The public has expressed growing concern for the well-being of fishes, including popular pet species such as the Siamese fighting fish (Betta splendens). In captivity, male Bettas behave aggressively, often causing injuries and death if housed together. As a result, they are typically isolated in small fishbowls, which has been widely criticised as cruel. To investigate the impact of keeping Bettas in these conditions, we recorded the behaviour of individual males in containers of different sizes that were either bare or enriched with gravel, large rocks, and live plants. When male Bettas were housed individually in small bowls (0.5 L) they spent less time swimming than they did when they were kept in larger aquaria (10, 38, and 208 L). Fish that were kept in enriched containers exhibited more instances of swimming. To determine if two male Bettas housed together might coexist peacefully if given enough space and cover from plants and large rocks, we quantified the behaviour of pairs of male Bettas in bare or enriched aquaria of different sizes (10, 38, 208, 378 L). Fish performed fewer approaches and aggressive displays, but not attacks, and more bouts of foraging, when in larger aquaria. This study shows that the small fishbowls typically used in pet stores suppress swimming behaviour in male Bettas and at least a 10-L aquarium is required to ensure full expression of swimming behaviour. Furthermore, even the use of very large aquaria cannot guarantee peaceful cohabitation between two males.

Finfishes are caught from the wild for food, feed (often in the form of fishmeal and oil) and bait. According to the Food and Agriculture Organisation of the United Nations (FAO), between 74 and 83 million tonnes (averaging 77 million tonnes) were caught annually in 2000–2019. Although fishes are now widely recognised as sentient beings, capture is still quantified as biomass rather than number of individuals (in contrast to wild-caught marine mammals and crocodiles; and farmed mammals and birds). Here, we estimate global numbers of wild-caught finfishes using FAO capture production (landing) tonnages (2000–2019 data) and estimates of mean individual weight at capture, based on internet-sourced capture and market weights. We estimate that between 1,100 and 2,200 billion (1.1–2.2 × 1012), or 1.1–2.2 trillion, wild finfishes were caught annually, on average, during 2000–2019. Anchoveta (Engraulis ringens) comprised 28%, by estimate midpoint. Estimated numbers in 2019, totalling 980–1,900 billion, were lower due to reduced anchoveta landings, but still represented 87.5% of vertebrate numbers killed for food or feed, as obtained or estimated from FAO data. These figures exclude unrecorded capture such as illegal fishing, discards and ghost fishing. Estimated finfish numbers used for reduction to fishmeal and oil represented 56% of the total 2010 estimate (1,000–1,900 billion), by midpoint. It is recommended that the FAO reports fish capture numbers. The welfare of wild-caught fishes, which is generally very poor during and after capture, should be addressed as part of sustainable utilisation of aquatic resources.

Global farmed finfish production increased from 9 to 56 million tonnes between 1990 and 2019. Although finfishes are now widely recognised as sentient beings, production is still being quantified as biomass rather than number of individuals (in contrast to farmed mammals and birds). Here, we estimate the global number of farmed finfishes slaughtered using FAO aquaculture production tonnages (1990–2019 data) and estimates of individual weight at killing (determined from internet searches at species and country level where possible). We relate these numbers to knowledge on humane slaughter, animal welfare law, and certification schemes. Since 1990, farmed finfish numbers killed annually for food have increased nine-fold, to 124 billion (1.24 × 1011, range 78–171 billion) in 2019. This figure does not represent the total number farmed (due to mortalities during rearing and non-food production) and is expected to increase as aquaculture expands. Our estimates indicate that farmed finfishes now outnumber the 80 billion farmed birds and mammals killed globally each year for food. The majority are produced in Asia. Inhumane slaughter practices cause suffering for most farmed finfishes. Most, 70–72%, have no legal welfare protection, and less than 1% have any fish-specific legal protection, at slaughter. The main global certification schemes in 2013–2015 accounted for 2% of slaughtered farmed finfishes. Fishes for which species-specific parameters for automated humane stunning are published comprise 20–24%. As the dominant taxa of farmed vertebrates, finfishes would benefit from better welfare if species-specific humane slaughter was defined and incorporated into laws and certification schemes.

Atlantic salmon (Salmo salar) are an economically and ecologically important fish species that interact with humans during farming, fishing and research operations. Routine handling in nets exposes fish to mesh and causes scale loss. To promote welfare and experimental refinement, a study was performed in a controlled environment to investigate the effect of net mesh type (rubber-coated or standard knotless, both bag volumes circa 7 l; mesh size: 6 mm) and the number of fish per net (capture density) on scale loss. Up to three large adult salmon (mean weight: 900 g) or 15 small smolts (mean weight: 145 g) were briefly captured in hand-nets during routine immersed-stock movement between tanks. Scales were recovered and counted from transportation containers, to establish a simple and rapid methodology. For both size grades, scale loss was generally proportional to capture density. For large adult salmon, scale loss significantly increased with capture density when knotless mesh was used, however the increase was less marked and not statistically significant for adults handled in rubber mesh. Small smolts also demonstrated significantly reduced scale loss when handled with rubber mesh, which increased gradually with capture density. In contrast, small smolts handled in knotless mesh showed greater scale loss as capture density increased. An overall reduction in scale loss with increased capture density was not shown, although the biomass loading per net used in this study were intentionally low (< 3.5 kg). This method suggests a low-tech and rapid approach to quantitatively compare net types and husbandry techniques and suggests a fundamental but simple improvement to salmonid handling in recreational and commercial operations. However, any correlation to conventional stress assays or behavioural observations remains to be established.

An estimated 0.9 to 2.5 trillion individual wild finfish, destined for human or animal consumption, are captured globally each year by commercial fisheries. The majority of wild-caught fish die either from asphyxiation or evisceration; slaughter methods considered inhumane. Humane stunning or stun/killing may improve the welfare of wild-caught fish at the time of slaughter by rendering fish immediately insensible to pain, stress and suffering. Reducing fish stress during slaughter may also improve product quality. This study systematically mapped and narratively evaluated scientific research regarding humane stunning or stun/killing of wild finfish caught for food and feed on a commercial scale. Multiple knowledge gaps were identified: Key humane stunning parameters have only been scientifically tested, in terms of fish welfare and/or flesh quality, for a minority of wild-caught fish species and stunning methods. More species-specific protocols and use of brain activity measurements to accurately assess state of consciousness on application of the stunning method are required. More scientific testing of stunning methods in commercial wild-capture settings is needed to validate findings of controlled laboratory studies and ensure the methods are practical in fisheries. Research is needed to better understand the potential economic benefits of humane stunning for fishermen. From a donor perspective this review highlights the need to support more primary research into the feasibility and implementation of humane stunning or stun/killing in wild-capture fisheries.

Globally, Siamese fighting fish (Betta splendens) continue to be sold and kept in small, barren jars or tanks, with little concern for their welfare. This study aimed to examine the impact of housing size and furnishings (i.e. live plants, refuges) on the behaviour of Siamese fighting fish, to understand optimal tank conditions. Thirteen male Siamese fighting fish were rotated between five different housing conditions: ‘jar’ (1.5 L); ‘small’ (3.3 L); ‘medium’ (5.6 L); ‘large’ (19.3 L); and ‘large-barren’ (19.3 L). All tanks had gravel and furnishings, except the large-barren tank which was devoid of these. Overall, tank size influenced behaviour. Fish were significantly more active and spent significantly less time resting and performing ‘abnormal’ behaviours (hovering and stereotypic swimming), in the large tank compared to the smaller tanks. Tank furnishings also influenced behaviour. Fishes in the large-barren tank performed more ‘abnormal’ behaviours (hovering, stereotypic swimming, interaction with the walls), compared to the large tank which had furnishings. These results suggest that the small, barren jars and tanks that Siamese fighting fish are often housed in are detrimental to their welfare, and larger, furnished tanks are more optimal. Behavioural variations were observed between the fish in this study, highlighting individual fish personality. We recommend a minimum tank size of 5.6 L for the display and sale of Siamese fighting fish, and tanks larger than this for keeping Siamese fighting fish at home. All tanks should contain gravel, live plants and refuges.

The aim of this study was to investigate the effects of gillnet soak time to gain a better understanding of fish welfare, mortality, stress, and quality (as measured as muscle haemoglobin) during experimental gillnet fishery of Atlantic cod (Gadus morhua). An experimental study was conducted in a large-scale tank at a research facility with 131 wild-caught fish in four groups with gillnet soak times of 0, 2, 12, and 24 h (23–34 fish per soak time). Longer soak time caused higher mortality, with a mortality rate of 0, 7, 18, and 25% in the 0-, 2-, 12- and 24-h groups, respectively. Blood lactate levels were significantly affected by soak time, peaking at 2 h (with the widest confidence interval) and showing their lowest concentrations at 0 and 24 h. Soak time also significantly increased blood glucose and serum cortisol levels. Magnesium, creatinine, and iron increased significantly in all groups compared with control levels, but there was no significant difference between soak times. Haemoglobin content in the loin increased significantly only after 24 h of soak time for live fish. There was no significant increase in haemoglobin in the belly as a function of soak time. However, for all soak times, the belly had significantly more haemoglobin than the loin. Physiological evidence of traumatic injuries and stress were noted prior to increased muscle haemoglobin, meaning that good quality did not necessarily equate to good welfare. However, a higher level of muscle haemoglobin is a strong indication of poor welfare.

Opercular deformity is a common morphological abnormality of laboratory and other cultured fishes, observed in a wide variety of species but with an unclear aetiology. Following observations of short opercula in stocks of Atlantic salmon (Salmo salar) reared in our laboratory, we developed a photographic key to score individual fish on a scale of 1 to 5. Inter-rater reliability was assessed as ‘almost perfect’. This visual method is quick and simple to use, can be used to score live fish in situ in tanks as well as sampled fish, does not require sophisticated equipment and provides quantitative information to investigate the aetiology of short opercula. Opercular size was scored for a cohort of in-house reared Atlantic salmon, in a time series of random samples of ≥ 30 fish (mean weights ranging from 0.8 to 299 g) over 14 months. Short opercula were first recorded during the parr stage, prevalence and severity increased as the fish grew, and the deformity was asymmetrical, occurring predominantly on the left side. Therefore, among the many potential causal factors, nipping is suggested as the primary cause of short opercula within our culture system, with asymmetry due to the clockwise current. We also present evidence that short opercula are associated with gill damage which supports this deformity being a welfare issue that merits assessment.

Humans interact with fish in a number of ways and the question of whether fish have the capacity to perceive pain and to suffer has recently attracted considerable attention in both scientific and public fora. Only very recently have neuroanatomical studies revealed that teleost fish possess similar pain-processing receptors to higher vertebrates. Research has also shown that fish neurophysiology and behaviour are altered in response to noxious stimulation. In the light of this evidence, and in combination with work illustrating the cognitive capacities of fish, it seems appropriate to respond to a recently published critique (Rose 2002) in which it is argued that it is not possible for fish to experience fear or pain and that, therefore, they cannot suffer. Whilst we agree with the author that fish are unlikely to perceive pain in the same way that humans do, we believe that currently available evidence indicates that fish have the capacity for pain perception and suffering. As such, it would seem timely to reflect on the implications of fish pain and suffering, and to consider what steps can be taken to ensure the welfare of the fish that we exploit.

Delousing treatment for salmon sea lice (Lepeophtheirus salmonis) is considered a significant welfare concern in farming of Atlantic salmon (Salmo salar), where both industry and legislative bodies prompt for better methods. Currently, the most common method is thermal delousing, where fish are crowded, pumped into a vessel and exposed to ~28-34°C for ~30 s. Physical collisions occurring as a result of a loss of behavioural control lead to acute stress. Crowding triggers vigorous escape behaviour as salmon respond not only to treatment but also to being channeled to and from the treatment zone. A sequence of events considered to cause mortality and poor welfare. The present case study was motivated by an urgent need for delousing in groups of small salmon post-smolts in experimental research. For this purpose, a simple, small-scale system for thermal delousing was constructed, including anaesthesia to alleviate behavioural responses. The anaesthetised fish showed little behavioural response to thermal treatment, strong appetite within hours, and negligible mortality. The described method is regarded as a welfare-friendly alternative to industrial delousing in smaller fish groups, for example, in experimental research. We would encourage detailed research aimed towards gaining a deeper understanding of the welfare effects of anaesthesia prior to treatment for delousing.

Slaughter is considered an important fish welfare issue. For a slaughter method to be considered humane, effective stunning that lasts until death as well as the reduction of pain and fear throughout all procedures are essential. Our objective was to investigate current techniques for slaughtering fish in Brazil. A digital questionnaire with six multiple-choice and three open questions was distributed to companies registered as fish slaughterhouses and to fish farms in 2019, addressing fish species, quantities processed and slaughter procedures adopted. From 62 facilities in 15 Brazilian states that answered the survey, nine slaughterhouses and 30 fish farms slaughtered fish, totalling 452 tonnes per day. All reported pre-slaughter stunning, and live chilling to be the most commonly cited method (82.0%), followed by electronarcosis (18.0%). Slaughtering techniques included exsanguination (38.5%) and decapitation (2.5%). For the remaining companies, no slaughter method was declared, suggesting death by asphyxiation or by further processing (59.0%). Twenty-nine companies adopted pre-slaughter fasting, lasting from 10 to 48 h. All sites worked with tilapia (100%) and 24 (82.7%) reported that they worked exclusively with this species. Other species reported were: pacu (25.6%), tambacu (17.9%), tambaqui and carp (15.3%), jundiá (12.8%), pintado and pirarucu (7.7%), matrinxã and pangassius (5.1%). Asphyxia and live chilling are not considered humane, as animals remain conscious and, thus, suffer. Results show that most establishments do not perform humane slaughter. This scenario highlights the urgent need for development and enforcement of humane fish slaughter techniques, with routine supervision and normative requirements.

This paper reviews the impact on fish welfare of a wide range of slaughter methods used commercially around the world. Because the end result of the slaughter is a food product, and because of the well-known relationship between an animal's welfare and subsequent meat quality, the effects of the slaughter methods on the quality of the flesh are also reviewed where possible. Fish slaughter methods are incredibly diverse, but fall into two broad categories: those that induce loss of sensibility slowly, and those that achieve this rapidly. This paper shows that, in general, the methods that induce loss of sensibility over a long period of time tend to impinge more on the welfare of the animal and are detrimental to the overall quality of the carcass. Methods that cause a rapid loss of sensibility result in the best welfare, providing that they are carried out correctly. They may also produce the highest quality product from the stock offish.