Addressing the welfare needs of farmed lumpfish: Knowledge gaps, challenges and solutions

Lumpfish ( Cyclopterus lumpus L.) are increasingly being used as cleaner fish to control parasitic sea lice, one of the most important threats to salmon farming. However, lumpfish cannot survive feeding solely on sea lice, and their mortality in salmon net-pens can be high, which has welfare, ethical and economic implications. The industry is under increasing pressure to improve the welfare of lumpfish, but little guidance This Sena De Silva paper. The Sena De Silva paper is an honorific title dedicated to the memory of Professor Sena De Silva, who was the founding editor of Reviews in Aquaculture and a globally renowned aquaculture scholar, pioneer and advocate. The title is awarded to high quality articles that excel in one, or more, of the following qualities: i) Novelty and originality; ii) Likelihood of direct positive impacts for the aquaculture sector, with keen focus on any of, or all three: environmental sustainability, economic viability, and social responsibility iii) Overall quality of scientific reasonings coupled with real- world applicability. Participants highlighted fin erosion and body damage as the most useful and practical operational welfare indicators, and blood parameters and behavioural indicators as the least practical. Species profiling revealed profound differences between Atlantic salmon and lumpfish in relation to behaviour, habitat preferences, nutritional needs and response to stress, suggesting that applying a common set of welfare standards to both species cohabiting in salmon net- pens may not work well for lumpfish. Our study offers 16 practical solutions for improving the welfare of lumpfish and illustrates the merits of the Delphi approach for achieving consensus among stakeholders on welfare needs, targeting research where is most needed and generating workable solutions.


| INTRODUC TI ON: THE NEED FOR WELFARE IND IC ATOR S FOR LUMPFIS H
There are ~250 species of fish farmed globally 1 which account for 52% of fish used for human consumption. 2 Yet, despite the scale of the fish farming industry-and evidence that fish are sentient, very little is known about the welfare needs of most farmed fish. [3][4][5] Specific welfare standards and Animal Health Codes have been developed for some intensively farmed species such as the Atlantic salmon (Salmo salar) [6][7][8] and the rainbow trout (Oncorhynchus mykiss), 9,10 but only generic guidelines exist for most farmed fish. 11,12 For most farmed fish, knowledge on their welfare needs typically lags behind advances on production.
The Atlantic lumpfish (Cyclopterus lumpus L.) is a case in point.
Lumpfish are a novel species to marine aquaculture whose farming has increased exponentially over the last 10 years and represents one of the fastest-growing aquaculture sectors in Europe. 13 Unlike most farmed fish, lumpfish are not farmed for human consumption, but are used (along with some temperate wrasse) as cleaner fish to control parasitic sea lice (Lepeophtheirus salmonis). [13][14][15] Sea lice are one of the major threats to salmon farming 16 as it causes huge economic losses 17 and compromises the welfare of wild and farmed salmon alike, 18,19 tarnishing public's perception of salmon farming. 20,21 Sea lice have developed resistance to most anti-parasitic chemical compounds, 22 prompting an interest in the use of cleaner fish as a 'green' alternative to chemotherapeutants. 13 However, lumpfish survival in salmon net-pens is often poor, and there is increasing concern regarding their welfare. 23,24 Studies have shown that lumpfish mortalities can reach 27% or more shortly after deployment in salmon net-pens [25][26][27][28][29][30][31] although the reasons for this are not clear. 30 Emaciation, stress, diseases and poor knowledge of the specific nutritional and habitat requirements have been flagged as some of the main welfare challenges for the species. [32][33][34] Many of these problems stem from the fact that salmon farming has evolved to suit the needs of Atlantic salmon, not those of cleaner fish. For example, exposed sites that may be appropriate for salmon may have currents that are too strong for lumpfish. 35 The soundness of using cleaner fish to control sea lice is also being questioned on efficacy grounds, 36,37 as delousing efficacy varies enormously among studies, from 9% to 97% in lumpfish. 37,38 Yet, until more powerful preventive methods are developed, ie vaccines, artificial selection, 39 the judicious use of cleaner fish will continue to offer the industry an attractive and cost-effective means of controlling sea lice, 38,40 but only if cleaner fish welfare is not compromised. 13 Most Europeans (79%) want better welfare in the salmon farming industry 41 and are willing to pay more for the salmon they eat, 42 but it is essential that solving one problem for salmon farming (sea lice control) does not create a welfare problem for lumpfish. The cost of poor fish welfare in salmon farms has recently been estimated at $4.6 billion, 43 but sacrificing one species for another species' welfare cannot be the way forward. 44 The welfare of lumpfish cannot be improved without welfare indicators. These need to be practical and easy to use, or they will not be used by fish farmers. 45,46 Operational welfare indicators (OWIs) are designed to address this practical need (as opposed to laboratory-based indicators) and should provide an objective assessment of the welfare of the fish that can easily be scored at the farm, 6,47 can be used to benchmark farming operations and can serve to identify areas in need of improvement and develop best exists on how this can be achieved. We undertook a knowledge gap and prioritisation exercise using a Delphi approach with participants from the fish farming sector, animal welfare, academia and regulators to assess consensus on the main challenges and potential solutions for improving lumpfish welfare. Consensus among participants on the utility of 5 behavioural and 12 physical welfare indicators was high (87-89%), reliable (Cronbach's alpha = 0.79, 95CI = 0.69-0.92) and independent of participant background. Participants highlighted fin erosion and body damage as the most useful and practical operational welfare indicators, and blood parameters and behavioural indicators as the least practical. Species profiling revealed profound differences between Atlantic salmon and lumpfish in relation to behaviour, habitat preferences, nutritional needs and response to stress, suggesting that applying a common set of welfare standards to both species cohabiting in salmon net-pens may not work well for lumpfish. Our study offers 16 practical solutions for improving the welfare of lumpfish and illustrates the merits of the Delphi approach for achieving consensus among stakeholders on welfare needs, targeting research where is most needed and generating workable solutions.

K E Y W O R D S
cleaner fish, Delphi expert assessment, feeding rations, habitat preferences, operational welfare indicators, Salmon farming practices, as done for Atlantic salmon, 6 ballan wrasse 48 and rainbow trout. 9 However, assessing the welfare of lumpfish poses particular challenges, caused mostly by the lack of agreed guidelines. Some welfare indicators have recently been developed for lumpfish, 27,49,50 but only a few have been validated and can easily be used by fish farmers. 49,51 The benefits of having agreed welfare standards and guidance for improving the welfare of lumpfish are multiple. For example, operational welfare indicators are now stipulated in quality assurance schemes, without which certification cannot be made. Achieving high welfare makes economic sense, as fish with deformities and low welfare do not survive as well. 51,52 Reducing the incidence of lumpfish with deformed suckers, and skeletal deformities, 53 may improve delousing efficacy and consequently reduce the number of cleaner fish required by industry. Higher welfare should also result in less stressed lumpfish which might reduce the risk of pathogen crosstransmission from one species to the other. 13,54 Ultimately, improving the welfare of lumpfish will help increase the sustainability, social acceptance and reputation of the salmon farming industry, but only if agreed, evidenced-based welfare standards are used to demonstrate improvements.

| AIMS
The aims of this study were threefold: (1) to identify the main challenges and knowledge gaps surrounding lumpfish welfare, (2) to offer potential solutions and identify opportunities for improving lumpfish welfare and (3) to assess the degree of consensus among different stakeholders on the value of different welfare metrics.

| US ING THE DELPHI APPROACH TO IDENTIF Y CHALLENG E S AND PRIORITIS E SOLUTIONS
We hosted a workshop dedicated to Lumpfish Welfare (Swansea, 14 May 2019-https://www.welfa reaqu acult ure.com/1st-sympo sium) with the participation of 53 experts from three main stakeholder groups: fish farming, academia and animal welfare. We employed a Delphi approach to identify the main challenges, opportunities and potential solutions for improving lumpfish welfare and for making their use in salmon farming more sustainable. The Delphi approach seeks to harness the value of expert judgment through consultation rounds and is based on the tenet that the views of a group are more authoritative (and thus more likely to gain support and become adopted) than the views of individuals. This approach is increasingly being used to prioritise knowledge needs in fields where opinions may differ among knowledge users, such as aquaculture 55 and animal welfare. 56 The classic Delphi approach is entirely anonymous, but we used a 'modified Delphi approach' that benefitted from group discussions, as in other applications to animal welfare. [56][57][58] Consultation followed by open discussion does not guarantee consensus, or can be a substitute for research, but it can help identify what experts consider important and reduce the risk that key issues are not being missed.
A three-step process was implemented ( Figure S1), similar to that described in other prioritisation assessments. 55 In the first step, participants were divided into 10 pre-allocated tables consisting of 4-5 people representative of the three stakeholder groups, and a facilitator. Each participant was asked to write (in coloured notes and independently of each other) a list of the main knowledge gaps, opportunities and solutions for improving the welfare and sustainability of lumpfish. In the second step, each table was asked to find common answers and to reach a consensus on the three most common answers. In the third step, each table reported their answers to the whole group, all colour coded notes were displayed and the most popular answers were identified and compiled. Steps 2 to 3 were repeated for the three aspects pertinent to lumpfish welfare

| PERCEIVED UTILIT Y OF D IFFERENT WELFARE INDIC ATOR S
To assess the perceived utility of different welfare indicators for lumpfish (i.e. their usefulness under farm conditions), a close-ended questionnaire was given to participants (Table S1)

| D IFFEREN CE S IN THE NI CHE OF LUMPFIS H AND ATL ANTI C SALMON
Lumpfish are deployed with Atlantic salmon but conditions that may be suitable for one species may not be adequate for the other.
We, therefore, compiled data on 23 traits that define the habitat TA B L E 1 Challenges (C), solutions (S) and opportunities (O) for improving the sustainability and welfare of farmed lumpfish identified by a focus group, weighted by their relative frequency (LF = lumpfish).

Challenges (n = 40) Weight
Knowledge  51 In terms of husbandry and logistics, participants highlighted problems surrounding the optimal timing and frequency of grading, the correct use of shelters 73  Other challenges highlighted by the focus group included lack of standard operating procedures (SOPs) and guidance for collecting data, particularly in sea cages (including data on mortalities and delousing efficacy), the need for staff training and guidance on monitoring and assessing body condition and optimal weight as condition factor is unreliable for this species owing to its round shape. 51 Lumpfish are often injured or stressed unnecessarily when salmon are treated and harvested at sea, 7,29 and these, along with their humane slaughter, are the aspects that would also benefit from having clear SOPs.
The impacts of noise pollution, potential water quality issues due to high microbial loads 76,77  were the third most cost-effective control measure (£0.14 per fish per unit of effectiveness), after in-feed medication and use of skirts as physical barriers (£0.10 cost-effectiveness).

| Solutions
The focus group identified 40 potential solutions to the challenges highlighted above ( Table 1)

| Opportunities
Participants identified 43 opportunities to improve the welfare of lumpfish and to make their use as cleaner fish more sustainable, mostly through research and development and through technological improvements ( Table 1) Stratification may confine fish to some parts of the net-pens Sit and wait visual feeders Require eyes in good condition for feeding Food particle density must be high and pass close by May not be able to feed at low light levels Need to adjust feeding ration frequently

| S PECIE S PROFILING : HOW UNI QUE ARE LUMPFIS H?
A search of the literature was undertaken to bridge some of the knowledge gaps highlighted by the focus group (Table 2) and to explore the potential welfare implications and consequences for production ( Gaps in knowledge relevant to welfare are also evident with regards to optimal densities, tank flows, nutritional requirements and recommended feeding rations. Some studies suggest that densities of <60 kg/m 3 or even <40 kg/m 3 should be maintained, 27 Similarly, there is uncertainty about optimal feeding rations. Larvae (<0.5 g) are being fed at 5-10% body weight, 10 g juveniles at 3-4% and juveniles just before deployment at 1.5-2%. 13,104 This level of feeding should result in specific growth rates of 1.5-3.5% per day in hatcheries. 103 However, traditional ways to detect underweight fish in fish farming, like the use of Fulton's condition, do not work well in lumpfish owing to their round shape and different growth stanzas, and there is little information on feed conversion ratios. It is suspected that current feeding rations may be too high and lead to wastage. Overfeeding has been associated with a higher incidence of cataracts in lumpfish, 13,108 so more precise information on appropriate feeding levels is obviously needed. Feeding in lumpfish depends on prey density and metabolic rate 109 , but how this translates into guidance on feeding management merits further investigation.
Ensuring lumpfish are fed adequately is particularly important in sea cages, as the species is at risk of malnutrition 51,110 and cannot survive grazing solely on sea lice. The lack of agreed standards for feeding lumpfish in sea cages was highlighted as one of the main knowledge gaps and is made worse by the difficulty of obtaining accurate data on the number of lumpfish actually present in sea cages.
In this sense, advances in tracking methods and in fish image recognition may help to obtain more accurate estimates of fish biomass, both in hatchery tanks and in sea cages. 100,111

| D IFFEREN CE S B E T WEEN ATL ANTI C SALMON AND COHAB ITING LUMPFIS H
One of the problems highlighted by the focus group was the tendency by farmers to prioritise the welfare needs of Atlantic salmon over those of lumpfish. We, therefore, asked whether conditions that favour salmon might also benefit lumpfish. Although uncertainties exist, our comparative analysis reveals profound differences between the two species that will likely have welfare implications (Table 4). Compared with salmon, lumpfish have a lower cortisol response and metabolic rate, are solitary, have a weaker swimming ability and a reactive response to threat; they also prefer colder F I G U R E 6 Species differences in habitat preferences and selected life-history traits between lumpfish and Atlantic salmon (% difference in log 10 scale) [Colour figure can be viewed at wileyonlinelibrary.com] and deeper waters, are much less domesticated, live longer, feed lower in the food web, grow more slowly, and are generally closer to the r end of the r-K life-history continuum (sensu Pianka 112 ) than salmon, that is they have a life strategy characterised by high fecundity, rapid development, broad niche and density-independent mortality, better suited to living in highly variable and unpredictable habitats. Some of these differences vary by over three orders of magnitude ( Figure 6). This means that a common 'one-size-fits-all' approach to ensuring high welfare in salmon net-pens will not work for lumpfish, particularly in relation to habitat preferences (water current, depth), feeding and the response to stress. Yet, we trust that the approach shown in our study can be used to prioritise the welfare needs of other cleaner fish species and identify workable solutions.

| 16 PR AC TI C AL WAYS TO IMPROVE THE WELFARE OF LUMPFIS H
Based on the advice of the expert group, and our comparative analysis, we suggest the following practical ways of improving the welfare of lumpfish used as cleaner fish to control sea lice in salmon farming: 1. Adopt welfare guidelines specifically developed for this species 49

Slaughter lumpfish humanely
Many of the recommendations listed above will also apply to cleaner wrasse used to control sea lice in salmon farming, but it must be remembered that cleaner fish differ in their behaviour and habitat requirements, 65 and likely also in their welfare needs. The approach shown in our study can be used to prioritise such needs and propose workable solutions.

ACK N OWLED G EM ENTS
We wish to thank all the speakers and participants in the First Symposium on Welfare in Aquaculture and the Welfare of Lumpfish Workshop. The financial support of the ERDF SMARTAQUA Operation, INTERREG Atlantic Area Access2Sea, the Animal Welfare Research Network, BBSRC ARCH-UK, BioMar, Swansea University