Introduction
The fishery and aquaculture sectors are essential to satisfy the global food demand, offering a high-quality animal protein (FAO et al. 2024. However, their activity could conflict with the EU ‘Farm to Fork’ strategy, which aims to create sustainable food production sources (European Commission: Directorate-General for Health and Food Safety 2020). The sustainability strategy includes the improvement of animal welfare, including those in aquatic environments. The latest Eurobarometer (Eb) polling instrument revealed that a majority of respondents expressed a strong concern about farmed animal welfare (FAW) issues (Riuzzi et al. Reference Riuzzi, Contiero, Gottardo, Cozzi, Peker and Segato2025), suggesting that welfare regulations are increasingly driven by consumer expectations. Still, welfare considerations for aquatic species remain inconsistently addressed across species, primarily due to limited knowledge about the needs of many farmed groups (Franks et al. Reference Franks, Ewell and Jacquet2021). Despite not being predominantly bred in captivity, lobsters are maintained in holding facilities for extended periods (up to two months) prior to live trade, introducing a managed phase analogous to late production stages in aquaculture (Hinchcliffe et al. Reference Hinchcliffe, Agnalt, Daniels, Drengstig, Lund, McMinn and Powell2022), before live trade.
For crustaceans, the welfare standards are not yet especially well developed, partly due to the limited number of studies supporting their capacity to experience pain or suffering (Crump et al. Reference Crump, Browning, Schnell, Burn and Birch2022). Animal welfare is understood as the biological, behavioural and emotional way of an animal to cope with the environment (Dawkins Reference Dawkins2017). Live-traded animals may be exposed to various types of stress that directly impact their health and well-being, such as hypoxia, poor water quality, handling, starvation, disease, transport, cannibalism, and slaughter methods (Barreto et al. Reference Barreto, Rey Planellas, Yang, Phillips and Descovich2022; Wuertz et al. Reference Wuertz, Bierbach and Bögner2023). Humane slaughter methods aim to minimise pain and suffering by inducing a loss of sensibility, typically through anaesthesia or stunning (Conte et al. Reference Conte, Voslarova, Vecerek, Elwood, Coluccio, Pugliese and Passantino2021; Rotllant et al. Reference Rotllant, Llonch, García del Arco, Chic, Flecknell and Sneddon2023).
Electrical stunning is identified as one of the most acceptable methods for humane slaughter of crustaceans, as it can render animals unresponsive almost (< 10 s) immediately (Neil et al. Reference Neil, Putyora and Albalat2024). Some electrical stunning devices, such as the CrustastunTM, have been designed for dispatching cold water decapod crustaceans, aiming to halt all nervous activity (Neil & Thompson Reference Neil and Thompson2012). However, depending on the intensity of the electrical current, the heart activity may resume even if the animal seems irresponsive (Weineck et al. Reference Weineck, Ray, Fleckenstein, Medley, Dzubuk, Piana and Cooper2018, Neil et al. Reference Neil, Putyora and Albalat2024). To evaluate the effectiveness of stunning, a behavioural scoring system can be used, assessing the response of the eyes, appendages, and mandibles to an external stimulus (Roth & Grimsbø Reference Roth and GrimsbØ2016). In this case, the stunning effect is translated into a reflex impairment.
Reflex movements have been extensively studied in crustaceans, and they include the claw reflex, rhythmical movements of the abdominal appendages, ‘tail-flip’ and the righting response (Shuranova & Burmistrov Reference Shuranova and Yuri Burmistrov2006). The righting response (as the ability of the lobsters to ‘return’ to an upright position) was used as the measure to determine the duration to recovery, seen as the regain of full-body coordination (Davis Reference Davis1968). This response can be affected by stress factors, such as temperature (Young et al. Reference Young, Peck and Matheson2006), salinity (Denisse Re et al. Reference Denisse Re, Díaz and Valdez2006), CO2 levels (Zittier et al. Reference Zittier, Hirse and Pörtner2013), and seismic damage (Day et al. Reference Day, McCauley, Fitzgibbon, Hartmann and Semmens2019). Some of these stressors also affect metabolic homeostasis.
Metabolic stress in crustaceans may be revealed by changes in crustacean hyperglycaemic hormone (CHH) concentrations (Elwood et al. Reference Elwood, Barr and Patterson2009), as this group lacks the cortisol equivalent of vertebrates (Sadoul & Geffroy Reference Sadoul and Geffroy2019). During emersion stress, increased CHH levels have been correlated with glucose and L-lactate concentration (Chang Reference Chang2005), establishing these metabolites as potential biomarkers. In fact, the monitoring of L-lactate, glucose, and haemocyte numbers (where feasible) has been proposed as part of operational welfare assessment (Conneely & Coates Reference Conneely and Coates2024). While stress responses are usually adaptative, chronic and/or severe, stress can have irreversible negative consequences, leading to poor growth, disease, infection, and even death (Stoner Reference Stoner2012).
Some of the changes in the animals’ health can directly affect their muscles, and consequently, the intrinsic composition of the meat (Barrento et al. Reference Barrento, Marques, Vaz-Pires and Nunes2010, Lorenzo et al. Reference Lorenzo, Tomac, Tapella, Yeannes and Romero2021). Potential impacts on the product quality involve the accumulation of undesired substances (e.g. L-lactate, ammonia) or tissue damage caused by proteolysis or bacteria proliferation (Gornik et al. Reference Gornik, Albalat, Atkinson, Coombs and Neil2010). Few studies have assessed the impact of electrical stunning on the meat quality of crustaceans (Albalat et al. Reference Albalat, Gornik, Muangnapoh and Neil2022), as boiling live animals is still a common practice (Conte et al. Reference Conte, Voslarova, Vecerek, Elwood, Coluccio, Pugliese and Passantino2021). Boiling or steaming live crustaceans may induce death rapidly, especially for small crustaceans (Roth & Øines Reference Roth and Øines2010); however, larger individuals take longer to die (circa 2.6 min), resulting in avoidable suffering and raising welfare concerns (Fregin & Bickmeyer Reference Fregin and Bickmeyer2016; Lauridsen & Alstrup Reference Lauridsen and Alstrup2024).
This study aimed to test the impact of electrical stunning on the highly commercialised American lobster (Homarus americanus) with the CrustastunTM device to: (i) determine the effect of the electrical stunning duration on the recovery and behaviour; (ii) assess the physiological disturbances of the different stunning protocols; and (iii) evaluate the impact of electrical stunning with CrustastunTM on meat quality in two retail scenarios: direct from market and domestic cooking. We expect that, given sufficient time, lobsters will recover from electrical stunning, since their heart rate (HR) does not seem to be impacted by this procedure (Neil et al. Reference Neil, Putyora and Albalat2024). Also, we do not expect a negative impact on the overall meat quality by applying the CrustastunTM before dispatch, as seen in other crustacean species (Albalat et al. Reference Albalat, Gornik, Muangnapoh and Neil2022).
Materials and methods
Experimental animals and husbandry
The American lobsters used in the present study were obtained from Maresmar SLU wholesaler at Mercabarna (Barcelona, Spain) between September 2022 and May 2023. All lobsters were transported refrigerated (< 4°C) to either the Institute of Marine Sciences (ICM-CSIC, Barcelona) or the Institute of Agrifood Research and Technology (IRTA, Girona, Spain), depending on each institute’s area of expertise. At ICM male lobsters (n = 60) were acclimatised for 4 weeks at the Aquaria facilities (ZAE), in tanks with running seawater (13°C and 38 ppt) and under a photoperiod of 14L:10D. During that period, animals were kept under blue light and fed with frozen mussels and squid. The day prior to the onset of the experiment lobsters were transferred to 200-L glass tanks (130 × 38 × 40 cm; length × width × height; Figure 1A) with running seawater, where animals were filmed to study their behaviour. Lobsters taken to the IRTA (males and females; n = 60) were stored in aquaria filled with artificial seawater (Instant Ocean, UK) in refrigerated chambers (4°C and 30 ppt) and subsequently distributed among the different treatments, see Experimental design. All lobsters were in intermoult, weighing 497 (± 5) g, and measuring 83.2 (± 0.9) mm carapace length.
American lobsters (Homarus americanus) in running seawater aquaria (A), image from video analysis distinguishing animals on their upright (bottom aquaria) and the upside down (top and middle) position; lobster placed inside the CrustasunTM (B); and sampling scheme of the abdominal section in lobsters, being (1) exudation %, (2) instrumental texture, (3), (4) and (5) samples for sensory evaluation and (6) instrumental colour and pH (C).
Figure 1. Long description
Panel A is a photo of a vertical rack holding three seawater aquaria. Computer vision bounding boxes identify lobsters. The top aquarium shows a lobster labeled H O A M minus D O W N 0.93. The middle aquarium, outlined in yellow, shows a lobster labeled H O A M minus D O W N 0.94. The bottom aquarium shows a lobster labeled H O A M minus U P 0.61.
Panel B is a photo of an American lobster with green bands on its claws, placed on a perforated metal tray inside a Crustasun T M machine.
Panel C is a line drawing of a lobster tail section divided into six sampling zones.
- Zone 1 is at the top for exudation percentage.
- Zone 2 is below it for instrumental texture.
- Zone 3 is in the middle for sensory evaluation.
- Zones 4 and 5 are side-by-side below zone 3 for sensory evaluation.
- Zone 6 is at the bottom right for instrumental colour and p H.
Experimental design
Experiments carried out in ICM and IRTA used the commercially available device CrustastunTM (manufactured by Mitchell & Cooper Ltd; Figure 1B) to apply electrical stunning to unanaesthetised animals, following manufacturer’s instructions. Briefly, the chamber of the stunning device was filled with a brackish water solution (3 ppt), which was absorbed by a sponge positioned over the electrode. Once the device was securely closed, an electrical shock of 1.0–1.3A was delivered to the animals. The device is set to apply a 100-V shock for either 5 s (short cycle) or 10 s (long cycle).
At ICM, after acclimatisation, lobsters were randomly assigned to the following treatments (Table 1): Control (lobsters were taken individually from tanks, placed inside the CrustastunTM for 10 s with no electric shock and returned to the filming tanks); Short cycle (lobsters were taken individually from tanks, placed inside the CrustastunTM for 10 s – and electroshocked for 5 s [short cycle] – and then returned to the filming tanks); and Long cycle (lobsters were taken from tanks, placed inside the CrustastunTM for 10 s [long cycle] and returned to the filming tanks). After the CrustastunTM, animals were placed back into the aquaria on their backs (upside down on the dorsoventral axis) to measure the righting response (see Behavioural recordings). Animals were filmed 1.5 h before the experiment and for 12 h after the treatment was conducted (see Behavioural recordings). For the physiological determinations, Control and Short treatments (but not Long, given the recovery time) were repeated with a different set of animals, taking haemolymph samples immediately before the stunning and at 1-h intervals (to ensure all animals were rightened), for a total of 12 h. Also, initial physiological effects of electrical stunning were assessed by repeating the Control and Long treatments with a separate group of animals, from which a single sample was obtained 30 min after passage through the stunner.
Summary of treatment names and description in study evaluating the impact of electrical stunning on the welfare and meat quality of the American lobster (Homarus americanus)
Table 1. Long description
Summary of treatment names and description in study evaluating the impact of electrical stunning on the welfare and meat quality of the American lobster (Homarus americanus)
In a parallel study, the lobsters in IRTA were randomly assigned to one of four treatments (Table 1), before being cooked alive in a steam oven (100°C, 5 min). The treatments consisted of: T1 Retailer/No stunning (lobsters removed from tanks and then cooked); T2 Retailer/Stunning (lobsters stunned by applying a 10-s cycle and then cooked); T3 Domestic/No stunning (lobsters emerged for 30 min at RT and 2 h at 4°C for purchase and transport simulation, and finally cooked); and T4 Domestic/Stunning (lobsters removed from tanks and electrically stunned 10 s) followed by transport simulation (30 min RT and 2 h at 4°C) and then cooked. Electrical stunning was performed with the CrustastunTM using the long (10 s) cycle only, following the manufacturer’s guidelines (Mitchell & Cooper Ltd 2024).
Behavioural recordings
Videos were analysed using computer vision tools involving supervised machine learning and quantitative evaluation methods (Viola & Jones Reference Viola and Jones2001). Briefly, lobsters were annotated by fitting polygons and labelling the orientation (‘UP’; upright or ‘DOWN’; upside down) on images extracted from recorded experimental videos using the LabelMe software (Russell et al. Reference Russell, Torralba, Murphy and Freeman2008). After labelling 2164 ‘.JSON’ images, the latter were converted into YOLO-compatible ‘.txt’ files containing the position and orientation coordinates, and the dataset description file (‘.yaml’) was created. To reduce the dataset bias, we performed data-augmentation (translation, blurring, and rotation up to 90°) and, together with the videos, we trained a YOLO neural network using the Ultralytics (8.2.25) framework (Sohan et al. Reference Sohan, Sai Ram and Rami Reddy2024) and by tailoring hyperparameters (Table S1; Supplementary material), to enhance the accuracy of object detection and tracking throughout the analytical workflow (Figure 1A and Figures S1–3 [Supplementary material]). Data extraction (and creation) and model training were carried out in Python (V3).
Finally, the algorithm creates an output file containing time-stamp information, coordinates (x, y), probability scores and class identifiers, designating the centroid of the detected animal’s bounding box. Thereafter, the activity is measured by calculating the squared difference on the animal’s position (and converting pixel measurement to cm) at 1-s intervals and subsequently averaged for every 10 min for graphical representation. The time taken to return to an upright position (righting response) was defined by a change in orientation registered in the class identifier (transitioning from ‘DOWN’ to ‘UP’).
Physiological assessments
Haemolymph was extracted with a 1-mL syringe using a 21-Ga needle (previously kept on ice) from the abdominal sinus; 300 μL were extracted and added in an Eppendorf tube containing 300 μL perchloric acid (PCA) 6%, then the tubes were centrifuged at 10,000 g for 10 min at 4ºC. The supernatant was mixed with 100 μL K2CO3 2.5 M, centrifuged once more and the supernatant frozen until the further metabolite analyses. The isolated metabolite fraction was used to measure glucose and L-lactate. For glucose analyses we used the method described by Webster (Reference Webster1996). Briefly, 20 μL of samples were supplemented with 180 μL of the working reagent (0.2 M Phosphate buffer containing glucose oxidase). After 20 min of reaction at RT, the plate was read at 410 nm in a microplate reader (Infinite M200 PRO, Tecan, Switzerland). L-lactate dehydrogenase method (Babson & Babson Reference Babson and Babson1973) was used to measure L-lactate. Briefly, 20 μL of sample were added to the plates containing 210 μL of a hydrazine/glycine buffer (0.4 M and 0.5 M). The plate was incubated for 5 min at 37°C. Following this, a 1-μL aliquot of the L-lactate dehydrase enzyme was added to each well and the plate was incubated again for 30 min at 37°C. After this last incubation, the absorbance was read at 340 nm in the microplate reader (Infinite M200 PRO, Tecan, Switzerland).
Meat quality
The cooking of the lobsters from all meat quality treatments (T1–T4) was carried out simultaneously. Later, each animal was weighed, and the abdominal/caudal muscular area was sampled, removing the exoskeleton, and the sample was sectioned with a knife following transverse segments indicated in the sampling scheme (Figure 1C). With the samples obtained, the exudation percentage (%), instrumental texture, sensory evaluation, instrumental colour and pH, were measured (regions 1–6; Figure 1C) on each lobster.
Exudation was determined in meat samples of 25-mm diameter by obtaining the drip loss percentage by gravimetry on a drop collector tube (112-mm length) with a funnel cup (Sasted AG & Co KG; Germany), for 24 h at 4˚C (Rasmussen & Andersson Reference Rasmussen and Andersson1996). The instrumental texture involved the calculation of cohesiveness, elasticity and chewiness, evaluated with the TA.HDplus texture meter (Stable Microsystems Ltd, UK) by applying a Texture Profile Analysis (TPA), considering a 50% compression of the sample, and using a 50-kg load cell and a P/50mm probe (Mittal et al. Reference Mittal, Nadulski, Barbut and Negi1992). The colour parameters were measured using a spectrophotometer (CM-600d; Konica Minolta Holdings, Inc, Japan), calculating in the CIELAB space (L*, a* and b*; Table S2; Supplementary material) using the D65 Illuminant and a 10° observer (MacAdam Reference MacAdam1974). Finally, the pH was measured using a portable pH meter (PH 25+; HACH SLU, Spain) with a xerolyt electrode.
Sensory evaluation of cooked meat samples was conducted by three trained panellists following preliminary training sessions. After the evaluation of each sample presented in blind conditions, a consensus score between panellists was obtained for each attribute. Since the number of panellists is low, the results of the sensory panel should be interpreted with caution; however, they still provide a useful preliminary insight into the sensory characteristics of the meat across the experimental groups. The sensory attributes and scoring system (following ISO 8586-1&2; ISO 2023) were those presented in Table 2. The primary attributes assessed included texture (hardness, juiciness, fibrosity, and chewiness) and freshness, defined by taste intensity and the presence of abnormal odours in the lobster meat segments.
Sensory attributes and scoring system used in the sensory taste panel conducted in cooked meat sample of American lobster (Homarus americanus)
Table 2. Long description
Sensory attributes and scoring system used in the sensory taste panel conducted in cooked meat sample of American lobster (Homarus americanus).
Ethical considerations
Ethical approval for experimentation with decapod crustaceans is not required by law in the European Union. All lobsters used were treated with care (both husbandry and during experimental work) in order to minimise discomfort and stress. We used the minimum required number to obtain a reasonable threshold of certainty and replication, needed for consistent results. Experimental design followed the 3Rs principle (Replacement, Reduction and Refinement). When animals were removed from the study, they were electrically stunned prior to being immediately frozen for euthanasia.
Statistical analysis
The activity before and after the CrustastunTM was analysed with a Linear Mixed-effects Model (LMM), using each animal as a random effect. To assess changes in righting times, a Kruskal-Wallis rank sum test was carried out (McKnight & Najab Reference McKnight and Najab2010). Outliers were identified by accepting the null hypothesis of the Grubb’s Test (Grubbs Reference Grubbs1969), and were removed according to quartile threshold. For the physiological changes during recovery, both glucose and L-lactate in the recovery were analysed with a Mixed Model, considering each animal as a between-subjects variable (Pinheiro et al. Reference Pinheiro, Bates, DebRoy and Sarkar2019). In the case that the interaction treatment × time was significant, a post hoc test was done to determine which times the treatments differed significantly. For the immediate response, a one-way ANOVA was carried out to analyse the differences. Pearson’s correlation tests were done for glucose and L-lactate maximum concentrations and the righting time. In order to visualise possible clustering of data and differences between experimental groups, a principal component analysis (PCA) was applied to the meat quality parameters (only for continuous parameters). Descriptive statistics were performed for these parameters, followed by a mixed linear model to find differences among treatments, with each animal included as random effect. For sensory evaluation, a non-parametric Kruskal-Wallis rank sum test was performed (McKnight & Najab Reference McKnight and Najab2010), with pairwise Wilcoxon test for post hoc differences. All statistical analyses were performed using R (version 4.2.3), except the descriptive statistics for meat quality that were analysed using SAS statistical software (SAS V.9.4, SAS Institute Inc, Cary, NC, US).
Results
Behavioural responses to CrustastunTM cycles
In order to validate the neural network’s result on righting time, videos from Short and Long treatments were reviewed. A comparison was made between the first time there was a change in orientation (‘AI-first’), the last time (‘AI-last) and the time registered by the observer for a given video (Figure S4; Supplementary material). Even though there were no differences among the three approaches (Table S3; Supplementary material), we chose the ‘AI-first’ as the righting time, to avoid underestimation.
The average activity (movement) of lobsters was similar among treatments before the animals were taken from the tanks and placed in the CrustastunTM (Figure 2A, Table S4; Supplementary material). Once animals recovered to an upright position (and were able to move), the activity levels were increased for Short and Long treatment (Figure 2B, Table S4 [Supplementary material], returning to control levels after 60 min.
Average activity of American lobsters (Homarus americanus) (A) before stunning and (B) up to 80 min after recovering upright position, and (C) righting time in the different treatments: Control/sham (green), Short/5 s (blue) and Long/10 s (red) cycle. Continuous lines denote LMM adjustments with confidence intervals as shadings. Different letters (a-c) indicate statistical differences (P < 0.05, Tables S4, S5; Supplementary material) (9 < n < 11).
Figure 2. Long description
Panel A and B are box plots with overlaid L M M adjustment lines and confidence interval shading. The y-axis represents Activity in centimeters per second (cm/s) and the x-axis represents Time in minutes (min).
* Panel A shows pre-stunning activity from -30 to -10 minutes. The green (Control), blue (Short), and red (Long) groups show overlapping activity levels between 0.2 and 0.7 cm/s with slight fluctuations.
* Panel B shows post-recovery activity from 0 to 80 minutes. At time 0, statistical markers ‘a’, ‘b’, and ‘c’ indicate significant differences, with the Long treatment (red) starting highest, followed by Short (blue) and Control (green). Over time, all groups converge toward a similar activity level near 0.5 cm/s.
* Panel C is a standalone box plot showing Righting time in minutes on the y-axis for the three treatments on the x-axis. The Control group (green) is near zero (labeled ‘a’). The Short group (blue) shows a slight increase (labeled ‘b’). The Long group (red) shows a massive increase in righting time, ranging from approximately 120 to 380 minutes (labeled ‘c’).
A legend to the right of Panel B identifies the treatments: Control (green), Short (blue), and Long (red).
Righting was immediate (~1 s) for the Control treatment (Figure 2C), which was significantly different (Table S4; Supplementary material) from the Short treatment which took circa 31.2 (± 4.8) min (Figure 2C). In turn, the Long treatments’ righting time was significantly longer (Figure 2C, Table S4; Supplementary material) than the Short treatment, taking circa 4.5 (± 1.0) h and presented higher variability.
Physiological assessments
The haemolymph glucose concentration of lobsters was similar 30 min after the CrustastunTM long cycle (Long 10 s) when compared with those in the sham (Control) treatment (Figure 3A, Table S5; Supplementary material). Contrarily, lobsters’ L-lactate levels were significantly higher (P < 0.05) in animals that underwent a Long (10 s) CrustastunTM cycle (Figure 3C, Table S3; Supplementary material).
Initial glucose (A) and lactate (C) concentration in haemolymph of American lobsters (Homarus americanus) 30 min after being placed in CrustastunTM (n = 16) under a 10-s cycle (Long, red) or sham (Control, green), and before (time 0 h) and through recovery phase (1–12 h) of lobster’s haemolymph glucose (B) and lactate (D) concentration on the sham (Control, green) and the 5-s cycle (Short, blue) of the CrustastunTM (n = 14). Different letters indicate differences between treatments (P < 0.05) and asterisks indicate differences between treatments at a given time-point (P < 0.05).
Figure 3. Long description
A four-panel figure labeled A through D.
Panel A: Box plot of initial Glucose in m M. The y-axis ranges from 0 to 0.5. Two boxes are shown: Control (green) with a median near 0.25 and Long (red) with a median near 0.18. No significant difference letters are present.
Panel B: Time-series box plot of Glucose in m M over 12 hours. The x-axis is Time in h. Control (green) shows a stable trend near 0.5 m M. Short (blue) shows a curved increase peaking around 6 hours at approximately 1.5 m M before a slight decline. The Short treatment is labeled with ‘b’ and Control with ‘a’ at the 12-hour mark.
Panel C: Box plot of initial Lactate in m M. The y-axis ranges from 1 to 4. Control (green) has a median near 1.9 labeled with ‘a’. Long (red) has a significantly higher median near 3.1 labeled with ‘b’.
Panel D: Time-series box plot of Lactate in m M over 12 hours. The y-axis ranges from 0 to 15. Control (green) remains relatively low, peaking near 4 m M. Short (blue) shows a sharp increase, peaking between 3 and 6 hours at approximately 8 to 9 m M. Asterisks above the Short treatment boxes from hours 1 to 10 indicate significant differences from the Control. At 12 hours, Short is labeled ‘b’ and Control is labeled ‘a’.
Both time-series panels (B and D) include shaded error bands around the trend lines.
Glucose and L-lactate concentration were overall significantly higher in stunned lobsters who underwent a Short (5-s) CrustastunTM cycle (Figure 3B, Figure 3D, Table S6; Supplementary material) when compared with Control (sham) ones throughout the recovery period. Moreover, lobster’s L-lactate concentration of the Short (5-s) treatment showed a significant interaction with time (Table S6; Supplementary material), presenting significant differences 1 h after stunning, which were maintained for 10 h post-stunning (Figure 3C). During the recovery, we checked for correlations between righting times and glucose (or L-lactate) maximum concentration (Figure S5; Supplementary material). There was no significant correlation between the righting time of lobsters from the Short (5-s) treatment and their maximum glucose or L-lactate concentration (Table S8; Supplementary material) during the recovery phase.
Meat quality under stunning and transport
Animals from both sexes were in intermoult (presenting hard shells) at the time of cooking. There were only two animals showing movements 1 h after electrical stunning in the Domestic/Stunning treatment (T4; Table 1). The freshness parameters (pH, cooking loss, colour and exudation) were similar among treatments (Tables 3 and S9; Supplementary material). There were differences regarding the instrumental texture analysis: the hardness (1st and 2nd cycle force) and chewiness were higher (Tables 3 and S9; Supplementary material) in treatments without stunning (T1) compared with lobsters stunned with the CrustastunTM (T2 & T4).
Meat quality parameters of American lobsters (Homarus americanus) cooked meat in the different experimental treatments (T1: Retailer/No stunning, T2: Retailer/Stunning, T3: Domestic/No stunning and T4: Domestic/Stunning). Values show the mean (± SEM) and the P-value of the mixed linear model on the evaluated parameters (n = 15). Values in bold indicate statistical significance at P < 0.05 and superscripts (a,b) indicate differences between treatments
Table 3 Long description
Meat quality parameters of American lobsters (Homarus americanus) cooked meat in the different experimental treatments (T1: Retailer/No stunning, T2: Retailer/Stunning, T3: Domestic/No stunning and T4: Domestic/Stunning). Values show the mean (± SEM) and the P-value of the mixed linear model on the evaluated parameters (n = 15). Values in bold indicate statistical significance at P < 0.05 and superscripts (a,b) indicate differences between treatments.
The principal component analysis (PCA) to study meat quality parameters of the 4 treatments (T1–T4) accounted for 47.1% of the variability in the first two components (Figure 4A, Table S10; Supplementary material). The highest separation is seen between T1 and T4, yet they overlap, not being distinguishable by any separation. The first principal component (PC1) accounted for a mainly positive correlation with Hardness (1 & 2) and Chewiness; while the second principal component (PC2) showed a negative correlation with Cohesiveness, a* and b*. The secondary principal components (PC3 & 4) added only 25.7% of the variability (Figure S6, Table S10; both Supplementary material), with no further separation among treatments (T1–4).
Principal components analysis of meat quality parameters for American lobster (Homarus americanus) stunned and simulated transported (A) and taste panel results (B) in the different experimental groups (T1: Retailer/No stunning, T2: Retailer/Stunning, T3: Domestic/No stunning and T4: Domestic/Stunning) of the abdominal section in the evaluated lobsters (n = 15). Different letters indicate differences between treatments (P < 0.05).
Figure 4. Long description
A two-panel figure labeled A and B.
Panel A is a P C A biplot. The x-axis is Standardised P C 1 (30.4 percent) and the y-axis is Standardised P C 2 (16.7 percent). Four overlapping ellipses represent treatment groups T 1 (red), T 2 (green), T 3 (blue), and T 4 (purple). Vectors originate from the center (0,0). Vectors for Hardness, Hardness 2, Chewiness, and Elasticity point toward the right (positive P C 1). Vectors for Exudation and p H point upward. Vectors for Cooking underscore loss point left. Vectors for Cohesiveness, a star, and b star point downward.
Panel B is a radar chart with six axes: Off Odour (top), Fresh Taste, Chewiness, Fibrosity, Hardness, and Juiciness. The scale ranges from 1 at the center to 5 at the perimeter. Four lines represent the treatments:
- T 1 (solid red line) scores highest in Fresh Taste and Chewiness.
- T 2 (dashed green line) scores highest in Off Odour (labeled with ‘a’).
- T 3 (dotted blue line) and T 4 (dash-dot purple line) generally cluster lower on Fresh Taste and Chewiness (labeled with ‘b’).
Small letters ‘a’, ‘b’, and ‘ab’ indicate statistical significance at various vertices.
The sensory evaluation by a specialist panel (n = 3) showed a clear separation among treatments that were stunned with the CrustastunTM (T2 & T4) and those that did not (T1 & T3) (Figure 4B, Table S9; Supplementary material). The differences were seen both on the freshness and on most of the textural parameters evaluated (Table S9; Supplementary material).
Discussion
In the present study, after stunning, lobsters were unable to right themselves for several minutes to hours, depending on the duration of the electric stunning. Lobsters subjected to electric stunning were initially unresponsive, unable to move, and their claws were inactive. Only small movements of some pleopods were observed, which could be an autonomous response. In contrast, control/sham animals placed in the aquaria tanks immediately (~1 s) righted themselves and began to move.
To induce an immediate ‘unconsciousness’ before killing (i.e. stunning), CrustastunTM, an electric stunner prototype has been used in lobsters, langoustines, crayfishes and crabs (Fregin & Bickmeyer Reference Fregin and Bickmeyer2016; Albalat et al. Reference Albalat, Gornik, Muangnapoh and Neil2022; Kells et al. Reference Kells, Perrott and Johnson2023; Neil et al. Reference Neil, Putyora and Albalat2024). The first and last studies evaluated the nerve activity, which was monitored through extracellular recordings (in vitro), while the other studies (Fregin & Bickmeyer Reference Fregin and Bickmeyer2016; Kells et al. Reference Kells, Perrott and Johnson2023) analysed neural activity in situ. Irrespective of the method, the four studies demonstrate that electrical stunning with the CrustastunTM can arrest spontaneous activity within the central nervous system (CNS), yet the recovery or behaviour after this arrest has not been studied.
We were also interested in the recovery, since after a ‘simple’ stunning method animals should remain unconscious until dispatch/slaughter (More et al. Reference More, Bicout, Bøtner, Butterworth, Calistri, Depner, Edwards, Garin‐Bastuji, Good, Gortázar Schmidt, Miranda, Saxmose Nielsen, Velarde, Thulke, Sihvonen, Spoolder, Stegeman, Raj, Willeberg, Winckler, Marano, Verdonck, Candiani and Michel2018), hence the duration should be recorded to inform the cook or consumer the time they have before cooking it without regaining consciousness. We left the lobsters to recover (while checking for their righting response) in a recirculating aquarium overnight, observing their response up to 11 h after the electrical current application, while previous studies stopped their observation at 4 h (Albalat et al. Reference Albalat, Gornik, Muangnapoh and Neil2022; Neil et al. Reference Neil, Putyora and Albalat2024). Lobsters that underwent a 10-s electroshock rightened on average after 5 h 33 min, with an animal taking 10 h 44 min to righten. On average, the righting response of H. americanus took approximately 10 times longer using the long (10-s) CrustastunTM cycle, compared to the short (5-s) cycle, suggesting a potential exponential relationship between stunning duration and righting time. According to EFSA (2004), electrical stunning of farmed fish is considered humane when current, voltage, method, and exposure time ensure immediate and sustained insensibility, ideally achieved in < 1 s.
One way to ensure the reversibility of any procedure that aims to render an animal unconscious is by checking the recovery, as the restoration of the ‘awake’ state (de Souza Valente Reference de Souza Valente2022). Behaviourally, the average activity (movement) of an animal could provide information about the recovery to an ‘awaken’ stage. In our experiment, we saw comparable activity levels before their passage through the CrustastunTM, but increased activity levels after the animals were righted. For activity levels after the CrustastunTM, we computed them only once the animals had righted, since prior to that they remained motionless on their backs or exhibited negligible movements during righting attempts. The increase in activity after both CrustastunTM cycles (Short and Long) could indicate either an attempt at escaping the actual environment or a renewed exploratory phase after mobility is regained. Although circadian activity is well characterised as a predictor of behaviour, we do not consider it as the main driver of the observed activity shift, as there is less than an hour separating the activity onset of Control (sham) animals and those exposed to the 5-s CrustastunTM cycle. In other words, we interpret the stunning procedure as having a greater influence on activity than the circadian rhythm at the time of sampling.
The use of machine learning and computer vision in aquaculture has grown exponentially, especially for automating size and weight measurements, quality grading, disease detection and species classification (Vo et al. Reference Vo, Ko, Huh and Kim2021). Additionally, behaviour monitoring has been characterised for different crustacean species (Li et al. Reference Li, Liu, Song and Wang2021), enabling the identification of behaviour such as moulting (Asif et al. Reference Asif, Gao and Zhou2018), mating (Oishi et al. Reference Oishi, Tsagaan and Nakatani2012), feeding (Aguzzi et al. Reference Aguzzi, Costa, Menesatti, García, Chiesa and Sardà2009), posture detection (Yan & Alfredsen Reference Yan, Alfredsen, Wang, Pham, Vozenilek, Zhang and Xie2017) and locomotion (Garcia et al. Reference Garcia, Sbragaglia, Masip and Aguzzi2019). Nonetheless, this technology continues to advance, allowing for fine-tuning to address critical problems, such as ensuring welfare standards. In our case, the controlled conditions facilitated the training of the neural network, achieving maximum accuracy, and allowing for a precise detection for the activity (as well as the behaviour) measurement.
As introduced, a righting response involves the activation of multiple muscle groups, which vary across crustacean species (Neil Reference Neil1985), as well as numerous interneurons connected to the statocysts (Shuranova & Burmistrov Reference Shuranova and Yuri Burmistrov2006). While we acknowledge that this is not a direct measure of consciousness (since the animals are aware of their environment prior to the response) it does indicate the recovery of full-body awareness. The righting time should not be interpreted as a ‘safe’ time window for dispatch, even if the animal appears completely immobilised. Ideally, dispatch (whether by slaughter or cooking) should occur immediately after electrical stunning. If animals show any signs of recovery, such as an attempt to right themselves, stunning should be reapplied before dispatch, as recommended by the CrustastunTM manufacturer (Mitchell & Cooper Ltd 2024).
The changes in glucose and L-lactate concentration in the haemolymph of stunned lobsters throughout the experiment indicate that they were under physiological stress, compared to control ones. Similarly, L-lactate concentration was increased by 3-fold in the haemolymph of H. gammarus and Cancer pagurus 10 min after stunning (Neil & Thompson Reference Neil and Thompson2012). However, there was a similar increase in the sham treatment, not quite indicating if the stress was due to the CrustastunTM or the manipulation of the animals during the stunning procedure (Neil & Thompson Reference Neil and Thompson2012). We consider the stress caused by the repeated haemolymph sampling in our experiment to be negligible, as glucose concentration in the control treatment remained similar to basal levels (time 0). In contrast, L-lactate levels increased over time, although to a lesser extent than in animals that were stunned. Interestingly, there was no correlation found between the glucose or L-lactate levels in the haemolymph of lobsters with righting times, meaning that the stress accumulated on these metabolites is not necessarily related to the ability of the animal to turn upright. This further supports the findings of Davis (Reference Davis1968) in H. americanus, which describes the simple neural model by which the CNS integrates the righting response. Additionally, it supports Stoner’s (Reference Stoner2009) findings that righting behaviour is not a reliable mortality predictor in the cold-stressed tanner and snow crabs. In other words, the lack of correlation found suggests that the metabolic stress observed does not directly influence the recovery of an upright position, nor that the ability to recover is linked to lower metabolic stress.
Stress responses in crustaceans are associated with changes in glucose and L-lactate concentrations. In C. pagurus the amputation of claws (de-clawing) has a short- and long-term effect on glucose and L-lactate levels (Patterson et al. Reference Patterson, Dick and Elwood2007). This increase could be explained by elevated concentration of crustacean hyperglycaemic hormone (CHH) in the haemolymph, which has been associated with various stress responses, such as emersion (Webster Reference Webster1996), exposure to heavy metals (Lorenzon et al. Reference Lorenzon, Edomi, Giulianini, Mettulio and Ferrero2004) and environmental factors, such as salinity (Camacho-Jiménez et al. Reference Camacho-Jiménez, Díaz, Sánchez-Castrejón and Ponce-Rivas2018) and temperature (Chang et al. Reference Chang, Keller and Chang1998, Prymaczok et al. Reference Prymaczok, Pasqualino, Viau, Rodríguez and Medesani2016). Since emersion stress would be present in both the sham and stunned treatment, the increases in glucose and L-lactate could be attributed to another factor: temperature. Electro-stunned Nephrops norvegicus showed an increase in body temperature from ~13°C to ~26°C after a stunning cycle of 10 s (Albalat et al. Reference Albalat, Gornik, Muangnapoh and Neil2022). Body temperature was not measured in this study; however, during post-stunning handling, a distinctive warmth was noted around lobsters’ dorsal cephalothorax when performing the experiments.
The rise in cephalothorax temperature in N. norvegicus following electro-stunning, increased melanosis development, but the overall Quality Index Method (QIM) score remained unaffected (Albalat et al. Reference Albalat, Gornik, Muangnapoh and Neil2022). The authors concluded that the increased temperature was not sustained long enough to promote accelerated microbial growth, disregarding a significant effect of stunning towards spoilage (Albalat et al. Reference Albalat, Gornik, Muangnapoh and Neil2022). Although we calculated CIELAB space (L*, a* and b*), no significant differences among treatments were detected in the abdominal muscle of H. americanus, suggesting that melanosis was absent at the time of sampling. Other parameters were consistent with the findings from N. norvegicus electro-stunning: similar pH values and comparable juiciness/moistness between electro-stunned and control (ice in their case) treatments (Albalat et al. Reference Albalat, Gornik, Muangnapoh and Neil2022). Moreover, sensory analysis of the cooked meat revealed lower scores for hardness/firmness and chewiness in electro-stunned animals; however, these differences were not statistically significant for N. norvegicus (Albalat et al. Reference Albalat, Gornik, Muangnapoh and Neil2022).
Similar to the sensory analysis, the textural properties of the lobster meat indicated that the CrustastunTM treatments resulted in greater tenderness. Crustacean meat textural properties can be affected by several factors, such as: nutrition (Luo et al. Reference Luo, Monroig, Zhou, Tocher, Yuan, Zhu, Lu, Song, Jiao and Jin2021); capture method (Albalat et al. Reference Albalat, Gornik, Beevers, Atkinson, Miskin and Neil2012); transportation (Lorenzo et al. Reference Lorenzo, Tomac, Tapella, Yeannes and Romero2021); and cooking methods (Yang et al. Reference Yang, Zhang, Jiang, Lu and Lin2023). The textural properties depend on the cross-linking of the muscle fibres (Yang et al. Reference Yang, Jiang, Zhang, Xia, Tian, Liu, Yang, Yu, Wu, Zhu, Xu, Gu, Wang and Chen2024), how much water is retained (Ling et al. Reference Ling, Xuan, Xu, Ding, Lin, Cui and Liu2021) and, in the post-harvest phase, the protein autolysis (Zheng et al. Reference Zheng, Ping, Xu, Li and Guo2022). In broiler chickens, a successful and efficient stunning process improved meat tenderness (Hayat et al. Reference Hayat, Ismail-Fitry, Kaka, Rukayadi, Kadir, Radzi, Kumar, Nurulmahbub and Sazili2024). This improvement was attributed to reduced energy expenditure by the birds during slaughter, which helped maintain a normal glycolysis rate and, consequently, minimised protein denaturation (Hayat et al. Reference Hayat, Ismail-Fitry, Kaka, Rukayadi, Kadir, Radzi, Kumar, Nurulmahbub and Sazili2024). On one hand, lobsters’ haemolymph glucose levels remained similar 30 min after the CrustastunTM shock, while their haemolymph L-lactate levels increased, potentially indicating higher protein and carbohydrate metabolism (Santos & Keller Reference Santos and Keller1993; Li et al. Reference Li, Chiu and Lee2019). However, this L-lactate accumulation may not be reflected in the muscle of lobsters, as pH levels were similar among treatments. Alternatively, compounds such as total volatile basic nitrogen (TVB-N) may accumulate and contribute to pH buffering (Lorentzen et al. Reference Lorentzen, Skuland, Sone, Johansen and Rotabakk2014). This could affect the shelf life, as odour differences are already apparent immediately after slaughter, with the stunned animals’ meat presenting with a more neutral scent. Therefore, further analyses are required to clarify the mechanisms underlying the observed changes in meat texture after stunning.
Animal welfare implications and conclusion
The present study demonstrates that the behavioural and physiological alterations after the use an electrical stunner (CrustastunTM) are reversible in both long (10-s) and short (5-s) cycles. This indicates that the technique can be applied in situations where animals are to be dispatched shortly, or when recovery after a procedure is required, such as in certain research protocols. For lobsters, the dispatch method should be applied well before the onset of righting, as this response reflects whole-body coordination and implies that the animal is aware of its orientation. Although the electrical stunning method induced metabolic stress, most of the post-harvest quality parameters remain unaffected, with the exception of textural properties. The use of CrustastunTM for stunning may represent a more humane method prior to lobster dispatch, yet further work is needed to confirm that insensibility is achieved within a 1-s stunning duration, as recommended by EFSA. Despite this limitation, the device is particularly relevant for small-scale businesses such as restaurants and retailers, due to its user-friendly interface and fixed operating settings. Furthermore, the recovery of animals when returned to seawater suggests that the stunning method is effective within a finite time window, reinforcing the need for timely dispatch to meet animal welfare standards.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/awf.2026.10112.
Acknowledgements
We thank Oscar Chic, Xavier Leal and Mercedes Blazquez for their help and assistance while mounting the experiment in the experimental aquaria zone, ZAE. We are also grateful for Loris Mahevas’ assistance while fitting and labelling images with lobsters’ polygons and trailing the AI model.
This study was supported by a company that preferred to remain uncited and co-funded by the European Union’s Horizon Europe Project 101136346 EUPAHW, the Spanish government though the “Severo Ochoa Center of Excellence” accreditation (CEX2024-001494-S). Pol Llonch received funding from the Ramon y Cajal programme of the Spanish Ministry of Science (Spain; Grant Ref. RYC2020- 029067-I). The authors also acknowledge the support of the Consolidated Research Group (2021 SGR 00461) and CERCA Programme (Generalitat de Catalunya).
Competing interests
None.
Data availability
Data and scripts are available from the CSIC repository, https://doi.org/10.20350/digitalCSIC/17520.
Open access
