Assessment - Introduction
The objective determination of animal welfare relies on the selection, collection and interpretation of different parameters. Such systems need to be sensitive, practical and robust, as well as comprehensive and meaningful. Although there is currently no consensus on the most appropriate means of assessing fish welfare, systems such as the health and condition profile (Goede & Barton, 1990) provide possible guidelines for ways to assess individual fish.
The welfare status of an animal can be described to lie somewhere on a continuum between good and poor welfare (Appleby & Hughes, 1997), although this is undoubtedly an oversimplification of welfare, which is a multidimensional phenomenon. There are numerous examples of potential poor fish welfare, many of which were identified in the 126 recommendations made by the UK Farm Animal Welfare Council’s report on the Welfare of Farmed Fish (Anon., 1996). However, other than the absence of obvious signs of poor welfare, our understanding of exactly what constitutes good welfare for a farmed fish is very limited. In light of this, a pragmatic approach could be to focus on safeguarding fish welfare by identifying risk factors and indicators that are associated with an increased risk or prevalence of poor welfare.
Schematic representation of the continuum between good and poor welfare
One of the major challenges associated with assessing the welfare of fish compared with terrestrial animals is the difficulty associated with observing fish underwater. Several recent studies have used hydroacoustics to monitor things such as swimming depth and elective stocking densities of Atlantic salmon in production cages ( Juell & Fosseidengen, 2004 ; Johansson et al. 2005), although the very nature of aquaculture systems means that incorporating fish behaviour in a quantifiable manner is always likely to be limited.
In terrestrial animal welfare research there has recently been a movement away from complex and expensive analysis of numerous different parameters towards systems that place an emphasis on behavioural observations incorporating preference testing and demand curves (i.e. how hard an animal will work to obtain a commodity e.g. food, company, bedding material). One simple yet innovative system method that relies heavily on this type of approach proposes that animal welfare can be assessed by asking two questions: Are the animal healthy? Do they have what they want? (Dawkins 2004).
This leads on to an interesting question. What does a fish want? Or perhaps more importantly, what does a farmed fish want? We can gain some idea regarding what a fish might want from knowledge of behaviour displayed by wild populations of fish such as the types of habitat that they occupy, what they eat and whether they are shoaling or solitary. However, the constraints of commercial aquaculture systems will invariably mean that there is a limit to how far these ‘wants’ can be addressed. There are also challenges that are likely to be impossible to meet in commercial systems e.g. Atlantic salmon are highly motivated to reproduce to the extent that they will swim vast distances to return to the river from which they originated and usually kill themselves in the process.
We currently have only a limited understanding with regard to what a fish wants, or what constitutes good fish welfare. Until we know more about these two key questions the most logical approach would seem to focus on the removal or minimisation of risk factors that we know are associated with poor welfare.