Slack regulation compliance in the Mediterranean fisheries: a paradigm from the Greek Aegean Sea demersal trawl fishery, modelling discard ogives

Abstract

Annual discard ogives were estimated using generalised additive models (GAMs) for seven demersal fish species (or taxa) and deep-water rose shrimp, Parapenaeus longirostris (Lucas). Analysis was based on data collected on board commercial bottom trawlers in the central Aegean Sea from 1995 to 2008. Length of specimens and fishing depth (along with year) were the variables that had the most profound effect on the proportion of fish discarded. Compliance with the established minimum catch size of marine organisms (MS) was very low, a fact attributed to the low selectivity of currently used mesh sizes, the market demand for undersized fish, as well as the low control and enforcement effort.

Introduction

In 2003, the European Commission (EC) proposed a Compliance Work Plan for the Common Fisheries Policy (CFP), intended to ensure a more effective, uniform and equitable application of the CFP (EC 2003). Accordingly, Member States are responsible for control and enforcement within their own waters. The basic regulation states, ‘unless otherwise provided for in Community law, Member States shall ensure effective control, inspection and enforcement of the rules of the Common Fisheries Policy’ (Council of the European Union 2002b, Art. 23(1)). The basic regulation also outlines guidelines as to how the inspection should take place and what elements it should contain. The CWP was accompanied by a scoreboard, which ‘provided an indication of the level of compliance of different regulatory provisions by the Member States’ control and enforcement activities and their level of compliance with the rules of the CFP’. The scoreboard contributed to transparency regarding compliance within the EC. The editions published up to 2006 made clear that control and enforcement has been an Achilles heel of the CFP from the beginning (EC 2006a,b; Lutchman et al. 2009).

The sizeable CFP agenda embraces among others the discard problem. Discards refer to that part of the catch not retained on board during commercial fishing operations and are returned to the sea. Discards can include non-commercial species, non-marketable commercial material and marketable organisms. A recent estimate of annual global discarding was 7.3 million tonnes (Kelleher 2005). Reduction or elimination of discards is a primary target of the CFP and is one of the core aspects of an ecosystem-based management of fisheries (Hilborn 2011). Designing a management regime that limits or prevents discarding while meeting multiple social, economic and biological objectives and enforcing regulations designed to prevent or minimise discards is a challenging task. The 2012 CFP reform aspires to provide a new legal framework, removing all compulsory discarding and establishing objectives and time limits for a gradual phasing out of unwanted catches (EC 2011a,b). This simplistic approach sums up this very complex problem in one sentence: ‘the future CFP should ensure that discarding no longer takes place’.

Among worldwide fisheries, Mediterranean fisheries stand out as notoriously difficult to monitor, assess and manage (Swan & Gréboval 2003). Fisheries in the region are considered to be in crisis with many fish stocks characterised by declining landings in the last few decades. Specifically, during the last 10 years, catches from demersal resources have decreased by 20%. According to recent assessments, the most important demersal species – Merluccius merluccius (L.), Mullus barbatus (L.) and some valuable crustaceans – appear to be overexploited (CIESM 2000; Lleonart 2008; GFCM 2010a,b). This crisis is attributed to excess fishing effort and failure of the present management schemes. The Mediterranean fisheries are generally managed through effort control rules and technical measures, such as closed seasons, closed areas, limited issue of new licences, minimum catch sizes (MS), mesh size regulations and maximum size of fishing gears (TAC – Total Allowable Catches apply only to bluefin tuna, Thunnus thynnus L.). However, such restrictions differ between countries or even between regions and/or fisheries of the same country. Furthermore, Mediterranean fisheries are strongly multispecies in nature, practising numerous metiers. Metiers can be considered as a group of fishing operations targeting similar (assemblage of) species, using similar gear, during the same period of the year and/or within the same area and that are characterised by a similar exploitation pattern. The situation gets more complicated because currently it is recognised that the Mediterranean fisheries management is practised through technical, anachronistic and often non-scientific measures that are far from being precautionary (Stergiou et al. 2004). Discard practices in the Mediterranean vary significantly among fishing techniques, with the higher rates being produced by the bottom trawl fisheries (average 45–50% – Kelleher 2005; STECF 2006), although variability in discard quantities and rates originates also from factors such as market prices, legal measures, (e.g. minimum catch sizes – MS), technical characteristics, low gear selectivity, environmental conditions, species composition and size distribution of the community (Rochet & Trenkel 2005).

Discarding can be expressed as the probability of an individual being discarded at a given length. For commercial species with a set minimum catch size, this probability generally decreases with increasing length and is zero after a certain length, because the specimen will eventually attain a commercial size and will be landed. Such data will have an S-shaped curve and have previously been described as logistic function ogives. This study estimates discard ogives for a set of commercial marine species (or group of species), using generalised additive models (GAMs). In the process, the factors that might influence the probability of a species being discarded were determined. The aim was also to unravel the low level of compliance with the national and EU-CFP regulations for catch of undersized fish using data from the Greek bottom trawl fleet in the central Aegean Sea.

Methods

A total of 485 records of bottom trawl hauls, carried out between 50 and 365 m depth, were collected on board commercial vessels three times a year (winter, spring and autumn; the fishery is regulated by a summer closure) during 1995–2000 and 2003–2008 in the central Aegean Sea, a major trawl fishing ground (Table 1; Fig. 1). According to the official Hellenic fleet registry (IMAS-Fish 2007), the number of otter bottom trawlers in the central Aegean Sea during 1995–2008 fluctuated between 53 and 70. Ten vessels were selected as the most representative vessels in the active fleet and monitored during this study. These vessels comprise the same ten vessels on which the EU DCF was carried out (Data Collection Framework – former DCR, applying the EU CFP), providing data to the annual national reports submitted to the EC (DG MARE), as well as all data calls in the framework of SGMED/STECF1 and FAO (1996)/GFCM2 workgroups. Operational characteristics of the vessels monitored were similar (mean length 28 m, min. 23 m, max. 33 m; mean gross tonnage 125, min. 97, max. 141; mean horse power 322, min. 293, max. 367). The gear deployed was a typical otter bottom trawl used in Greek waters (Adamidou 2007). The entire length of the trawl net was 58 m (average), and the mouth circumference (stretched circumference at the fishing line) was 61 m (average). The vertical opening of the gear was 1.3 m and the horizontal opening (at spreaders) 14 m (min 12, max 16). The last part of the net, the cod-end, was about 7 m long; its overall width was mostly 300 meshes, which is the same lengthwise, and the netting stretched mesh size was 28–32 mm until 2001 and 40 mm afterwards. The mesh size modification was dictated by a European Union regulation (EC 2000), enforced officially on 1 October 2001, in the Greek bottom trawl fishery. The towing speed was approximately three nautical miles per hour (min. 2.4, max. 3.6), and the average tow duration was 148 min (min. 45 min, max. 330 min.).

Figure 1. Map of the studied area with the sampling locations indicated as solid black triangles.

Table 1. Sampled hauls by depth stratum and year

Depth (m)	1995	1996	1997	1998	1999	2000	2003	2004	2005	2006	2008	Total
0–100	30	21	33	21	25	11	2	19	15	10		187
100–200	12	19	4	5	11	2	10	30	44	43	18	198
200–300	6	7	15	11	9	4		19	9	9	6	95
300–400				1				1	1	2		5
Total	48	47	52	38	45	17	12	69	69	64	24	485

Regarding minimum catch size of marine organisms (MS), as dictated by Article 15 in the most recent regulation (EC 1967/2006), the legislator demands that undersized marine organisms shall not be caught and that discards of species managed through minimum size are no longer allowed (‘…a marine organism which is smaller than the minimum size specified in Annex III (hereinafter “undersized marine organisms”) shall not be caught, retained on board, transhipped, landed, transferred, stored, sold, displayed or offered for sale..’). Article 3 of the same regulation, concerning the even more sensitive case of non-commercial protected species, states, ‘…the deliberate catching, retention on board, transhipment or landing of marine species referred to in Annex IV to Directive 92/43/EEC shall be prohibited except when a derogation has been granted according to Article 16 of Directive 92/43/EEC…’. In the latter case, the legislator defines ‘deliberate’ catching, which differentiates a catch from being ‘incidental’ or ‘by-catch’ or ‘unintentional’. According to Article 3, a fisher is considered a violator only if he targets a protected species, and he is not prosecuted if he catches it unintentionally and discards it. On the other hand, according to Article 15, a fisher is a violator if he catches (and discards) an undersized commercial species, even unintentionally. Ideally, a very selective gear would not allow for undersized fish to be caught, and the regulation includes provisions for such mitigating measures (e.g. gear configuration, mesh sizes, protected habitats). However, these measures do not ensure the absence of undersized specimen in the catch. If Article 15 of the regulation is interpreted literally, EU Mediterranean fisheries would come to a standstill.

Regardless, the Greek enforcement authorities (as well as most, if not all, Mediterranean EU member states authorities) do not interpret the regulation literally and do not apply sanctions for catching and discarding undersized fish, but only for landing or marketing them. This is confirmed by the annual statistical bulletin of inspections in the Hellenic fisheries sector (2009 – available at: http://www.yen.gr/wide/yen.chtm?prnbr=35640, in Greek), which showed all 2995 inspections for undersized specimens were undertaken either at the auction market or at the vessels’ landing locations. No on-board controls included minimum size inspection, because officers will not take legal actions against a fisher for just catching undersized fish (but they will do so if he attempts to market, land or tranship them).

Data collection

Data were collected on board by scientific personnel who did not interfere with the normal fishing practices of the crew. After the sorting operation by the crew, the catches were separated into marketable and discard groups. The discards included species without commercial value, species of low commercial value and undersized individuals of commercial species. The total weight of each of the two fractions (marketable and discard) and for each species was determined for the total catch. Additionally, observers recorded fishing operational data such as date, position, depth and haul duration. When a certain species was represented by many specimens, only a representative sample was measured. For each species, the number and the weight of the sample, as well as the size of each animal (total length for fish and carapace length for crustaceans, mm) were recorded. Data were processed to estimate the length frequency distribution for the marketable and the discarded fraction of each commercial species: (1) the marketable and discard samples standardised in each haul were weighed for the total marketable and discarded yield of the catch, respectively; (2) the weighed samples were transformed to numbers of specimens caught per one hour of haul (# h⁻¹); (3) the length frequency distribution was estimated for each species at total length class intervals of 1 mm (carapace length for crustaceans) (see Machias et al. 2004; for sampling details and raising procedures). Annual discard ogives were based on the length frequencies of discards (Fig. 2).

Figure 2. Annual catch length frequencies used in the analyses, for the selected species of commercial interest in the Greek Aegean demersal trawl fishery. Lines represent minimum and maximum lengths sampled, boxes the 25th and 75th percentiles and horizontal lines inside the boxes the median length.

The analyses were carried out for seven species or taxa: bogue, Boops boops (L.); anglerfish, Lophius spp. (including L. budegassa Spinola and L. piscatorius L.); European hake, Merluccius merluccius (L.); red and striped red mullet, Mullus spp. (including M. barbatus L. and M. surmuletus L.); common pandora, Pagellus erythrinus (L.); horse and jack mackerels, Trachurus spp. (including T. picturatus Bowdich; T. mediterraneus Steindachner; T. trachurus L.); and deep-water rose shrimp, Parapenaeus longirostris (Lucas). These species were chosen because of their commercial importance, because they are largely caught by the demersal bottom trawl fleet (more than 30% of total landings in the region is attributed to these species – NSSG 2006) and because they have different fisheries management strategies (e.g. modification of MS – Table 2).

Table 2. Minimum catch sizes (MS in mm)1 for selected species of commercial interest in the Greek Aegean demersal fishery

Statistical analyses – modelling

Assuming that nonlinearities are most likely to occur in the functional relationships between discarding probabilities and explanatory variables, generalised additive models (GAMs – Hastie & Tibshirani 1990) were applied to identify the probability of a fish being discarded at a specific length and to reveal the form of the relationships. In a GAM, the expected values Y_i of the response variable are related to the predictor variables Z_mi according to the following general formulation:

• 

where the response variable is allowed to follow any distribution from the exponential family, f is the link function, LP is the linear predictor, c is the intercept, s_m() is the one-dimensional smooth function of covariate Z_m and Z_mi is the value of covariate m for the i-th observation (Wood 2006). The smooth function s_m() is represented using penalised regression splines, estimated by penalised iterative least squares (Wood 2006). The optimum degree of smoothing was defined by the generalised cross-validation (GCV) criterion using the mgcv package (Wood 2006) in R, version 2.12.0 (R Development Core Team 2010).

In this study, the response variable was the binary variable Discard, assigned a value of 0 if no fish was discarded at a specific length and 1 otherwise (Bernoulli-type 0/1 measurements). The best underlying probability distribution was binomial, because of the discrete nature of the dependent variable. The explanatory variables considered in the analysis were (1) Length of fish, (2) Fishing Depth, (3) Year, (4) Month, (5) Mesh size and (6) minimum catch size (MS) of the given species or taxon.

A series of 15 candidate models (plus the null model: a model with no predictor variable) was constructed including combinations of the six parameters under investigation that plausibly influenced species discarding practices (Table 3). The best model was selected based on the lowest GCV criterion value (Wood 2006). Length was included in all candidate models as it is usually the driving factor of discarding. Additionally, Year was present in all models, because one of the aims was to investigate the effectiveness of management schemes by comparing annual discarding trends and compliance to introduced MS.

Table 3. The set of candidate models of discard probability for seven commercial species in the central Aegean Sea bottom trawl fishery

Model	Linear predictor (LP)
g0	Null model
g1	s(Length) + Year
g2	s(Length) + Year + Month
g3	s(Length) + Year + S(Depth))
g4	s(Length) + Year + MS
g5	s(Length) + Year + Mesh size
g6	s(Length) + Year + Month + S(Depth)
g7	s(Length) + Year + Month + MS
g8	s(Length) + Year + Month + Mesh size
g9	s(Length) + Year + S(Depth) + MS
g10	s(Length) + Year + S(Depth) + Mesh size
g11	s(Length) + Year + MS + Mesh size
g12	s(Length) + Year + Month + S(Depth) + MS
g13	s(Length) + Year + Month + S(Depth) + Mesh size
g14	s(Length) + Year + Month + MS + Mesh size
g15	s(Length) + Year + S(Depth) + MS + Mesh size
g16	s(Length) + Year + Month + S(Depth) + MS + Mesh size

Results

Annual discard proportions for the total of species caught by the Greek bottom trawl fishing vessels operating in the central Aegean Sea fluctuated between 31 and 64% by weight and 40 and 87% by number (Table 4). Highest discard rates were observed for Trachurus spp. By contrast, Mullus spp. for the whole period and P. longirostris for most of the period were almost exclusively retained on-board for marketing.

Table 4. Annual summary table for the selected species of commercial interest in the Greek Aegean demersal trawl fishery: proportion of catch with specimens below MS (in numbers); proportion of catch discarded (in numbers and in weight)

Species	1995	1996	1997	1998	1999	2000	2003	2004	2005	2006	2008
n.d., no data; –, no MS.
Boops boops (L.)
Proportion < MS (in No)	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.1	0.0	0.0
Proportion discarded (in No)	55.1	3.5	21.9	48.7	82.4	64.8	98.7	67.2	67.8	59.6	39.3
Proportion discarded (in W)	42.6	1.5	11.3	32.3	73.2	46.3	99.6	59.1	57.0	51.1	31.7
Lophius spp.
Proportion < MS (in No)	98.4	82.2	100.0	100.0	87.5	100.0	44.0	78.6	70.0	–	–
Proportion discarded (in No)	93.4	85.6	46.3	56.8	49.2	48.4	56.1	80.3	69.4	67.5	30.7
Proportion discarded (in W)	38.4	23.2	7.2	3.9	4.9	1.3	5.9	19.9	24.1	13.0	7.2
Merluccius merluccius (L.)
Proportion < MS (in No)	36.8	48.5	36.9	27.3	31.2	36.8	42.9	55.1	61.3	57.0	76.4
Proportion discarded (in No)	9.0	78.8	76.7	26.3	26.7	3.4	22.7	37.0	37.1	57.2	50.9
Proportion discarded (in W)	2.5	28.9	23.6	4.1	4.4	0.4	2.4	10.6	11.1	19.4	16.4
Mullus spp.
Proportion < MS (in No)	7.8	0.5	2.1	0.6	5.1	0.1	0.2	0.2	0.3	0.3	0.7
Proportion discarded (in No)	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.5	2.0	1.2	2.3
Proportion discarded (in W)	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.4	1.1	0.6	0.6
Pagellus erythrinus (L.)
Proportion < MS (in No)	1.5	0.8	1.6	0.0	0.0	0.0	9.6	0.0	13.8	4.2	11.1
Proportion discarded (in No)	0.0	0.0	7.7	0.0	0.0	0.0	66.5	2.3	45.6	0.0	0.0
Proportion discarded (in W)	0.0	0.0	0.7	0.0	0.0	0.0	40.5	1.3	8.1	0.0	0.0
Parapenaeus longirostris (Lucas)
Proportion < MS (in No)	–	–	–	–	–	–	–	–	–	–	51.9
Proportion discarded (in No)	n.d.	0.0	0.0	0.0	0.0	0.0	1.2	37.2	24.0	12.7	42.1
Proportion discarded (in W)	0.0	0.0	0.0	0.0	0.0	0.0	0.9	18.1	12.0	6.7	24.6
Trachurus spp.
Proportion < MS (in No)	8.6	6.5	5.9	3.8	10.6	1.2	16.6	13.0	21.5	34.4	30.2
Proportion discarded (in No)	93.0	98.7	95.8	53.9	54.6	92.90	99.9	91.5	97.5	93.8	63.0
Proportion discarded (in W)	54.7	93.4	88.1	39.0	29.9	78.3	99.4	79.8	87.5	78.1	28.8
All other species
Proportion discarded (in No)	61.0	97.9	95.8	94.7	92.0	77.7	76.7	94.4	89.7	87.6	83.0
Proportion discarded (in W)	55.6	64.8	73.9	69.7	69.3	54.9	55.5	69.0	56.9	62.1	39.7
Total
Proportion discarded (in No)	67.6	87.1	86.7	81.8	78.5	61.0	47.2	62.9	52.1	40.4	64.0
Proportion discarded (in W)	46.6	57.1	64.3	52.2	51.8	41.4	42.9	41.0	38.7	42.3	31.0

Compliance with the set MS was low for Lophius spp. and M. merluccius, as can be deduced by the discrepancy between the proportions of undersized fish caught and discarded (Table 4). Conversely, no deviations appeared in the MS for B. boops. The introduction of a MS for P. longirostris after 2006 was partially respected in the period following (few specimens < MS were directed to the market). The persistent very high levels of undersized catches for Lophius spp. (44–100% of specimens caught < MS) were resolved by the removal of the MS in 2006.

GAM analysis

Model g13, that is, logit(Discard probability) = c + s(Length) + Year + Month + s(Depth) + Mesh size, was the most frequently included in the set of substantially supported models (based on GCV criterion), in three of seven cases (Table 5). The deviance explained by the best models varied between 36% for B. boops and 82% for P. erythrinus.

Table 5. Summarised results for the ‘best’ model selected for the seven commercial species (or group of species), caught in the bottom trawl fishery of the central Aegean Sea. Pr(F) refers to the P-values from an anova F-ratio test. Edf are the estimated degrees of freedom

Species	‘Best’ model based on generalised cross-validation criterion	Edf	Pr(F)						Percentage of Deviance explained
			Length	Year	Depth	Month	Mesh size	MS
Boops boops (L.)	g10	14.9	<0.001	0.021	<0.001		0.170		36.0
Lophius spp.	g13	20.8	<0.001	<0.001	<0.001	<0.001	<0.001		67.5
Merluccius merluccius (L.)	g13	24.5	<0.001	<0.001	<0.001	<0.001	0.076		72.5
Mullus spp.	g13	14.4	<0.001	0.011	<0.001	0.033	0.062		47.4
Pagellus erythrinus (L.)	g9	14.3	<0.001	<0.001	<0.001			<0.001	82.8
Parapenaeus longirostris (Lucas)	g6	19.9	<0.001	<0.001	<0.001	<0.001			52.3
Trachurus spp.	g16	19.8	<0.001	0.012	<0.001	0.022	<0.001	0.017	73.9

Generalised additive models demonstrated that Length and Depth (along with Year) had the most profound effect on the proportion of fish discarded annually in all groups of species. The effect of Length was a monotonical decrease on the discard likelihood, whereas Depth had a nonlinear diverse effect among species (Fig. 3; Table 5). Different discard behaviours were exhibited annually for the different species, and therefore, dissimilar annual discard ogives were obtained (Fig. 4).

Figure 3. Generalised additive models (GAM)-derived effects of Length and Depth on the discard probability by species. Dashed lines indicate two standard errors above and below the estimates. Relative density of data points is shown by the ‘rug’ on the x-axis.

Figure 4. Predicted discard ogives based on the generalised additive models of discard probability vs predictor variables (length, year, month, depth, mesh size, MS) for selected species of commercial interest. MS is indicated by a vertical line (for some cases, dotted lines indicate past MS and solid lines current MS).

Although discarding probability by length varied among years for B. boops, it was consistently near or above the MS for the whole period studied (Fig. 4, top left). The opposite was found in the annual change in probability of discarding by Length for Lophius spp.; a sharp drop followed the removal of the MS in 2006. For this last period of the study, discarding length decreased significantly, and this is noticeably demonstrated in the discrepancy of ogives; the last period estimates shifted to the left (smaller lengths) by almost 100 mm (Fig. 4, top middle). Merluccius merluccius was one of the species exhibiting very low compliance to the MS. Here, MS did not determine discarding length, because throughout the study period, the likelihood of discard was higher than 50% only at lengths (100 mm) significantly lower than the MS (200 mm) (Fig. 4 – top right). In a similar way, Mullus spp. marketing size was not dictated by regulation limitations (Fig. 4 – mid-left). There was no apparent change in discarding length for P. erythrinus despite the introduction of a higher MS (150 mm in 2006, compared with the previous 120 mm) (Fig. 4 – mid-middle). Parapenaeus longirostris (Lucas) was a species for which no MS was set for most of the study period. During the 1 year (2008) that a 20-mm-carapace-length limitation was set, compliance was low (Fig. 4 – mid right). Discarding lengths were higher than the MS for Trachurus spp., although discarding probability by Length varied considerably among years (Fig. 4 – bottom).

To avoid confusion, Table 4 results must not be confused with Figure 4 estimates. The former refer to actual lengths sampled, while the latter to the discard probability of the full-length spectrum; for example, very few Mullus spp. below 100 mm were observed, and as a result, the proportion of fish lower than the MS was low (Table 4). However, very few of these undersized specimens were discarded (Table 4), and as a consequence, discard ogives curves were positioned to the extreme left, well below the set MS (Fig. 4).

Discussion

Discarded proportions, as well as length at discard, are key parameters in most stock assessment methodologies, and failure to estimate them adequately represents a major deficiency in the assessment, invalidating outcomes and leading to ineffective management measures (ICES 2004; Dickey-Collas et al. 2007). Herein, using data from the central Aegean Sea bottom trawl fishery, the magnitude of regulation non-compliance in a Mediterranean fishery was estimated based on catch of undersized marine organisms and discard practices.

Unlike many other marine areas, very few species are assessed within the eastern Mediterranean region. Although GFCM and SGMED/STECF have produced a series of assessments on various Mediterranean demersal species, none has been undertaken in the Aegean Sea, although some basic stock parameters (F, MSY) have been provided for M. merluccius, M. surmuletus and M. barbatus (SGMED/STECF 2010; Maravelias et al. 2011). This shortage of long-term scientific advice regarding the status of the stocks has resulted in a situation where periodical evaluation of the effectiveness of management measures adopted in the past (MS being one of them) cannot be performed. The observed disrespectful attitude of fishers towards regulation limitations emanates from the low recognition of scientific advice accompanied by loose enforcement by the control authorities. It is documented that compliance in European fisheries is worst when fishers have no faith in the science underpinning management decisions and particularly when catches are plummeting (Young 1998; Wilson et al. 2003).

Regulation non-compliance is a considerable issue striking at the foundations of fisheries management. This is caused by damaging marine resources, giving merit to illegal practices, belittling the image of control authorities, undermining prices paid for legitimate fish, weakening consumer confidence and causing reputational damage to the industry. Most of all, it penalises fishers who play by the rules by giving an unfair advantage to those who ignore rules. A report on illegal, unreported and unregulated (IUU) fishing in EU fisheries estimated a total cost to EU Member States of 825 million € per year, about 15% of total fishery value (EFTEC 2008). Monitoring, control and surveillance (MCS) are key components in the fishery management (Arnason 1999). Wallis and Flaaten (2000) calculated MCS costs as a fraction (on average 6%) of revenues in the EU fisheries. Germany, Sweden and Ireland displayed values above 20%, while EU Mediterranean countries exhibited very low values, apparently investing little in regulation enforcement (France 10%, Greece 9%, Italy 4%, Spain 1%). As an example, the Greek bottom trawl fleet comprises of 326 vessels each exerting an effort of approximately 200 days at sea per annum (Katsanevakis et al. 2010), with a total annual fishing effort of about 65 000 days. In 2009, the Greek authorities have made 2995 inspections regarding minimum catch size of marine organisms. It must be underlined that this number considered all fishing gears and vessels (approximately 18 000 boats – 4 000 000 days at sea) and not just trawlers; thus, the likelihood of facing an inspection is extremely low, and the very low compliance levels are not a surprise to the scientific community.

Compliance with the MS was low for highly commercial species, particularly Lophius spp., M. merluccius and Mullus spp., possibly due to the strong market demand even for individuals smaller than the respective MS, while there were no deviations from the MS in B. boops. The latter species, as well as Trachurus spp. presenting the highest discard rates owing to high grading, are low-value species, habitually discarded to make space for more profitable fish. However, in cases when catches of target species are below expectations, they constitute a supplementary source for the fishers, who usually tend to retain the relatively large individuals. The latter practice is quite common in multispecies fisheries in Greece (Machias et al. 2001; Tzanatos et al. 2007; Tsagarakis et al. 2008). On the other hand, it should be pointed out that low compliance with MS regulations produces less discards in the already high-discarding portions generated by the Aegean trawl fisheries. Rochet et al. (2002) and Borges et al. (2006), studying trawler discarding patterns in the Irish and Celtic Seas, respectively, found that marketing at lengths below MS took place and that introduction of a lower MS has not triggered the anticipated change for a series of fleets. They suggested that a reduction in MS will not be effective in reducing discarding and indicated that the management measures adopted in the Irish Sea and West of Scotland have not been effective in protecting juvenile haddock, Melanogrammus aeglefinus (L.), and whiting, Merlangius merlangus (L.).

Selected GAMs may serve as an indication of discard preference or association for the selected variables under study. Species Length and fishing Depth had the most profound effect on the proportion of species discarded. Information on landed catch and fishing depth per trip, available through fishers’ logbooks, could become useful in estimating discarded quantities during certain time periods and in specific areas. This could be achieved based on indirect identifications of the potential discarding hotspots, delivered through GAM analyses.

In the past, GAMs were applied to study the abundance of certain species such as M. barbatus, M. merluccius and Lophius spp. in Greek waters in relation to bathymetric and ocean environmental conditions, and strong correlations were found (Maravelias & Papaconstantinou 2003; Maravelias et al. 2007; Katsanevakis et al. 2009). The latter suggests that certain areas with high proportions of juveniles of specific species during certain periods could be designated and proposed for closing. Such targeted spatiotemporal closures might be an effective managerial tool to protect juveniles in Greek waters, taking also into account the peculiarity of the Greek fisheries organisational structure with numerous landing sites, which are difficult to be supervised. It is widely accepted that spatiotemporal closures in predictable hotspots of unwanted catches may effectively reduce unwanted catches (Hall et al. 2000; Gilman et al. 2006; Poos et al. 2010; Dunn et al. 2011; Gilman 2011). Spatiotemporal closures/refugia are one of the main types of marine protected areas (MPAs), which, according to Stergiou et al. (1997, 2009), might be a more effective fishery management tool in relation to static, non-adaptive technical measures, such as the MS.

The entrenched prevailing governance approach, to reduce by-catches and discards, remains the selectivity of fishing gears. Catchpole (2009) reviewing methods and options for fishers and managers advocated that improvement in trawl selectivity could be an ideal solution. Concurrently, and to face up to the alarming situation of several Mediterranean fish stocks, several scientists and regional projects have underlined the need to bring about a strong effort in the improvement in the Mediterranean trawl selectivity (GFCM 2007). Studies carried out in the Mediterranean Sea showed the potential benefits of different selectivity systems (Sacchi 2008). Mediterranean demersal trawl fisheries traditionally operate using small diamond-shaped meshes in the cod-end (Caddy 1990), which tend to retain almost all animals (Stergiou et al. 1997; Stewart 2002). To reduce mortality rates for juveniles and discards of dying marine organisms by fishing vessels, Council Regulation (EC) No. 1967/2006, concerning management measures for the sustainable exploitation of fishery resources in the Mediterranean, contemplated that it would be appropriate to proceed with increasing mesh sizes of trawl nets and making the use of square-meshed netting mandatory. Sala et al. (2008) and Bahamon et al. (2006) demonstrated that substantial improvements in size selectivity of commercially important species are achieved by switching from the conventional diamond mesh cod-end to square mesh cod-end. More specifically, studies carried out in the central Aegean Sea involving a species present in this study (M. merluccius) found that M. merluccius catch length (L₅₀: length at which the probability of a fish to be caught is 50%) increased by 15% when a diamond mesh cod-end was replaced by a square mesh cod-end in an experimental bottom trawl survey (Petrakis & Stergiou 1997). However, by concentrating fishing mortality on a narrow subset of an ecosystem's components through gear selectivity might alter ecosystem structure and function, impacting ecosystem services (Rochet & Trenkel 2009; Zhou et al. 2010).

To summarise, landing of undersized marine organisms and discarding practices in Greek waters were dictated by market demands, while regulation compliance was weak. MS is not respected by the fishers because of incentives to violate the law (market demand for undersized fish) and the low probability of imposing sanctions (low monitoring, control and surveillance effort by the authorities). Solving the discard problem is complex, particularly in the essentially multispecies demersal fisheries of the Mediterranean Sea. Policies aiming to reduce catching of undersized organisms as well as discards should be a result of careful analysis (Bellido et al. 2011), without disregarding cultural characteristics (Johnsen & Eliasen 2011), and consider that an integrated approach that will remove or at least reduce incentives to catch undersized specimens and discard them might be more effective (Gezelius 2008).

Acknowledgments

We wish to thank all fishers (skippers and crew) who assisted observers in data collection on-board the fishing boats.

1. 1

   Scientific, Technical and Economic Committee for Fisheries, Subgroup on the Mediterranean (http://stecf.jrc.ec.europa.eu/)

2. 2

   Food and Agriculture Organization, General Fisheries Commission for the Mediterranean (http://www.gfcm.org/gfcm/en)