An analysis of fishing gear competition . Catalan fisheries as case studies

An asymmetric index was developed to measure the competition relationships among fishing fleets (or gears or métiers) in a multispecies fishery. This index can be used to measure the degree of dominance of each fleet and its level of independence from competition. To illustrate the concepts, the index is applied to two case studies using two datasets, both from Catalonia, NW Mediterranean. The results show that in both case studies the dominance of bottom trawl over most other gears (especially small-scale ones) is evidenced and quantitatively measured. Bottom trawl is also highly independent of the others. Purse seine appears to be quite independent, but not dominant over the other gears. A practical use of these asymmetric indices is to assist fisheries managers in the decision-making process to optimize the allocation of fishing effort, including energy efficiency, and to reduce environmental impact.


INTRODUCTION
Most Mediterranean fisheries are characterized by the activity of several different gears working on a multispecies environment (métiers, Pelletier and Ferraris 2000).Each gear exerts a different selectivity pattern on each species, so different métiers may affect a fish population in different stages of its life history.This pattern may be due to the technical aspects (i.e.mesh or hook size) of specific gear, and also to the environment (depth, habitat) where a particular gear is deployed or the skipper's knowledge or local tradition (Maynou et al. 2011).
The management of multispecies, multigear fisheries is a problem typical of mixed fisheries.This problem has often been approached from the perspective of technological interactions in the capture of particularly important species (e.g.Aldebert et al. 1993 and Aldebert and Recasens 1996 for hake; Demestre et al. 1997 for red mullet;Stergiou et al. 1996 andErzini et al. 2003 for multispecies fishery).These studies show, for instance, that trawling with low selectivity and high catches of juvenile European hake (Merluccius merluccius) negatively affects the profitability of set gear fleets (longliners and gillnetters) (Lleonart et al. 2003, Merino et al. 2007).The technical interactions between fishing gears is also important in the study of fisheriesinduced evolution (Kuparinen et al. 2009).
Here we propose a multivariate approach based on an asymmetric index to identify and quantify the relationships of dominance and competition among fishing fleets or métiers sharing resources in a mixed fishery.
Competition is an asymmetric relationship because competitors are unequal.There are generalist (or opportunistic) vs specialized fleets and gears; large fleets and small ones; highly mobile and less mobile fleets; and gears targeting large fish and gears targeting small fish.In ecology, different types of asymmetric indices have been developed either to assess the predator-prey relationships in trophic webs (Gallopín 1972) or to measure niche overlap among species (Mouillot et al. 2005, Pledger andGeange 2009).These indices are usually appropriate to field collected data in which the relative abundance of species in their environment is usually related to preference of trophic resources or gradients of environmental variables.Their applicability to fisheries landing data is limited because these data do not inform about the environment in which a fish was caught and the species fished are a collection of items selected based on their commercial importance, regardless of their ecological role.
The objective of this paper is to introduce an asymmetric index that is easily applicable to fisheries landing data and can be used to quantify the competition relationships among fleets or métiers for common fish resources.This index introduces the concept of dominance of one fleet over another, meaning the impact of the first fleet on the common resources of the two fleets.It can also be used to calculate overall dominance and dependence indices (a measure of shared resources) for each fleet.
The approach is illustrated with two case studies from data of Catalan fisheries for the 10-year period 2000-2009.

Data source
Two different data sets were used, one for each case study.The first one was the official statistics of Catalonia from 2000 to 2009 (the daily fish sales database of the General Fisheries Directorate of the Catalan Government, not published).The data come from the daily auctions by boat and species.Boats are classified according to fishing gear into five classes, following the official types of license issued by the local Fisheries Department: bottom trawl, purse seine, bottom longline, surface longline and small-scale vessels.These five types correspond to otter bottom trawl, purse seine, set longline, drifting longline and passive gear for vessels 0-12 m length, according to the classification used in the EU Data Collection Framework (Commission Regulation (EC) No. 665/2008).The total number of species (or species groups) reported in the statistics is 198, although 37 make up 90% of the catch.As shown in Appendix 1, the trawl fleet and the purse seine fleet produce the highest catches, although the former is more diverse and depends on a higher amount of species than the latter.
Taxa such as Osteichthyes and Invertebrata could be included or excluded from the analysis, although it is not clear whether this inclusion or exclusion would be advantageous.This uncertainty reflects the inaccuracy of species identification of commercial landing reports.In this paper these species groups have been included in the analysis.In the data set there are 26 groups in the 198 taxa.Of these groups, 10 are genera without species definition (such as Lophius sp. or Mullus sp.), 7 are families (such as Gobiidae or Labridae) and 9 represent higher taxonomic groups (such as Crustacea or Chondrichthyes).Since there is no objective criterion for including or excluding species groups, we chose to follow the FAO reporting standards and include them.
The second case study refers to a single harbour, Vilanova i la Geltrú, during the same period (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009), for which data were taken from the daily fish sales database of the General Fisheries Directorate of the Catalan Government (not published).The raw data also include the daily catch by boat and species, but in this case the vessels were divided into 11 métiers according to the multivariate analysis published by Maynou et al. (2011).The number of species, or taxonomic units, reported was 132.The two main fishing types, in terms of catches, are bottom trawl and purse seine, as in the previous case, but nine other small-scale métiers were revealed in the multivariate analysis: trammel nets targeting cuttlefish (a métier known as "cuttlefish"), gillnets targeting European hake ("g-hake"), bottom longlines targeting European hake ("l-hake"), clay pots targeting octopus ("octopus"), trammel nets targeting striped red mullet ("redmullet"), trammel nets targeting sole ("sole") and trammel nets targeting striped seabream and other sparids ("varied"), towed dredges targeting coastal bivalves ("dredge") and toothed beam trawl targeting sea snails ("beam trawl") (Appendix 2).
The data used in this second case study show some apparent inconsistencies, such as the fact that the octopus clay pot fishery accounts for only 78% of octopus.This is not because of data errors but rather inaccuracies in the commercial reports, which have a minimum resolution of one day.Small-scale fishers often use different gears (i.e.pots and small longlines) in the same day and they report the whole catch for only one gear.These inaccuracies have been respected in order to keep the approach realistic with real inaccurate data.
The reason for using these two data sets is to analyse different scales.The first case study covers a wide geographical range with few great gear groups, while the second one is much finer in grouping gears (actually métiers) although it only refers to a single harbour.
Although the original data were collected at a daily temporal scale, they have been aggregated at an annual scale to illustrate the application of the metrics described herein.Figure 1 shows the study area and the annual catches of the fleets studied.

A metrics for measuring the competition among fleets or métiers
We are dealing with a multispecies fishery in which a number of vessels using several gears are acting.Let us consider a "fleet" to be a group of vessels practising the same métier.
Let n be the total number of species (or taxa) and m the total number of fleets working on the fishery.Let [D] be an n×m matrix where each element d ij is the total weight of species i caught by fleet j during a period of time (e.g. one year).
Two new matrices [P] and [Q], both n×m, can be obtained from [D] according to the following definitions of their elements: is the relative weight that fleet j represents for species i.The effect of species abundance is eliminated.
q ij represents the relative weight of the species i in the catch of fleet j.The effect of fleet size is therefore eliminated.
Given two fleets, A and B, the following index can be defined: The meaning of r iAB is as follows: p iA is the proportion of the total catch of species i caught by fleet A. If fleet A captures most of species i it deprives other fleets of this resource, so p iA represents the importance of A in the capture of i. q iB is the proportion of species i in the total catch of fleet B. The higher the value of q iB , the more important this species is for fleet B. If p iA and q iB are both high, this means that A captures most of i, which is an important resource for B, so A is a strong competitor against B regarding species i.If p iA is low and q iB is high, the species i is important for B, and A is not a heavy competitor for this resource.If p iA is high and q iB is low, A captures a good deal of i, but this species is not very important for B. Finally, if both p iA and q iB are low, the species i is not important either for A or B, and is probably the target of other fleets.
This index measures the competition of A with B concerning only species i.This index is asymmetric because the competition of A with B is different than that of B with A (r iAB ≠r iBA ).
The interaction of A with B considering all species is then s r where s AB is the index expressing the competition of fleet A with fleet B. It ranges between 0 and 1.For instance, if a gear A is getting high catches on most of the target species of gear B, s AB will be close to 1, so A must be considered as a strong competitor of, or dominant over, B. If the set of species targeted by A were different from the set caught by B, s AB would be close to 0, so A would be an irrelevant competitor of B. As expected, the index s AB is asymmetric (s AB ≠s BA ), as are the competition relationships.
s AB is an element of a new m×m matrix called [S].This matrix is computed as [S] is the matrix of interactions, or competition, between fleets.
The main diagonal of [S], or competition of a fleet with itself, can be interpreted as the level of independence of the corresponding fleet.If a fleet captures only species not caught by any other fleet, its value is 1 (the only competitor of a fleet is itself) and there is no interaction with others.If the species composition of a fleet is the same as that of another fleet or fleets but its catch is much lower, that value approaches 0 and the catch composition of this fleet is highly dependent on others.
An overall measure of the dominance of a fleet (A) over all the other fleets (minus itself) could be calculated as Asymmetric indices are not as easy to represent as symmetric ones, which can be managed through a series of multivariate statistical tools, mainly cluster analysis and diagonalization methods such as principal component analysis and similar procedures, which allow meaningless dimensions to be removed and cluster structures to be synthesized.As far as we know, there is no statistical tool for representing asymmetric relationships; only "sociograms", a type of non-quantitative graphs showing the asymmetric relationships between "individuals" used in social sciences, appear to be commonly used in such studies.
A sociogram consists in representing the elements (fleets) as circles.The diameter could represent any feature of the fleet.In the present case we use two concentric circles: the independence and the overall dominance.The circles are connected by asymmetric arrows with the width of the ends proportional to s AB , near to fleet A, and s BA , near to fleet B. When the pairs s AB , s BA are both smaller than a given value (i.e.0.05), they are not represented to improve readability.This simplification was applied only to the case of 11 fleets, involving 55 pair relationships.The software used for sociogram representation can be found in Lewejohann (2005).

Catalonia five-fleet data set
Catches with purse seine and bottom trawl were far greater than those with the other three gears (Fig. 1) throughout the time series.
The structure of matrix [S] was very similar across years.Linear correlation between pairs of years was 0.98 or higher, so [S] mean is shown, as an example, in Table 1.
This table also shows the standard deviation matrix of the ten annual [S] matrices.The standard deviation of the matrix elements across years confirms the relative temporal stability, with a possible single exception It must be noted that the first three higher values are found in the main diagonal, and correspond to purse seine, trawl and drifting longline, which means that they were the least dependent fleets in terms of species composition, though drifting longline was really very small in terms of catches.Bottom trawl is a strong competitor of the bottom longline (0.53) and small-scale fleets (0.40).The rest of the interactions were quite small, although some dominance of the small-scale fleet over longlines and of bottom trawl over drifting longline were observed.
The analysis of the trends of the main diagonal, as an indicator of independence (or self-dependence) among fleets (Fig. 3), shows that purse seine is the most independent fleet.This is because its target species, mainly sardine and anchovy, are not significantly targeted by any other fleet.Bottom trawl is also quite independent, while passive gear and bottom longline fleets have a fairly low independence level.This means that their catch composition is shared with other fleets, but bottom trawl takes a higher share.Among those   fleets, it is interesting to observe that drifting longline displays a temporal trend of increasing independence (from approximately 0.25 to 0.85 in Fig. 3), possibly related to its progressive specialization.In terms of dominance, bottom trawl appears to be the most dominant, thus being the most competitive (highly independent and dominant) of the fleets, followed by the small-scale fleet.Set longline is the most affected by competition because it is highly dependent and less dominant (Fig. 3).

Vilanova i la Geltrú, 11 métier data set
As in the previous case study, the correlation between matrices [S] of successive years is higher than 0.95.Table 2 shows 5).Dredge shows a trend to become more independent although it is not dominant because it is very specialized, targeting only bivalves.Beam trawl shows the opposite trend towards increasing its dependency on competitors.The remaining seven gears are both dominated and dependent and do not show significant trends over time.

DISCUSSION
The analyses carried out have quantified a set of relationships that are more or less qualitatively known by fisheries experts.However, it is also known that the lack of quantification could mislead the interpretation of relationships because of prejudices or preconceptions.The asymmetric index presented here helps the analyst to better understand the quantitative relationships between fishing gears or métiers in competitive multispecies fisheries.For instance, in the fisheries analysed here (as elsewhere in the Mediterranean) bottom trawl fleets are clearly dominant, in the sense that they produce large catches of species which are shared with other fleets, and they are independent of smallscale fleets that cannot compete in terms of landings.On the other hand, the Mediterranean purse seine fleet has low dominance and high independence because of its specialized target resource (small pelagics).In general, small-scale fisheries in the Mediterranean show an asymmetric relationship with bottom trawl and their long-term viability can be ensured only if they become more specialized (e.g. more independent, like surface longline in Catalonia and dredges in Vilanova i la Geltrú) or if market differentiation of the fish product can be enhanced (e.g.added value of high-quality large fish from small-scale métiers in Vilanova i la Geltrú).
Computing the indices over a time series should help to display trends and changes in dominance among fleets and their independence level over time.A practi- cal use of these asymmetric indices is to assist fisheries managers in decision-making to optimize the allocation of fishing effort, i.e. to increase the independence of fleets and decrease the dominance, thus reducing a kind of competition that yields benefits in the short term and to the less selective fleets and also helping to minimize social conflict.For instance, Lleonart et al. (2003), under a bioeconomic perspective, show that reducing the effort allocated to bottom trawl would enhance the productivity of small-scale longliners, which compete with bottom trawl for European hake (Merluccius merluccius).
Although the purpose of this paper is limited to studying competition between gears in a strict technical interaction scenario, these asymmetric indices could also be useful for examining other aspects of fisheries, such as unwanted by-catch that is ultimately discarded, fisheries employment and efficiency in fuel consumption (Suuronen et al. 2012).In particular, the application of such indices to the study of economic competition could be useful and the object of possible future expansion.Indeed bottom trawl, the dominant gear, uses large amounts of fuel and its competitiveness and even economic sustainability is dependent on the fuel tax exemption.
In the NW Mediterranean the classification of fleets as industrial or artisanal (or small-scale) is not as clear as that in other parts of the world.Following the multivariate definition of "artisanal fishery" proposed by Coppola (2006) and Griffiths et al. (2007), not a single fleet can be identified as fully artisanal (or industrial) in this area of the Mediterranean.The equipment of these small-scale fleets is "modern" or "intermediate" (sensu Misund et al. 2002) and in some cases (some longliners or netters) the small-scale fleets are technically more advanced than the dominant industrial bottom trawl or purse seine fleets.
Different gears often target different sizes of a species.For example, trawl and longline fleets can compete for the same species but over different size ranges.In Mediterranean fisheries, a classic example is European hake: juvenile hake (10-30 cm TL) are mainly caught by trawlers, while large juveniles and adults (larger than 30 cm TL) are caught by longliners or gillnetters, raising the problem of technical interaction within the same species (Lleonart et al. 2003).In cases in which the commercial catch reports are disaggregated according to size (the commercial categories "small hake" and "large hake"), the different sizes could be taken as different species for competition analysis using the index developed here.However, our index is not dynamic and does not take into account delayed competition (i.e. the effect of present catches of juveniles on future abundance of spawning adults).In another possible expansion, the use of these indices can be extended to assess the competition (dominance/independence continuum) of fleets from different countries or world regions or to assess the relationship between different seafood producing sectors (fisheries vs aquaculture) from readily available data such as the FAO fish production data sets.

Fig. 1 .
Fig. 1. -Location of the study area.The lower panels show: left, the evolution of catches of the five fishing fleets operating in Catalonia according to the official typologies of license: OTB (otter bottom trawl), PS (purse seine), LLS (set longline), LLD (drifting longline) and PG (passive gear for vessels of 0-12 m length); right, the evolution of the catches of the 11 métiers practiced in the Vilanova i la Geltrú fishery.

Fig. 2 .
Fig. 2. -Sociogram of five fleets in Catalonia.The white circle is proportional to T (dominance) the grey one to the main diagonal (independence).

Fig
Fig. 3. -Independence vs dominance plot of five fleets in Catalonia, showing the annual trends.The first year of the series (2000) is indicated by a solid dot.
the average and standard deviation of the elements of the annual [S].The sociogram of [S] is presented in Figure 4. Regarding independence, four fleets show values above 0.5: bottom trawl, purse seine, beam trawl and dredge.The other seven are quite dependent, below 0.4.Bottom trawl is by far the most dominant fleet with T values around 4, purse seine has T values around 2 and all the other fleets have T values below 1 (Fig.

Fig. 4 .
Fig. 4. -Sociogram of 11 fleets in Vilanova i la Geltrú.The white circle is proportional to T (dominance) the grey one to the main diagonal (independence).The arrows corresponding to values smaller than 5% of the interval of [S] have been eliminated for clarity.Of the 55 total pair relationships only 25 have been retained.

Fig. 5 .
Fig. 5. -Independence vs dominance plot of 11F Catalonia, showing the annual trends.The first year of the series (2000) is indicated by a solid dot.In A the whole plot to note the positions of bottom trawl, purse seine, beam trawl and dredge is presented.The other seven fleets are presented in plots B and C to discriminate them.

tAble 1 .
-Five-fleet results forCatalonia.2000Catalonia.-2009average [S]   average [S]and standard deviation matrices.For each element S AB , A correspond to the rows and B to the columns (for instance S OTB, PG =0.40).The main diagonal of the matrix corresponds to the level of "independence" of each fleet.T is the measure of dominance.OTB, otter bottom trawl; PG, passive gear; PS; purse seine; LLS, stet longline; LLD, drifting longline.
in the main diagonal element for surface longline.A sociogram of the average [S] is presented in Figure2.

tAble 2
. -Eleven-fleet results for Vilanova i la Geltrú.2000-2009 average [S] and standard deviation matrices.For each element S AB , A correspond to the rows and B to the columns (for instance S bottom trawl, cuttlefish =0.43).The main diagonal of the matrix corresponds to the level of "independence" of each fleet.T is the measure of dominance.