Content uploaded by Colin E. Adams
Author content
All content in this area was uploaded by Colin E. Adams on Oct 04, 2018
Content may be subject to copyright.
REVIEW PAPER
Current issues in fish welfare
F. A. HUNTINGFORD*†, C. ADAMS*, V. A. BRAITHWAITE‡,
S. KADRI*, T. G. POTTINGER§, P. SANDØE{AND
J. F. TURNBULL**
*Fish Biology Group, Institute of Biomedical & Life Sciences, Graham Kerr Building,
University of Glasgow, Glasgow, G12 8QQ, UK., ‡Institute of Evolutionary Biology,
University of Edinburgh, Kings Buildings, Edinburgh, EH9 3JT, U.K., §NERC
Centre for Ecology & Hydrology, Lancaster Environment Centre, Library Avenue,
Bailrigg, Lancaster, LA1 4AP, U.K., {Centre for Bioethics and Risk Assessment,
Royal Veterinary and Agricultural University, Copenhagen, Denmark and
**Institute of Aquaculture, University of Stirling, Stirling, FK9 4LA, U.K.
(Received 30 September 2005, Accepted 29 November 2005)
Human beings may affect the welfare of fish through fisheries, aquaculture and a number of
other activities. There is no agreement on just how to weigh the concern for welfare of fish
against the human interests involved, but ethical frameworks exist that suggest how this might
be approached.
Different definitions of animal welfare focus on an animal’s condition, on its subjective
experience of that condition and/or on whether it can lead a natural life. These provide
different, legitimate, perspectives, but the approach taken in this paper is to focus on welfare
as the absence of suffering.
An unresolved and controversial issue in discussions about animal welfare is whether non-
human animals exposed to adverse experiences such as physical injury or confinement experi-
ence what humans would call suffering. The neocortex, which in humans is an important part of
the neural mechanism that generates the subjective experience of suffering, is lacking in fish and
non-mammalian animals, and it has been argued that its absence in fish indicates that fish
cannot suffer. A strong alternative view, however, is that complex animals with sophisticated
behaviour, such as fish, probably have the capacity for suffering, though this may be different
in degree and kind from the human experience of this state.
Recent empirical studies support this view and show that painful stimuli are, at least, strongly
aversive to fish. Consequently, injury or experience of other harmful conditions is a cause for
concern in terms of welfare of individual fish. There is also growing evidence that fish can experience
fear-like states and that they avoid situations in which they have experienced adverse conditions.
Human activities that potentially compromise fish welfare include anthropogenic changes to
the environment, commercial fisheries, recreational angling, aquaculture, ornamental fish
keeping and scientific research. The resulting harm to fish welfare is a cost that must be
minimized and weighed against the benefits of the activity concerned.
Wild fish naturally experience a variety of adverse conditions, from attack by predators or
conspecifics to starvation or exposure to poor environmental conditions. This does not make it
acceptable for humans to impose such conditions on fish, but it does suggest that fish will have
†Author to whom correspondence should be addressed. Tel.: þ44 (0) 141 330 5975; fax: þ44 (0) 330
5971; email: f.a.huntingford@bio.gla.ac.uk
Journal of Fish Biology (2006) 68, 332–372
doi:10.1111/j.1095-8649.2005.01046.x,available onlineathttp://www.blackwell-synergy.com
332
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles
mechanisms to cope with these conditions and reminds us that pain responses are in some cases
adaptive (for example, suppressing feeding when injured).
In common with all vertebrates, fish respond to environmental challenges with a series of
adaptive neuro-endocrine adjustments that are collectively termed the stress response. These in
turn induce reversible metabolic and behavioural changes that make the fish better able to
overcome or avoid the challenge and are undoubtedly beneficial, in the short-term at least.
In contrast, prolonged activation of the stress response is damaging and leads to immuno-
suppression, reduced growth and reproductive dysfunction. Indicators associated with the
response to chronic stress (physiological endpoints, disease status and behaviour) provide a
potential source of information on the welfare status of a fish. The most reliable assessment of
well-being will be obtained by examining a range of informative measures and statistical
techniques are available that enable several such measures to be combined objectively.
A growing body of evidence tells us that many human activities can harm fish welfare, but
that the effects depend on the species and life-history stage concerned and are also context-
dependent. For example, in aquaculture, adverse effects related to stocking density may be
eliminated if good water quality is maintained. At low densities, bad water quality may be less
likely to arise whereas social interactions may cause greater welfare problems.
A number of key differences between fish and birds and mammals have important implica-
tions for their welfare. Fish do not need to fuel a high body temperature, so the effects of food
deprivation on welfare are not so marked. For species that live naturally in large shoals, low
rather than high densities may be harmful. On the other hand, fish are in intimate contact with
their environment through the huge surface area of their gills, so they are vulnerable to poor
water quality and water borne pollutants.
Extrapolation between taxa is dangerous and general frameworks for ensuring welfare in
other vertebrate animals need to be modified before they can be usefully applied to fish.
The scientific study of fish welfare is at an early stage compared with work on other vertebrates
and a great deal of what we need to know is yet to be discovered. It is clearly the case that fish, though
different from birds and mammals, however, are sophisticated animals, far removed from unfeeling
creatures with a 15 s memory of popular misconception. A heightened appreciation of these points in
those who exploit fish and in those who seek to protect them would go a long way towards improving
fish welfare. #2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles
Key words: aquaculture, fisheries, ornamental fish, pain, stress, welfare.
INTRODUCTION
The aim of this review, which arose from a briefing paper prepared for the
Fisheries Society of the British Isles (http://www.le.ac.uk/biology/fsbi/brief-
ing.html), is to give a broad overview of the current understanding of a number
of issues relating to fish welfare, an area of increasing public concern. The term
‘fish’ includes animals of very different taxonomic status and in this review we
mostly consider teleost fish, since these have been the subject of almost all recent
research into fish welfare. A broad approach necessarily precludes in-depth,
exhaustive coverage of all the relevant issues, but many of these issues have
been the subject of recent published reviews and we cite these in the relevant
sections. We briefly address what welfare means, why it matters and how welfare
science relates to the philosophical discipline of ethics, before considering human
activities that may compromise fish welfare and how welfare might be measured.
We concentrate on the impact of human activity on welfare at the level of
individuals, as opposed to populations, species or ecosystems and address the
ISSUES IN FISH WELFARE 333
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
experiences of living animals (up to and including the point of slaughter) and not
the question of whether it is right to kill animals.
To discuss animal welfare objectively, we need a definition and this is not easy
to produce because the concept is complex and the word is used in a number of
different ways (Dawkins, 1998; Appleby, 1999). Most definitions fall into one of
three broad categories (Duncan & Fraser, 1997; Fraser et al., 1997), none of
which is right or wrong from a scientific point of view; rather they express
different ideals about what we should be concerned about in our dealings with
animals:
Feelings-based definitions are set in terms of subjective mental states. Here, the
requirement for good welfare is that the animal should feel well, being free from
negative experiences such as pain or fear and having access to positive experi-
ences, such as companionship in the case of social species. This use of the term
welfare obviously depends on the animal concerned having conscious subjective
experiences and our ability to interpret such experiences, controversial points
(Dawkins, 1998) that are discussed below.
Function-based definitions centre on an animal’s ability to adapt to its present
environment. Here good welfare requires that the animal be in good health with
its biological systems (and particularly those involved in coping with challenges
to stasis) functioning appropriately and not being forced to respond beyond their
capacity. This definition is based on things that are relatively easy to observe and
measure.
Nature-based definitions arise from the view that each species of animal has an
inherent biological nature that it must express. Here good welfare requires that
the animal is able to lead a natural life and express its natural behaviour. This
approach, which reflects a view that what is natural is inherently good, focuses
on something we can measure, namely what animals do in the wild and in
captivity.
Because suffering, health problems and impairment of natural behaviour often
accompany each other, in many cases the above mentioned three approaches will
reach the same conclusions. Chickens (Gallus domesticus) are strongly motivated
to build nests (as opposed to having access to a completed nest) and will work
hard for the opportunity to build (Hughes et al., 1989); arguably then, nest-
building reflects a behavioural need that must be met if the chicken’s welfare is
not to be compromised. In some cases, however, different conclusions about
whether welfare is compromised will follow from the different definitions. For
example, much behaviour of wild animals is shown in response to adverse
conditions (as when fleeing from a predator), but it is hard to argue that feelings
of suffering will occur if these responses are not evoked. In other cases, animals
may be highly motivated to perform an action independent of its consequences.
Their welfare may be compromised if they are deprived of the opportunity to do
so, but this is not necessarily the case and it may be difficult to decide whether
the different approaches lead to the same conclusion. For example, wild Atlantic
salmon (Salmo salar L.) migrate long distances at sea. If this happens because
fish leave an area when the local food supply is poor and stop swimming when
they find food, there is no reason to believe that farmed salmon suffer ill effects
when they are prevented from migrating, provided they have plenty of food. If
they are simply motivated to swim, then swimming in large circles may be
334 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
sufficient. On the other hand, if they have an instinctive drive to move to new
areas regardless of food supply, confinement in cages might well lead to suffer-
ing, even though fish are able to swim continuously.
It is, therefore, important to state clearly the definition of animal welfare that
is being used (Appleby & Sandøe, 2002). In this article, we adopt a feelings-based
approach that focuses on animal suffering that is on more-or-less intense unpleas-
ant mental or physical states felt by the animal. One important complicating
factor is that the occurrence of unpleasant states does not by itself imply suffer-
ing. Such states are an unavoidable part of normal animal life and often serve as
signals or behavioural prompts that help the animals satisfy their biological
needs. Sometimes, negative experiences are compensated for by corresponding
positive experiences, so suffering may be defined as prolonged experience of
unpleasant mental states.
SCIENCE, ETHICS AND WELFARE
Humans may affect the welfare of fish in many ways, through fisheries,
aquaculture, sports fishing, scientific research or the keeping of fish as a hobby
activity, all of which have associated benefits (see Section 8). There is therefore
every reason to seek a better empirical understanding of fish welfare and to give
careful thought to how we should weigh the welfare of fish against the interests
of humans when these are in conflict. Their disciplinary training gives biologists
a special role to play in the first of these aims; but not in the second, which is the
job of the moral philosopher and ethicist. To put it another way, biologists may
be able to tell us whether the welfare of fishes is compromised by a certain
human activity and even perhaps by how much, but normally they have neither
the expertise needed for nor the responsibility of deciding whether that human
activity is justified. Welfare biology and the ethical analyses of animal use,
however, share the goal of elucidating the relationship between humans and
other animals (Fraser, 1999). We describe very briefly some frameworks that
ethicists, typically scholars with a background in philosophy, have developed to
guide clear thinking on the complex moral issues of whether and when humans
have the right to make use of animals for food production, sports and hobbies
(e.g. Rollin, 1993; Sandøe et al., 1997; Heeger & Brom, 2001). It should be noted
that, like the biologist, the ethicist does not have the right to decide what is right
or wrong; rather their expertise allows them to deliver thoughts that may be
relevant when we try to think clearly about these issues.
Thinking clearly is not always the fashion when it comes to animal issues.
Often feelings without much thought seem to prevail and this has its problems.
The first problem of being led by one’s feelings rather than approaching matters
through ethical theory is simply that people’s feelings about animal use are often
unstable or ambivalent and so cannot be relied upon as a rational guide. This
immediately leads to a second problem, namely that ambivalence encourages
double standards that are both morally objectionable and logically indefensible.
The third problem is perhaps the most serious. It is clear that, at present, we are
engaged in Europe and North America in an increasingly serious debate about
the rights and wrongs of animal use. As long as they merely press their intuitively
held beliefs, people on either side of the debate about making use of fish as a
ISSUES IN FISH WELFARE 335
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
resource will not be able to communicate effectively. These beliefs are often
sincere and strongly held, but they can be extremely difficult to understand and
highly resistant to change. The ideal of meaningful and transparent discussion
leading to mutual understanding is attainable, however, because people’s feelings
about matters are very often based on underlying ethical assumptions and
theories, which are more susceptible to rational assessment than the individual
beliefs to which they give rise. The suggestion we wish to make here, then, is that
if lay people and scientists are willing to think a little about fundamental ethical
theory, they will have a much greater prospect of communicating with one
another effectively, articulating their convictions in a coherent manner, and
perhaps even reaching a compromise upon which all can agree.
Moral philosophers distinguish a number of types of ethical theory, and in
principle any of these might underlie a person’s views about animal use by
humans. Here we will discuss three prominent theoretical positions: contractar-
ianism, utilitarianism and rights views. These have been selected because they
have direct and obvious implications for the ongoing debate over animal use.
CONTRACTARIANISM
Why should we act morally? This is a central question in moral philosophy,
and one to which the contractarian gives a straightforward answer: one should
act morally because it is in one’s self-interest to do so. The outlook underlying
contractarianism is egoism. According to the egoist, when one is obliged to show
consideration for other people this is really for one’s own sake. In general, by
respecting the rules of morality one contributes to the maintenance of a society
that is essential to one’s own welfare. The moral rules are thus those that best
serve the self-interest of all members of the society. Contractarian morality is
confined to those individuals who can ‘contract in’ to the moral community, so it
is important to define who these members are. On this topic, Narveson (1983)
writes: ‘‘On the contract view of morality, morality is a sort of agreement among
rational, independent, self-interested persons, persons who have something to
gain from entering into such an agreement . . .’’
A major feature of this view of morality is that it explains why it occurs it and
who is party to it. This Morality occurs for reasons of long-term self-interest,
and parties to it include all and only those who have both of the following
characteristics: 1) they stand to gain by subscribing to it, at least in the long
run, compared with not doing so, and 2) they are capable of entering into (and
keeping) an agreement. Given these requirements, it will be clear why animals do
not have rights, for there are evident shortcomings on both scores. On the one
hand, humans have nothing generally to gain by voluntarily refraining from (for
instance) killing animals or ‘treating them as mere means’. On the other, animals
cannot generally make agreements with us, even if this was deemed desirable.
On this view there is clearly a morally relevant difference between one’s
relationship to other human beings and one’s relation to animals. We are
dependent on the respect and co-operation of other people. If we treat our fellow
humans badly, they will respond by treating us badly. By contrast, the animal
community will not strike back if, let us say, we use some of its members to hunt
or fish for the fun of it. From an egoistic point of view we need only treat the
336 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
animals well enough for them to be fit for our own purposes. In any case, as
Narveson (1983) points out, non-human animals cannot enter into a contract, or
agreement, governing future conduct, so they cannot join the moral community.
For the contractarian, since neither animal suffering nor the killing of animals
is an ethical problem per se, any kind of animal use is in itself ethically accep-
table. The lack of standing of animals in the moral community does not neces-
sarily mean that the way animals are treated is irrelevant from the contractarian
point of view: if people like animals, for example, and dislike the practice of their
being used in this or that way, animal use can become an ethical issue, because it
is in a person’s interests to get what he or she likes. Nevertheless the contrac-
tarian view of animals is highly anthropocentric, since any rights to protection
afforded to animals will always be dependent on human concern. Inevitably, we
tend to like some types of animal more than others and are more troubled by the
suffering of our favourite types of animal. Hence, levels of protection will differ
across different varieties of animal.
The contractarian view accords with certain attitudes to animal treatment that
are prevalent in our society. Thus it serves to explain why legislation, allegedly
for the protection of animals, usually protects the animals that matter most to
humans, such as cats and dogs. Contractarianism can, however, seem inade-
quate. Can it really be correct to hold that causing animals to suffer, even for a
trivial reason, or for no particular reason, is morally straightforward as long as
no human being is bothered by the relevant conduct? Many would want to insist
that it is immoral as such to cause another to suffer for little or no reason,
whether one’s victim is a human being or an animal. An ethical theory that
captures this insistence is utilitarianism.
Typically people have less strong feelings about fish and therefore fish are less
well protected than other kinds of animals. According to this approach, one
might have legitimate concern for the welfare of fish in an ornamental tank, since
the pleasure one gets from them might be reduced if they were in a poor state of
welfare. Fish biologists might have a concern for the welfare of their subjects,
since poor welfare may well equate to poor science. In both cases we need to be
able to recognize and measure good v. poor welfare.
UTILITARIANISM
According to the utilitarian, the interests of every individual affected by an
action count morally and deserve equal consideration. In utilitarian writings the
notion of an interest is usually defined in terms of ‘‘the capacity for suffering
and/or enjoyment or happiness’’ (Singer, 1989). Thus individuals have an interest
in acts that will enhance their enjoyment or reduce their suffering. From this it
follows, of course, that all sentient beings, human and non-human, have inter-
ests. And since for the utilitarian all interests count morally and deserve equal
consideration, this implies that the impact of one’s actions on all sentient
creatures, including animals, is a matter of moral concern. Thus Singer (1989)
writes: ‘‘Many philosophers have proposed the principle of equal consideration
of interests, in some form or other, as a basic moral principle; but . . . not many
of them have recognized that this principle applies to members of other species
as well as to our own. . . . If a being suffers, there can be no moral justification
ISSUES IN FISH WELFARE 337
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
for refusing to take that suffering into consideration. No matter what the nature
of the being, the principle of equality requires that its suffering be counted
equally with the like suffering –in so far as rough comparisons can be made
–of any other being.’’
For the utilitarian, then, ethical decisions require us to strike the most favour-
able balance of benefits and costs for all the sentient individuals affected by what
we do. Doing the right thing, according to the utilitarian, is not only a matter of
doing what is optimal. It is also essential to do something rather than nothing: if
something can be done to increase well-being, we have a duty to do it. This
utilitarian duty to act always to bring about improvements has important con-
sequences for society. In contemporary European and North American society
we have a general tendency to give ourselves priority over animals. A utilitarian
will regard this tendency as essentially wrong. However, the anthropocentric
outlook is obviously well established, and in view of this it may well be that, for
the time being at least, any attempt to ensure that sentient animals are accorded
the same status as human beings is bound to fail. It may be that the best thing a
utilitarian can do is to secure higher levels of animal welfare within the current
system. It may here be relevant to mention that the Journal of Fish Biology
requires its authors to refer to the Association for the Study of Animal
Behaviour’s Guidelines for the Use of Animals in Research (2001), which are
firmly based in such a semi-utilitarian approach.
The overwhelming majority of domestic animals are kept for food production.
Most are kept under restrictive conditions in which basic behavioural or phy-
siological needs are thwarted. Laying hens, for example, are commonly kept in
battery cages where they cannot perform strongly motivated nesting behaviour
before egg laying and where the restriction of their movement results in bone
brittleness and a high incidence of broken bones. Farmed fish too are kept at
high densities. Whether or not their physiological and behavioural needs are
thwarted remains unclear. Naturally, such costs must be weighed against the
benefit, to human beings, of access to cheap meat and eggs. But given that the
average citizen in the developed world consumes far more protein than is
physiologically necessary, and often more animal fat than is healthy, low-cost
meat cannot be considered a vital human interest.
What all utilitarians agree on, however, is the methodological precept that
ethical decisions concerning animal use require us to balance the harm we do to
the affected animals against the benefits we derive for humans and other ani-
mals. Interestingly, this very precept (i.e. the notion that we can work out what is
ethical by trading off one set of interests against another) has been attacked by
some moral philosophers. The allegation is that such trade-offs violate the rights
of the individuals whose interests are in the moral balance. To clarify this point it
is necessary to turn to rights theories.
RIGHTS VIEWS
There is an obvious sense in which, in focusing on overall improvements in
welfare, the utilitarian treats sentient beings as mere instruments. The utilitarian
believes that it is ethically justifiable to sacrifice the welfare of one individual
where this sacrifice is outweighed by connected gains in welfare. Rights theorists
338 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
object to this, holding that it is always unacceptable to treat a sentient being
merely as a means to obtain a goal.
Historically, rights theory is associated with the eighteenth-century German
philosopher, Immanuel Kant. In Kant’s view, human beings have ‘‘an intrinsic
worth, i.e. dignity’’ and should therefore be treated ‘‘always as an end and never
as a means only’’ (Kant, 1785). Clearly this view is at variance with the utilitar-
ian’s willingness to sacrifice one individual’s welfare where this leads overall to
welfare gains. Kant himself confined the right to be treated as an end to human
beings, but later rights theorists, such as the American philosopher Tom Regan
(1989), have argued that the principle of dignity should be extended to animals.
Thus ‘‘. . . attempts to limit its scope to humans only can be shown to be
rationally defective. Animals, it is true, lack many of the abilities humans
possess. They can’t read, do higher mathematics, build a bookcase, or make
baba ghanoush. Neither can many human beings, however, and yet we don’t (and
shouldn’t) say that they (these humans) therefore have less inherent value, less of
a right to be treated with respect, than do others. It is the similarities between
those human beings who most clearly, most non-controversially have such value
(the people reading this, for example), and not our differences that matter most.
The really crucial, basic similarity is simply this: we are each of us the experien-
cing subject of a life, a conscious creature having an individual welfare that has
importance to us whatever our usefulness to others. We want and prefer things,
believe and feel things, recall and expect things. All these dimensions of our life,
including our pleasure and pain, our enjoyment and suffering, our satisfaction
and frustration, our continued existence or our untimely death –all make a
difference to the quality of our life as lived, as experienced, by us as individuals.
As the same is true of those animals that concern us (the ones that are eaten and
trapped, for example), they too must be viewed as the experiencing subjects of a
life, with inherent value of their own.’’
What implications does the rights view have for animal use? Obviously, the
answer to this question will depend on whether we are prepared to go along with
Regan (1989) and ascribe rights to animals. If we refuse to take this step, rights
theory will have little to tell us about animal use. If we allow that animals possess
intrinsic dignity and have rights, however, various things will follow. To begin
with, the balancing of human benefits against animal suffering that has been
central in our discussion so far becomes to some extent a background issue. No
benefit can justify disrespect for the rights of an individual, human or animal.
Categorical abolitionism of this sort probably goes further in its attempt to
limit the utilitarian trade-offs than most of us would consider necessary. After
all, weighing costs against benefits and seeking what is best overall, in private
decisions is part of our daily life. We expect others (for example, employers and
government bodies) to do the same. In all this, we accept that we are not treated,
and do not treat others, purely as ends. On the other hand, most people would
presumably allow that certain rights are sacrosanct, and that there are limits to
the extent to which an individual can be sacrificed for an overall benefit. Only
(what we might call) a moderate rights view is likely to command widespread
acceptance.
How would such a moderate view apply to animal use? The detail would
depend on what rights we take to be fundamental. The right to life (or more
ISSUES IN FISH WELFARE 339
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
accurately, the right not to be killed) is often regarded as basic. Curiously,
however, this does not appear to be a basic right that people would ascribe to
animals: after all, most of us happily eat animals that have been killed just for
this purpose. Something like a right to protection from suffering, or significant
suffering, seems to be much more promising. We might agree that all animals
should be protected from suffering if this involves intense or prolonged pain or
distress that the animal cannot control. Fish biologists have a part to play in this
debate, since they can help to develop husbandry practices that do not infringe
the rights of fish, on this moderated rights view.
An assumption underlying the previous discussion is, of course, that fish are
sentient beings. This matters both for the utilitarian, for whom the ability to suffer
and feel pleasure is the key criterion of moral consideration and for the adherents
of the rights view, who would typically claim that only sentient beings can be
bearers of rights. This assumption, however, is not uncontroversial especially in
the case of non-mammalian animals, as will become clear in the next section.
WELFARE, SUFFERING AND THE PERCEPTION OF PAIN AND
FEAR IN FISH
In the light of the definition of welfare used here (based on absence of suffering),
a major unresolved issue is whether and to what extent an animal can feel pain and
experience other forms of suffering. Do events that compromise health or interfere
with natural behaviour generate the mental state of suffering? In this context,
suffering can be defined as conscious experience (i.e. based on awareness of
internal and external stimuli, Chandroo et al., 2004a) of something as very
unpleasant (Dawkins, 1998). If non-human animals have no capacity for suffering,
then arguably it does not matter that animals are exposed to such events
(Bermond, 1997). A plant may be dying, but as it has no nervous system to
generate mental experiences the possibility that it might be suffering does not
arise. This question can arise for any group of animals, including invertebrates and
larval fish, but we concentrate here on whether adult fish are capable of suffering
and we approach this by considering the controversial issue of whether they
experience physical damage as pain (Rose, 2002). To anticipate, our view and
that of several other commentators (Chandroo et al., 2004a,b) is that adult fish
probably do experience some of the adverse states that humans associate with pain
and emotional distress, even if they do not have the capacity for self awareness
necessary for conscious suffering in the full human sense (Braithwaite &
Huntingford, 2004). On this basis, if fish are injured or exposed to other harmful
conditions, this is a cause for concern not just in terms of responsible stewardship
of fish populations (Rose, 2002), but also in terms of the welfare of individuals.
People arguing on either side of this debate have used a number of kinds of
evidence, none of them entirely satisfactory. Thus it has been argued that the longer
the life span of a given species of animal and the more sophisticated its general
behaviour, the greater its need for complex mental processes similar to those that in
humans generate the conscious experience of suffering. In this context, therefore, it
is relevant that the longest-living vertebrates are found among the fishes and that
fish behaviour is rich, complicated and far from stereotyped. For example, we know
that some species form mental representations of their environment and use these
340 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
for quite complex feats of navigation (Reese, 1989; Rodriguez et al., 1994). Also,
many fish live in social groups and some can recognize individual companions (e.g.
Swaney et al., 2001). Fish can remember negative experiences. For example, para-
dise fish (Macropodus opercularis Linnaeus, 1758) avoid places where they have
experienced a single attack by a predator and continue to do so for many months
(Czanyi & Doka, 1993) and carp (Cyprinus carpio L.) learn to avoid bait for up to 3
years after they have been hooked just once (Beukema, 1970). Several fish species
are capable of learning complex spatial relationships and forming mental maps
(Odling-Smee & Braithwaite, 2003) using an homologous forebrain structure to
that responsible for spatial memory in birds and mammals (Broglio et al., 2003);
some are capable of forming hierarchical associations about order or sequence of
spatial information (Burt de Perera, 2004). Furthermore, techniques used by
experimental psychologists have demonstrated that different types of information,
such as the timing of an event or the experience of a noxious stimulus, are processed
in different areas of the forebrain, yet somehow these experiences can be integrated,
and enable the fish to generate appropriate avoidance responses (Portavella et al.,
2004; Yue et al., 2004). Thus, while animals could show these kinds of associative
learning without necessarily having conscious awareness (Rose, 2002), clearly
experiences such as exposure to a predator or tissue damage can be strongly
aversive for a fish. Indeed, current literature on fish cognition indicates that several
fish species are capable of learning and integrating multiple pieces of information
that require more complex processes than associative learning (e.g. Braithwaite,
2006; Sovrano & Bisazza, 2003). Thus, we conclude that where there is evidence of
fish species with sophisticated cognitive and behavioural processes, the experience
of suffering may be a real possibility.
On the specific point of whether fish experience physical injury as pain, it is
helpful to consider current knowledge of pain perception pathways in mammals.
In this context, the sensory structures that detect harmful (or noxious) stimuli
are called nociceptors rather than pain receptors, to stress the fact that detecting
and responding to noxious stimuli is not necessarily the same as feeling pain
(Broom, 1998). What do we know of these systems in fishes? As far as the
possession of receptors that detect harmful stimuli is concerned, more primitive
fish such as lampreys (Petromyzon marinus Linnaeus, 1758) have free nerve
endings in the skin that respond physiologically to mechanical pressure and
heat, but behavioural reactions associated with nociception were not recorded
(Matthews & Wickelgren, 1978). It is also difficult to determine whether the
mechano-receptors in lamprey are truly nociceptive-specific or simply pressure-
specific (Christenson et al., 1988). In at least one teleost fish (the rainbow trout,
Oncorhynchus mykiss (Walbaum)), however, anatomical and electrophysiological
examination of the trigeminal nerve (which is known to convey pain information
from the head to mouth in terrestrial vertebrates) has identified two types of
nociceptor, A-delta and C fibres (Sneddon, 2002; Sneddon et al., 2003a).
In terms of the anatomy that generates the conscious experience of pain in
humans, the brain of a fish is clearly far smaller relative to body size (some 300
times smaller by volume) and simpler in structure than that of a human
(Kotrschal et al., 1998). In particular, the forebrain, or telencephalon, is rela-
tively undeveloped compared with humans and fish lack cortical structures such
as the neocortex, part of the brain with a key role in the subjective experience of
ISSUES IN FISH WELFARE 341
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
pain in humans (Rose, 2002). There may be a greater degree of homology,
however, between the forebrain of fish and mammals (e.g. Broglio et al., 2003;
Portavella et al., 2004) and even if this is not the case, we know that the same job
can be done by different parts of the brain in different kinds of animals. For
example, the brain of cephalopods is built on an entirely different plan from that
of vertebrates, yet it generates highly complex behaviour (Hanlon & Messenger,
1996). Likewise, visual stimuli are processed by part of the cerebral cortex in
mammals, but in birds some visual information is processed extensively by the
midbrain optic tectum (Shimizu & Karten, 1993). It is not impossible that parts
of the brain other than the cerebral cortex have evolved the capacity for gen-
erating negative emotional states/suffering in non-mammalian vertebrates.
Jawed fish are known to produce some of the natural opiates that modulate
nociception in mammals (Substance P, enkephalins and B-endorphins,
Rodriguez-Moldes et al., 1993; Zaccone et al., 1994; Balm & Pottinger, 1995).
This does not necessarily mean that these substances serve the same function in
fish as they do in mammals, although the behavioural response of goldfish to
analgesics is similar to that of a rat (Rattus norvegicus, Ehrensing et al., 1982). In
mammals opiates act at neural levels below the neocortex (Rose, 2002), but this
does not preclude their having a pain-suppressing effect. In support of this point,
recent behavioural experiments have demonstrated effects of noxious stimula-
tions around the mouth of rainbow trout (Sneddon et al., 2003a). In contrast to
control treatments, fish administered with a weak acetic acid solution or bee
venom showed dramatic and prolonged increases in opercular beat rate and
suspension of feeding. In addition, the trout given the noxious stimulation
were observed to rest on the substratum, rocking from side to side. Trout treated
with acetic acid were also observed rubbing their snouts on the base and walls of
the test tank. Similar studies have also shown that the adverse behaviour of fish
under noxious stimulation can be mitigated if an analgesic (morphine) is admin-
istered (Sneddon et al., 2003b). Taken together, these findings suggest that fish
have the sense organs and the sensory processing systems required to perceive
harmful stimuli and, probably, the central nervous systems necessary to experi-
ence at least some of the adverse states that we associate with pain in mammals.
Hence our working position that adult fish have the capacity to perceive painful
stimuli and that these are, at least, strongly aversive.
COSTS AND BENEFITS OF HUMAN INTERACTIONS WITH
FISHES
Table 1 outlines a number of human activities that may potentially compro-
mise the welfare of individual fish and so cause the harm against which any
benefits must be weighed. The word ‘potentially’ is used deliberately because at
this point we wish to simply identify areas of possible concern. Harmful effects
on welfare can be indirect, as when humans inadvertently alter natural habitats
or expose fish to harmful chemicals, or direct, for example through commercial
fisheries, through sports fisheries, through intensive production, through keeping
fish as pets or in public aquaria or through scientific research. This review does
not aim to make judgements about what is acceptable and what is unacceptable,
342 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
TABLE I. Some human activities that could potentially compromise fish welfare
Activity Examples of potential effects on welfare
Environmental
degradation
Reduced availability of natural food.
Introduction of exotic species into existing fish communities.
Habitat modification, creating (e.g.) sub-optimal
hydrological regimes.
Loss of or displacement from natural habitats.
Reduced population densities (or crowding) and abnormal
social experiences.
Disturbance through tourism.
Acute and chronic exposure to pollutants and litter.
Commercial and
sports fisheries
Injury during trawling.
Tagging/fin clipping during stock assessment.
In both, tissue damage, physical exhaustion and severe
oxygen deficit during capture.
In both, pain and stress during slaughter.
In angling, pain and stress in tethered fish when live bait is used.
In angling, release of reared fish inappropriately equipped
for survival in the wild.
In angling, stocked fish introduced to lakes may be denied
the opportunity to migrate.
Aquaculture High densities in simple and constraining conditions,
both in normal rearing conditions and for husbandry.
Poor water quality.
Aggressive interactions, which can cause damage and constrain
access to food.
Food deprivation (e.g. during disease treatment
and before harvest).
Handling and removal from water during routine
husbandry procedures
Unnatural light-dark regimes, to control breeding.
Handling, constraint and, sometimes, low oxygen levels
during transportation.
Permanent adverse physical states and possibly increased
levels of aggressiveness due to selection for fast growth.
Increased exposure to predators, attracted to fish farms or used
to grade out smaller fish (in extensive tilapia aquaculture).
Transmission of disease between wild and farmed stocks.
Crowding, handling, removal from water and pain during slaughter.
Keeping ornamental
fish and display fish
in public aquaria
For ornamental fish, capture by sublethal poisoning.
For ornamental fish, permanent adverse physical states
due to selective breeding.
For ornamental fish, release or escape of exotic species.
Inappropriate temperatures, poor water quality and
physical constraint during transport.
Confined space and poor water quality once housed.
Inappropriate physical conditions.
ISSUES IN FISH WELFARE 343
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
but instead seeks to identify potentially harmful effects of human activities on
fish welfare, as far as possible on the basis of documented evidence.
Having said that, it is worth stressing the benefits that humans derive from such
activities. An estimated 150 million people worldwide rely on fisheries, aquaculture
and associated activities for their livelihoods, with 38 million directly engaged in
fishing, and 10 million in aquaculture as of 2002 (Food and Agriculture
Organisation of the United Nations, 2004). Many of those engaged in fishing live
in the world’s poorest countries and remain landless, hence fisheries are the only
means of support for whole communities. Aquaculture is making an increasing
contribution as fisheries production from 39% of the total in 1970 to 319% in
2003. This has been the result of increased aquaculture production against a back-
drop of declining wild stocks. Between 1993 and 2003 there was an average annual
increase of 94% in aquaculture production, with 423 million tonnes of aquatic
animals produced in 2003 with an estimated farm-gate value of US$610 billion
(Lowther, 2005). Total world trade of fish and fisheries products was US$582
billion (export value) in 2002 (Food and Agriculture of the United Nations, 2004)
greater than that of rice, coffee, sugar and tea combined (World Bank, 2005).
Sports fisheries and ornamental fish keeping are major recreational activities. For
example, in England and Wales there are an estimated 29 million freshwater
anglers, equating to c.3
5% of the population. Of these, 23 million are coarse
(non-salmonid) anglers making an average of 43 trips per year (National Rivers
Authority, 1995). In 2001 there were 341 million anglers in the U.S.A. or c. 1 in 6 of
the population over 16 years of age, these people spent an estimated US$346
billion on fishing, this is direct expenditure excluding employment and ancillary
industries associated with fishing (USA Department of the Interior, Fish and
Wildlife Service U.S. Department of Commerce and U.S. Census Bureau, 2001).
Ornamental fish are the third most common pet after dogs and cats, with 35 to 40
million fish entering the US per annum, with a retail value of the fish and acces-
sories ranged between 189 and 305 million US$ (e.g. Mintel, 1991). Whether or not
any of these activities do indeed harm fish is considered later.
TABLE I. Continued
Activity Examples of potential effects on welfare
Inappropriate social conditions, with shoaling fish at
low densities and predators with prey.
Inappropriate diets.
Scientific research Genetic-modification induced for scientific research may have
detrimental effects on welfare.
Fish used in the laboratory for experimental purposes are often
confined and may be exposed to a range of deliberately-imposed
adverse physical, physiological and behavioural states.
Fisheries research often involves electrofishing, tagging,
fin clipping or otherwise marking fish, which potentially
cause pain and injury.
In both cases, handling during research procedures may
cause injury.
344 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
NATURAL THREATS TO WILD FISH
It is not self-evident that natural is necessarily good in any general sense.
Additionally, in nature fish regularly make adaptive trade-offs between different
needs. For example, like other animals they accept heavy costs in order to
reproduce, in terms for example of physical injury and serious depletion of
nutrient reserves/impairment of body condition; these are all conditions that
humans would regard as evidence of impaired welfare. This poses a problem
when considering welfare, since what is natural conflicts with what is good in
terms of physical condition, and there may be no real situation where all aspects of
welfare are ideal. There is a moral difference between deviation from an optimal
state that is caused by natural events and suffering caused by human activity
(especially when fish have no choice of environment). Arguably, the term ‘welfare’
is not relevant to adverse experiences that are not anthropogenic; even so, an
understanding of the natural threats encountered by wild fish and how frequently
these occur (Table 2) can help to clarify our thoughts on fish welfare. Wild fish
experience injury, poor environmental conditions and stressful events due to
encounters with potential predators and fish of the same species, restricted food
supplies, parasitic infection and disease and natural environmental change. One
implication is that fish are likely to have mechanisms for dealing with the adverse
conditions that they encounter naturally and that these will respond (up to certain
limits) during their interactions with humans. Following from this, such natural
responses might provide a means of assessing fish welfare.
HOW FISH RESPOND TO NATURAL THREATS TO THEIR
WELFARE
STRESS RESPONSES IN FISH
Much of our understanding of how fish respond to adverse conditions comes
from the extensive literature on the biology of stress. In common with all
vertebrates, fish possess a suite of adaptive behavioural and physiological strat-
egies that have evolved to cope with destabilizing challenges, or stressors.
Although there are some differences in detail arising from the contrast between
the aquatic and terrestrial environment and from minor differences between the
endocrine systems of fish and higher vertebrates, overall the stress responses of
fish equate closely to those of other animals (Barton, 1997; Wendelaar Bonga,
1997). This parity extends to the behavioural elements of the stress response.
Recent work with salmonid fish has shown that the integrated behavioural and
physiological mechanisms that comprise the distinct ‘coping strategies’ believed
to be present in mammals (Koolhaas et al., 1999; Wingfield, 2003; Huntingford
& Adams, 2005) are also evident in fish, with heritable reactive and proactive
traits demonstrated in rainbow trout (Øverli et al., 2005).
BEHAVIOURAL RESPONSES TO STRESS
Behavioural responses are an animal’s first line of defence against adverse
environmental change, predators, and social conflict, often being triggered by
ISSUES IN FISH WELFARE 345
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
TABLE II. Natural threats to the welfare of wild fish
Stressor Comment and selected examples
Predators Predation rates can be very high. e.g. excluding
predators reduces mortality by 26% in wrasse
Thalassoma hardwicke (Shima, 2002).
Unsuccessful predatory attacks may cause wounding
and an increased risk of disease. e.g. 30% of wild
sticklebacks (Gasterosteus aculeatus Linnaeus) showed injuries
due to failed predatory attacks (Reimchen, 1994).
The threat of predation may suppress feeding and
may cause fish to forage sub-optimally (Hart, 1997).
Conspecifics Many species live naturally in groups of the same
species, which provide protection against predators.
Obligate shoaling fish separated from companions will
strive to join a shoal (Pitcher & Parrish, 1993).
In many other species (or in shoaling species under
particular circumstances) conspecifics fight over
resources and this can cause physical damage and
depletion of energy reserves (Neat et al., 1998). Many
mature wild Atlantic salmon parr have wounds from
attacks by larger fish (Garcia de Leaniz, 1990). Losers
may be deprived of resources and/or exposed to chronic
social stress (Abbott & Dill, 1989; Alanara, 1997).
Food availability
and body condition
Wild fish often experience periods of food shortage
(Dutil & Lambert, 2000), though many species have
flexible metabolic systems to cope with periods of
prolonged food deprivation (O’Connor et al., 2000).
Growth rates in fish held captive with excess food
consistently and markedly exceed those achieved by fish
in the wild [e.g.inCynolebias viarius Steindachner, 1876
(Errea & Danulat, 2001)].
Lipid deposition rates and mineral content of body
tissues may also differ between wild and captive reared
fish (e.g. for sea bass, Orban et al., 2002).
At least 50% of larvae of the common Japanese goby
Rhinogobius brunneus Temminck & Schlegel, 1845
die through starvation prior to obtaining their first food
(Iguchi & Mizuno, 1999).
Extensive migration Daily vertical migration by pelagic fishes results in
slower growth (Lima, 1998).
Energy reserves in spawning salmon can be reduced by >90%
following upriver migration (e.g. Jonsson et al., 1991).
Parasites and disease In the wild most fish carry a parasite burden that
impairs their health (Margolis et al., 1982). High gill
parasite loads in fish from the Salton Sea,
California caused gill damage, depressed respiration
and osmoregulation and juvenile mortality in several
species (Kuperman et al., 2001).
346 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
the same stimuli that initiate a physiological stress response. As noted above, it is
becoming clear that in fish, as in other animals, individual’s exhibit distinct
behavioural strategies when faced with potentially threatening circumstances,
and the type of behavioural response initiated, and the magnitude of the neu-
roendocrine response to the stressor, can be expressed as individual traits
(Schjolden et al., 2005). The range of behavioural responses exhibited by fish
to deal with stressors of varying magnitude is diverse. Altered patterns of
swimming (changes in speed and direction) are shown in response to many
stressors (e.g. Juell & Fosseidengen, 2004). After an attack by another fish of
the same species, fish may flee and hide or take up a submissive posture, often
with altered body colour (e.g. O’Connor et al., 2000; Sutor & Huntingford,
2002). When attacked by a predator, fish may respond by shoaling (Pitcher &
Parrish, 1993), freezing (e.g. Goodey & Liley, 1985) or taking shelter (e.g. Brown
& Warburton, 1999) and may change colour in this context as well (Endler,
1986). Feeding may be suppressed following an encounter with a predator, or
inefficient feeding strategies may be adopted (Hart, 1993) and fish may avoid
areas in which they have been attacked (Lima, 1998). Specific adaptive beha-
viour patterns are observed in response to parasitic disease (Furevik et al., 1993)
and to tissue damage (for example, carp that are hooked in the mouth show
rapid darting, spitting and shaking of the head (Verheijen & Buwalda, 1988) and
rainbow trout injected with acetic acid in their lips rub their snouts against the
substratum (Sneddon et al., 2003).
ACUTE PHYSIOLOGICAL STRESS RESPONSES
The neuroendocrine stress response in fish is virtually identical to that of
mammals (Wendelaar Bonga, 1997) and is mediated by the hypothalamic-pitui-
tary-interrenal (HPI) axis. Perception of a stressor by the fish initiates a rapid,
neurally-stimulated release of catecholamines (adrenaline/epinephrine and nora-
drenaline/norepinephrine; Perry & Bernier, 1999) from the chromaffin tissue,
homologous to the mammalian adrenal medulla. This is accompanied by release
of corticotropin-releasing hormone (CRH) from the hypothalamus which in turn
promotes the release of corticotrophin (adrenocorticotrophic hormone, ACTH)
by the pituitary and subsequent synthesis and secretion of cortisol by the inter-
renal tissue, the homologue of the mammalian adrenal cortex (Weld et al., 1987;
In Orange roughy Hopostethus atlanticus Collette, 1889
from New Zealand, parasite loading was negatively
correlated with growth (Gauldie & Jones, 2000).
Suboptimal
environmental
conditions
Most environmental variables fluctuate naturally,
so wild fish will experience conditions that deviate from
optimal for the species concerned, and that may be near
to or beyond their limits of tolerance.
Fish can avoid or adapt to sub-optimal environmental
conditions (at an energetic cost), but exposure to conditions
beyond their limit of tolerance is, by definition, lethal.
ISSUES IN FISH WELFARE 347
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
Okawara et al., 1992; Sumpter, 1997). Cortisol concentrations return to pre-
stress levels within hours of exposure to a brief stressor (Pickering & Pottinger
1989; Waring et al., 1992), but elevated cortisol levels generally persist during
continuous, chronic stress (Pottinger & Moran, 1993; Pottinger et al., 1994). If a
repeated stressor is not inherently damaging, acclimation, or habituation, can
occur (Pickering & Pottinger, 1985). The neuroendocrine stress response is
responsible for coordinating and stimulating adaptive adjustments to respiratory
and metabolic function and is endocrinologically more complex than indicated
by this brief overview.
CHRONIC PHYSIOLOGICAL STRESS RESPONSE
Where fish cannot escape a stressor, or where the stressful stimulus is episodic
or intermittent, prolonged activation of the stress response has deleterious con-
sequences. These include loss of appetite, impaired growth and muscle wasting,
immunosuppression and suppressed reproduction. Clearly, observing such
changes provides strong indications that the well-being of the fish has been
significantly compromised. Many of the adaptive elements of the acute response
described above affect energy intake and increase energy utilization, so pro-
longed activation of the HPI axis is likely to reduce growth indirectly through
a negative effect on energy balance. In addition, secretion of growth hormone is
reduced in fish during periods of stress (Pickering et al., 1991; Farbridge &
Leatherland, 1992), so there are also direct effects on the mechanisms that
control growth. Poor growth has also been reported in wild fish as a result of
environmental stressors such as altered pH (e.g. Puste & Das, 2001), reduced
dissolved oxygen (Kramer, 1987) and salinity (Brett, 1979). Since growth and
reproduction are functionally linked (Thorpe et al., 1998), stress-induced impair-
ment of growth may indirectly interfere with maturation. Additionally, repro-
ductive activity is suppressed directly during periods of stress, via a wide range of
mechanisms (Pottinger, 1999).
In teleost fish, defence against disease is mainly based on a non-specific
immune system that does not depend on prior disease challenge. The main
components are chemicals in the body fluids that destroy or inactivate invading
organisms and circulating and tissue-dwelling phagocytes that engulf or destroy
invading organisms. The specific immune system, which has a memory compo-
nent that can adapt to different invading organisms (e.g. Press, 1998) is less well
developed than in birds and mammals, comprising circulating lymphocytes
responsible for antibody production and phagocytic cells, which have an addi-
tional role in presenting antigens to the specific immune system. Chronic stress
has a generally immunosuppressive effect in fish, mediated in particular by the
actions of cortisol (Weyts et al., 1999), and increased mortality due to fungal and
bacterial pathogens (e.g. Pickering & Pottinger, 1989; Plumb, 1994) is the com-
mon outcome.
FUNCTIONAL CONSIDERATIONS
The stress response has evolved to assist the survival of the animal under
demanding conditions in the natural environment. Natural stressors, however,
348 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
tend to be brief and/or avoidable. In contrast those stressors that are imposed
upon fish by anthropogenic agents may be unavoidable and prolonged or
repetitive. Under such circumstances, chronic or repeated activation of behav-
ioural and physiological stress responses is maladaptive and potentially harmful.
Ideally, indicators of welfare should signal problems in advance and allow
intervention before this state has been reached.
ASSESSING FISH WELFARE
WAYS OF MEASURING FISH WELFARE
Based on knowledge of the natural responses of fish to adverse conditions, the
physiological, health and/or behavioural status of individual fish have been used
as indicators of compromised welfare, though the link between components of
the stress response and welfare is not simple.
Stress responses represent an animal’s natural reaction to challenging condi-
tions and these are often used as indicators of impaired welfare, so studies of
physiological stress feature prominently in welfare research. It is important,
however, to recognize that physiological stress is not synonymous with suffering
(Dawkins, 1998). There is no particular reason to suggest that the temporary
physiological activation that prepares fish for activity is detrimental to welfare
and in some contexts short-term stress responses (for example, in anticipation of
feeding) may well be beneficial (Moberg, 1999). Indirect effects such as sup-
pressed reproduction may well be adaptive responses to poor conditions in the
wild, but even so it seems reasonable to assume that in captive fish they indicate
exposure to chronic, unavoidable stress, which may have compromised welfare.
Thus although the concept of stress does not fully capture the complexities of
animal welfare, monitoring stress responses may give us an important part of the
picture. In particular, where several components of the stress response (including
up-regulation of particular genes, Ribas et al., 2004) are all influenced in a
similar way by the same condition, this suggests that there is cause for concern
about the welfare of the fish involved.
The link between health and welfare is also complex. If an individual fish
shows disease symptoms, it seems reasonable to infer that it is in a poor state of
welfare, as a direct result of the disease. The converse, however, is not necessarily
true since the welfare of a healthy fish may be compromised, for example
through inappropriate social environments. In addition, because stress can sup-
press immune function and increase risk of infection, a high incidence of disease
and mortality in a population may indicate that there is an underlying problem
with the fish’s environment. It would be overly simplistic to assume that disease
is invariably the result of poor living conditions or that the occurrence of disease
inevitably implies that the problem is due to human mismanagement. Even fish
experiencing optimal conditions may suffer from disease and serious epidemics
occur in populations of wild fish (e.g. Epizootic Ulcerative Syndrome, Lillie &
Roberts, 1997).
Behavioural studies have been important in welfare research for a number of
reasons. Since altered behaviour is an early and easily observed response to
adverse conditions, specific responses to natural stressors (such as ‘freezing’ in
ISSUES IN FISH WELFARE 349
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
the presence of a predator or rubbing to remove ectoparasites) can be used as an
indicator of impaired welfare. Likewise, since animals pay attention to those
stimuli that are currently important for fitness, changes in attentional state can
be used to highlight welfare problems. For example, trout exhibit strong avoid-
ance responses when exposed to a novel object (Sundstrom et al., 2004). Such
responses are suppressed if the fish has been exposed to a noxious stimulus, but
not in fish treated with some form of analgesic. The fact that exposure to
noxious stimuli interferes with the normal neophobic responses suggests that
fish give a high priority to such stimuli (Sneddon et al., 2003b). Additionally,
since animals may suffer if prevented from performing their full behavioural
repertoire, behavioural deficits have been used to identify conditions that
compromise welfare (Mench & Mason, 1997). Behavioural techniques such as
choice tests that give insights into the priorities that animals place on different
options have proved valuable in welfare research on birds and mammals,
even though the underlying assumption that animals choose what is good for
them is not always valid (Dawkins 1998, 2004). Choice tests have often been used
on fish, though the aim is not usually related to fish welfare; for example, fish may
be required to chose between different temperatures (Bevelhimer, 1996), between
schools of different composition (Metcalfe & Thomson, 1995) or between water
with different concentrations of potentially lethal pollutants (Giattina & Garton,
1983). There is clearly scope for more work along these lines directly aimed at
identifying conditions that promote good welfare in fish.
INTEGRATING DIFFERENT MEASURES OF WELFARE
There are many potential signs of impaired welfare, hence the most reliable
assessment of well-being will be obtained by examining a range of informative
measures. This raises the question of how such a battery of measures can be
combined objectively to give an overall impression of welfare, and there are
various methods of multivariate analysis that might be used in this context. For
example, Principal Components Analysis was used to integrate four commonly
used measures of fish welfare reflecting different functional systems (condition of
body and fins and plasma concentrations of glucose and cortisol) into a single
welfare score for farmed Atlantic salmon (Turnbull et al., 2005). As well as
reflecting coherence within the data, this score was consistent with the evaluation
of welfare by experienced farmers and had a significant and negative relationship
to, among other things, stocking density and cage position. In a related study
using experimental tanks, a similar multivariate score was used to relate welfare
to disturbance (among other factors), welfare being best in more frequently
disturbed tanks. Behavioural studies suggested that this counter-intuitive result
arose because aggression during feeding was suppressed by human disturbance
(Adams et al., unpublished data).
SENSITIVE AND EASILY APPLIED WELFARE INDICATORS
FOR FISH
Data on fish physiology, biochemistry and behaviour are informative, but
collecting them is time consuming, technically complex and involves handling or
350 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
killing fish in order to collect blood or other tissue. Non-invasive methods exist,
such as measuring cortisol levels in the water in which fish have lived (Ellis et al.,
2004) or in their faeces (Oliveira et al., 1999, Turner et al., 2003), but these
sometimes lack the precision of direct measurements made on individual fish.
Such intensive work is necessary in scientific research, but is impractical for
everyday use, in pet shops or on working fish farms, for example. What is needed
for practical management of welfare here is a set of simple, non-intrusive signs or
danger signals that can be used easily without complicated laboratory analysis.
A number of possible welfare indicators can and have been used to assess the
welfare of individual fish (Table 3). Some of these are based on assessments of
health that can be used on dead fish (see Tierney & Farrell, 2004), but others are
based on behaviour and production. These are well known to people with a
practical interest in fish welfare, such as owners of ornamental fish and respon-
sible fish farm workers (Turnbull et al., 2005).
How well these signs work in any given case will depend on the species
concerned (for example, eye colour may be a good indicator of social stress
in salmon, but not in sticklebacks), on circumstances (for example, depleted
energy reserves might be cause for concern in an immature salmon, but not in
one that has just bred) and also on individual status (failure to feed may be a
sign of poor welfare in a juvenile salmon in the summer, but not necessarily
in the winter when they may show adaptive natural anorexia). The potential
for using the full range of available indicators will vary with the context in
which fish welfare is to be assessed; fish farmers may have to rely on a few
signs, but people keeping ornamental fish are well placed to use many of them,
on all their fish. In addition, on farms, sick or damaged fish may be more
conspicuous than those in good condition, so it is important to develop sampling
protocols that give an accurate picture of the welfare status of the whole
population.
HOW HUMAN ACTIVITIES AFFECT FISH WELFARE
The scientific study of fish welfare lags behind that of the welfare of other
vertebrates (reflecting the pressure of public concern), but there is still an
extensive literature on the subject, using several of the welfare indicators outlined
above. It is beyond the scope of this review to provide an exhaustive account of
this literature, but in this section we comment briefly on the subject, using just a
few examples and concentrating on the areas of potential concern identified in
Table 1.
ENVIRONMENTAL DEGRADATION
As far as environmental degradation is concerned, this is clearly a cause of
poor welfare in very large numbers of fish (Montgomery & Needleman, 1997).
Heavy metals cause extensive gill damage in acidic water, but are non-toxic in
hard, alkaline water. (see Wedermeyer, 1997). We now know that heavy tourism
can cause detectable stress response of reef fish (Oliveira et al., 1999). Pollution
of water bodies as a result of direct or indirect input of industrial emissions and
effluents is widespread through natural fish habitats and can severely affect fish
ISSUES IN FISH WELFARE 351
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
TABLE III. Possible easily-measured indices of welfare and examples of their use. We list
only those based on direct observations of fish rather than the systems in which they are
held
Possible index Examples
Changes in colour Stress-induced changes in skin or eye colour (which have a
complex neural and hormonal background) have been
reported in a number of fish species, including ornamental
species (Etscheidt & Manz, 1992) and so could be a sign of
exposure to adverse events, e.g. eye colour as an index of social
stress/subordinate status in salmonids (O’Connor et al., 2000;
Sutor & Huntingford, 2002).
Changes in ventilation
rate
A high oxygen demand is reflected by rapid irrigation of the
gills. The rate of opercular beats is therefore increased by
stress and can be counted automatically or by eye. This,
together with a visual assessment of gill status, is used as a
sign of incipient problems in ornamental fish (Etscheidt &
Manz, 1992) and ventilation rate has been used to monitor
exposure to pollutants (Handy & Depledge, 1999).
Changes in swimming
and other behaviour
patterns
Fish may respond to unfavourable conditions by changing
swimming speed and space use (Morton, 1990; Etscheidt & Manz,
1992; Kristiansen et al., 2004). Abnormal swimming has been
used as a sign of poor welfare in farmed fish (Holm et al., 1998).
Behavioural responses to adverse conditions (or lack of
responsiveness to specific stimuli) are signs of both general and
specific trouble (Morton, 1990). These include excessive activity or
immobility (Etscheidt & Manz, 1992), body positions that protect
injured fins, escape attempts and rubbing to dislodge ectoparasites
(Furevik et al., 1993).
Reduced food intake There are many reasons why a fish might not eat, but the
fact that feeding is suppressed by acute and chronic stress
means that an unexpected loss of appetite is a sign of
potentially impaired welfare.
Loss of condition Fish change shape and/or lose weight for many reasons, but
because reduced feeding and mobilization of reserves are
secondary stress responses, where fish are regularly weighed
and measured, or where body shape can be assessed by eye
(for example by the visibility of the vertebrae, Escheidt & Manz,
1992) loss of condition can indicate possibly impaired welfare.
Slow growth Growth rates in fish are flexible and naturally variable, but
provided we have an estimate of expected growth prolonged
low rates of growth may be indicative of chronic stress. Thus
where fish are regularly weighed or where size can be assessed
by eye (or by underwater camera) slow growth can be used as
a possible sign of trouble.
352 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
welfare. To give just a few examples, reduced growth rates in fish has been
reported in response to low pH (Puste & Das, 2001) and reduced dissolved
oxygen (Kramer, 1987). Eutrophication can impair growth of fish even when
food is available in excess, but at the same time remedial measures to reduce
impacts of human activities may also have adverse effects on fish. For example,
following policy-induced reductions in phosphorus input and subsequent ‘oligo-
trophication’ of the waters, growth rates were reduced in largemouth bass in a
Morphological
abnormalities
Because adverse conditions can interfere with normal
development, the occurrence of morphological abnormalities
can be used as an indicator of poor larval rearing conditions
(Boglione et al., 2001; Cahu et al., 2003), although whether
this represents problem for welfare depends on the degree of
sentience of the larvae concerned.
Injury, including
fin damage
Injury may be a direct consequence of an adverse event, in
which case, a high frequency of such injuries is a sign of poor
welfare. For example, dorsal fin injury in salmonids is often
caused by attacks from conspecifics (Turnbull et al., 1998) and
scales that are dislodged with blood visible rather than lying
flat, are a sign of poor welfare in ornamental fish (Etscheidt &
Manz, 1992). In addition, because immune responses can be
suppressed by cortisol, slow recovery from injury (or a high
incidence of injury) may be a sign of generally poor conditions.
As well as acute damage, healed injuries may result in
long-term abnormalities (e.g. in salmon healed fin injuries may
cause permanently short fins) that potentially compromise
performance and welfare.
Disease states Since the causes of most aquatic diseases are complex and
dependent on environmental conditions, the presence of
disease can indicate an underlying problem with the
environment or management. Increased incidence of disease in
any population of fish should be treated as a warning that
there may be other underlying problems. However,
interpreting the welfare implications of an observed disease
requires a detailed understanding of the natural history of the
disease. In some cases, diseases are not sufficiently well
understood to interpret their implications for welfare. Even
records of treatment can be difficult to interpret since they
may either indicate that the owner is responding appropriately
to disease outbreaks or the fish are being exposed to a
predictable endemic disease.
Reduced reproductive
performance
For many farmed species, reproduction is prevented or avoided
in growing stock. Where this is not the case, for example, in
broodstock or in ornamental fish, because chronic stress
impairs reproductive function, failure of adult fish to breed or
to display normal patterns of reproductive development when
feed, light and temperature regimes are appropriate is a
possible sign of poor welfare.
ISSUES IN FISH WELFARE 353
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
well established fishery, due to reduced primary productivity with consequences
further up the food chain (Maceina & Bayne, 2001).
COMMERCIAL AND SPORTS FISHERIES
The welfare of fish caught by commercial fisheries (still the largest area of
human-fish interactions) is also a cause for serious concern. Fish are harmed by
capture (e.g. cortisol levels increase in sea bream captured by trammel net and
many fish are mortally injured, Chopin & Arimoto, 1995) and slaughter methods
(especially asphyxia) are highly stressful (Poli et al., 2002). In addition, non-
target species captured as by-catch are often injured or killed (Pronovi et al.,
2001). A growing scientific literature has shown that several aspects of sports
fisheries have negative effects on fish welfare. (Table IV). Compared to aqua-
culture very few studies have addressed fish welfare in commercial fisheries.
AQUACULTURE
Because there is growing public concern for well-being farmed fish, there has
been a considerable amount of research into the impact of many aspects aqua-
culture practice on fish welfare, some of which is reviewed by Conte (2004).
More examples are given in Table V.
KEEPING ORNAMENTAL FISH
Various bodies are concerned with ethical issues arising from the keeping of
ornamental fish, whether in private homes or in public aquaria. These issues
include conservation of species used by the aquarium trade and their habitats as
well as the welfare of the individual fish themselves. Table VI gives examples of
some recent scientific studies of the impact of various practices in ornamental
fish keeping, on the welfare of individual fish.
SCIENTIFIC RESEARCH
Scientific research (including research by welfare scientists) raises various
concerns about its impact on welfare of its subjects, which have been the subject
of a number of reviews and guidelines (Nickum, 1988; Borski & Hodson, 2003;
Jackson, 2003). Such research (including studies of fish) is strongly regulated in
many countries to ensure that harm in terms of compromised welfare is out-
weighed by benefits in terms of enhanced knowledge on important issues (e.g.
the UK Animals (Scientific Procedures) Act, 1986), and is not considered further
in this review.
WHAT ARE THE WELFARE ISSUES?
A number of general points emerge from the brief literature review presented
in the previous section:
354 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
1. Some categories of human activity do indeed compromise fish welfare: If one
accepts disturbed behaviour, chronically elevated cortisol levels, increased
incidence of disease and poor growth (for example) as indicators of welfare
in fish, then there are many examples of harmful effects of human activities
on fish welfare (Table 2).
2. The same factor can have variable effects on fish welfare depending on
circumstances: Experiments in which the same environmental factor (for
example, stocking density) is manipulated often give variable results because
different species of fish, and different life-history stages of the same species,
require different environments for good welfare. Additionally, even within
the same species and age group, inherited individual differences in strength
of response to a standardized stressor have been reported, for example in
TABLE IV. Examples of scientific studies of the impact of various aspects of angling on
fish welfare
Practice Some demonstrated effects on welfare
Capture – hooking Injury and mortality following hooking is common,
but primarily associated with deep-hooked fish
(DuBois et al., 1994; Hulbert & Engstrom-Heg, 1980;
Muonehke & Childress, 1994).
Capture – playing/landing Capture of fish of various species by rod and line
elicits a stress response of short duration (Gustaveson
et al., 1991; Pankhurst & Dedual, 1994; Pottinger,
1998). Estradiol levels are suppressed in rainbow trout
within 24 h of capture by rod and line (Pankhurst &
Dedual, 1994). Capture and playing in largemouth
bass (Micropterus salmoides Lacepe
`de, 1802) produces
marked increases in heart rate (Cooke & Philipp, 2004).
Capture – handling Exposure of exercised fish to air can have severe
metabolic effects (lactate increase and altered acid-base
balance) that may be greater in larger fish
(Ferguson et al., 1993). Capture and handling
suppresses reproductive function in brown trout
(Melotti et al., 1992).
Retention/constraint/release Retention of fish postcapture in either keepnets or
stringers induces physiological stress responses, but
recovery following release can be rapid (Pottinger,
1998; Sobchuck & Dawnson, 1988). Hooking and
handling for release can increase scale damage by
16% (Broadhurst & Barker, 2000), possibly making
released fish liable to infection. Behavioural
modification can occur following release
(Mesa & Schreck, 1988; Olla & Davis, 1989;
Cooke & Philipp, 2004). Livewell confinement
increased mortality in walleye and largemouth bass
used in live release tournaments (Suski et al., 2005).
ISSUES IN FISH WELFARE 355
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
TABLE V. Examples of scientific studies of the impact of various aspects of aquaculture
on fish welfare
Practice Some demonstrated effects on welfare
Transportation Transportation induces physiological stress requiring
prolonged recovery (Bandeen & Leatherland, 1997;
Iversen et al., 1998; Rouger et al., 1998; Barton, 2000;
Sandodden et al., 2001; Chandroo et al., 2005).
Handling and netting Physical disturbance evokes a neuroendocrine stress response
in many species of farmed fish (reviewed by Pickering, 1998)
and reduces disease resistance (Strangeland et al., 1996).
Handling stress increases vulnerability to whitespot in
channel catfish (Davis et al., 2002).
Confinement and
short-term crowding
Physical confinement in otherwise favourable conditions
increases cortisol and glucose levels and alters immunological
activity in various species (Garcia-Garbi, 1998). Carp
(Cyprinus carpio) show a mild, physiological stress response
to crowding that declined as the fish adapted, but crowded
fish are more sensitive to an additional acute stressor
(confinement in a net; Ruane et al., 2002). Crowding
during grading increases cortisol levels for up to 48 h in
Greenback flounder Rhombosolea tapirinia, Gunther
(Barnett & Pankhurst, 1998).
Inappropriate densities High densities may impair welfare in some species (trout and
salmon, Ewing & Ewing 1995; sea bass, Dicentrarchus labrax
L., Vazzana et al., 2002; red porgy, Pagrus pagrus, Rotllant
& Tort, 1997; seabream Sparus auratus, Montero et al., 1999),
but enhance it in others (Arctic charr, Jergensen et al., 1993).
Halibut suffer less injury at high densities (Greaves, 2001)
but show more abnormal swimming (Kristiansen & Juell,
2002; Kristiansen et al., 2004). The relationship may not be
linear (in salmon negative effects begin to kick in at a critical
density, Turnbull et al., 2005) and density interacts with
other factors such as water quality (Ewing & Ewing, 1995;
Scott et al., 2001; Ellis et al., 2002). Genes coding for heat
shock proteins are over-expressed in sea bass held at
high densities (Gornati et al., 2004). An enolase gene is
up-regulated in sea bream held at high densities
(Ribas et al., 2004).
Enforced social contact Aggression can cause injury in farmed fish, especially
when competition for food is strong (Greaves &
Tuene, 2001). Subordinate fish can be prevented
from feeding (Cubitt, 2002), grow poorly and are
more vulnerable to disease (reviewed by Wedermeyer,
1997).
356 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
rainbow trout (Pottinger & Carrick, 1999) and in carp (Cyprinus carpio L,
Tanck et al., 2001). Finally, the complex nature of fish welfare means that
the various factors that impact on it may interact. For example, in aqua-
culture a given stocking density may generate signs of poor welfare under
Water quality deterioration Many adverse effects of poor water quality have been
described, with different variables interacting, e.g.
undisturbed salmonids use c. 300 mg of oxygen per kg of
fish per hour and this can double if the fish are disturbed.
For these species, access to aerated water is essential for
health (Wedermeyer, 1997). Immunoglobulin levels fall in
sea bass held at low oxygen levels (Scapigliati et al., 1999).
Poor water quality mediates density effects on welfare in
rainbow trout (Ellis et al., 2002).
Bright light and
photoperiod manipulation
Atlantic salmon avoid bright light at the water surface,
except when feeding (Ferno
¨et al., 1995; Juell et al., 2003).
Continuous light is associated with increased growth in
several species (e.g. cod: Puvanendran & Brown, 2002).
Food deprivation Dorsal fin erosion increases during periods of fasting
in steelhead trout (Winfree et al., 1998). Plasma glucose
increased in Atlantic salmon after 7 days without food,
but other welfare indices were unaffected (Bell, 2002).
Atlantic salmon deprived of food for longer periods
(up to 86 days) lost weight and condition, but this
stabilized after 30 days (Einen et al., 1998). Farmed
Atlantic salmon swim slower and fight less during
feeding bouts when fed on demand (Andrew et al., 2002).
Disease treatment Therapeutic treatments themselves may be stressful
to fish (e.g. Yildiz & Pulatsu, 1999; Griffin et al., 1999,
2002; Thorburn et al., 2001; Sørum & Damsga
˚rd, 2003).
Unavoidable contact
with predators
Brief exposure to a predator causes increased cortisol
levels and ventilation rate and suppressed feeding
(e.g. Metcalfe et al., 1987). Mortality and injury due to
attacks by birds and seals can be high among farmed
fish (e.g. Carss, 1993).
Slaughter All slaughter methods are stressful, but some are less
so than others (Robb et al., 2000; Southgate & Wall,
2001). Sea bass killed by chilling in ice water had lower
plasma glucose and lactate levels and showed less marked
behavioural responses than those killed by other methods,
in particular by asphysia in air and electro-stunning
(Skjervold et al., 2001; Poli et al., 2002), see Robb &
Kestin, 2002; Lines et al., 2003; Van de Vis et al., 2003.
ISSUES IN FISH WELFARE 357
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
TABLE VI. Examples of scientific studies of the impact of various aspects of ornamental
fish keeping on fish welfare
Practice Some demonstrated effects on welfare
Capture by sublethal poisons Marine tropical fish captured by sodium cyanide
suffer very high mortality for several weeks after
capture (Hignette, 1984). Clove oil is a better
alternative (Erdmann, 2002).
Transportation Mortality during capture of ornamental fish from
South America ranges may be as high as 30%.
A further 5–10% mortality is estimated to occur
during transportation and at the holding facilities
(Ferrez de Olivera, 1995). During the acclimation
period following importation mortalities can be up
to 30% (FitzGibon, 1993). Shipping of zebra fish
(Brachydanio rerio Hamilton, 1822) by road in oxygenated
bags elevated cortisol levels, but recovery is rapid on
transfer to aquaria (Pottinger & Calder, 1995).
Constraint in a confined space See above, under aquaculture.
Handling See above, under aquaculture.
Inappropriate densities/
species combinations
Lack of appropriate social environment (wrong
species or inappropriate numbers) is an important
cause of poor health in ornamental fish
(Etscheidt, 1995).
Poor water quality 81% of ornamental fish are held outside the optimal
pH range, 36% at inappropriate temperatures
(Etscheidt & Manz, 1992). Poor water quality is
the commonest cause of mortality in ornamental
fish (Schunck, 1980).
Deprivation of social contact Angelfish transferred singly to a new tank take
longer to resume feeding than those transferred in
groups of 3 or 5 (Gomez-Laplaza & Morgan, 1993).
Inappropriate food levels Inappropriate range and types of food can cause
poor health in ornamental fish (Etscheidt, 1995).
Inappropriate feeding is not usually a direct cause
of mortality in ornamental fish, but can be a
contributory factor (Schunck, 1980).
Unavoidable contact
with predators
In 19% ornamental tanks prey were housed in small
tanks in direct contact with predators (Escheidt &
Manz, 1992; Foggitt, 1997). See above under
aquaculture.
Disease treatment See above under aquaculture.
358 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
one flow regime (Ellis et al., 2002) or level of disturbance (Turnbull et al.,
2005), but not under other conditions.
3. It is therefore not possible to specify conditions that guarantee fish welfare:
The fact that the effect of varying one factor (such as density) frequently
depends on the status of many other factors (such as disturbance and water
quality), leads to the important conclusion that, even for a particular species,
gender and age of fish, we cannot guarantee the welfare by defining a simple
set of husbandry conditions. This in turn emphasizes the need for sensitive
on-the-spot indicators of welfare.
4. Fish are different from other vertebrates in ways that have important
implications for welfare: The effects of human activity on fish welfare are
not always what one would predict by extrapolating from birds and mam-
mals. One way to emphasize this point is to consider the influential frame-
work for animal welfare based on the so-called five freedoms (used by the
UK Farm Animal Welfare Council, 2005) or domains in which welfare may
be compromised (Mellor & Stafford, 2001) and how this framework might
be applied to fish.
Domain 1. Water and food deprivation, malnutrition. Animals should have ready
access to clean water and an appropriate diet in sufficient quantities and with a
composition that maintains full health and vigour.
Fish allow their body temperature to fluctuate with that of the environment
(i.e. they are ectothermic) and also show striking natural variation in appetite
and the evidence suggests that food deprivation is not such a critical aspect of
their welfare. Wild fish show marked changes in appetite (some temperature-
based and others depending on life-history events) that determine the effect of
food deprivation on welfare. In the winter juvenile salmon may become naturally
anorexic, eating little for weeks (Metcalfe et al., 1988). These fish will feed when
their energy reserves fall to a critical level, but up to this point, low rations would
not compromise welfare. On the other hand, maturing salmon show a spontan-
eous peak in appetite in spring, when nutrient reserves for migration and
spawning are accumulated (Kadri et al., 1996) and food deprivation at this
point may well compromise welfare. This is not to say that it is acceptable to
starve fish for long periods; they certainly have mechanisms that motivate them
to feed when their stomachs are empty and their nutritional reserves are low and
restricted food may have other effects such as increasing levels of aggression.
Under appropriate circumstances, however, periods of food deprivation may not
cause welfare problems and may well mitigate adverse effects of other husbandry
activities.
The natural diet of wild fish varies markedly between species and as with other
vertebrate groups, it is important to ensure that captive fish are given a nutri-
tionally appropriate diet, although in most species we do not know exactly what
this should be. The nutritional requirements of established farmed species are
well known, the industry has invested substantially in developing appropriate
diets and farmed fish no doubt enjoy better nutrition than their wild counter-
parts. However, for less well established species, there may be nutritional pro-
blems. For example, diets lacking in critical micronutrients impair welfare in
many species, according to a range of indicators such as high mortality,
ISSUES IN FISH WELFARE 359
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
morphological abnormalities, poor immune function, abnormal behaviour, poor
feeding, impaired sensory function and slow growth (De Silva & Anderson,
1995). As far as water deprivation is concerned, juvenile salmon transferred to
sea water too soon in the smolting process can become severely dehydrated, and
may die as a consequence (Southgate & Wall, 2001).
Domain 2. Environmental challenge. Animals should have a suitable environment,
including shelter and a resting area, whether outdoors or indoors.
Since fish are in intimate contact with their environment through the huge
surface of their gills, water quality (in terms of dissolved oxygen, ammonia and
pH) and the presence of contaminants (organic and inorganic pollutants) are
probably the most critical aspects of the environment for fish welfare and also
the best defined. Optimal conditions vary markedly between species. For exam-
ple, cyprinid fish are very tolerant of low dissolved oxygen levels whereas
salmonid fish are not (Wedemeyer, 1996). Within the cichlids, some species are
found naturally in waters with a pH as low as 4, whereas others flourish at pH as
high as 9 (Lowe-McConnell, 1991; Ross, 2000). The flow characteristics of the
fish’s natural habitat are also of importance, some species preferring static water,
others tolerating or preferring relatively high flow rates. The extent to which the
nature of the substratum is important for welfare, particularly in bottom-dwelling
species, is not known, but several species have been shown to grow better when
shelters are available (e.g. Gwak, 2003). In general, except for the high densities
involved, the environment experienced by farmed fish is markedly less complex
than that experienced by their wild counterparts. Several behavioural deficits in
hatchery-reared fish (for example, inability to handle live prey and impaired
antipredator responses) have been ascribed developmental effects of such defi-
ciencies and various kinds of environmental enrichment have been used to
reverse these deficits (reviewed by Huntingford, 2004).
Domain 3. Disease, injury and functional impairment. Disease should be prevented
or rapidly diagnosed and treated.
Diseases frequently indicate an underlying environmental problem, so diag-
nosing and controlling a disease must always take account of the whole system
and not consider the fish alone. Diseases of fish are mostly species and system
specific and many are poorly understood, but methods of prevention are avail-
able for an increasing number (Biering et al., 2005; Ha
˚stein et al., 2005).
Domain 4. Behavioural/interactive restriction. Animals should have sufficient
space, proper facilities and where appropriate, the company of the animal’s own
kind.
Many species form dense schools in the wild and this is important when
assessing the welfare of such species if held at high density in captivity, since
being held at densities that are too low rather than too high may impair welfare.
This is a case where it can be misleading to extrapolate from birds and mammals
to fish. As discussed earlier, we do not know whether fish such as salmon are
motivated to migrate by a particular route (as opposed to finding food or
swimming long distances, which they can do in farm cages for example); if this
were the case, their behavioural needs might not be met in a sea cage. The
360 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
concept of ‘facilities’ may be inappropriate for fish, though some species need
shelter or cover, some may require nesting material when breeding, some need
tough structures on which to chew (Etscheidt, 1995).
Domain 5. Mental and physical suffering. Conditions that produce unacceptable
levels of anxiety, fear, distress, boredom, sickness, pain, thirst, hunger and so on
should be minimized.
Domain 5 is critical since it relates adverse experience to emotional response.
As discussed in section 2, the subjective experiences of fish are very hard to
understand, so it is not easy to identify ‘‘conditions that produce unacceptable
levels of anxiety . . . and so on.’’ Most of the cues that are employed to identify
distress and fear responses in other vertebrates are simply not accessible for fish.
For example, there are no direct parallels for facial or vocal signalling in fish.
Greater understanding of cognitive processes in fish is needed before we can
make the link between welfare and suffering in this group of organisms.
CONCLUSIONS
There are >30, 000 species of teleost fish and we know only a little about
conditions that promote welfare for just a few of these, but even so a picture is
beginning to emerge (partial and blurred at present) of how various human
activities impinge on the welfare of fish and therefore of what might be done
to improve matters. This review highlights the need for better knowledge and a
fuller understanding of fish welfare. Some of these areas of ignorance concern
issues that are fundamental to the whole concept of fish welfare, what it means
and how it might be measured. These areas are listed below:
1) The single most important area of ignorance is a lack of understanding of
the mental capabilities of fish and how objectively measurable responses to
challenges (physical damage and physiological and behavioural responses)
are associated with subjective states of well-being or suffering.
2) After that, given the importance of the concepts of natural being good, for
each exploited species, it is important to discover whether there are actions
that the fish are highly motivated to perform and that, like nest building in
domestic hens, may be described as ‘behavioural needs’.
3) We also need to know more about diseases in fish, about the links between
stress, immune function and disease states and therefore about the relation-
ship between health and welfare.
4) In practical terms, a better array of welfare indicators (for example, easily
observed morphological and behavioural cues) is requires for everyday use
in circumstances where time consuming scrutiny of fish using laboratory-
based tests is impossible.
5) Other gaps in existing knowledge are also important, but will be somewhat
easier to fill because they involve expanding the information already avail-
able for some species and in some contexts.
6) A certain amount is known about the effect of angling and aquaculture
practices on fish welfare, but there is very little information on the welfare of
ornamental fish, especially from capture to point of sale. Questions also
ISSUES IN FISH WELFARE 361
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
remain about conditions within aquaria and ornamental ponds. What are
the effects of being confined in a small, exposed space, of social isolation or
of frequent interactions with a predator?
7) As with the welfare of ornamental fish there is very little information on the
welfare of fish in the context of commercial fisheries
8) Where both angling and aquaculture are concerned, we know a reasonable
amount about welfare of salmonids, but very little about other kinds of fish
that are reared commercially or caught by anglers
9) Even for the well studied species and well documented effects, the exact
mechanism by which the adverse effects come about are unknown. For
example, there is plenty of evidence of poor welfare in salmon and trout
held at very high densities, but it is not clear whether this is the result of poor
water quality, high levels of aggression, simple physical damage or some
other process (Ellis et al., 2002). It is important to clarify such issues in order
to decide on appropriate remedial measures.
By spelling out current understanding on the welfare of fish, we hope that this
review will contribute to debate on the subject. This is a difficult area to review,
because many academic disciplines have an interest in it, because complex
concepts are involved that are hard to define and because there are large areas
of ignorance and, consequently, of disagreement. In this document a pragmatic
working position has been taken on a number of important questions (whether
fish suffer and whether this matters, for example), recognizing that this position
may have to be changed in the light of facts that emerge in the future. In spite of
these difficulties, a great deal of painstaking research has shown how fish
respond to the adverse events that they experience in nature and how these
could be used to probe their welfare.
The authors thank two anonymous referees for constructive comments and the numer-
ous individuals that have read, commented and provided information relevant to the
preparation of this manuscript.
References
Association for the Study of Animal Behaviour (ASAB) (2001). Guidelines for the
treatment of animals in behavioural research and teaching. Animal Behaviour 61,
271–275.
Abbott, J. C. & Dill, L. M. (1989). The relative growth of dominant and subordinate
juvenile steelhead trout (Salmo gairdneri) fed equal rations. Behaviour 108,
104–111.
Alanara, A. (1997). Feeding behaviour and growth performance in fish within a
dominance hierarchy. In Voluntary Food Intake in Fish (Houlihan, D., Kiessling, A.
& Boujard, T., eds), pp. 10. Oxford: Blackwell Science Ltd.
Andrew, J. E., Noble, C., Kadri, S., Jewell, H. & Huntingford, F. A. (2002). The effects
of demand feeding on swimming speed and feeding responses in Atlantic salmon
Salmo salar L., gilthead sea bream Sparus aurata L. and European sea
bass Dicentrarchus labrax L. in sea cages. Aquaculture Research 33, 501–507.
doi:10.1046/j.1365-2109.2002.00740.x
Appleby, M. C. (1999). Tower of Babel: Variation in ethical approaches, concepts of
welfare and attitudes to genetic manipulations. Animal Welfare 8, 381–390.
362 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
Appleby, M. C. & Sandøe, P. (2002). Philosophical debate on the nature of well-being:
implications for animal welfare. Animal Welfare 11, 283–294.
Balm, P. H. M. & Pottinger, T. G. (1995). Corticotrope and melanotrope POMC-derived
peptides in relation to interrenal function during stress in rainbow trout
(Oncorhynchus mykiss). General Comparative Endocrinology 98, 279–288.
Bandeen, J. & Leatherland, J. F. (1997). Transportation and handling stress of white
suckers raised in cages. Aquaculture International 5, 385–396.
Barnett, C. W. & Pankhurst, N. W. (1998). The effects of common laboratory and
husbandry practices on the stress response of greenback flounder (Rhombosolea
tapirinia,Gu
¨nther 1862). Aquaculture 162, 113–329.
Barton, B. (1997). Stress in sunfish: past, present and future – a historical perspective. In
Fish Stress and Health in Aquaculture, Society for Experiment Biology, Seminar
Series 62 (Iwama, G. K., Pickering, A. D., Sumper, J. P. & Schreck, C. B., eds),
pp. 1–31. Cambridge: Cambridge University Press.
Barton, B. A. (2000). Salmonid fishes differ in their cortisol and glucose responses to
handling and transport stress. North American Journal of Aquaculture 62, 12–18.
Bermond, B. (1997). The myth of animal suffering. In Animal Consciousness and Animal
Ethics (Dol, M., Kasanmoentalib, S., Lijmbach, S., Rivas, E. & van den Bos, R.,
eds), pp. 125–143. Assen: Van Gorcum.
Beukema, J. J. (1970). Angling experiments with carp: decreased catchability through one
trial learning. Netherlands Journal of Zoology 20, 81–92.
Bevelhimer, M. A. (1996). Relative importance of temperature, food and physical
structure to habitat choice by smallmouth bass in laboratory experiments.
Transactions of the American Fisheries Society 125, 274–283.
Biering, E., Villoing, S., Sommerset, I. & Christie, K. E. (2005). Update on viral vaccines
for fish. Developmental Biology 121, 97–113.
Boglione, C., Gagliardi, F., Scardi, M. & Cataudella, S. (2001). Skeletal descriptors and
quality assessment in larvae and post-larvae of wild and reared gilthead bream
(Sparus aurata L.). Aquaculture 192, 1–22.
Borski, R. J. & Hodson, R. G. (2003). Fish research and the institutional animal care and
use committee. Ilar Journal 44, 286–294.
Braithwaite, V. A. (in press). Cognitive ability in fish. Fish Physiology.
Braithwaite, V. A. & Huntingford, F. A. (2004). Fish and welfare: do fish have the
capacity for pain perception and suffering. Animal Welfare 13, 587–592.
Brett, J. R. (1979). Environmental factors and growth. In Rainbow Trout Growth in
Circular Tanks. Consequences of Different Loading Densities. US Department of
the Interior, Fish and Wildlife Technical Report 86, 16.
Broadhurst, M. K. & Barker, D. T. (2000). Effects of capture by hook and line on plasma
cortisol, scale loss and survival in juvenile mulloway (Agrysomus hololepidotus).
Archive of Fishery and Marine Research 48, 1–10.
Broglio, C., Rodriguez, F. & Salas, C. (2003). Spacial cognition and its neural basis in
teolost fishes. Fish and Fisheries 4, 247–255. doi:10.1046/j.1467-2979.2003.00128.x
Broom, D. M. (1998). Welfare, stress and the evolution of feelings. Advances in the Study
of Behaviour 27, 371–403.
Brown, C. & Warburton, K. (1999). Social mechanisms enhance escape responses in
shoals of rainbowfish (Melanotaenia duboulayi). Environmental Biology of Fishes
56, 455–459.
Burt de Perera, T. (2004). Fish can encode order in their spacial map. Proceedings of the
Royal Society London B 271, 2131–2134.
Cahu, C., Infante, J. Z. & Takeuchi, T. (2003). Nutritional components affecting skeletal
development in larval fish. Aquaculture 227, 245–258.
Carss, D. N. (1993). Cormorants Phalacrocorac carbo at cage fish farms in Argyll,
western Scotland. Seabird 15, 38–44.
Chandroo, K. P., Cooke, S. J., McKinley, R. S. & Moccia, R. D. (2005). Use of
electromyogram telemetry to assess the behavioural and energetic responses of
rainbow trout, Oncorhynchus mykiss (Walbaum) to transportation stress.
Aquaculture Research 36, 1226–1238. doi:10.1111/j.1365-2109.2005.01347.x
ISSUES IN FISH WELFARE 363
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
Chandroo, K. P., Duncan, I. J. H. & Moccia, R. D. (2004a). Can fish suffer?: perspectives
on sentience, pain, fear and stress. Applied Animal Behaviour Science 86, 225–250.
Chandroo, K. P., Yue, S. & Moccia, M. D. (2004b). An evaluation of current perspectives
on consciousness and pain in fishes. Fish and Fisheries 5, 281–295. doi:10.1111/
j.1467-2679.2004.00163.x
Chopin, F. S. & Arimoto, T. (1995). The condition of fish escaping from fishing gears – a
review. Fisheries Research 21, 315–327.
Christenson, J., Boman, A., Langerback, P. & Grillner, S. (1988). The dorsal cell, one
class of primary sensory neuron in the lamprey spinal cord I. Touch, pressure but
no nociception – a physiological study. Brain Research 440, 1–8.
Conte, F. S. (2004). Stress and the welfare of cultured fish. Applied Animal Behavioural
Science 86, 205–223.
Cooke, S. J. & Philipp, D. P. (2004). Behavior and mortality of caught-and-release
bonefish (Albula spp.) in Bahamian waters with implications for a sustainable
fishery. Biological Conservation 118, 599–607.
Cubitt, F. S. (2002). Individual growth patterns in farmed Atlantic salmon: the role of
social interactions. PhD Thesis, University of Glasgow.
Czanyi, V. & Doka, A. (1993). Learning interactions between prey and predator fish.
Marine Behaviour and Physiology 23, 63–78.
Davis, K. B., Griffin, B. R. & Gray, W. L. (2002). Effect of handling stress on susceptibility
of cannel catfish Ictalurus punctatus to Ichthyophthirius multifiliis and channel
catfish virus infection. Aquaculture 214, 55–66.
Dawkins, M. S. (1998). Evolution and animal welfare. Quarterly Review of Biology 73,
305–328.
Dawkins, M. S. (2004). Using behaviour to assess animal welfare. Animal Welfare 13, 3–7.
De Silva, S. S. & Anderson, T. A. (1995). Fish Nutrition in Aquaculture. London:
Chapman and Hall.
DuBois, R. B., Margenau, T. L., Stewart, R. S., Cunningham, P. K. & Pasmussen, P. W.
(1994). Hooking mortality of northern pike angled through the ice. North
American Journal of Fish Management 14, 769–775.
Duncan, I. J. H. & Fraser, D. (1997). Understanding animal welfare In Animal Welfare
(Appleby, M. C. & Hughes, B. O., eds), pp. 19–47. Wallingford, Oxon: CAB
International.
Dutil, J. D. & Lambert, D. (2000). Natural mortality from poor condition in Atlantic
cod (Gadus morhua). Canadian Journal of Fisheries & Aquaculture Science 57,
826–836.
Ehrensing, R. H., Michell, G. F. & Kastin, A. J. (1982). Similar antagonism of morphine
analgesia by MIF – 1 and naloxone in Carassius auratus.Pharmacology,
Biochemistry and Behaviour 17, 757–761.
Einen, O., Waagan, B. & Thamassam, M. S. (1998). Starvation prior to slaughter in
Atlantic salmon (Salmo salar) – Effects on weight loss, body shape, slaughter- and
fillet- yield, proximate and fatty acid composition. Aquaculture 166, 85–104.
Ellis, T., James, J. D., Stewart, C. & Scott, A. P. (2004). A non-invasive stress assay
based upon measurement of free cortisol released into the water by rainbow trout.
Journal of Fish Biology 65, 1233–1252. doi:10.1111/j.0022-1112.2004.tb00499.x
Ellis, T., North, B., Scott, A. P., Bromage, N. R., Porter, M. & Gadd, D. (2002). The
relationships between density and welfare in farmed rainbow trout. Journal of Fish
Biology 61, 493–531. doi:10.1111/j.1095-8649.2002.tb00893.x
Endler, J. A. (1986). Defence against predators. In Predator – Prey Relationships.
(Feder, M. E. & Lauder, G. V., eds), pp. 109–134. Chicago: University of
Chicago Press.
Errea, A. & Danulat, E. (2001). Growth of the annual fish Cynolebias viarius
(Cyprodontiformes) in the natural habitat compared with laboratory conditions.
Environmental Biology of Fishes 61, 261–268.
Etscheidt, J. (1995). Kriterien zur Beurteilung der Haltung von Susswasseraquarienfischen
im Zoohandel. Tierarztl Prax 50, 196–199.
364 F. A. HUNTINGFORD ET AL.
#2006 The Authors
Journal compilation #2006 The Fisheries Society of the British Isles, Journal of Fish Biology 2006, 68, 332–372
Etscheidt, J. & Manz, D. (1992). Suswasseraquaristik und tierarzliche Praxis. Teil 2:
Untersuchungen zur artgerechten Haltung von Zierfuschen. Tierarztl Prax 20,
221–226.
Ewing, R. D. & Ewing, S. K. (1995). Review of the effects of rearing density on the
survival to adulthood for Pacific salmon. Progressive Fish-Culturist 57, 1–25.
Food and Agriculture Organisation of the United Nations (2004). The State of World
Fisheries and Aquaculture (Sofia), p. 153. Rome: Food and Agriculture
Organisation of the United Nations.
Farbridge, K. J. & Leatherland, J. F. (1992). Plasma growth hormone levels in fed and
fasted rainbow trout (Oncorhynchus mykiss) are decreased following handling
stress. Fish Physiology and Biochemistry 10, 67–73.
Ferguson, R. A., Kieffer, J. D. & Tuffs, B. L. (1993). The effects of body size on the
acid-base and metabolite status in the white muscle of