Annex 1. Contribution to the terms of reference of the Workshop on fleet-fishery based approach
Classification of fishing activities combined to a new fleet segmentation : proposition to build an operational dataframe for the collection of fisheries data.
Joël Vigneaua, Fabienne Dauresb, Sebastien Demanecheb, Christian Dintheerd, Patrick Lespagnole, Paul Marchalf, Claude Merrieng, Alain Tétarda
a Ifremer, DCMMN-HMMN Port-en-Bessin
b Ifremer, DCB-EM Brest
d Ifremer, DCN-EMH, Nantes
e Ifremer, DCM-HMT, Sète
f Ifremer, DCMMN-HMMN Boulogne/Mer
g Ifremer, D, Lorient
The concept of métier and fleet disagregation tree, or how to link biological and economical sampling to build an operational dataframe
Joël Vigneaua, Fabienne Dauresb, Sebastien Demanechec, Alain Tétarda, ……………..
a Ifremer, DHMMN Port-en-Bessin
b Ifremer, SEM, Brest
c Brest
Abstract
The inclusion of economic parameters in fisheries management has proved to be more problematic than initially thought, especially when combined to the biological parameters. The reason is that the two estimation process are sampling different populations, one is sampling the population of fishing trips while the other samples the population of vessels. This document proposes first to stratify fishing activities regarding the gear used and the species targeted, this stratification is aimed to enhance the precision of the biological sampling by shifting from stock sampling to métier sampling. The fleet segmentation based on dominance criteria has shown instability of the vessels remaining in one segment during several years. The new segmentation proposed here is more stable and is supposed to empower the use for management purpose. In the last part of the document, both stratification will be linked together to form a data collection grid authorising the association of both biological and economic information. This three step process has been drawn from the French experience in order to start the reflection on the enhancement of the European gathering program for fisheries data.
1.Introduction
For numerous years, fFisheries catch-at-age data have traditionally been collected onat a species stock basis level for the purpose of stock assessments and forecasts. Some attempts to turn to a fishery/métier based sampling have been undertaken in the European framework with projects like FIEFA (FIEFA, 1995), SAMFISH (Anon, 2000) in Atlantic waters and COPEMED (Coppola, 2000) in the Mediterranean. Sampling for biological parameters like discards, length/age structure implies to refer to a population of fishing trips (Anon, 2003 and ICES 2004b). Fishing trip is defined as the sampling unit and the population can be split into strata like quarter, gear types or more generally into strata of similar exploitation pattern. The idea behind regrouping the fishing trips according to their exploitation pattern is to improve gain inthe sampling precision by integrating the dynamics and stick to the realityof fishing in an explicit way.
With regards to economic data, the sampling unit is the vessel throughout the year and the population is derived from the fleet register. The segmentation proposed in appendix III and IV of the EC Regulation No 1639/2001 (hereafter called Data Collection Regulation, DCR) will be tested against two criteria: the stability of the vessels and the variability of gross revenue in each segment. Based on the outcomes of this analysis, a new segmentation will be proposed, and linked to the vessels’ annual fishing activity and capacityan indicator of the vessel size.
A major challenge t the momentof undertaking bio-economic analysis modelling and also of or management strategies evaluations, scenarii, the difficulty arisesis to be able to link input data collected under different sampling schemes and derived comingfrom different populations. From the construction of both sampling plans, the The common item of both the biological and the economic sampling scheme is not the species stock, but the fishing activity. The fishing activity at a yearly basis defines the economic strata and the fishing activity at the trip level defines the exploitation pattern to sample.
This working document will try to proposessensible waysa generic approach to split the fishing trips into groups of similar exploitation pattern. These groups will be clustered asin a hierarchical tree, which will represent theto enable different levels of biological sampling. The same concept will be applied to the fleets for economic sampling purpose. Those two concepts will then be linked into a stock-based matrix where the different sampling levels could be positionedwill help the reader to understand the overall concept.
2.The population of fishing trips
By sampling or by census, the collection of data in fisheries science is often linked to a fishing trip. Catches, discards, length or age structure, effort, CPUE can be linked to variables of interest like gear used, geographical area, time of the year and species targeted for example. Grouping fishing trips into homogeneous sampling units raises three issues. The first issue idea to define the fishing trip as a sampling unit is to state ensure that this sampling unitsis are homogeneous considering the variables of interest and that all the variables are known at the population level (Cochran, 1977). The second issueThe second difficulty is to define fishing units with similar exploitation pattern so that samplingforof biological parameters achievesassuresminimum minimal of bias and maximum maximal of precision. Finally, the ultimate purposeThe third issue is to arrange cluster the different fishing units into a hierarchical tree to enable sampling at different levels of disagregation
2.1.Homogeneity of a fishing trip
Within a fishing trip, some vessels will use only one gear, target one (assemblage of) species into one geographical area, some but others will nothave a more flexible behaviour. Looking Analysing at the landings of one vessel can prove difficult, become very tricky when theyse landings result fromare a combination of gears and/or areas. AlthoughObviously, the universal most appropriate way to sample biological parameters is in theory to sample onboard, there are practical difficulties in implementing such a sampling procedure (ICES, 2005) but everyone knows the inherent difficulties of such a sampling program.
A good candidate for sampling unit would be the fishing operation, as a fishing operation is done with one gear, targeting one (assemblage of) species in one area. However, there is currently no data sources (e.g. log-books) documenting comprehensively total landings at the scale of the fishing operation. Therefore, it is currently not possible to raise age-structured landings sampled by fishing operation to the total landings.
For that reason, the sampling unit will remain the trip, and the problem of vessels using several gears during the main trip will have to be simplified considering only the main een through the principal gear used. The problem of multi areas is more difficult as it can touch different stocks could be caught during the same tripfor a same species. One way to deal with this problem is to avoid sampling these problematic voyages. The age structure of the landings of such voyages will be estimated using biological samplings collected from single-area voyages.
2.2.Stratification based on fishing trips : the concept of métier
Fishing trips from vessels of similar characteristics, using one gear, targeting the same (assemblage of) species in one area is the agreed definition of métier (ICES 2003 and 2004a, SGECA, 2004). There are hundreds of different gears if one looks at their precise characteristics as there are an infinity of areas depending on the definition of their frontiers. The notion of exploitation pattern is important here to discriminate two different métiers or to aggregate others.
In France, hundreds of different métiers have been listed, as different as clam fishing on the beach, scallop dredging, fixed net targeting crustaceans or saithe offshore trawling. There is no theoretical limit in the number of métiers, as long as the fishing trips within the métiers strata do not overlap. For example, a vessel targeting pelagic fish during the day with anappropriate pelagic trawl, andwaiting for the night to targeting sole during the night with a bottom beam trawl will not bepractice both in pelagic trawling and sole trawling fishing métiers. To consider which is the principal métier practised, the rule taken here is based on the most valuable species and the remaining species will be seen as by-catch. Another idea would be to define a mixed pelagico-flat fish trawling metier if this is characteristic of the local habits, but this would add a lot ofcause confusion, especially in the construction of the hierarchical tree (Cf. bellow).
2.3.Hierarchical model
For obvious practical reasons, it is not possible to sample At the end we have the numerous French métiers at the most disaggregated level, which has been identifiedto sample in France.! The main aim of a sampling plan is, (i) to estimate parameters with a given precision and, is designed (ii) to get the best compromise between the quality and the cost. This search of compromise implies requires to sample at another a more flexible level than the métier, which we refer to as let’s call it the “métier family” or “super métier” level. For example, the vessels targeting cod with demersal trawls during one fishing trip belong to the same family as vessels targeting haddock and/or whiting and/or gadoid with demersal trawl if they fish in the same area. For the purpose of length sampling, the population to sample will be the family of fishing trips were gadoids were targeted. Other vessels in the same area can use demersal trawls to fish non-anything but gadoid species, like such as squids or red mullet. In the purpose of discard sampling, it is more convenient to regroup all the gadoids and non- gadoids fishing trips into a family of fishing trips using a demersal trawl. Table 1a to 1c present a generic hierarchical tree, which could be applied to cluster a broad variety of for France fisheries.
For a given year, one vessel may have, (i) fished, (ii) been employed for the purpose of another activity than fishing and, (iii) stopped any activity for reparation. This split into three categories is referred to as the the level 0 (table 1a). TheSome sampling of some biological parameters, which are purely stock-based (e.g. and free from any gearmaturity, weight at length/age and the age/length key) could be envisaged at level 0.
During the time were the vessel has a fishing activity, this activity can be undertaken in the open sea, ashore or in an estuary and this corresponds to the second level of the tree (level 1 of table 1a).
A vessel fishing in the open sea will operate mainly with one class of gear given the rule of paragraph 2.1. This vessel may use this class of gear during a period of the year, and choose another one in a different period. Switching between gear classes reflects the opportunistic behaviour of fishermen. The exhaustive list of classes of gears available for fishing in the open sea is given on level 2 of table 1b. These classes of gears are then split into more precise categories, gears families (level 3) and gears (level 4). The gears catalogue and coding corresponding to level 4 (table 1b) are fully consistent with the international gear classification used by FAO (Nedelec, 1982, Anon. 1994, Le Gall, 2004). Level 4 is a possible generic candidate for sampling discards.
With a specific gear (level 4), a vessel may target different species, depending on the fisher’s choice regarding fishing grounds, fishing season and gear attributes. In the purpose of sampling for biological parameters, it is impractical to split level 4 on the basis of all possible commercial species. For the sake of simplicity, groups of species were constructed. Level 5 is then constructed by splitting the level 4 activities into gear specific targeted groups of species (Shellfish, crustaceans, molluscs, benthic, demersal and pelagic fish). Level 5 is a possible generic candidate for the biological sampling of the landings length distribution by stock.
The level 6 splits the species groups of level 5 into species families. This level is not meant to be generic and it is presented here for the French case study. Level 6 is a possible candidate for the definition of Regional métiers at the scale of a RAC.
Finally, level 7 is country specific and may contain as many métiers as necessary. Only one métier is given as an example in table 1b.
Concerning the shore and estuary fishing (table 1c) only a sketch of hierarchical tree has been undertaken. Coppola (2000) in the framework of COPEMED stresses the difficulty to carry out this kind of work in such an heterogeneous population. This prototype hierarchical tree should be further examined by the forthcoming STECF workshop on small-scale fisheries. Nevertheless, it has been thought important to propose at that stage a basis structure relevant to shore and estuary fishing, in order to ensure consistency between the classification of the open sea fishing activities (table 1b) and of the shore and estuary fishing activities (table 1c).
3.The population of vessels
For each country, the population of vessels is derived from the fleet register on a yearly basis. The vessels are individually split into strata (or fleet segments defined by the DCR) on the basis of homogeneous fishing activity and similar physical characteristics. Economic information must be collected for this population by sampling for each segments defined by the DCR.
Theoretically, the vessel’s activity during a year can be defined with regards to the different métiers carried out from January to December. The concept of métier has been presented previously (combination of gear/target species/fishing areas) but for practical reasons, only the gear(s) used by the vessel was considered as selection criterion. However, it must be recalled that the same gear can have different use depending on target species and/or fishing areas.
The appendix III of the DCR is a little ambiguous, as the definitions of both fleet segment and of métier are based on the gear used. Considering the fleet segment, the vessels are gathered in the same stratum when they devote more or less time to one specific (type of) gear and when their physical characteristics are more or less the same. This “more or less” rule is very important as it leads to some instability that will be discussed. Indeed, the question is related to the efficiency of this stratification to (i) define an optimal sampling plan for the collection of economic information and (ii) provide a stable stratification to analyze the evolution of economic indicators over time. At last, the practical use of this economic information for the purpose of bio economic modeling of fisheries must be discussed.
Briefly, the DCR considers the gear as the key point to gather vessels into fleets. The “bottom trawler’s fleet” gather the vessels which have devoted the majority of their annual fishing time using the bottom trawl. Apart from 6 “well known” categories of fleet (bottom trawl, pelagic trawl, beam trawl, netter, potter and hook) , 3 categories of polyvalent fleets exist (polyvalent mobile, polyvalent fixed gears and polyvalent mobile and fixed) and some categories are created considering national cases (eel fleet in France for example). On the other hand, the DCR considers the difference in physical characteristics of the vessels on the basis of 4 length categories (less than 12 m., 12 to 24 m.; 24 to 40 m.; up to 40 meters). The crossing between fleet and length category defines the segment or the stratum.
The objective of this part is to propose a new methodology for the definition of the stratum based on the combination of gear and not on the dominance (for the fleet part) and on more classes of length categories for the capacity part. Then, the two segmentations (First DCR segmentation, referred to as DCR_0, and a new generic DCR segmentation, referred to as DCR_1) will be compared regarding their stability over the time and the variability of economic indicators. Finally, a last DCR segmentation (referred to as DCR_2), more detailed than the DCR_1 will also be presented, as a candidate for economic sampling in France.
3.1.From DCR_0 to DCR_1 fleet segmentations
The collection of information on fishing activity for each vessel of the French fleet register is done annually on the basis of a questionnaire (see explanation later). Each vessel gives information on each métier (combination of gear/target species and fishing areas) it practised for each month of the given year. The combination of gears used during a year (disregarding when and for how long they have been used) is considered as a reasonable proxy of the fishing activity, and this combination has been used as stratification criterion (IFREMER detailed segmentation). Based on 2003 data, 30 large groups of vessel can be defined regarding the gear their use over the year (Table 2).
Table 2 shows a large diversity of activities (defined on the basis of gear combination) represented in each first DCR fleet. The bottom trawl fleet is composed with vessels using bottom trawl exclusively, mix bottom and pelagic trawl, but also vessels which combined trawl and dredge or trawl and eel gear over the year. The time allocated to each gear could be used as a criterion for another first DCR fleet. Finally, two vessels with the same size can belong to the same Ifremer’s detailed segment but to different first DCR fleets and this is problematic as the IFREMER detailed segment reveals a kind of homogeneity in economic indicators especially in terms of investment and gross earnings.