Ecological importance of survival of unwanted invertebrates discarded in different NW Mediterranean trawl fisheries

There is currently very little information on the survival of discards of unwanted and unregulated catches of invertebrates after the stresses caused by capture. A great number of the unregulated invertebrate species form the basis of essential fish habitats for important fisheries resources such as hake, red mullet and cuttlefish. Thus, data on their survival after discarding may help to interpret the role of these species within the benthic ecosystems. Furthermore, descriptor 6 of the Marine Strategy Framework Directive (EU Directive 2008/56/E) foresees maintaining sea floor integrity at a level that ensures that the structure and functions of the ecosystems are safeguarded, and Article 7(d) of the Common Fisheries Policy (EU Reg. 1380/2013) foresees the implementation of management measures for fishing with low impact on the marine ecosystem and fishery resources. Survival measurements by direct recovery of tagged discarded species are not effective in bottom trawl fisheries, for which alternative studies such as semi-quantitative measures obtained on board prior to discarding can be considered as appropriate for mortality estimation. The present work assessed the survival of unwanted species using a semi-quantitative assessment on the deck of trawlers and at the laboratory for a period of 96 hours in two Mediterranean areas (the Catalan coast and the Ligurian and Northern Tyrrhenian seas). A high number of discarded invertebrates showed a high percentage of survival (>70%) in both assessments. The results can be used to provide information that can help to achieve higher survival levels of discarded specimens and enhance the productivity of fishing grounds by increasing the health of benthic ecosystems.


INTRODUCTION
Mediterranean fisheries are characterized by a high rate of unwanted catches and a great number of marine organisms that are discarded at sea (Lleonart 2015, Tsagarakis et al. 2014. One of the fishing methods that produces most discards is otter bottom trawling, which is also one of the least selective fishing gears. The discards include both species with non-commercial value and marketable species that are undersized or of low value. Technical regulations, such as the introduction of the 40-mm square mesh or the 50-mm diamond mesh in the cod-end (EC 1967(EC /2006 can reduce discards to some extent, but cannot solve the impacts of bottom trawling on habitats and benthic communities. The investigated areas were the Catalan coast, corresponding to FAO division 37.1.1, Geographical Sub-Area 6 (GSA06), and the Ligurian and northern Tyrrhenian seas, corresponding to FAO division 37.1.3, Geographical Sub-Area 9 (GSA09), both comprising chronically exploited fishing grounds. In the last ten years the demersal fisheries carried out mainly by bottom trawl fleets in the two areas accounted for about 40% of the total landings and 70% of the economic value (STECF 2016).
A large fraction of this discarded biomass (30%-50% of the total biomass caught) is composed of species of commercial interest (small-sized or damaged specimens), while the remaining fraction is composed of species with low or no economic value (Machias et al. 2001, Sánchez et al. 2004. Furthermore, trawl fleets operate in a great variety of soft habitats (e.g. muddy-sand, sandy-muddy, mud, sandy-gravel, sand), so discards are characterized by extremely high species diversity with a high percentage of non-commercial species, some of which are macroinvertebrates (echinoderms, crustaceans, poriferans, ascidians, cnidarians, bryozoans, bivalves and gastropods). In many cases the discarded species belong to sensitive benthic habitats, such as maërl or crinoid beds.
The impact of bottom trawling on benthic habitats and communities and demersal species is little known. The impact depends on the fishing activity (Martín et al. 2014) and can create changes in the ecological functioning of benthic components that have important repercussions on the exploited populations (de Juan et al. 2007, Frid 2011, Hewitt et al. 2008. The European Marine Strategy Framework Directive (EU 2008/56/E) encourages member states to move towards an ecosystem-based fishery management in order to protect the goods and services that marine ecosystems provide. Therefore, it is important to take into account the link between benthic communities and habitats and fisheries resources, because a great number of ecological interactions may be adversely impacted by fishing. The capture of benthic invertebrates and their discarding at sea will impact benthic habitats to a certain degree, depending on the post-release survival of each species. The survival of habitat-structuring invertebrates, such as crinoids and echinoderms, can help to maintain the good status of the essential fish habitats where the most important commercial resources, such as European hake, red mullet, spiny lobster and cuttlefish, use them as areas of nursery, recruitment or growth (Abella et al. 2008, Colloca et al. 2009).
There is currently very little information on the survival of unwanted and unregulated invertebrates after the stresses of being captured, handled and discarded. The specific biological characteristics make an organism more or less vulnerable to different stressors of the capture method and release process (de Juan and Demestre 2012). Other factors affecting the survival of released animals are related to the handling practices during the sorting and release processes and to the environmental conditions during capture, hauling on board and sorting, such as hypoxia and temperature , Giomi et al. 2008, Tsagarakis et al. 2017.
Some unwanted and unregulated invertebrates such as crinoids and ophiuroids form the basis of essential fish habitats for commercial species such as hake and red mullet. Robust information on discard survival after fishing and release to the seabed can improve the interpretation of the role of unregulated invertebrates on the benthos (Benoît et al. 2012).
The main objective of this paper was to estimate survival rates of invertebrates discarded from trawlers working in two northwestern Mediterranean areas, the Catalan coast and the Ligurian and northern Tyrrhenian seas. The study focused on unwanted invertebrates which belong to the unregulated species and are likely to continue to be released after capture. To estimate the vitality rates, a vitality assessment on the captured organisms was carried out under normal fishing activity of the trawl fleets in both selected areas. The approach was developed using a semiquantitative assessment (SQA) (ICES 2014) according to Benoît et al. (2012). Two different procedures of survival estimation were developed, considering first the survival on the deck of trawlers and second the long-term survival at the laboratory with captive observations. The survival rate of the discarded fraction in the trawl catch can be used to propose man-agement based on better control of discards or spatial restrictions to trawling that may accompany specific management plans based on conserving the functionality and health of benthic ecosystems.

Study areas and sampling activities
The study was carried out in two NW Mediterranean trawl fishing areas, the Catalan coast (GSA 6), from March 2016 to February 2017, and the Ligurian and northern Tyrrhenian seas (GSA 9), from November 2016 to February 2017. Data were collected during fishing trips on board commercial vessels performed on a monthly basis.
Trawl sampling in the Catalan coast area was performed on board four different trawlers in five fishing grounds (Garotes, Las 40, Capets, Planassa and Malica) adjacent to the port of Blanes. The depth range was between 50 and 494 m, with a total of 23 hauls ( Fig. 1A  At the end of each haul the trawl gear was retrieved on board and the cod-end was opened on the deck following the normal commercial fishing practices. After that, prior to sorting the catch into commercial and discard fractions, the net was shot for a new haul. Depending on the depth, this process took 10 to 25 minutes. During the fishing trips, there was no interference by the researchers on board with the habitual modus operandi of the fishermen in the daily fishing activity (position, duration, sorting, etc.) and the sorting processes of the catch, which lasted 20-30 minutes depending on the capture.

Survival experiments
The hauls considered for survival analysis were 4 on the Catalan coast and 19 in the Ligurian and northern Tyrrhenian seas (Supplementary material Table  S1A). The average depth range was between 99 and 362 m on the Catalan coast and between 84 and 470 m in the Ligurian and northern Tyrrhenian seas. The hauls were carried out on both the continental shelf and the upper slope, with a standard deviation of 117.66 and 152.89 m, respectively. The complexity of the experiment forced us to limit the number of hauls. While the catch was being sorted manually on deck by the fishermen, in both areas the vitality of the unregulated, non-commercial invertebrates was assessed just before the species were discarded. The SQA (ICES 2014) was performed by means of indicators of state of vitality according to mobility, injuries and lesions suffered by the organisms due to the fishing activity. To estimate the vitality of each individual, a categorical assessment scale of four vitality levels (VL) was applied: 1 (excellent), 2 (good), 3 (poor) and 4 (dying or dead) (Benoît et al. 2012, ICES 2017. A detailed explanation of each VL is presented in Table 2. Two approaches to performing the survival experiments were developed: i) direct survival estimation on deck and ii) survival estimation at the laboratory for a period of 96 hours. In both study areas the studied invertebrates are only captured by trawling. Because of the challenges of obtaining viable control samples with other gears, the individuals who were classified in excellent state of vitality (Table 1) at the start of the captive experiment (time T0, just as it was being released on deck) were taken as pseudo-controls for each haul (ICES 2014).

Survival estimation on deck
In both areas the immediate survival on deck was estimated following the same methodology. For the selected species, as many individuals as possible were taken and one of the four VLs was assigned to each one using the SQA ( Table 2). The selection process was carried out only during the first 30 minutes after opening the net on the deck, starting before the gear was shot for a new haul. It was done as quickly as possible to decrease as far as possible the time of air exposure on deck.

Survival at the laboratory
The long-term survival rate was estimated for 96 h to achieve a deeper knowledge of the actual survival of the invertebrates discarded. It was only performed in the Catalan coast area. To analyse survival at the laboratory, the individuals of the selected species were sampled from the last daily haul, just when the catch was laid on deck and prior to sorting, but only during the first 30 minutes, as in the previous case. The VL was assessed according to the SQA ( Table 2). The four possible VLs were observed for each specimen, which were individually introduced in four white containers (one for each VL), with running sea surface water to avoid hypoxia. The maximum number of selected animals was that which could be introduced in each container without causing stress. A total of 324 individuals from 22 species of invertebrates (10 echinoderms, 8 crustaceans, 2 cnidarians and 2 ascidians) were sampled but only the 6 most abundant species were selected for this survival assessment.
Survival at the laboratory was estimated for 96 hours in an aquarium at the ICM laboratory by applying an SQA. A total of ten time survival observations (T) were carried out during the experiment (from T0 to T9). The first observation, time T0, just as animals were being released on deck, was executed on board. Individuals were selected for a maximum time of 30 minutes, and each one was introduced successively into the containers, as explained above. The second observation, T1 (6 h), was performed just before each individual was transferred to the aquariums at the laboratory. The individuals that were in VL 4 (dead or moribund) were removed to avoid contamination in aquariums, but were accounted for. The following eight observation times, from T2 (18 h) to T9 (96 h), were made on the specimens placed in the aquariums, with a periodicity of 12 h until the study had lasted for 96 h. The aquariums were divided into three sections for each VL, 1, 2 and 3, and no more than ten individuals were introduced per section. The time of the transport from the sea to the aquariums was a maximum of 2 hours and the animals were on the white plastic containers with oxygen pills during the whole transportation time.
The natural environment conditions were simulated in the aquariums through an open seawater circuit and water temperature was maintained between 13°C and 14°C, similar to the in situ temperature in the northwestern Mediterranean fishing grounds. The photoperiod was adapted to the natural luminosity with black canvas to dim the light. Controls of salinity, nitrates, nitrites and silicates were periodically performed. The whole process was carried out under food abstinence conditions.

Data analysis
The survival on deck was estimated by applying the survival index, which was calculated as the ratio between the number of specimens (VN) with a vitality level of 1 to 4 and the total number of discarded individuals (DN) and was expressed as a percentage.
A Wilcoxon test between the exploratory variables recorded (Supplementary material Table S1) and the survival of invertebrates on deck for the first 30 min on board was carried out. The test analyses the relationship of each variable with survival, comparing the data related to live individuals (VLs 1, 2 and 3) with data of dead individuals (VL 4). Finally, if significant differences were observed between the variable and the survival, the group mean was calculated. The Wilcoxon test was implemented in R 3.4.3 (R Core Team 2017).
Because a seasonal sampling was not performed, variables related to seasonality such air and water temperature were not taken into account. We therefore conducted the analysis only with Depth and Haul Time.
To calculate survival rate for each experiment over 96 h, the Kaplan-Meier analysis was used (Kaplan and Meier 1958). The Kaplan-Meier survival curve is a function of the data only, and in the absence of censored values it follows the proportion of individuals alive at each time interval during the holding phase of the experiment and seeks a point in time when survival stabilizes. Plots were made on six selected species for the three live VLs, showing their 95% confidence intervals. Analyses were conducted with the "survival" package in R 3.4.3 (R Core Team 2017).

Survival on deck
The VLs of discarded invertebrates were identified for each individual of each selected species. Table 3 shows the results of each VL analysed in each study area, according to the continental shelf and slope and the corresponding habitat. From the discarded fraction, the species presented in Table 3 were scored with vitality levels in both study areas. Among the invertebrates captured, only N. norvegicus and P. longirostris were subject to minimum conservation reference sizes (MCRS) (Council Reg. EC 1967, while the rest of the discarded invertebrates were non-regulated species.
A survival index of 100% was shown by 14 species in the Catalan coast area and 6 in the Ligurian and northern Tyrrhenian seas area, and Astropecten aranciacus and the genus Ophiura were common in both areas. The species Munida intermedia, Goneplax rhomboids and Macropipus tuberculatus showed a lower survival index in the Ligurian and northern Tyrrhenian seas area, where the processed number of individual was higher and the results were probably more reliable. On the other hand, N. norvegicus gave more robust results in the Catalan coast area, where more individuals were analysed.
The percentages of the four vitality levels (VL) of Norway lobster and deep water rose shrimp, the two commercial species subjected to MCRS, are presented in Table 4 for both areas. Values of each VL represent the percentage of the total number of individuals for each level from the total analysed hauls. The Catalan coast area showed the highest percentages of live VLs.
The Wilcoxon test was carried out only with the variables Haul Time and Depth. The results showed no significance between depth and survival. However, the test found significant differences in survival due to Haul time with a p-value <2.2e-16 and W=7898500. The mean duration of hauls with live animals was 217.39 min, while the mean duration with dead animals was 236.12 min (i.e. a 9% time increase).

Survival at the laboratory
The survival estimation at the laboratory was carried out only for the specimens collected in the Catalan coast study area. The analyses were undertaken only with the invertebrate species with a higher number of individuals scored with VL on deck previous to the discard (Table 3A); a total of six species were analysed. Of the six invertebrates analysed, the three in Figure  2 showed 100% survival for 96 h in the aquarium experiment.
The species Leptometra phalangium showed more than 90% survival, and only after 30 h did the percent-  age of mortality increase ( Fig. 3; Table 5A, B) in the specimens with the VL 3. VLs 1 and 2 showed no evidence of mortality, but VL 3 showed an evident nonstability of survival. Similar results were shown for the species Ophiura texturata. In this case the first evidence of mortality appeared after 66 h in VL 3, but the percentage of survival was still high (>79 %). VLs 1 and 2 showed no evidence of mortality (Fig. 4, Table 5C, D).
The last species studied in the aquarium was Echinaster sepositum, which showed mortality at 6 h of the experiment in VL 3, but maintained a steady survival rate >90% until the end of the experiment (Fig. 5 and Table 5E).

DISCUSSION
This study was carried out with those individuals that showed signs of vitality when arriving on board, which means they were still alive. In fact, there was a low number of specimens that could be assessed, and this may indicate the severe impact of trawling on the seabed and benthic communities Spencer 1995, Jennings et al. 2001). The preservation of exploited resources is probably the main goal of fishery management, but the perturbation of chronic fishing activity on fishing grounds has negative ecological effects leading to high levels of mortality (DFO 2006, van Denderen et al. 2013. Several studies have evidenced an improvement in the health of exploited resources when effort limitation and seasonal or temporal closures of trawl fishing activities are implemented (Demestre et al. 2008, Pipitone et al. 2000, but the effects at the level of benthic communities remain less well known. The by-catch of invertebrates in bottom trawling yields a high amount of epifauna or infauna that have important functions for the sea floor ecology. For instance, echinoderms or gastropods are important bioturbators and comprise several feeding guilds, such as deposit or filter feeders, or predators (e.g. Echinus melo, Spatangus purpureus, Echinaster sepositus, Ophiura texturata, Chlamys opercularis, Calliostoma granulatum and Aporrhais serresianus). These organisms play an important role in ecosystem function by maintaining  or enhancing secondary marine production. They are very sensitive to disturbance and easily destroyed by fishing impact, and their decrease could have lasting consequences for benthos-pelagic processes (Lohrer et al. 2004, Demestre et al. 2017, de Juan et al. 2011, because the good status of the habitats in which the fisheries resources live depends to a large extent on these organisms. In order to maintain the good status of the sea bottom, one of the priority actions to be taken is to determine the mortality levels of routinely discarded species. A study carried out near the Catalan coast area related the effects of trawl fishing and feeding of the red mullet Mullus barbatus , showing negative effects due to changes in benthic functional components in the fishing ground. In areas where there was no fishing (fishing closure areas) the macroinfauna which constitutes the food base for M. barbatus was significantly more abundant than in areas disturbed by the trawl. Changes in the habitat structure (homogenization) and functionality of benthic communities caused by fishing can alter the normal supply of food (e.g. polychaetes and crustaceans) for both adult and juvenile red mullet . Furthermore, as a consequence of the habitat alteration, the characteristics of the seabed that serves as a nursery, spawning or growing habitat could be modified, with possible negative consequences on future recruitment of the species.
The rates of survival shown by invertebrates in both areas investigated in this study showed great variability between VLs of the same species once the individuals had been captured and deposited on the deck of the trawler. Mortality levels also vary from one species to another, depending mainly on the biological and functional traits of each species, such as fragility, emergent or surface position, filter feeding and sedentary motility (Costello et al. 2015. External protection is one of the most relevant traits for increasing survival, as evidenced by the monitoring of VLs on deck to analyse immediate mortality. In both areas the majority of crustaceans remained alive, even reaching percentages of 100%. Invertebrates with regeneration traits such as echinoderms also have a high level of survival. We went one step further in estimating discarded invertebrate mortality by attempting to identify and separate the injuries of each individual on deck according to its VL. Individuals with VL 1 and 2 at time T0 (time of release on deck) survived on deck until they were released into the sea in a maximum time of 30 min, but those with VL 3 showed low survival on deck. The experiments at the laboratory to analyse survival at 96 h confirmed this behaviour for all analysed species. At the laboratory it was evident that when the survival was not 100% it was because the organism was at VL 3 when released on deck, and in fewer cases at VL 2.
The results of the Wilcoxon test indicated that Haul Time was an important factor for improving organisms' survival on deck. Injuries increased and VL decreased when invertebrates arrive on deck after long hauls, as was observed in the continental shell hauls, which showed a higher survival of species of crinoids and crustaceans . Consequently, failure of individuals to survive for a long time in the laboratory experiment is due to their low VL when they were left on deck. It is therefore important to handle the organisms on deck quickly and safely to increase their survival when they are discarded back into the sea. During fishing operations on deck, it is recommendable to keep the organisms under a wet cover to avoid drying. Another easy method for improving the survival of discards could be a direct operating system such as a duct with water from the deck for throwing animals back into the sea.
It must be taken into account that, in addition to the unhealthy state of the invertebrates who died during the experiment, the mortality may also have been due to the captivity conditions, where no food was available and the environment was only similar to the most appropriate habitat. However, the possibility of discarded invertebrates escaping predators or obtaining food is low because of the injuries they suffer during capture (Ramsay et al. 1996, In-gólfssonIng et al. 2007). Therefore, our experimental results can be assumed as a proxy to the level of survival of discards at sea.
According to the Common Fisheries Policy and the landing obligation to prevent discarding of regulated species (MCRS, Council Reg. EC 1967.4b), a high level of post-capture survival can be adduced by member states to include an exemption from the landing obligation in their discard management plans. Our results for the survival of N. norvegicus can be considered a starting point with information focused on the aim of a possible species exemption but, obviously, more studies based on larger samplings need to be carried out before this exemption can be recommended. Conversely, a second crustacean species regulated by MCRS, Parapenaeus longirostris, showed low survival and would not be a good candidate for exemption.
The species selected for survival estimations were the most representative of different taxonomic levels and were of ecological importance in their habitats. In view of this, the 100% survival of the crinoids A. mediterranea and even L. phalangium, whose increase in post-release mortality started at VL 3, shows that the impact of trawling on crinoid beds may be less serious than assumed until now, as most crinoid individuals would survive the encounter with trawlers and postcatch release. Furthermore, in many cases crinoid beds are essential habitats for nursery and spawning areas of some commercial species (Colloca et al. 2004). The other two echinoderms that were assessed, O. texturata and E. sepositum, gave similar results, both starting mortality at VL 3. Therefore, the results may suggest again an optimistic possibility for maintaining a good environment status and a sustainable structure on the soft-sediment habitats that form the majority of trawl fishing grounds (Piet and Hintzen 2012). However, to maintain this optimistic perspective and a good environment status on the Mediterranean fishing grounds, it is mandatory to contain the current exploitation levels, especially in other types of habitat that may be even more sensitive to trawling than crinoid beds, such as maërl and Isidella (Kamenos et al. 2004, Mastrototaro et al. 2017. To achieve this, fishing activity and fishing effort must be reduced, temporal and spatial closures and even permanent closed areas must be implemented, and the measures regulating the reduction of discards must be implemented (FAO 2011).
The results of the present work offer some new knowledge on the survival of discarded invertebrates that may be useful for improving ecosystem health and productivity. Nevertheless, it should be regarded as a starting point, because mortality after discards at sea depends on many factors, such as susceptibility to predation and lower competitiveness for obtaining food , Demestre et al. 2000, Kaiser et al. 2006. Knowing levels of survival of discarded invertebrates helps to obtain a more realistic image of the state of the benthic ecosystem, and consequently of the fishing grounds. The sustainability of the exploited populations depends on the conservation of these habitats, because a large part of their life cycle takes place in them.