Small-scale fisheries in the Murcia Region (SE Spain) are carried out by a large number of vessels (129 licensed units in 2016 or 72% of the local fleet: General Directorate of Agriculture and Fisheries, Region of Murcia) with length overall (LOA) between 6 and 13 m (engine power 50-100 hp) operated by a crew of two or three persons. The main fishing gears deployed during the year are trammel nets, gillnets and boat seines used to catch a variety of demersal resources. The fleet rotates the métiers practised during the year, depending on the availability of the main target species. In the case of trammel nets, the métiers are defined by the target species: for instance, in winter and spring the target species is the cuttlefish (Sepia officinalis), but it changes to finfish in summer and to striped red mullet (Mullus surmuletus) in late autumn and early winter. The fishing activity with trammel nets is carried out close to home ports, with daily fishing trips of less than 12 h. The trammel nets employed by this fishery are made of 50-m sheets for a total length of 1500 to 2000 m, and they are 1.5 to 2.5 m high.

Two local fishing boats practising the cuttlefish métier were used to carry out the experiments with the modified trammel nets. In the first boat (coded “SP”) two trammel nets of 1500 m length and 1.6 m high with the same technical characteristics were employed (inner panel 80 mm nylon (polyethylene) mesh, 40 meshes high; outer panel 200 mm nylon mesh, 8 meshes high; hanging ratio 0.82) following the usual professional configuration. One trammel net (the experimental one) was fitted with a guarding net made of 2.5-mesh-high polyethylene mesh (200 mm, twine thickness 4.2 mm) and positioned between the trammel net and the footrope (weighed rope 10 mm diameter). The other trammel net was employed without modification and served as a control (Fig. 1). From 2 to 23 March 2017, the two fishing gears were set on sandy bottoms between 24 and 30 m depth at approximately 7 p.m. and were recovered the following day at around 9 a.m. The second experiment was carried out concurrently with the first: a standard trammel net was divided into three sections of 500 m each. Two sections were provided with artificial lights fixed on the floatline (12 mm float nylon rope): one section was provided with white lights, the other with green lights (Fig. 1). The artificial lights were units employed in tuna longlining fisheries, commercialized as “LED fishing light, deep sea drop light” manufactured by Ningbo Solars Lighting Electrics (Zhejiang, China). Each unit used two AA batteries whose duration in the laboratory was in excess of one week (nominal manufacturer’s specification 300 h), but the batteries were replaced every two days in each unit to avoid decreasing light intensity. The position of the three sections along the trammel net was changed randomly in each fishing operation to randomize the effect of light position. The second fishing vessel (coded “AG”) only carried out the part of the experiment testing artificial lights from 10 April to 16 May 2017.

Each experimental setup was fished on ten consecutive days, on the basis of operational constraints due to weather conditions and permitted working days. For each fishing operation, the entire catch was analysed for species composition (number and weight). Non-commercial organisms were classified in three categories, D, K1 and K2, where “D” are undersize specimens of commercial species (e.g. belonging to species included in Annex III of EU Reg. 1967/2006 and therefore included in the Landing Obligation provisions), “K1” are damaged specimens of legal size that cannot be commercialized, and “K2” are marine organisms without commercial interest (usually epifaunal invertebrates).The most abundant commercial species were also measured [mantle length (ML) in cm for cuttlefish; total length (TL) in cm for fishes].

Catch data were standardized in number or weight per 100 m h (soaking time). The commercial species were analysed on the basis of standardized weight data because the possible benefits of using the experimental fishing gear will be assessed by fishers as increases in catch in weight (or in economic value). Conversely, the metric used for discards was standardized number of individuals because trammel net fishers sort the catch specimen by specimen and any reduction of unwanted catches should appear as a reduction in the number of unwanted specimens to be sorted. The immediate expected benefit of the modifications should be reduced sorting time.

The standardized catch rates in weight of commercial species (kg/100 m h) were compared by means of linear mixed models using the treatment as a fixed factor (green light, white light, no light) and the boat as a random factor. The assessment of the effect of the guarding net was carried out similarly, but without the random factor, as only one vessel was used; in this case, the linear mixed model reduces to classical ANOVA. The models were fitted with function lme of the R library nlme using the restricted maximum likelihood algorithm. The posterior difference in means for factors with more than two levels was tested by post-hoc analysis using library lsmeans. The effect of the modification was assessed by calculating its magnitude (1 = no effect) and its confidence interval (significant effect when the CI does not cross 0). The standardized number of discarded weight (ind./100 m h) was subject to a similar statistical analysis, adding the factor type of discard (D, K1 and K2).

The species composition of catches (commercial and discards fraction) was also assessed by means of descriptive multivariate analysis, comparing the species composition in the different treatments (control; lights; guarding net). PRIMER^® (Clarke and Gorley 2006) routines were used to examine the ordination in multivariate space using non-metric multidimensional scaling (nMDS). The significance of factors (treatment guarding net or lights against control) was tested by means of analysis of similarities (ANOSIM).

During the experiments, 61 species of commercial interest were caught (Table 1). The target species, cuttlefish, was present in all nets sampled, with a yield of 1.43 kg/100 m h on average. Other important commercial species in the catch were different species of sparids (e.g. the pandora Pagellus erythrinus, the gilthead seabream Sparus aurata, the dentex Dentex dentex, the seabass Dicentrarchus labrax, the scorpion fish Scorpaena scrofa, the octopus Octopus vulgaris and, due to their individual weight though they were not frequent, the rays Torpedo marmorata and Raja asterias) (Table 1). The hermit crab Dardanus arrosor was very frequently caught; this species is locally commercialized as bait for hook and line fisheries, not for human consumption.

The number of taxa discarded was 53, including damaged or undersize specimens of commercial species that could not be commercialized because they were undersized or damaged (Tables 2 and 3). The unwanted catches of non-commercial species (mainly epifaunal in vertebrates) were dominated by sea cucumbers (Holothuria sp.) and seastars (Astropecten sp. and Echinaster sepositus). Within commercialized species, the main reason for discarding was damage (type “K1”), which was particularly important for five species: D. labrax, M. surmuletus, P. acarne, P. erythrinus and S. officinalis (Table 3). Only specimens of Pagellus acarne and P. bellottii were discarded because of small size (Table 3).

The standardized yield of commercial catch species was lowest in both weight and number in the control trammel net, whereas that of unwanted catches was lowest in trammel nets fitted with the guarding net and highest in the trammel nets with lights on the head rope (Table 4). By typology of unwanted catches, the control trammel net had the highest amount of undersize fish and green lights the lowest. In terms of unwanted catches of categories K1 and K2, the trammel nets with the guarding net always had the lowest amount of unwanted catches and green lights the highest.

As regards the commercial species with frequency of occurrence higher than 30%, the yields due to the different net configurations were compared. The total commercial catch was 32% higher in the trammel nets with the guarding net than in the standard configuration (Fig. 2). This was due mainly to the increased catches of the target species Sepia officinalis (increase of 95%), Octopus vulgaris and some sparids (Diplodus vulgaris, Dentex dentex and Pagellus erythrinus) (Fig. 2). Notably, the catch of the gastropods Bolinus brandaris and Hexaplex trunculus showed no difference, while the catches of Dardanus arrosor were significantly lower with the guarding net (Fig. 2, see also Supplementary Material, Fig. S1).

The yields of the most frequently caught commercial species were tested also for the effect of artificial lights (Fig. 3). Overall, the commercial catch was significantly higher (around 13%) in the trammel nets with head rope lights, but no statistical differences were observed between green or white lights. This result was primarily due to the increased catches of the target species Sepia officinalis (37% higher combining both lights; no statistical difference between the two colours). As regards fishes, only the sparids Dentex dentex and Spaurus aurata showed a positive effect of lights, but in these two species the effect was significant for white light, not for green light. In other species, the effect of light was usually positive (except for Symphodus tinca), although not statistically significant (Fig. 3, see also Supplementary Material, Fig. S2).

The catch rates of unwanted catches were always significantly higher in the control nets (mean of 2.74 ind./100 m h compared with 1.37 ind./100 m h; Figs 4 and 5, Table 4). The differences in unwanted catches between standard nets and nets with a guarding net varied depending on the reason of discarding, showing no interaction with the experimental treatment. A post-hoc analysis showed that undersize unwanted catches (category “D”) were not different between the control and modified configuration, while damaged individuals (K1) and non-commercial organisms (K2) were significantly lower in the modified configuration.

The amount of unwanted catches (ind./100 m h) was not significantly different in the nets modified with lights on the head rope compared with the control nets (Fig. 5). However, there were significant differences in the type of unwanted catches captured in the modified net (p=2.3 10–6), with a significantly higher amount of category “K2” non-commercial invertebrates.

The species composition of unwanted catches did not vary significantly according to treatment, either in the modified nets with a guarding net or in the nets fitted with lights (Fig. 6). The ANOSIM yielded a global correlation r=0.062 for the comparison of faunal assemblages in the control against guarding net experiments (p=0.016). The ANOSIM results for the comparison of lights yielded a correlation r=–0.004 (p=0.68), with partial correlations lower than 0.01 in all cases and p-levels higher than 0.1. In the commercial fraction, although the difference between the configuration with guarding net and the other nets was not significant, examining the species lists showed that the importance of fishes (especially sparids and seabass) was higher in the trammel nets with a guarding net than in the other configurations, where commercial invertebrates such as purple dye murex or commercial hermit crabs came second to the target species (cuttlefish) in abundance.