432. Nimon, A.J. and Broom, D.M. 1999. The welfare of farmed mink (Mustela vison) in relation to housing and management : a review. Animal Welfare, 8, 205-228. Pre-publication version THE WELFARE OF FARMED MINK (MUSTELA VISON) IN RELATION TO HOUSING AND MANAGEMENT: A REVIEW A J Nimon and D M Broom Department of Clinical Veterinary Medicine, University of Cambridge, Madingley Road, Cambridge CB3 DES, UK Abstract Early research on farmed mink was predominantly concerned with increased productivity; however, in recent years there have been an increasing number of studies related to welfare. The biology offeral mink has also become better understood, and such knowledge can aid in the assessment of welfare on farms, or in the interpretation of problems related to captivity. This paper is a comprehensive review of research pertinent to the welfare of farmed American mink, Mustela vison, in relation to their housing and management. It indicates how housing conditions might be changed to improve welfare, and where our present knowledge is insufficient. Many significant aspects of mink behaviour in the wild, such as their lack of social contact, their tendency to travel long distances and use several den sites, and regular swimming and diving, are denied them in captivity. Farmed mink also show high levels of stereotypy, suggesting that their welfare is not good. Welfare may be improved by appropriate environmental enrichment and changes in the social environment of farmed mink. In general, studies aimed at improving housing conditions have been limited in scope and outlook. Keywords: animal welfare, housing, management, mink, Mustela vison, recommendations, stereotypies Introduction There is debate about whether or not animals should be killed for their fur. The personal decision whether to wear or not to wear a fur garment may be different in a very cold region, where alternatives to fur garments are difficult to [md, from that in regions where alternatives are readily available and the fur is a luxury worn to impress others. However, the assessment of the welfare of fur-bearing animals is independent of such considerations. Welfare refers to the state of an animal at a specific time and can be good or poor irrespective of what people think about the morality of such usage of animals. If the individual animal is having difficulty in coping with its environment, or is failing to cope, then its welfare is poor; but if strongly preferred resources and opportunities for behaviour are available, and normal behaviour can be shown, then good welfare is indicated (Broom & Johnson 1993; Broom 1996). A wide range of indicators of welfare can and should be used when welfare is being assessed. The selection and interpretation of welfare indicators should Thewelfare offarmedmink least 5or6weeks ofage, yet in the wild they remain with their mothers for longer. In astudy of a wild population in Scotland, the first occasion on which a kit was observed outside the den alone was at 6-7 weeks of age (Dunstone 1993); in another study, kits were weaned, leaving their natal territory at 11-12 weeks (Gerell1970). On farms, they may be weaned as early as 5 weeks (Joergensen 1985). This is postulated to be the cause of behavioural maladaptions, aswill be discussed below. Aquatic lifestyle Wherever mink occur in the wild there will be water close by, in fact most mink activity in North America and Europe occurs in water, or within 100-200 m of it (Dunstone 1993). A radio-tracking study of 19mink in southern Finland found that both sexes swam distances of 250m almost daily, sometimes twice a day (Niemimaa 1995), and a large proportion of the wild mink's diet is generally derived from aquatic sources (Day & Linn 1972; Poole & Dunstone 1976; Birks & Dunstone 1985; Dunstone & Birks 1987; Niemimaa & Pokki 1990, cited in Niemimaa [1995]; Dunstone 1993). Laboratory studies have found that farm-bred mink can be trained to retrieve objects from atank with no reinforcement other than from the moving cork or cotton reel itself (Poole & Dunstone 1976). Although mink do not see particularly well underwater (Sinclair et a/1974; Dunstone & Sinclair 1978a), they tend to retrieve prey by locating it from the surface, then making brief dives of between 5 and 20 seconds and sometimes longer (Poole & Dunstone 1976; Dunstone & Sinclair 1978b). Despite the fact that mink seem better adapted to locomotion on land than in water, they have partially webbed feet, and their swimming speed is much greater than that of terrestrial mammals and half that of the otter, Lutra /utra (Dunstone 1983; 1993). It seems clear that swimming and diving are highly significant aspects ofthe mink lifestyle. Other behaviours and activity patterns Mink have been noted for their agility and flexibility (Rice 1967) and for their ability to climb trees (Burton 1979). Radio-tracking studies of 20 mink (9 male, 11female) in Britain determined that mink spent 5 per cent to 20 per cent of the 24h period (on average, approximately 3h) actively foraging outside their dens; male mink spent just under 2h travelling, and female mink travelled for around half this time (Dunstone & Birks 1985). Mink travel within a definable home range. Home range lengths of individual radio-tracked feral mink have been found to vary from 0.5-5.94 km (Gerell 1970; Birks & Linn 1982), although Birks and Linn (1982) considered these upper estimates to have been affected by the behaviour of males with unstable social environments (such males travel distances up to 30km [Niemimaa 1995]). They estimated mean home range length to be between 1and 3km. A recent study of five mink in eastern Tennessee measured mean home range lengths of between 5.6 and 11.1km in autumn and early winter (Stevens et a/1997). Mink use anumber of dens within these ranges: Gerell (1970) recorded mink using between two and five dens, changing dens on successive nights to aden usually 500m distant (although sometimes as far as 2000m); Stevens et a/ (1997) recorded the use of between 8 and 24 dens per mink, with overnight trips between them of up to 4300m. Dens are not excavated by the mink themselves but tend to derive from rabbit holes or natural crevices, for example around tree roots. Mink favour specific dens rather than apparently similar sites for reasons that are not evident (Dunstone 1993). Travelling and the use of several den sites are, therefore, significant components of feral mink behavioural biology. The fact that physical activity is important to mink has also been suggested by experimental studies showing that caged mink willrun on awheel fornoreward (Zielinski 1986). Animal Welfare 1999, 8: 205-228 207 Nimon andBroom It is not known for certain that mink actively defend their home ranges as territories, however male territories never overlap (Dunstone 1993). Females may tolerate the intrusion of kits for part of the year, and can on occasion have territories partially overlapping with males, especially during the breeding season (Dunstone & Birks 1983; Dunstone 1993). Severe aggression has been reported among unfamiliar adults forced together by captivity (Rice 1967; Heller & Jeppesen 1986; Dunstone 1993). However, it is possible to keep captive mink which are familiar with one another together (eg those housed together since weaning orreared without weaning; see, Social conditions oj housing). Sensory biology The mink's anal gland isamajor source of specialized odour compounds (Brinck et aI1978). Mink perform marking by dragging this gland over the ground, or by depositing faeces (Dunstone 1993). Brinck et al (1978) suggested that mink may be able to discriminate between individuals, and that such an ability may help in detecting intruders in the home range. Unpublished research (Robinson 1987, cited in Dunstone [1993]) supports minks' use of scent to identify one another, to distinguish between known and unknown individuals, and between males and females. Experimental studies have shown that ferrets, Mustela Jura, have similar capabilities (Clapperton et al 1988). Gerell (1968, cited in Dunstone [1993]) first noted that mink leave scats in prominent positions within their home ranges, where the scent islikely to have carried furthest. Dunstone (1993 p 139) states 'the violent reaction of a mink to the scent of an intruder leaves one inno doubt about the value of ascat'. Mink are sensitive to lighting conditions, as the processes of fur growth and sexual development are dependent on exposure to an appropriate photoperiod. Melatonin is the photoperiodic signal for the autumnal weight increase and moult (Valtonen et aI1995). The shorter day length controls such changes via melatonin excretion. Other experimental studies have demonstrated that short day lengths during autumn stimulate spermatogenesis in October to November, and this effect is enhanced by additional month-long illumination before the short-day period (Klotchkov et alI985). Jallageas et al (1994) and Gulevich et al (1995) have further demonstrated that artificial photoperiods can cause significant changes in mink gonadal function and activity. The extent of modification in relation to captivity There is anecdotal evidence suggesting that mink on farms have become less fearful and tamer in response to contact with humans. Hansen (1996) noted that early reports (eg in the 1940s, see Shackelford [1984]) described mink fleeing into the nest box when people were close by. Hansen (1996) stated that mink now often respond to the presence ofa human with curiosity. He suggested that nervous individuals might have been eliminated by natural selection. Hansen's own study (Hansen 1996) involved 3000 farm mink over 6 years in which he selected and bred mink for exploratory, aggressive or fearful behaviour in the presence of ahuman. He found that itwas possible to select for fearful behaviour, but not for exploratory behaviour; in fact those mink selected and bred for exploratory responses showed a significant increase in fearful behaviour. Thus, while this, and earlier work with similar results (eg Hansen [1991b]); Houbak [1990], both cited in Braastad [1992]), suggested that certain behavioural traits may be inherited, no results were achieved which suggested that farmers could selectively breed for less fearful, and potentially better adapted, animals. 208 Animal Welfare 1999. 8: 205-228 Thewelfare offarmedmink Standard farm housing conditions Details of standard housing conditions on mink farms appear in Joergensen (1985), Mason (1991a), M0ller (1991b), and Hansen et al (1994). These indicate that mink cages have approximately 0.27m2 (90x30 cm) of floor space, and a height of 30--40 em. A single nest box, with a floor area of 0.06m2, is attached to the cage via a circular hole. The cages are made of square, galvanized steel mesh, allowing faeces to fall through, and the nest boxes are generally made of wood, but may have a mesh roof. Drinking water is continuously available via an automatic system: adrinking nipple inside the cage isattached to ahose. The water is protected from freezing via a circulation or heating system. Mink are fed once or twice aday with anutritious puree that isplaced on top oftheir cages. Cages are made in six- or eight-cage sections (joined side-by-side) and placed in rows in sheds. Sheds generally contain two ormore parallel rows, with aroof above and walls which are either left open or filled in with wire mesh (Mason 1991a) or a screen made of heavy plastic or sacking stretched on wooden frames (Joergensen 1985). Joergensen (1985 p 38) recommends that a hedge be grown around the sheds for 'cold does not affect mink, but draught istheir worst enemy'. After weaning (usually at approximately 7 weeks of age [Mason 1996]), mink may be housed individually, or placed in male/female sibling pairs in one cage. They stay this way until pelting, at approximately 8 months of age. At this point, some individuals may be selected forbreeding purposes and rehoused. It is clear that the housing conditions of mink on farms differ markedly from those of mink in the wild in terms of the rearing of kits, the proximity of other mink, and the opportunities for swimming and diving, roaming, using different dens, and engaging in physical activity. Undoubtedly, the olfactory environment and lighting conditions on farms also differ greatly from conditions in the wild. Health and disease on mink farms Mink have been described as 'extremely hardy' (Rice 1967 p 73) and 'healthy vigorous animals [that] rarely get sick if they have adequate food, water and clean conditions' (Dunstone 1993pp 183-4). The incidence of disease appears to be relatively low: Wahlstrom (\987, cited in Harri et al [1995]) reported that the summer mortality rate on 16 farms in Sweden averaged 1.3per cent for adult males, 1.5per cent for adult females and 2.1 per cent for kits. Joergensen (1985) recommended that good housing conditions include well-spaced sheds and that good hygiene should involve both cleaning and disinfection to prevent manure and dirt from becoming chronic reservoirs for viruses and bacteria. It is further recommended that farmers be vigilant and regularly (eg daily) inspect their mink for any signs ofdisease. Disease is always a threat. Rice (1967) considered that there are three diseases which, if introduced, will seriously deplete a colony: botulism, distemper and Aleutian disease. Botulism is said to occur only infrequently, but when it does occur, the powerful toxin produced by the anaerobic bacterium Clostridium botulinum can cause widespread mortality throughout a colony. Only a small portion of infected food may kill a mink. Distemper is also very infectious and will easily spread from one infected mink to the majority of the colony during the 8-14 day incubation period. One-time vaccination can protect animals fromthese diseases for life; however, Joergensen (1985) noted that vaccination is costly, and the actof vaccination may risk spreading disease via infected syringes. He recommended Animal Welfare /999, 8: 205-228 209 NimonandBroom frequent cleaning under cages, and disinfecting of sheds, while the decision to vaccinate is consIdered and made on an annual basis. Aleutian disease can be a major cause of mortality, and there is no treatment for infected animals. Investigations involving over 5500 autopsies in Argentina indicated that Aleutian disease was the most important cause of premature death (Martino et al 1991). Airborne transmission may be substantial, although mechanical transmission is probably more to blame (Jackson et al 1996a). Jackson et al (1996b) reported that transmission may be enhanced by the use of contaminated toenail clippers for blood collection. Farmers can, however, test forthe disease and cull infected animals from the stock (Dunstone 1993). Nursing disease is also reported among adult female farmed mink in Europe (M011er 1991b). A study of the pattern and relative frequency of diseases in adult female mink during the lactation period on 48 farms in southern Ontario found that nursing disease was the most common diagnosis (56%), the mortality rate among this group ranging from 0.2 per cent to 10.1per cent, with a median of 1.9per cent (Schneider & Hunter 1993a). Variation between farms was associated with the type of water source, size of the farm and source of feed (Schneider et al 1992). It has also been suggested that nursing disease may result from energy depletion due to lactation (Schneider & Hunter 1993b). While farmed mink may be relatively hardy, there exist no published reports of the incidence, causation and appropriate prevention of disease among farmed mink in Europe. Harri et al (1995) examined the possibility that farmed mink experience a high incidence of stomach ulcers. An incidence of ulcers in 35 per cent to 40 per cent of kits, and in 55 per cent of adults, was reported by Wahlstrom (1987, cited in HaITiet al [1995]). This result was then cited to demonstrate that approximately half of all farmed mink have ulcers, and to conclude that ulcers were the result of stress and an inability to cope with farm conditions (Kollberg & Bjorkland 1989, cited in Harri et al [1995]). Examining singly housed mink under standard conditions, Harri et al (1995) found a lower incidence of stomach ulcers (24%). Most of these ulcers rated only one on a five-point scale of severity. However, the study also found that mink subject to aversive treatments (such as frequent, regular immobilization in a restrainer) failed to develop significantly more ulcers than control animals. They concluded that the incidence of stomach lesions was not a useful indicator of poor housing conditions. The nest box As discussed above, wild mink use a number of den sites, change dens daily, and prefer particular dens for as yet unidentified reasons. This would suggest that the presence of at least one nest box is essential. Results from Hansen (1988) suggested that male and female mink housed in standard-size cages without nest boxes had poorer fur quality than those with access to nest boxes. This was thought to result from their lying against the wire mesh, and from the inhibition of blood circulation and nourishment of the hair follicles due to additional heat loss. Reduced fur quality can be considered a sign of poor welfare - and an indication that an animal is having difficulty in coping. Results also suggested that the absence of a nest box might lead to an increased metabolic rate and higher adrenal weights. The latter effect may indicate chronic stress, reflecting the adrenal gland's capacity for cortisol secretion (Mason 1992). Determining the effect(s) of captivity without access to a nest box on physiological parameters in mink has been complicated by results which could support contradictory hypotheses (Hansen & Brandt 1989). However, later investigations (Hansen & Damgaard 210 Animal Welfare 1999,8: 205-228 Thewelfareoffarmedmink 1991a) showed that male and female mink in standard cages without nest boxes had a lower level of circulating eosinophil leucocytes than mink with nest boxes - a reaction seen in mink subjected to acute immobilization (Heller & Jeppesen 1985). The effects of the absence of a nest box were comparable with those produced by 30-min immobilization sessions. Furthermore, female mink housed without nest boxes also had higher cortisol levels than those housed with nest boxes. In another experimental study, female mink deprived of nest boxes during the whelping period were found to lose significantly more kits than other experimental groups, and weight gain among their kits was significantly lower (Moller 1990). Hansen et al (1992; 1994) found that female mink in cages without nest boxes showed more stereotypic behaviour than those housed with nest boxes. They considered that this might arise from frustration at the inability to avoid contact with neighbours. These mink also had agreater food intake which did not result in more rapid, or greater, weight increase. Hansen et al (1992) stated that the growth rate was higher in mink housed with nest boxes, although it is unclear how this was measured. These results accord with those of Hansen (1988), iethat mink housed without nest boxes had ahigher metabolic rate. The importance of the nest box for the health and welfare of farmed mink has now been amply demonstrated for both male and female animals. No further experimentation should deprive mink of this basic necessity. Rather, any further experiments could examine how the provision of nest boxes might be improved: for example, work by Moller (1990) suggested that a 'drop-in' bottom (ie a false bottom) in the nest box was associated with better kit survival than in standard nest boxes. He supposed that this made it easier for amink to keep her kits from wandering and, thus, protect them from hypothermia. An obvious further issue forinvestigation isthe preference ofmink formore than one nest box (see below). Drinking water Moller (1991a) identified three points with regard to the provision of drinking water for mink. i)Water must be available 24h aday, for captive mink drink frequently, day and night. Standard farm conditions meet this recommendation (as above). ii) While water is often circulated so as to prevent it freezing and becoming inaccessible, it may reach high temperatures in summer. iii) The water intake of mink and kits during the female's lactation period is an important issue. Moller's observations showed that kits began to eat food at 4 weeks of age, but did not begin to drink water until 6weeks of age: therefore, their moisture requirements must be met via food, milk, and licking the mother's saliva. During this period, Moller (1991a) noted that females may experience dehydration, and weight loss (particularly during hot summers), and that 'nursing disease in the female and cannibalism among the kits' (Moller 1991ap 5) occurs. He concluded that the earlier kits begin to drink, the greater therelief forthe mother and her kits. Supplementary watering systems It was not possible to determine how widespread or serious the problem of potential kit and mother dehydration may be in Europe. However, it is clear that water intake during the lactation period is an important issue that warrants investigation. Moller (1991a) noted that mink will drink from an open water surface but have no natural inclination to drink from the valve in the cage, and that various devices have been used to teach kits how to drink water. Hetested the effect of a supplementary watering system - a 'drip' watering system, which is described by Joergensen (1985) as a tube dripping water into drinking cups. Experiments over 2 years, involving two groups of 60 mink families each year, indicated that kits Animal Welfare /999, 8: 205-228 211 Nimon andBroom provided with this supplementary system made significantly fewer unsuccessful attempts to drink than those provided withjust the standard nipple system. The behavioural observations conducted in the second year indicated that kits with the drip watering system began to lick water from around 40 days of age. In the first year, there was no difference in weight gain between experimental and control groups; but in the second year, kits in the experimental group were reported to gain weight significantly faster, and females to lose significantly less weight, than those in the control group. Joergensen suggested that the difference in results between these two years might be due to the higher average temperature in the second year (2.1°C higher). The benefits of supplementary watering systems may, therefore, be greatest when the summer is especially hot. M0ller (l991a) concluded that it is important to provide supplementary systems for helping kits to drink before they are able to activate nipple systems. M0ller and Hansen (1993) examined minks' use of a spray watering system which, when activated on a timer, produces ajet of water. This is said by the manufacturer to provide a shower for the mother and extra water on her pelt for the kits to lick (thus supplementing their water intake). In experiments involving two groups of almost 100 animals, the spray watering system was switched on for 30-60 s at a time, four times each day. The results showed that while the majority of the females used the spray for at least part of the time it was available, there was no positive effect on the weight of the females. This, they suggested, might have been due to the effect of low temperature, and the possibility that benefits are greatest when temperatures are highest. Kits did, however, gain weight more rapidly. This effect was probably because kits licked water from the cage and floor after spray watering: they did not lick water from the female's pelt. In another experiment (Hansen 1990), the provision of a 2cm-deep water tray in the cages of lactating females did not affect weight change in the mother. However, the lack ofpositive effect may have arisen because the female was reported to routinely empty the tray ofwater by getting into it. The provision of supplementary watering systems would appear to be problematic, given that commercially available products have produced varying results, and that certain claims made by manufacturing companies asto how mink will behave in relation to their equipment are incorrect (eg M0ller & Hansen [1993]). The issue of mother and kit dehydration, and how to provide readily available supplementary watering systems, isacontinuing problem. Drinking water temperature M0ller (1991a) also investigated minks' drinking water temperature preferences in experiments which gave the animals a choice between cool (6°C) or warm (40°C) temperatures. His results showed that mink accepted, and sometimes preferred, drinking water at temperatures up to 40°C. He reported that similar results have been found with rats. In fact, he argued, mink may prefer cold or warm water under different circumstances. A preference for warm water may occur when amink is dehydrated, as cold water is let out of the stomach only slowly, and the stomach is filled quickly. While the feeling of thirst is satiated, the animal is not as well hydrated. Water at body temperature passes through the stomach, allowing a greater volume to be ingested before stomach distention signals a satiation of thirst. On the other hand, cold water may be useful when the temperature ishigh. According to M0ller (1991a), mink may die from heat exhaustion in temperatures of30°C or more. Rats drinking cold water (eg at l20C) have been shown to lower their body temperature by up to 1.2°C (Deaux & Engstrom 1973):thus, the opportunity to cool the body may also be useful for mink. However, more experimentation is needed to provide conclusive evidence regarding water temperature preferences in mink, and whether these 212 Animal Welfare 1999, 8: 205-228 The welfare of farmed mink proposed benefits of hot and cold water might indeed be applied so as to benefit captive mink. Stereotypies and enrichment Stereotypies are repetitive, invariant behaviour patterns with no obvious goal or function (Mason 1991b). Their occurrence is often associated with barren and restrictive conditions, or environments which might be considered suboptimal, and they develop in animals faced with insoluble problems of frustration or conflict (Hinde 1970; Mason 1991b). Once developed, stereotypies can be elicited independently of the original stimulus, becoming part of an animal's behavioural repertoire. Some authors have argued that stereotypic behaviour may be an adaptive response to an aversive situation (eg, see Hansen et al [1992]), and some results suggest that farmed mink showing high levels of stereotypy are, in other respects, coping better with their environment. For example, in one study, it was concluded that highly stereotyping female mink had lower baseline levels of plasma cortisol than those showing low levels of stereotypy (Bilds0e et al 1991), although it is not entire Iy clear that their results supported such a conclusion. Also, signs of poor welfare, such as low body weight, do not necessarily correlate with stereotypy levels in adult mink (Mason 1991a; 1992; 1993), although in kits, the level of stereotypy was found to be negatively correlated with body weight, and positively correlated with the mean size of the adrenal gland (P <0.05 in both cases [Mason 1992]). While it is possible that those individuals performing stereotypies might be better off in some respects than those which do not (since individuals have widely differing ways of reacting to aversive stimuli [Broom & Johnson 1993]), the very existence of stereotypies indicates an inadequate environment in which mink are having to do much to cope. Furthermore, exposure to aversive stimuli, such as restricted feeding and daily immobilization sessions, has been shown to lead to an increase in stereotypies in farmed mink (Bilds0e etaI1991), indicating that stereotypies arise from aversive environments even if they are eventually elicited under different circumstances. Performance of stereotypies is also associated with negative consequences, such as slower growth in kits (Mason 1992; Mason et al 1995) and increased feed intake without a consequent increase in growth (Hansen et al 1992). In contrast, studies of both mink and other farmed or captive animals suggest that stereotypies can be ameliorated by improving the animals' environment, for example, via enrichment (eg increasing the size and complexity of the environment [Fraser 1975; Odberg 1987; Bryant eta11988; Markowitz eta11995; Cooper etaI1996]). Although stereotypies, and aberrant behaviour such as tail biting, appear to have a genetic component (eg Hansen [1993]; de Jonge [1988; 1989], cited in Mason [1994]), it is apparent that learning and environment can affect the occurrence of stereotypies. Stereotyped behaviour is widespread among farmed mink. In a study of 142 singly housed female mink, de Jonge et al (1986) found that 70 per cent of them performed stereotypies to a greater or lesser extent, and 50 per cent did so for greater than 25 per cent of the time they spent awake. In a study of 187 male and female mink, Bilds0e et al (1990a) found that stereotypies accounted for an average of 15.8 per cent of active behaviour. Levels of stereotypy vary between individuals, from farm to farm, and from season to season (Bilds0e etaI1990b), but are an ever-present feature of farmed mink behaviour. Wild mink, mink in zoos and mink in enriched laboratory conditions do not show these abnormal behaviours (eg Dunstone [1993]; Erlebach [1993]). This is a strong indication that standard farm conditions are inadequate and should be improved for the sake of mink welfare. The following sections Animal Welfare 1999, 8: 205-228 213 Nimon andBroom examine improvements to the cage environment, and note how these may be related to stereotypies. Feeding regimes It is often reported that stereotypies are at their peak in fanned mink prior to feeding (de Jonge et a11986; Bilds0e et a11990a; Mason 1991a; 1993), and it has been suggested that food deprivation (Bilds0e et al 1991) and minks' inability to access food at will are significant factors in producing stereotypies. In fact, Rushen and de Passille (1992) considered that most stereotypic behaviour reflects inadequacies in feeding regimes, rather than in housing. Hansen et al (1994) observed stereotypic activity among fanned mink which were fed ad libitum, and thus should have experienced no frustration at the lack of food: in these mink, stereotypy occurred mainly at night and after feeding. However, this study did not involve a control group which might have indicated whether ad libitum feeding reduced the incidence of stereotypies. Nor has research examined whether ad libitum feeding will prevent the occurrence of stereotypies inmink which have not yet learned to stereotype. If fanned mink stereotype because of their lack of control over their environment, particularly in relation to feeding, then measures aimed at increasing individual control may improve mink welfare and reduce stereotypies. For example, caged mink have learned to work for food, by running on a wheel and earning pellets at varied work:reward ratios (Zielinski 1988). Shepherdson et al (1993) found that providing food to zoo-housed small felids in a way that maximized the functional consequences of foraging behaviour (eg requiring cats to hunt for food) reduced the incidence of stereotypy and increased behavioural diversity. Evidently, such a change in feeding regime would require substantial changes in farm management in relation to the type of food delivered, but further research could explore the potential benefits of permitting caged mink greater control over their feeding regime, after which point the feasibility and practicality of such a change could be assessed. Stereotypies in farmed mink are, however, unlikely to derive from pre-feeding frustration alone, as post-feeding stereotypic behaviour occurs even in very young mink. Furthermore, Mason (1993) has pointed out that the form and timing of the occurrence of certain stereotypies (eg head twirling) suggests that they are more likely to be derived from attempts to escape the cage, thus suggesting that the physical conditions ofthe cage are inadequate. Cage size Research does not suggest that increases in cage size in the absence of further enrichment improve the welfare of farmed mink. Jonasen (1987, cited in Hansen [1988; 1991a]) conducted experiments using six- and eight-room cage sections (ie six or eight contiguous cages in a 2m section) and found no difference in the frequency of stereotypies associated with the different cage sizes. This result was replicated in arepetition of the same experiment with pairs of kits. Hansen (1988) studied the behaviour of 228 mink kits housed in pairs in bare wire cages with floor areas of 1.05m2 (large), 0.27m2 (standard) and 0.lm2 (small). He found that stereotypies did not occur significantly less often in the largest cage size, only that cage size influenced the type of stereotypy. Inaddition, he found significantly higher adrenal weights among the mink kept in the larger cages, suggesting that this treatment may have led to higher stress levels. Later work (Hansen et aI1992) on 60 male:female pairs in cages of these same sizes found that mink in the larger cages performed significantly more stereotypies, and had a significantly lower level of circulating eosinophil leucocytes, than 214 Animal Welfare 1999. 8: 205-228