigation. Although the reasons for these low values are not completely clear, it may be hypothesized that (i) either the typologies applied here to define métiers and catch profiles may not be totally adequate or (ii) recent changes in the relative distribution of fish stocks of the North Sea are such that fishing intentions rarely match catch profiles. Figure 3, and the outcome of recent investigations, may provide some support to hypothesis (ii). These suggest in particular that species considered in the past to be valuable bycatch, e.g. red mullet (Mullus surmuletus), sea bass, squid (Loligo sp.), have become in recent years an increasing component of the landings of that fleet, concurrent with the decline of traditionally targeted species (e.g. cod, whiting; Mahé et al., 2005; ICES, 2007).
More generally, one would have expected a priori the relative resource availability to be an important determinant of the linkage between fishing intentions and catch profiles. For instance, the decline of some traditionally important stocks might be regarded as a reason for the very poor forecasting scores for fleet FL2 (North Sea bottom trawlers). However, FL1 (Bay of Biscay pelagic trawlers) is an example of a fleet for which the traditional main target species (anchovy in ICES Subarea VIII) declined dramatically between 2001 and 2005 (according to the latest ICES advice available at http://www.ices.dk/committe/acfm/comwork/report/2007/oct/ane-bisc.pdf depleted), and yet for which the forecasting score is high. Our results indicate that the observed shift has been reflected by fewer trips dedicated to anchovy fishing (métier NEA27 in Figure 1c and d), and an increased number of trips dedicated to demersal species, mainly sea bass (métier NEA28 in Figure 1c and d), rather than by a discrepancy between the fishing intention and the catch profiles for those trips. The catch composition from vessels rigged to fish for anchovy is still dominated by anchovy, but fishers operate that métier less often than they used to. In other words, the results of this investigation suggest that fishers belonging to fleet FL1 have managed to adapt their strategy to prevailing stock and TAC conditions, and have made limited effort to target anchovy when they knew they had limited opportunity to catch this species in an economically efficient way. This is confirmed by both the number of vessels and total fishing effort remaining stable for fleet FL1 between 2001 and 2005 (Figure 1a and b).
Overall, it seems reasonable to forecast both catch profiles and métiers for gillnetters and pelagic trawlers building on past information, the risk of error being 10–25%. For Bay of Biscay bottom trawlers, the risk of error is higher (20–40% using information available at year n–1), but still lower than using past seasonal distributions. No forecast of either catch profiles or métiers should reasonably be attempted for North Sea bottom trawlers with the current typology, because the risk of error ranges between 70% and 95%. These results have practical implications. First, some management tools are applied to métiers (e.g. licensing systems). The efficiency of these depends heavily on the adequacy with which the métier is implemented and the resulting catch composition. Second, catch profiles are less onerous to calculate than organizing interviews to collect information on fishing intentions. Therefore, when catch profiles are seen as a good proxy for fishing intention (e.g. for fleets FL1, FL3, and FL7), the priority for collection of a priori information on fishing intentions could be regarded as lower than for the other fleets.
This study is subject to a number of limitations. First of these is the quality of data recorded in logbooks and activity calendars. The data included in logbooks include only part of the catches taken, not that part which is discarded or unreported. Therefore, the catch profiles used in the study are in fact only landings profiles, which may be a source of bias. In terms of activity calendars, the quality of the information collected on fishing intentions will depend to an unknown extent on the relationship between the interviewer (the scientist) and the interviewee (the fisher).
Second, the results of the investigation are sensitive to the métier typology used. Using métiers as sampling strata or management units, there is always a trade-off between maintaining a reasonably low number of categories and describing as comprehensively as possible the activity of the fleets. A range of four typologies, building on different sources of information (logbooks or activity calendars) and grouping procedures (partly or fully disaggregated) have been explored here. To test further the robustness of the results obtained, other typologies should be investigated and their outcomes compared with those achieved here.
Future research will ideally aim at evaluating the consistency between métiers and catch profiles over a longer period than considered here. Also, the extent to which the typology advocated here would influence the outcome of sampling strategies and management plans which would build on these métier categories remains to be tested.
|
Acknowledgements |
|---|
The work was funded through the CAFE project of the European Union (DG–Fish, contract 022644), for which support I am very grateful. I also thank Jan-Jaap Poos, Wageningen-IMARES (The Netherlands), for writing the programs used to carry out the cluster analysis. Finally, I am indebted to all the skippers for their cooperation during harbour enquiries, and to the reviewers of the draft manuscript for valued comment.
|
References |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
-
Anon. New Zealand Ministry of fisheries WAREHOU database documentation, catch and effort base views and fields, version 6.0 (2004) 42.
Biseau A. Definition of a directed fishing effort in a mixed-species trawl fishery, and its impact on stock assessment. Aquatic Living Resources (1998) 11:119–136.[CrossRef][Web of Science]
Biseau A., Gondeaux E. Apport des méthodes d'ordination en typologie des flottilles. Journal du Conseil International pour l'Exploration de la Mer (1988) 44:286–296.
Christensen A-S., Raakjer J. Fishermen's tactical and strategic decisions. A case study of Danish demersal fisheries. Fisheries Research (2006) 81:258–267.[CrossRef][Web of Science]
EC. Commission Staff Working Paper: Report of the Ad Hoc Meeting of Independent Experts on Fleet-Fishery based Sampling, Nantes (France), 23–27 May 2005. (2005) 34.
EC. Commission Staff Working Paper: Report of the Ad Hoc Meeting of Independent Experts on Fleet-Fishery based Sampling, Nantes (France), 12–16 June 2006. (2006) 98.
Goodman L. A., Kruskal W. H. Measures of Association for Cross Classification (1979) New York: Springer.
Holley J-F., Marchal P. Fishing strategy development under changing conditions: examples from the French offshore fleet fishing in the North Atlantic. ICES Journal of Marine Science (2004) 61:1410–1431.
ICES. Report of the Study Group for the Development of Fishery-based Forecasts. ICES Document CM 2003/ACFM: 08 Ref. D. (2003) 37.
ICES. Report of the Working Group on Assessment of New MoU Species. ICES Document CM 2007/ACFM: 01 Ref. LRC. (2007) 228.
Laurec A., Biseau A., Charuau A. Modelling technical interactions. ICES Marine Science Symposia (1991) 193:225–236.
Léwy P., Vinther M. Identification of Danish North Sea trawl fisheries. ICES Journal of Marine Science (1994) 51:263–272.
Mahé K., Destombes A., Coppin F., Koubbi P., Vaz S., Le Roy D., Carpentier A. Le rouget barbet de roche Mullus surmuletus (L. 1758) en Manche orientale et mer du Nord. IFREMER Document. (2005) 187. (in French).
Marchal P., Andersen B., Bromley D., Iriondo A., Mahévas S., Quirijns F., Rackham B., et al. Improving the definition of fishing effort for important European fleets by accounting for the skipper effect. Canadian Journal of Fisheries and Aquatic Sciences (2006) 63:510–533.
Marchal P., Horwood J. Long-term targets for the Celtic Sea mixed-species multi-métiers fisheries. Aquatic Living Resource (1996) 9:81–94.
Murawski S. A., Finn J. T. Optimal effort allocation among competing mixed-species fisheries, subject to fishing mortality constraints. Canadian Journal of Fisheries and Aquatic Sciences (1986) 43:90–110.
Neis B., Schneider D. C., Felt L., Haedrich R. L., Fischer J., Hutchings J. A. Fisheries assessment: what can be learned from interviewing resource users? Canadian Journal of Fisheries and Aquatic Sciences (1999) 56:1949–1963.
Pelletier D., Ferraris J. A multivariate approach for defining fishing tactics from commercial catch and effort data. Canadian Journal of Fisheries and Aquatic Sciences (2000) 57:51–65.
SAS/STAT. SAS/STAT User's Guide, version 8 (1999) Cary, NC: SAS Institute Inc.
Ulrich C., Andersen B. A. Dynamics of fisheries, and the flexibility of vessel activity in Denmark between 1989 and 2001. ICES Journal of Marine Science (2004) 61:308–322.
Ulrich C., Gascuel D., Dunn M. R., Le Gallic B., Dintheer C. Estimation of technical interactions due to the competition for resource in a mixed-species fishery, and the typology of fleets and métiers in the English Channel. Aquatic Living Resources (2001) 14:267–281.[CrossRef][Web of Science]
Vinther M., Reeves S. A., Patterson K. R. From single-species advice to mixed-species management: taking the next step. ICES Journal of Marine Science (2004) 61:1398–1409.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. Duarte, M. Azevedo, and M. Afonso-Dias Segmentation and fishery characteristics of the mixed-species multi-gear Portuguese fleet ICES J. Mar. Sci., April 1, 2009; 66(3): 594 - 606. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||