Discard survival of undersized European plaice caught with towed fishing gears in Danish waters

The reform of the European Union (EU)'s Common Fisheries Policy (CFP) in 2013 has led to significant changes in fisheries management, including an obligation to land all catches from regulated stocks, i.e. a ban on throwing unwanted catch back into the sea. The goal of the landing obligation is to promote more selective and targeted fishing that reduces unwanted bycatch. It is possible to put fish back into the sea instead of landing them if it can be scientifically proven that there is a high survival rate in a specific fishery. However, the criterion for what counts as "high" is not set and is evaluated for each individual case by the EU.

This report presents the work of DTU Aqua on discard survival of undersized European plaice (Pleuronectes platessa, below 27 cm in the North Sea and Skagerrak, below 25 cm in the Baltic) caught using towed gears in commercial demersal fisheries for human consumption. This work has focused on:
• estimate survival rates with the aim of obtaining exemptions from the landing obligation,
• improve our understanding of how operational, environmental, and biological stressors affect discard survival rates,
• explore the opportunity to create robust discard survival estimates from meta-analysis; iv) investigate the effect of the environment (temperature) on reflex impairment,
• develop and test the performance of an optimized reflex and injury index,
• use expert knowledge (Bayesian modelling) to predict discard survival.

We estimated survival rates as scientific documentation for seeking exemptions from the landing obligation in the EU CFP for three fleets: the Danish seine and bottom otter trawl fleets1 operating in Skagerrak, Kattegat, and the North Sea (ICES subdivision 3a and 4), and bottom otter trawls operating in the Baltic Sea (ICES subdivisions 22-25). Survival estimates considered the characteristics of the gear (gear types and designs), fishing practices (target species, seasonality, and handling practices) and ecosystem (e.g., hazards from hard-shelled prey items, or area-specific variability in oxygen conditions in the Baltic Sea), as required by the CFP.

The studies found the following survival rates in different situations:
• In a conservative scenario, i.e., discard survival during the warm water season (August-October) in Skagerrak, the discard survival in the demersal mixed fishery using an otter trawl with a 90 mm codend and a 120 mm SELTRA-panel fishing was 44% (95% confidence interval CI: 37%-52%).
• In comparison, it was 78% (95% CI: 67%-87%) in the Danish seine fishery fishing simultaneously. The discard survival rate for the otter trawl when targeting plaice improved to 75% (95% CI: 67%-83%) in the cold-water season (March).
• When fishing for Norway lobster (Nephrops norvegicus), the discard survival rate during winter was reduced to 40% (95% CI: 28%-57%) due to more injuries to the plaice when caught together with Nephrops.
• When we changed the design into a divided codend separating fish from Nephrops, the survival rate of plaice was higher with mean 94% (95% CI: 81%-100%) when caught together with fish in the upper panel than when caught in the lower compartment with Nephrops (61%, 95% CI: 48%-73%) or when mixed with Nephrops in the standard gear. The number of individuals was however low due to the high selectivity in the fish compartment.
• The lowest estimates of discard survival were observed when fishing plaice in the Baltic Sea with an otter trawl during a warm water period (autumn). A delayed survival rate of only 27% (95% CI: 9%-55%) was obtained when fishing with a T90 codend, and 14% (95% CI: 4%-29%) when using a Bacoma codend. In the cold-water season (November to April)2, the survival rate was 87% (95% CI: 82%-92%).

Based on the project's results, the EU Commission granted a year-round high survival exemption for plaice caught in the Danish seine fishery in Skagerrak and Kattegat (ICES Division 3a) and North Sea (ICES Subarea 4; EU, 2018, §16). The bottom otter trawl fishery was granted an exemption for the winter season (EU, 2018, §17). For the Baltic Sea, in line with Kraak et al (2018) in the German mixed demersal trawl fishery in ICES subdivision 22, the discard survival observed in our study might be considered “high” for plaice in winter. However, the High-Level Regional Group (BALTFISH) decided not to include a request for high survival exemption for plaice in the Joint Recommendation.

We investigated the effect of dissolved oxygen level at capture on delayed plaice mortality. Oxygen levels were related to seasons, with lower levels in autumn than in winter (confounding factors). Hypoxia-resistant priapulids were more common in stomachs from plaice caught in autumn, likely because a part of these stomachs comes from areas with severe hypoxia. The data on stomach contents indicate that plaice are performing excursions between areas or depths of different levels of hypoxia - a part of them probably feeds in severe hypoxia and returns to moderate hypoxia / normoxia to digest and recover, like it seems to be the case for cod in the eastern Baltic Sea. Fish discarded to hypoxic waters had a more severe stress response and a prolonged recovery period but recovered their measured biochemical indicators and oxygen consumption rates to pre-stress conditions within 24h, with no stress-related mortalities. Fish that were exposed to trawl simulation and discarded to hypoxic conditions showed no indications of trying to escape oxygen-poor conditions. Instead, they all burrowed in the sediment immediately following release. It remains unequivocal whether simulated trawl exerts the same magnitude of stress as experienced during commercial fishing.

An additional variable contributing to post-catch mortality may be the damage to the intestine during the catch and sorting processes due to sharp shell fragments in situations where plaice are feeding heavily on mussels. The sampled plaice were considered individual specialists, which means that the individual prey type in general is consumed only by a moderate part of the plaice and amounts to a significant part in the stomachs in which it occurs. There were fewer prey categories in wintertime (amphipods and mysids were missing), which is not surprising. There was no visible relationship between survival and shell content.

The common way to estimate survival rates is to observe in captivity fish that would be discarded under commercial conditions, until the mortality levels off. Such captive observation studies are labor-intensive, logistically challenging, and financially demanding. As an alternative, measures of impairment in fish condition can be used as an indicator for discard survival providing that they are calibrated with survival likelihood estimates from, e.g., captive observation studies. Promising indicators of fish condition as good survival proxies are external damages and reflexes (scoring the presence or absence of pre-determined attributes) pooled into an index. The optimization procedure aimed at finding the weighing of the reflex and injury attributes into the index (usually fixed to 0.1) with the best predictive performance. Bruising in the head and body were the most important contributors to the survival probability of discarded plaice with 90 and 95% of the best models showing coefficients higher than 0.10. Overall, none of the individual reflex or injury indicators were independent of Discard survival of undersized European plaice caught with towed fishing gears in Danish waters 9 biological, environmental, technical, and operational covariates when predicting plaice discard survival, both at fish and trip levels. The best models (based on AIC) for each vitality indicator all included the interaction between air exposure and sea temperature. The optimized index did not improve predictions markedly as both the reflex impairment and injury index as well as the less labour-intensive categorical vitality score were almost equally valuable proxies of plaice discard survival. When we compare observed and predicted survival ratio for each trip in the context of management purposes, i.e., assessing whether the survival ration is high, all vitality indicators could correctly predict high (>0.50) or low (<0.50) survival except for one trip.

In contrast to traditional (frequentist) methods previously used in our survival studies, the Bayesian network model approach can integrate expert knowledge regarding life-history traits and the prevailing operational, environmental, and biological conditions of fisheries to predict survival probability after release. This expert system may be suitable as a low-cost decision support tool for fisheries managers. The classification error of the ensemble approach was much lower than fitting a single naive Bayes model on multiple trips simultaneously or from causal network learned from data. Discretising all variables into three levels appeared to be a good trade-off between predictive accuracy, model complexity and predictive accuracy. Introducing the individual reflexes and injuries did not improve the predictive accuracy of the model. We also built an operational Bayesian Belief Network (BNN) model to estimate post-release survival potential of discarded plaice. The BBN model was constructed from a combination of historical data and subject matter expert knowledge. The typical user case would be to identify species-fisheries for which it would be meaningful to collect scientific documentation for a high survival exemption in the context of the CFP. The model output indicates the probability of a survival rate above 50% and can be used as a relative score to compare different scenarios.

While obtaining discard survival estimates have been a main aim of most studies for advisory purposes, investigations on factors affecting the survival rates have been made in parallel to reduce the need for demanding capture observation studies, and at the same time achieve more robust discard survival estimates and to inform how fishing operations can be changed to improve survival rates. Investigations on how various factors influence discard survival could to a higher degree be performed under controlled conditions in the laboratory. The link between observed vitality, reflex impairment, external damages, and survival, is still not well understood, specifically with respect to how quickly (or slowly) fish are able to recover from the capture process.