This study was conducted following the Guidelines of the European Union Council (2010/63/UE) and Portuguese legislation “The protection of Animals Used for Scientific Purposes” (DL 113/2013). All the procedures were approved by CCMAR Animal Welfare Committee (ORBEA CCMAR—Organization Responsible for Animal Welfare of CCMAR) and the Direção-Geral de Alimentação e Veterinária (DGAV) of the Portuguese Government. All animal protocols were performed under Group-C licenses from the DGAV, Ministério da Agricultura, do Desenvolvimento Rural e das Pescas, Portugal.

This study was conducted on a commercial crustacean bottom trawler in the South (37°-36°N; 9°-7.5°W) and Southwest (37.9°-36.7° N; 7.7°-9.6° W) coasts of Portugal (Figure 1). Ten fishing trips with a total of 77 hauls were performed from June 2020 to May 2022, summing up to 351 h of fishing effort. The target species were Norway lobster (Nephrops norvegicus), shrimps and prawns [Aristeus antennatus, Aristaeomorpha foliacea, Aristaeopsis edwardsiana (Johnson, 1868), Parapenaeus longirostris (Lucas, 1846), and Penaeus monodon, Fabricius, 1798] using net codend (i.e., the terminal section of a trawling net) mesh sizes of 70 mm (mean depth of 483 m) and 55 mm (mean depth of 636 m) respectively.

Fishing hauls lasted 2.3–9 h, with a velocity of 1.5–3.7 nm/h, and were conducted across all seasons of the year. Fishing effort was computed in hours from the moment the net reached the bottom of the ocean up to the moment it began to be hauled back to the vessel. Capture and handling time was computed in hours since the start of a haul until the start of the sorting of the catch by fishers. The weight of the net codend (kg) was a visual estimation given by the skipper when the net was lifted from the water at the end of every haul, or when the catch was already inside the “pond” (an area below deck where the codend is offloaded). Water temperature (°C) at the bottom/fishing depth was either collected by a Scanmar^® or by a mini DST-CTD logic^® Star-Oddi^® attached to the net (temperature was recorded at every 5 min for hauls < 1,000 m and every 10 min for hauls >1,000 m). Surface water temperature was collected by the DST-CTD whenever a reading was performed near the surface (< 1 m), otherwise, data from the Copernicus website on the sea surface temperature for the sampled regions, dates and hour of each haul (to ensure the precision of the estimate), was used instead. Technical (i.e., codend mesh size, codend weight, and fishing effort), environmental (i.e., water temperature, and fishing depth), and biological data (i.e., sharks' total length and maturity stage) were registered into the Electronic Logbook (eLog) Olrac Dynamic Data logger (Olrac DDL^®) where information per specimen was inserted. Capture and handling duration (h) was assessed from the moment each haul started up to the moment the sorting of the catch started.

The DSE caught during the fishing events, were sorted by species. Each specimen was identified following Ebert et al. (71) and Last et al. (72) and measured [total length (TL) to the nearest 0.5 cm: from the tip of snout to the tip of the caudal fin]. DSE's at-vessel condition was assessed following the criteria of Benoît et al. (73) and Catchpole et al. (74). Both methods include a rapid observation of possible injuries and reflex impairments (i.e., reduction or loss of responses to stimuli) to classify specimens into a small number of ordinal vitality categories (Table 1). After this evaluation, the alive specimens were kept inside acclimatized holding tanks until the moment they were discarded, i.e., after the end of all the analyses with the DSE from the same haul, which took up to 1 h in some instances.

Blood sampling was conducted over four field trips between May and August 2021 and February and April 2022 along the South coast of Portugal. Hauls were randomly selected for sampling, with five alive and five dead sharks from frequently caught species (Etmopterus spp., G. melastomus, and S. ringens) randomly chosen from each selected haul. Blood samples were collected as quickly as possible from dead sharks via the caudal peduncle using a 1 ml heparinized syringe, followed by sampling from alive sharks. Dead sharks were prioritized over live ones, as blood flow decreases over time post-mortem, making collection increasingly difficult. Because each fishing trip takes ca. three full days, plasma was obtained by centrifugation of blood samples on board within 2 h post-collection, stored at −20°C and kept frozen until laboratory analysis. The alive sharks were released after blood collection. At the laboratory, the concentrations of metabolites (i.e., glucose, lactate and urea) and electrolytes (i.e., phosphorus, chloride, magnesium, and calcium) were measured by colorimetric assays using Spinreact^® kits (Girona, Spain) with a Multi-Mode Microplate Reader BioTek Synergy™ 4 (BioTek^® Instruments, Winooski, VT, USA). Sodium and potassium concentrations were determined using a flame photometer (BWB-XP PerformancePlus, BWB Technologies, UK).

A total of 1,559 specimens belonging to 18 deep-sea species of sharks (15 species) and skates (3 species) were evaluated. Collectively they were mostly dead (n = 1,258) or in a poor condition (n = 224), with very low numbers of specimens in good condition (n = 70) and even lower in excellent conditions (7). All species presented a much higher number of deceased specimens (mean = 84) or specimens in poor condition (mean =15), than in excellent (mean = 2) and/or good conditions (mean = 6; Table 2). Species that presented the worst condition were Oxynotus paradoxus and Neoraja iberica with all specimens dead (however with low number of evaluated specimens), followed by Deania calceus and D. profundorum (Table 2). Species that presented comparatively better at-vessel condition (i.e., higher rates in good and excellent conditions) were the Centrophorus spp. with no deceased specimen. Mitsukurina owstoni also did not present immediate AVM but only one specimen was sampled (Table 2).

Stress analyses were conducted at 65 specimens of four deep-sea sharks' species. Alive E. pusillus and G. melastomus were larger, generally caught at greater depths, and presented greater duration of capture and handling procedures when compared with dead specimens (Table 5). For E. spinax the same was observed but that comparison was made considering only one dead specimen. Alive S. ringens presented larger size and were mainly caught at shallower depths. In this species, however, the capture and handling times between dead and alive specimens were similar (Table 5).

Plasma analyses revealed varying concentrations of metabolites and electrolytes for the different species (Table S1). Glucose and urea presented significantly higher concentrations for alive specimens of the species E. pusillus and G. melastomus and higher concentrations for alive S. ringens in comparison with dead specimens. However, urea presented higher values for deceased S. ringens in relation to alive specimens (Table S2). Lactate concentrations were not significantly different but were also higher for alive specimens of G. melastomus and S. ringens (Figure 2). For E. pusillus higher concentrations were found for deceased specimens in comparison with alive ones (Figure 2). Alive specimens of E. spinax presented the highest maximum values of all metabolites in relation to the dead specimen (Figure 2).

In relation to the electrolytes, significant differences were found for potassium and magnesium (Figure 3; Table S2). Potassium concentrations were significantly higher for dead specimens of E. pusillus and G. melastomus (Figure 3; Table S2) and higher for dead specimens of S. ringens (Figure 3). Magnesium concentrations were significantly higher for G. melastomus deceased specimens but higher for E. pusillus and S. ringens alive specimens (Figure 3; Table S2). Calcium was higher for alive specimens of E. pusillus and S. ringens but for G. melastomus it was higher for dead specimens (Figure 3). Chloride concentrations were higher for alive specimens of E. pusillus and G. melastomus whereas S. ringens deceased specimens presented higher concentrations (Figure 3). Sodium concentrations were higher in deceased G. melastomus and S. ringens but higher for alive E. pusillus (Figure 3). Phosphorus concentrations were higher in deceased E. pusillus and G. melastomus and a bit higher in alive S. ringens. The only dead specimen of E. spinax presented the highest value of sodium (449.7 mM), in comparison with alive specimens of the same species and all the other specimens of the other species (Table S2; Figure 3). Chloride, potassium and sodium were all greater for this dead E. spinax specimen when compared to the alive specimens, whilst calcium, and magnesium were greater for alive specimens in comparison with the dead specimen (Figure 3).

This study presents data on the at-vessel condition of DSE caught in bottom trawling operations, where most specimens were either dead upon retrieval (80.7%; n = 1,258) or in poor condition (14.4%; n = 224). Only a few were in excellent (0.4%; n = 7) or good condition (4.5%; n = 70). Specimens classified as in poor condition are unlikely to survive when released back into the sea [e.g., (23, 25, 32, 52, 78)]. Consequently, the PRM (including dead specimens or in poor condition) of the studied DSE species is expected to be extremely high, averaging 95% across all of the 18 DSE species sampled. This includes exceptionally high estimated PRM for the most numerous species like G. melastomus (93%), E. spinax (94%), E. pusillus (97%), S. ringens (98%) and D. profundorum (99%). Additionally, PRM is thought to be critically high for species considered endangered in Europe by the International Union for Conservation of Nature [IUCN; (78)], including Centroscymnus coelolepis and Deania calceus (100%), as well as Dalatias licha (93%). However, some species of conservation concern showed a higher proportion of specimens in relatively better condition compared to other DSE species. These include the critically endangered (78) Centrophorus granulosus (30% in good condition), the endangered Centrophorus squamosus (20% in excellent and 20% in good condition), and the least-concern but rare Chlamydoselachus anguineus (33% in good condition).

Most at-vessel condition studies on sharks and skates focus on demersal species caught by longline and gillnets presenting varying mortality estimates (0–100%) but generally lower than for bottom trawling (25, 53, 79, 80). Studies specifically on bottom trawling are scarcer (81) and mainly involve resilient demersal species like Scyliorhinus canicula, which showed relatively low mortality rates between 2 and 53% (30, 31, 57, 82). Only recently has research assessed the condition of a few DSE species in bottom trawling, as in a study of 12 demersal elasmobranch species in the Asinara Gulf, where specimens were classified into “active,” “inactive,” or “dead” conditions (26). Scacco et al. (26) reported high rates of inactive or dead specimens for Dipturus oxyrinchus (85%), E. spinax (88%), and G. melastomus (91%), aligning with our findings. Though our estimates of dead or poor-condition specimens were still slightly higher (98, 94, and 93%, respectively).

Using several technical, environmental and biological predictors, the GLM achieved a high prediction of 83.5% of sharks' AVM (dead and alive), which was specifically linked to several factors which included specimen size (TL), codend weight and mesh size, temperature differences, species, and fishing effort. Mortality rates were inversely related to shark size, with smaller specimens more likely to die in bottom trawling procedures than larger ones. Similarly, Ellis et al. (83) reported higher mortality in skates under 50 cm in the North Sea trawling fishery, and Talwar et al. (25) observed higher mortality in smaller Squalus cubensis in the Exuma Sound longline fishery. This trend is also consistent with findings for other elasmobranchs (24, 26) and bony fishes (84–86). Larger specimens generally exhibit greater swimming endurance and are less prone to injury. In contrast, smaller specimens are more vulnerable to physical damage and higher mortality rates when caught in trawling nets, primarily due to the increased volume and composition of the catch in the codend. Higher catch volumes intensify contact and abrasion among organisms, raising the risk of injury for smaller species (29). This agrees with the present study findings where heavier codend presented significantly higher probability of mortality (although with a small effect), than lighter codend. However, caution is needed when interpreting these results, as codend weight was estimated visually by the skipper, introducing the potential for under- or overestimation of the weight. Future studies could use a codend weigher, a tool that measures the codend's weight as it is brought on board (87). This would enhance the accuracy of codend weight measurements and contribute to a better understanding of how codend weight affects mortality rates in sharks.

Another important technical predictor on the sharks' mortality was codend mesh size, where hauls using smaller mesh sizes (55 mm) targeting prawns and shrimp had significantly higher mortality rates than those with larger mesh sizes (70 mm) targeting Norway lobster. Smaller mesh sizes increase retention, causing the net to clog more quickly and capturing a wider range of specimen sizes (88, 89). This congestion may cause mortality by compression and asphyxia. This would suggest that adjusting mesh sizes based on target species and fishing depth could help reduce mortality in non-target deep-sea shark populations. However, the relationship between codend mesh size and mortality is so far only supported in studies with bony fish. Studies with bony fish suggest that larger mesh sizes reduce injuries, and scale loss by allowing more fish to escape (90–92), although some research found no significant impact of mesh size on escapee survival (84, 85). In this study, depth—a key predictor of mortality—was excluded from the GLM due to collinearity with other factors, as smaller mesh sizes (55 mm) were predominantly used at greater average depths (623 m ± 286 m), while larger meshes (70 mm) were used at shallower depths (473 m ± 58 m). This does not rule out the potential impact of mesh size on shark mortality, but further research using varied mesh sizes at similar depths and conditions could provide clearer insights into its effects specifically on mortality rates in deep-sea sharks.

With larger temperature differences (between bottom and surface waters) there was an increase in mortality in the sharks analyzed for this study. Similar results were reported in deep-sea sharks caught by longliners in the Cantabrian Sea [NE Atlantic; (24)] and in Bahamian waters where the sea surface temperature reached 30°C (23, 25). Deep-sea species inhabit cold waters, and when brought to warmer surface temperature may increase the stress and mortality (33, 70, 93–95). Most sharks are ectothermic which means that they regulate their body temperature with the surrounding water and have a relatively narrow range of temperature in which they can thrive and reproduce (96). Water temperature was already suggested as a proxy to determine time/area closures when it reaches a certain threshold (94). This approach may become even more important as global water temperature rises due to climate change but will require an understanding of species-specific reactions to local conditions.

Fishing effort is generally positively related to mortality, as prolonged procedures elevate mortality rates (67, 69, 94). In static gear like longlines and gillnets, longer soak times increase mortality, particularly for species reliant on ram ventilation for oxygenation (81, 97). However, in this study, higher fishing effort was related with a slightly increased odds of finding alive specimens than dead ones. Unlike static gear, trawling does not allow for pinpointing the exact moment of capture, so fishing effort in this study is likely overestimated, measured from the start of each haul rather than the exact entry time of each specimen in the net. Hauls in this study ranged from 2:54 to 8:36 h, meaning that specimens were exposed to at least 2:54 h of fishing activity. It is possible that the alive specimens from longer hauls were captured closer to the end of the haul, increasing their chance of survival. This could suggest that shorter hauls, potentially < 2:54 h, may reduce shark mortality. However, longline studies have shown significant mortality in deep-sea sharks even with soak times under 3 h (24, 25). Additional controlled studies could help identify if there is an optimal haul duration that would minimize shark mortality.

Shark species exhibited markedly different mortality likelihoods, aligning with findings in other studies (26, 67, 68). Deep-sea squaliform sharks tend to have particularly vulnerable and conservative life histories which includes lower fecundity and smaller litter size, making them less resilient to the pressures exerted by fisheries (21, 22, 98). Indeed, the squaliform sharks (Etmopterus spp. and S. ringens) showed the highest mortality rates, contrasting with the carcharhiniform shark (G. melastomus). These differences may be partly attributed to the presence of an anal fin in G. melastomus and its absence in squaliform species. The anal fin is thought to enhance swimming stability (99), which could provide G. melastomus with improved swimming abilities, potentially reducing its mortality in trawling situations. However, this adaptation does not prevent its capture by trawlers. Further research is required to confirm this hypothesis.

In general, specimens appeared stressed by the conditions and procedures conducted during fishing activities, as indicated by plasma concentrations of secondary stress response indicators (e.g., metabolites and electrolytes). These concentrations were different when compared to baseline values for elasmobranchs from previous studies (100, 101) and stress indicators for demersal sharks caught by bottom trawlers [e.g., (102, 103)]. In this study, significant differences in glucose, urea, magnesium, and potassium concentrations were observed between deceased and alive specimens of E. pusillus and G. melastomus.

Potassium and urea levels could act as mortality markers for E. pusillus and G. melastomus, with deceased specimens showing significantly higher potassium and lower urea levels. For G. melastomus, significantly higher magnesium levels were also observed in deceased individuals. Although no statistical differences were noted for S. ringens and E. spinax due to low sample numbers, deceased specimens exhibited higher potassium levels. Post-mortem potassium and urea concentrations have previously been linked to mortality in Galeocerdo cuvier and Prionace glauca, alongside other markers like lactate, phosphorus, and calcium (63, 64). High potassium levels may lead to hyperkalemia, disrupting muscle cell membrane excitability, causing myocardial dysfunction in S. acanthias and muscle tetany in Mustelus antarcticus (47, 100, 104). Plasma potassium above 7 mM in this study consistently indicated poor at-vessel condition or death, supporting previous findings (105), and suggesting cell disruption leakage. Similarly, low urea can signal osmotic imbalance, possibly increasing ion exchange across osmoregulatory tissues—a stress response linked to capture in S. acanthias (50, 106).

Stressed fish generally exhibit significant increases in glucose levels (46, 51, 52), which is attributed to the mobilization of glucose for energy production [e.g., (107, 108)]. However, in the present study, glucose was lower for dead specimens of S. ringens and significantly lower for dead specimens of E. pusillus and G. melastomus in comparison with alive ones. Talwar et al. (25) also found that, lower blood glucose levels, resulted in a higher likelihood of mortality of deep-sea sharks in the Bahamas. Cliff and Thurman (47) observed that moribund or dead C. obscurus specimens had lower glucose levels compared to those that were alive. This might imply that, an inability to further mobilize glycogen stores and glucose depletion may contribute to metabolic failure and ultimately, death.

Lactate has been proposed as a mortality marker (51, 63, 64), given that increased lactate levels indicate a strong physiological response, often due to anaerobic respiration triggered by high stress or intense activity in low-oxygen conditions (46, 107, 108). In this study, lactate levels did not significantly differ between deceased and alive specimens of E. pusillus and G. melastomus, though across all four species studied, lactate levels were high when compared to the unstressed 5 mM level reported for species like Carcharhinus obscurus, C. plumbeus, and S. acanthias (47, 109, 110). Pelagic sharks with lactate levels above 16 mM often show high mortality (46). Notably, in this study, maximum lactate values for G. melastomus (33.12 mM) and S. ringens (27.27 mM) were among the highest recorded for elasmobranchs, comparable to species like Alopias vulpinus [27 mM; (111)] and Carcharhinus brachyurus [42 mM; (53)]. Elevated lactate levels in G. melastomus and some other alive specimens from this study suggest potential PRM, warranting further investigation into species-specific lactate responses (51).

Comparing interspecific responses, S. ringens displayed distinct concentrations of phosphorus, magnesium, and urea. This could indicate either a unique response to fishing procedures on the crustacean bottom trawler or naturally different baseline levels of these metabolites and electrolytes. Prohaska et al. (70) found species-specific and ecological differences in stress responses among deep-sea sharks, with deeper-dwelling species showing distinct reactions compared to those at shallower depths. Scymnodon ringens is a larger species and was generally found at greater depths than other studied species, particularly among live specimens. This size difference may relate to reduced fight intensity, potentially helping larger sharks better manage stress (25).

The observed differences in plasma secondary stress responses between alive and deceased and the high dispersion within a single species suggest individual variability in coping mechanisms (112) and that stress responses may be triggered at an individual level (63). Variability within parameters and species also likely reflect the differences each animal experiences while caught which are impossible to estimate in this study—i.e., actual time in codend, level of contact with the gear and other organisms, place within the organism's mass in the codend and opportunity to struggle or even access to oxygenated water. Studies on elasmobranchs indicate that plasma metabolites and electrolytes reach peak levels over different timeframes: lactate (~2 h), urea (~6 h), and sodium [~5 h; (47, 56, 100–102, 113)]. This highlights the importance of further species-specific studies where animals are exposed to various stressors in different fisheries, with blood sampling over extended time intervals to fully characterize stress responses following events like capture or air exposure (114). However, samples collected immediately after capture may not fully reflect physiological impact. Conducting controlled experiments with deep-sea elasmobranchs poses challenges due to the need to replicate their natural environment (low temperature, high pressure, darkness) and minimize captivity-related stress. Thus, endpoint concentrations of metabolites are often assessed in deceased specimens (63–66). In this study, only potassium, magnesium, and urea showed potential as markers for mortality in deceased E. pusillus and G. melastomus, suggesting that further studies should include additional species, sample sizes, and varied fishing conditions.

In the European Union, DSE are protected by regulations such as Regulation 2024/257, which prohibits the landing of several shark species, effectively setting a zero TAC to prevent targeted fishing. However, despite this protection, DSE continues to be caught as bycatch in bottom trawling operations and are subsequently discarded, often dead or dying, and under stress, as observed in this study. This is particularly concerning, as discarded bycatch is rarely recorded in fisheries logbooks, limiting our understanding of DSE's population dynamics and distribution. The ongoing mortality of DSE, combined with the absence of accurate catch and discard records, underscores a significant yet unrecorded impact on these vulnerable populations. This raises concerns about the long-term sustainability of DSE species in EU waters, as the true extent of their decline remains largely unknown. Given the unknown limits to which these species can withstand fishing pressures, a precautionary approach is necessary by avoiding the capture of DSE in the first place.

Studies on gear modifications provide a valuable foundation for mitigating bycatch. Turtle excluder devices and the Nordmøre grid have proven effective in reducing bycatch and are widely applied in elasmobranch studies [e.g., (115–119)]. These are angled panels with spaced bars, that allows smaller target crustaceans through while excluding larger, unwanted species such as sea turtles and elasmobranchs without impacting greatly the target catch rates (115–119). A modified version of the Nordmøre grid has been applied in Portuguese crustacean trawl fisheries, where it effectively reduced bony fish bycatch while maintaining the commercial catch of crustaceans (120). However, while effective for larger elasmobranchs, these devices still allow smaller DSE to pass through, meaning smaller-bodied DSE remain at risk. Therefore, further actions and gear modifications are encouraged to address the bycatch of smaller-bodied DSE species.

Adjusting the codend mesh size and weight in bottom trawling presents a viable approach to reducing bycatch and mortality of sharks. In Portugal, research on bottom trawler's codend mesh configuration suggests that increasing mesh size and switching from diamond to square mesh significantly enhances selectivity, allowing smaller specimens to escape (121–125). Diamond mesh tends to close under tension, trapping smaller species, while square mesh stays open, facilitating the escape of non-target species without reducing the catch of commercially valuable fish (121, 122, 132). Codend weight also plays a significant role in bycatch mortality as previously discussed. Heavier codend increase mortality not only directly, through the physical injuries caused by increased contact and abrasion among organisms (29), but also indirectly by prolonging sorting time onboard, thereby raising the risk of air exposure. Although air exposure was not measured in this study, the benefits of reducing fishing duration suggest that adjusting effort duration and codend weight could further reduce bycatch mortality. Shortening fishing hauls, when possible, could reduce both the codend's weight and handling time, minimizing air exposure and stress for non-target species.

Closures based on specific areas, seasons, and/or depths have proven effective in reducing bycatch, as they help to protect critical elasmobranchs' aggregations (126). For instance, a depth-based ban on bottom trawling below 800 m was implemented in the Northeast Atlantic in 2017 (Regulation 2016/2336) to protect deep-sea species, including DSE. Yet, for further spatial-temporal closures to be optimized, especially in Portuguese waters, a comprehensive understanding of DSE pupping and breeding grounds, would facilitate the establishment of targeted closures, maximizing conservation benefits for DSE populations. This can be achieved using a spatial distribution modeling approach (127, 128). While DSE distribution patterns have been modeled for the Azores (129), mainland Portugal still lacks crucial data on DSE distribution and habitat use. Integrating electronic monitoring programs, alongside onboard observers, can also support data collection on DSE bycatch, enhancing data on abundance and distributions, ensuring compliance and enabling adaptive management.

Implement onboard handling DSE protocols may aid in the mitigation of their mortality rates. Studies on elasmobranch PRM demonstrate that careful handling increases survival likelihood, even for species prone to stress-related mortality (126, 130, 131). Despite the high AVM rates in this study, improved handling techniques could be beneficial to minimize physical trauma, especially for conservation concern species like C. coelolepis, Centrophorus squamosus, C. granulosus, and D. calceus. Training fishers in these techniques, informed by the high mortality rates observed for DSE in bottom trawling, would be an actionable step toward reducing bycatch impacts.