Reduction of by-catch and discards in the Algarve small-scale coastal fishery using a monofilament trammel net rigged with a guarding net

1 Faculdade de Ciências e Tecnologia (FCT), Universidade do Algarve, 8005-139 Faro, Portugal. (MJS) (Corresponding author) E-mail: mszyn001@fiu.edu. ORCID iD: https://orcid.org/0000-0002-8243-7307 2 Centre of Marine Sciences (CCMAR), Universidade do Algarve, 8005-139 Faro, Portugal. (LB) E-mail: lbentes@ualg.pt. ORCID iD: https://orcid.org/0000-0001-6884-2886 (PM) E-mail: pmontei@ualg.pt. ORCID iD: https://orcid.org/0000-0003-1446-8256 (MR) E-mail: mrangel@ualg.pt. ORCID iD: https://orcid.org/0000-0001-5903-5687 (KE) E-mail: kerzini@ualg.pt. ORCID iD: https://orcid.org/0000-0002-1411-0126


INTRODUCTION
By-catch consists of marine species that are caught unintentionally alongside targeted species that live in the same environment, due to a lack of selectivity of fishing gears.By-catch may consist of undersized or juvenile target species as well as non-target commercial species and a wide variety of species of no value that are subsequently discarded.Commercial species may also be discarded for various reasons, including quota limitations, being undersize, or being unfit for sale because of damage or parasites (Hall et al. 2000).Although discarding in fisheries is considered a problem that should be reduced, the precise definition of by-catch and its significance in terms of the discarding and impact on ecosystems and populations is yet to be agreed upon (Borges et al. 2001).However, there is growing interest in by-catch and mitigation of by-catch by the fishing industry due to legislation concerning discarding (CEC 2007), increased public awareness and concern regarding impacts of overfishing (Gelcich et al. 2014), and more practical considerations such as the time expended in disentangling by-catch species from nets and the damage they cause to gear (Metin et al. 2009).
Trammel nets have discard rates ranging from 0% to 66% worldwide (Kelleher 2005) and a discard rate of about 13% in the Algarve region in Portugal (Borges et al. 2001).The monofilament trammel nets used in Portugal have low size selectivity compared with gillnets, catching wide size ranges of most species (Erzini et al. 2006).Currently, legislation such as that of the European Union Common Fishery Policy (CEC 2007) is seeking to encourage the development of technologies that can result in elimination or reduction of discards and unwanted by-catch in general.
There have been several successful approaches to the management of harvesting commercial catch while decreasing by-catch and discards.Gökçe et al. (2016), using a trammel net rigged with a guarding net in the northeastern Mediterranean, reported that the modified net exhibited 83% less by-catch and 16% less commercial catch than the standard commercial net.A 1.5% decrease in the catch rate of the main target species, green tiger prawn, Penaeus semisulcatus and a 66% to 85% decrease in the three main by-catch species were also reported.In Antalya Bay (eastern Mediterranean) Olguner and Deval (2013) found that a smaller mesh size of the inner panel (40 or 44 mm) provided higher amounts of commercial catch in both abundance and biomass for Pagellus acarne and a decrease in bycatch for Citharus linguatula.
In Izmir Bay (Aegean coast of Turkey), a guarding net was added to the common trammel net used for the commercial prawn Melicertus kerathurus (Metin et al. 2009).The study reported a 0.99% decrease in the prawn catch and a reduction in catch of three main by-catch taxa from 17% to 51%.In another study performed in Izmir Bay on Mullus spp.fishery (Aydin et al. 2013), discard rates decreased to about 55% to 63%, and the guarding net reduced the catches of the three main by-catch taxa (Hexaplex trunculus, Bolinus brandaris, Maja spp.).
The specific objectives of the present study were to compare a standard trammel net with a modified trammel net rigged with a guarding net, in terms of catch composition (commercial, by-catch and discard), economic yield, time needed to "clean" the nets (removal of by-catch and discard species), and net damage.

Net design
Two types of net were rigged: a standard trammel net (T) and a modified trammel net (M), a standard trammel net with a guarding net, consisting of a single layer of netting three meshes high, placed between the trammel net and the footrope (Fig. 1).The experimental nets consisted of fifteen 45-m sheets of nets per net type, with three standard and three modified nets interchanging five times, giving ten sections (T1...T5 and M1...M5) with two metres in between each section to reduce bias (fish guidance effect), for a total of approximately 1.5 kilometres of net.The small mesh inner panel consisted of 120-mm stretched mesh, 0.30-mm diameter monofilament, 40 meshes high.The large mesh outer panels were constructed of 600-mm stretched mesh, 0.60-mm diameter monofilament.The floatline was 52 m long and was made of 7-mm diameter polyester, while the leadline was 55.2 m long and made of 7-mm diameter polyester (braided line with lead core).The inner panel of the standard net (Fig. 1A polyamide) between the footrope and the trammel net (Fig. 1B).Both the standard net and the modified net had a vertical slack (inner panel stretched mesh height/ outer panel stretched mesh height) of 2.7.The trammel net used is characteristic of the nets used in the multi-species inshore fisheries of the Algarve, southern Portugal (Carneiro et al. 2006).Given the seasonality in abundance of cuttlefish (Sepia officinalis), this is the target species in late autumn-winter, while for the rest of the year trammel nets target a variety of species, especially soles (Soleidae).

Experimental fishing
Experimental fishing took place off the coast of Algarve (southern Portugal), using a fishing vessel belonging to the port of Quarteira (Fig. 2).Twenty experimental fishing trials were conducted from October 2016 to February 2017 at depths ranging from 10 to 30 m.The first ten fishing trials (autumn) targeted soles, while the other trials in late autumn and winter targeted cuttlefish (S. officinalis).The nets were set at dawn and hauled the next morning, usually before sunrise (for approximately 24 hours fishing time).Two GoPro filming cameras were set up on board to record any overlooked individuals in the net, and to evaluate the time required to clean the two different nets.The net was hauled using a hydraulic hauler, and each individual was removed manually by the fishermen.Each individual caught by each net was identified (to species level, if possible).In the case of the modified net, the catches were recorded whether they occurred in the upper trammel net part or in the guarding net.Individual size was taken as total length, carapace length and width.Each individual was registered as 'by-catch' or 'commercial' based on the decisions of the fishermen sorting the catch, and if considered 'commercial discard', the reason was noted (e.g.undersized, scavenged or with parasites).Some by-catch species were immediately killed by the fishermen to facilitate removal from the net and to continue hauling at a moderate pace.This was the case of Trachinus draco and Torpedo torpedo, which had their heads crushed to avoid onboard injuries and some invertebrates such as sea urchins and crabs, which were usually crushed.Special care was taken with species such as Rhizostoma pulmo, the barrel jellyfish, to avoid stings.

Net damage assessment
The damage occurring to the nets was assessed at the end of the experimental fishing trials by counting each visible "hole" or broken piece across the span of the entire 1.5-km net.Four observers checked each section of net for possible damage and recorded the damage according to the two types of net used.Upon observing a hole, a 20-cm ruler was placed within the hole and a picture was taken for later analysis of the diameter.Each hole was classified according to its position (e.g.guarding net, inner small mesh panel or outer large mesh panel; lower or upper half of the net) and size (small if <20 cm, large if >20 cm).

Data analysis
The weight of each fish was estimated using species weight-length relationships parameters.The weight of invertebrates was estimated by the fishermen directly on board.The total catch (kg) of each species was converted to value in euros using the average price per kg obtained from the mandatory first fish auction in Quarteira.The weight of each by-catch species was also recorded to compare the two net types, M and T. Trammel net catches were converted to catch per unit effort in numbers (CPUE n ) and biomass (CPUE kg ) and mean value (€) per 1000 m of trammel net.
Data were analysed using cluster analysis, multidimensional scaling (MDS) implemented in PRIMER 6 (Clarke and Gorley 2006) and the PERMANOVA+ add-on, in order to evaluate differences between the two net types (M and T) as regards catch compositions of discards, commercial catch and the combined catch data.Biomass and abundance data of each experimental trial were analysed according to target species (10 trips targeting sole, 10 targeting cuttlefish), season (15 trips in autumn, 5 in winter), and depth range (12 trips from 10 to 20 m and 8 trips from 20 to 30 m) as factors.
Net type (M and T) was also used as a factor, but only for the combined catch data and for the analyses with the two-way PERMANOVAs, aimed at evaluating interaction effects: net type vs. depth, net type vs. season, and net type vs. target catch.Cluster analysis, MDS, analysis of similarities (ANOSIM), and similarity percentage analysis (SIMPER) were also conducted.Two-way PERMANOVAs were used to evaluate the interaction between net type and the three main factors.Square-root transformation of data was applied when necessary to reduce skewness of data distribution.

Catch composition and discarding
The standard (T) and the modified (M) trammel nets caught a total of 40 commercial species, 19 of them in common with the two nets (Table 1).The CPUE n values for M and T nets were 16.2 and 28.3, respectively, while the corresponding CPUE kg values were 6.5 and 10.0.The values of the landings per 1000 m were €50.3 and €67.2 for the M and the T nets, respectively.The standardized values of discarded commercial species and discarded by-catch species (non-commercial) are shown in Table 2. Discard values were lower for M than for T nets: 38.4 and 68.2 for CPUE n and 71.7 and 124.0 for CPUE kg , respectively.The ratios of M to T net CPUE n and CPUE kg for discards were 0.56 and 0.57, respectively, meaning that the standard trammel net caught more than the modified one.As regards the three most important fish species of the by-catch, more Chelidonichthys obscurus (n=288) and Trachinus draco (n=202) were caught by the T than the M (M:T CPUE n ratio =0.4 and 0.7 respectively), while Scomber colias was caught in greater numbers by the M (M:T CPUE n ratio =1.3).
The M net caught 35% (n), 39% (kg) and 38% (value) of the total landed catch of the two nets.The most lucrative species caught in the M net were Homarus gammarus (€121.7),Microchirus azevia (€87.5), and S. officinalis (€77.3).The highest earnings per species for the T net were obtained for M. azevia (€237.3),Solea 3).The overall value of the catch of the M net was €647.2, while that of the T net was €1043.1 (ratio of 0.6).
The M:T discard ratios were 0.3 for commercial discards and 0.8 for non-commercial by-catch.The species with the highest discards in weight in M nets were Balistes capriscus (1.9 kg) and M. azevia (1.5 kg), while those in T nets were Raja undulata (5.7 kg), S. officinalis (5.5 kg), and M. azevia (3.6 kg).
Regarding by-catch species, 14.5 kg of C. obscurus was discarded from M nets and 33.3 kg from T nets.A total of 7.1 kg of T. draco was discarded from M nets and 11.2 kg from T nets.Finally, 6.5 kg of S. colias was discarded from M nets and 5.9 kg from T nets.
Combined catch (commercial catch, commercial discards and by-catch) composition showed high similarities within trips for both abundance and biomass data (Fig. 3).Two-way PERMANOVAs with net type, season, depth and target species as factors showed no significant interaction between any of the four factors, while net type was not significant and season, depth and target species were significant (Supplementary Material, Table S1).
The analyses based on the square root transformed data showed high significance levels in PERMANOVA and ANOSIM.The MDS plots for the abundance and biomass of discards with season as a factor are given in Figure 4.The results of ANOSIM for commercial species abundance and biomass provided global R val-ues of 0.438 and 0.308 and significance levels of 0.1% and 1.3%, respectively, showing significant seasonal effects on catch composition.The result is furtherly supported by the p-values of 0.002 and 0.016 obtained by PERMANOVA.The results of SIMPER (Tables S2  and S3 in Supplementary Material) showed strong dissimilarity between autumn and winter for both abundance and biomass.The four most important species contributing to the dissimilarities were S. colias, C. obscurus, T. draco and M. azevia, with fewer discards in autumn than in winter only in the case of M. azevia.
The MDS plots for the abundance and biomass of discards with target species as the factor are shown in Figure 5. ANOSIM, applied for abundance and biomass of commercial species, showed global R values of 0.368 and 0.154 with significance levels of 0.1%.This result is supported by p-values of 0.001 and 0.002 of PERMANOVA.The results of SIMPER (Tables S4  and S5 in Supplementary Material) show strong dissimilarities between trips targeting sole species and those targeting cuttlefish, both in terms of abundance and biomass.As regards abundance, only Porifera and M. azevia were discarded in greater numbers in the nets targeting cuttlefish.As regards biomass, only S. officinalis and M. azevia were discarded more in the nets targeting cuttlefish.
Figure 6 shows the MDS plots for the abundance and biomass data of the combined catch, with depth as the factor.ANOSIM gave global R values of 0.660 and 0.683, with significance levels of 0.1%, indicating significant differences between the two depth ranges.This is supported by a p-value of 0.001 obtained with PERMANOVA.SIMPER results (Tables S6 and S7 in Supplementary Material) also provide support for strong dissimilarity between the two depth ranges (10-20 m and 20-30 m) for both abundance and biomass.M. azevia was the species most contributing to the cumulative percentage, and was caught in greater numbers and biomass in the 20-30 m depth range.In terms of abundance, five of the six most important species contributing to dissimilarity between the two depth ranges were the discard species C. obscurus, S. colias, T. draco, Porifera n.id.and S. granularis.As regards biomass, five of the six most important species contributing to the cumulative dissimilarity were commercial (M.azevia, S. officinalis, Raja undulata, Octopus vulgaris and Solea senegalensis).
At a resemblance level of 25% there was similarity among depth ranges (10-20 m and 20-30 m) for both abundance and biomass of combined catch data (Fig. 6).

Discard removal times
Overall, 23 individuals per species/taxa were used to monitor the average time needed to remove the catch of the six most abundant by-catch species/taxa, which represented 85% of the by-catch in number for the M nets and 73% for the T nets.The removal time for R. pulmo was 18.0 s on average, followed by T. draco (14.0 s) and Cymbium olla (12.5 s).The average removal times for the other species/taxa (C.obscurus, S. colias and Porifera) were less than 10 s.

Net damage assessment
At the end of the fishing trials, the 1.5 km of trammel net had 127 holes, of which 84 were detected in the M net.Approximately 80% of the holes in the M net occurred in the lower half of the net, about 60% were larger than 20 cm, and 62% of the holes were found in the guarding net.Forty-three (34% of the total) holes occurred in the T net; about 90% of them were in the upper part of the net, 58% were larger than 20 cm in width/diameter, and 50% were in the inner layer and 50% in the outer layer.

Guarding net: a successful modification?
Mitigation of by-catch of sea birds, turtles and cetaceans in set nets has been widely studied (e.g.Melvin et al. 1999, Gilman et al. 2010, Wang et al. 2013, Martin and Crawford 2015).However, there have been relatively few studies on mitigation of non-commercial fish and invertebrate by-catch in set nets, especially trammel nets.
Compared with previous studies such as those performed in Izmir Bay and Antalya in Turkey (Metin et al. 2009, Olguner and Deval 2013, Aydin et al. 2013), the modified trammel net used in the present study did not provide similar successful outcomes; although there was a reduction in by-catch, there was also a loss of earnings of target species and commercial by-catch.Vecchioni et al. (2016) also reported decreased discard rates for modified trammel nets and decreased yield of commercial species in experimental trials with modified and standard trammel nets in the Egadi marine protected area in Italy, while Sartor et al. (2007) found that trammel nets with a guarding net from the Livorno coast (Italy) had lower catches of more benthic commercial species than standard trammel nets.
In the present study the greatest decrease in commercial catch was observed for the sole species, especially the bastard sole M. azevia.It should be noted that although S. officinalis was considered a target species, during the period of study an unusually poor cuttlefish season occurred, because water temperatures were warmer than usual.According to the records of the tuna trap net company Tunipex (http://www.tunipex.eu/pt/information_oceanic.php) in Olhão (southern Portugal), the average sea surface temperature recorded in the five months of the experimental fishing trials was 2.6°C higher than the average for the same five months over the past 14 years.
Because the majority of the trammel net sets targeted soles rather than cuttlefish, the landed catches of the modified nets had lower total value per unit effort than the standard trammel nets without the guarding net.This is a not unexpected result, because soles and bastard soles are more likely to come into contact with the nets near the footrope.Thus, the guarding net reduces catch rates of these species as the netting is thicker and more visible than the monofilament trammel net.However, we believe that in years of greater abundance of cuttlefish, which come into contact with and are caught higher in the net, when this species is truly the target species, there would be less difference in the catch value per unit effort between modified and traditional trammel nets.Excluding the three main sole species, the modified net caught nearly the same amount of commercial species as the standard net (ratio of 0.84:1).
In this study it was left up to fishers to decide the fate of the catches.Thus, Atlantic chub mackerel, Scomber colias, although commonly sold in the market, was discarded by the fishermen as it was caught in insufficient quantities to warrant sale at auction.Had this species been landed and sold rather than discarded, modified net discards in numbers would have been 46% those of the standard net, while the abundance of commercial landed species of the modified net would have increased from 54% to 70% that of the standard net.
By-catch in number of the guarding net was 41.8% less than that of the standard net.By-catch of noncommercial species is a problem in terms of taking up net area that could be catching commercial species, especially if the by-catch is being caught in relatively large quantities.Of the three most important by-catch species, the longfin gurnards (C.obscurus) have many spines that entangle in the net and cause damage, possibly explaining why nearly all of the holes found in the standard net were in the upper part of the net, especially in the inner fine mesh layer where this species was often caught.Atlantic chub mackerel (S. colias), the second most important discarded by-catch species, did not pose a threat of physical damage to the net and was the species that took the least amount of time on average to be removed from the net.The third most important by-catch species was the greater weever (T.draco), with 202 individuals (33.1% less than the standard net catch).The reduction in catch rates of this species is important due to the danger it poses to the fishermen, as its venomous sting can cause long-term impairment (Dekker 2001), and it was the second most time-consuming species to remove, with an average of 14.0 seconds per individual.

Net damage
Previous studies comparing standard trammel nets to modified trammel nets also assessed damage (Gökçe et al. 2016, Maccarrone et al. 2014).However, Gökçe et al. (2016) only recorded whether damage was in the upper or lower halves of the inner panel and did not take into consideration the size of the hole, while Maccarrone et al. (2014) simply measured the sizes of the holes in the net but not where the hole occurred.The holes in the guarding net were most likely due to the material not being strong enough against demersal species with spikes, or that fact that fishermen were less careful because they knew they would not be using the guarding net again after the experimental fishing trials.Given that the cost of the guarding trammel net (material and labour) was 105 euros per net (45 m), while that of the standard trammel net was only 58 euros per net, there were twice as many holes in the guarding net, and the catches of the modified net were worth less than those of the standard trammel net, there is little incentive to repair the damaged modified nets.

Savings in labour
As stated above, by-catch removal can be laborious and therefore reduction in catches of discard species would reduce time spent cleaning the net and removing catches.R. pulmo, the barrel jellyfish, was the most time-consuming to remove from the net, with the fishermen stopping hauling the net completely to release the jellyfish while it was still in the water.While the average times to remove a single individual may seem insignificant, the time accumulates if we take into consideration that commercial trammel netters fish many kilometres of nets in a single set (more than 10 km for the larger vessels), which means that considerable amounts of time and manpower are required to remove species such as the weever fish from the nets if we extrapolate the estimates obtained in this study based only on 1.5 km of trammel nets per set.

CONCLUSIONS
As in other studies (Sartor et al. 2007, Vecchioni et al. 2016), although the modified trammel nets reduced by-catch and discarding, commercial catches were less than those of standard trammel nets.However, we believe that there is a strong case from both ecologic and economic perspectives for promoting the use of modi-fied standard nets.The use of modified nets should be advocated in sensitive habitats and in marine protected areas.Likewise, in fisheries with high by-catch and discard rates, savings in time and reduced gear damage may compensate for loss of commercial catch.The effectiveness and suitability of modified trammel nets should be assessed by fishery/métier.For example, trammel net catch composition varies strongly with season and depth (Stergiou et al. 2006), so studies should be carried out in the spring and summer and include a greater depth range than that of this study.
Since this study was conducted on a commercial fishing boat, we received feedback on the guarding net from the fishermen as well as an understanding of their willingness to use modified trammel nets.While the fishermen were not convinced by the results, had the fishing trials been carried out during a normal cuttlefish season and had the Atlantic chub mackerel, S. colias, been considered a commercial species, the difference in earnings between the standard and modified nets would have been much smaller, and the benefits of using the guarding net clearer.Switching from guarding net to the standard net over the course of the year according to métier may also be an option.
Fig. 1. -Technical plans for the standard trammel net (A) and the modified trammel net rigged with a guarding net (B).PA, polyamide; PE, polyester.

Fig. 2 .
Fig. 2. -Location of the experimental fishing trials (the shaded area indicates the area off the port of Quarteira, south of Portugal) and the main ports of the Algarve region.

Fig. 3 .
Fig. 3. -Dendrogram (square root transformed data) of the abundance (A) and biomass (B) of all species caught in the 20 trips with the two types of net: M, modified trammel net; T, standard trammel net.

Fig. 5 .
Fig. 5. -MDS plot (square root transformed data) of the abundance (A) and biomass (B) of discarded species caught in 19 trips with the two types of net, with the target species (sole or cuttlefish) as factor.

Table 1 .
-Mean catches per 1000 m of trammel net in numbers (CPUE n ) and biomass (CPUE kg ) and mean value (€) per 1000 m and standard deviations (sd) of landed species for modified trammel net (M) and standard trammel net (T).

Table 2 .
-Commercial (C) and non-commercial (NC) mean discards per 1000 m of trammel net in numbers (CPUE n ) and biomass (CPUE kg ) and associated standard deviations (sd) for the modified trammel net (M) and the standard trammel net (T).

Table S2 .
-SIMPER analysis for the discards (commercial discards and by-catch) data for abundance with the factor season (autumn and winter).

Table S1 .
-Two-way PERMANOVA table with factors net type (Ne) and depth (De), net type (Ne) and season (Se), and net type and target catch (Ta) for abundance of combined catch (includes commercial, commercial discard and by-catch).

Table S3 .
-Simper analysis for the discards (commercial discards and by-catch) data for biomass with the factor season (autumn and winter).

Table S4 .
-Simper analysis for the discards (commercial discards and by-catch) data for abundance with the factor target catch (sole season and cuttlefish season).

Table S5 .
-Simper analysis for the discards (commercial discards and by-catch) data for biomass with the factor target catch (sole season and cuttlefish season).

Table S6 .
-Simper analysis for the combined catch (commercial catch, commercial discards, and by-catch) data for abundance with the factor depth (10-20 m and 20-30 m).

Table S7 .
-Simper analysis for the combined catch (commercial catch, commercial discards and by-catch) data for biomass with the factor depth (10-20 m and 20-30 m).