Scottish Animal Welfare Commission: welfare of isolated populations of feral animals - report

4. Outcomes of evidence gathering

4.1 Feral sheep populations and habitat

The islands of St Kilda are located 40 miles from the Uists in the Outer Hebrides of Scotland. Of these, the largest and most accessible of the islands is Hirta, with landing possible in Village Bay. Hirta is about 3.4 x 3.3 km and has an area of 6.3 km² and 15 km of coastline. There is a permanent Ministry of Defence contractor presence on the island, and seasonal presence of NTS and scientific staff as well as visits from the public. The sheep on this island, and in particular the population of sheep that live in the Village Bay area, are the most accessible and those for which most information is available. The sheep on the islands of Soay and Boreray are more remote (Appendix V) and, although it is occasionally possible to land from small boats on these islands, this is a very significant challenge. A census of the seabirds and sheep on these islands has been conducted by people landing on the islands, but can now be conducted more remotely using drone technology.

Village Bay on Hirta is a large bowl-shaped feature, containing the MoD base and the remnants of the black houses occupied by the St Kildans, some of which are inhabitable (Appendix IV and V). There are also some stone structures that may represent the earlier dwellings of St Kildans. The landscape is peppered with some existing drystone walls and more than a thousand cleitan (stone storage huts built of stacked stones). The Village Bay area is used by a resident population of around 600 Soay sheep, comprising hefts of ewes and their lambs and separate smaller groups of rams. The sheep are remarkably tolerant of human presence, although they reportedly exhibit greater vigilance with larger numbers of visitors, as seen in other situations^38,39. Outside Village Bay, as denoted by ‘head dyke’, a larger population of around 1000 Soay sheep use the rest of the island. These animals are significantly less habituated and maintain greater flight distances from humans compared to the Village Bay population. Their behavioural responses and the challenging terrain mean that these animals are not studied, although a count of total Hirta population size is conducted each year by SSP using telescopes. A similar count is also conducted for Soay and Boreray, but is considerably less accurate due to the inaccessibility of these islands.

The Orkney sheep of North Ronaldsay represent another primitive breed of sheep that exists largely outside of human management. Although the majority of the scientific literature relating to these sheep concerns their ability to survive on seaweed, and their movement patterns dictated by the tides, there is limited evidence surrounding their development. Ryder (1964), in his classification of British breeds of sheep, classifies the North Ronaldsay sheep with other primitive breeds, such as the mouflon, Soay and Hebridean. Investigation of the Neolithic site at Skara Brae suggests that primitive sheep were present and farmed then and seaweed may have been a source of nutrition even at that time¹¹. The confinement of the North Ronaldsay sheep to the shore occurred in 1832, when a 13-mile-long drystone ‘sheep dyke’ was erected to keep the sheep off the inland pasture to allow it to be used for more lucrative agricultural practices such as growing crops and farming cattle. This has led to their remarkable adaptation to a seaweed diet, which has been described extensively in the literature. From the time of the building of the dyke, the island’s ‘Sheep Court’ has determined the number of these animals which the island can sustain at 2,312 sheep, although these numbers may not always be achieved⁴⁰. The North Ronaldsay sheep do receive some human intervention: sheep are gathered three times a year for shearing (summer), at New Year when some animals are selected for market and in February when a census is conducted. There is little evidence in the scientific literature for population size and structure or other information on these sheep.

4.2 British Primitive Goat populations and habitat

Unlike the Soay and North Ronaldsay sheep, which can be clearly linked to a specific origin population, free-living and naturalised populations of goats, now recognised as British Primitive Goats by the Rare Breeds Survival Trust in 2024, exist in multiple pockets across island and mountainous regions of Scotland and the border areas with England. Far fewer studies have examined these populations, but there are data on the goat populations in the Cheviots^26,41 and on Rum⁴², as well as more generally on global populations of primitive or naturalised goats²⁸. As browsers, unlike sheep who predominantly graze, goats can exploit a wide range of habitats, but may also be perceived to be more destructive than sheep as they can consume a wide range of plants. In fact, goats are considered as one of the most invasive species and management in many places is centred on eradication²⁸. Most of the studies of primitive goats living in Scotland suggest they exist in a number of ‘hefts’ (see section 4.3) and in relatively small population sizes of around 200 animals.

On Rum, one of the larger Scottish islands (104.6 km²) situated in the Inner Hebrides, a few hundred primitive goats are found in eight hefts along the west coast of the island, occupying about 15% of the island. They share the island with a large population of about 1000 red deer⁴², which are believed to outcompete the goats for access to resources and so constrain their numbers to the western edge of the island. Goats have access to the beaches, sea cliffs, scree and cliff-top slopes. The female goats have high fidelity to their heft, whereas the male bands often move between neighbouring hefts. The goats are reported to make use of caves and sheltered rocky areas overnight (perhaps resembling the use of cleitan by Soay sheep). Less documented evidence is available for other goat populations, but we have been told that the number of goat herds across Britain have dwindled from a few hundred herds to the current position of 40-45 herds, often in small and isolated pockets of around 200 animals. Watt (1937)⁴³ enumerated many different pockets of wild goats across Scotland, often numbering no more than 30-40 individuals per herd, and comments that the goats on Oa on Islay ‘have been in existence since time immemorial’. Darling (1937)⁴⁴, in a companion piece, adds that goats can ‘go feral with no difficulty and in a very short time’, suggesting multiple locations where naturalised goat herds have arisen. Whitehead (1972)²⁹ listed 141 historic and contemporary goat populations in Scotland, of which 74 were extant at that time.

For the Cheviot goat herd, in the Northumberland National Park, there is good understanding of the herd structure, sex ratios and ages of the herd through an annual census undertaken on foot. Genetic analysis of this herd suggests no recent or historical additions of modern goat breeds and a close lineage to Irish goat breeds. The authors of this analysis suggest that these are genetically distinct and unique populations of goats derived from ancient goat husbandry practices²⁶.

Two PhD theses from 1982⁴⁵ and 1983⁴⁶ provide some information on the populations of goats in several locations around Moffat and the Southern Uplands. However, the origin of these goats is unclear, ranging from descendants of goats owned by monks in 1200 to escapes or abandonment around 1800s. In various censuses, the number of goats in this region range from 30-103 individuals, and in appearance resemble other primitive goats. The goats in this area are described as occupying a series of hills and rugged valleys, often overlapping fenced areas of sheep production and have been specifically excluded from an area of Forestry Commission land. In all locations there is mention of goats being occasionally culled or harvested by the Forestry Commission (Forestry and Land Scotland today) and others.

With the exception of the goats on Rum, where data collection seems to have continued to 2006⁴², there appear to be very few recent data on size and structure of herds, their use of the environment and constraints or otherwise on population size, with most work having been done in the 1980s.

4.3 Hefting

‘Hefting’ is a commonly observed social pattern reported in wild, naturalised and extensively managed small ruminants, although rarely seen in more intensively selected sheep breeds (although this may be due to the environments in which these animals are kept³²). Sheep and goats are not categorised as territorial, in that they do not defend their range, but do show home-range behaviour where the same area of land is occupied by related small ruminants. Home ranges are areas which contain all the resources that a sheep or goat may require (such as food, water, escape terrain), and where the edges are maintained by avoidance of unfamiliar animals in the neighbouring home range. The ‘heft’ then typically consists of the home range and the animals that live on that home range. As matrilineal species, hefts are held by grandmothers, mothers and daughters and there is cultural transmission of information about food sources, water and shelter between members of the heft. Home ranges can be large, especially in wild sheep populations (e.g., greater than 10 km² in mountain bighorn sheep, Ovis canadensis, in Canada), and sheep show seasonal use of their home range, often restricting themselves to smaller parts of the home range at lower elevations during the winter.

An important component of hefting is the complex social behaviours of small ruminants and their remarkable social recognition abilities. These have been most clearly studied in domestic sheep and goat breeds^47,48, but demonstrate sophisticated facial recognition patterns and an ability to remember more than 50 individuals for more than two years in sheep. Overt fighting between individual small ruminants, with the exception of male animals in the rut, is relatively uncommon and individuals rely on more subtle social cues of position, gaze and head position to assert dominance and displace subordinate animals from preferred locations. Movement about the home range is often led by an older, bolder and more experienced member of the heft, who has knowledge of the home range, although often not the most dominant animal.

4.4 Mortality data, longevity and changes in population size

Detailed data on the changes in population size over a number of years and mortality data are available for Soay sheep (restricted in detail to the sheep living in Village Bay, but with population censuses for the whole of St Kilda) via the Soay Sheep Project (SSP) from 1985 to the present⁷, and the goats living on Rum in some years from 1980 to 2006⁴². For other populations there are largely only historical records from short-term studies in the 1980s^45,46, which are mentioned where pertinent.

4.4.1 Soay sheep:

The changes in population size of Soay sheep are represented by the Village Bay population (approximately 500-600 animals) and by the wider annual census of the whole island (1854 animals recorded for 2023, the most recently publicly available data). A well described feature of this population has been the ‘boom and bust’ periods of population change, where large winter mortality occurred at approximately three-year intervals as recorded from 1985 to 2012, but losses of this magnitude have not been recorded since that time. These periods have been recorded in both the smaller Village Bay population and the wider Hirta census and can account for a reduction in whole island population size of as many as 1000 animals. Although not as comprehensively studied as the Village Bay population, there is also evidence for similar events occurring on Soay and Boreray⁴⁹. This demonstrates that adjacent populations show environmental synchronicity, which is accounted for by large scale weather variation between years. In addition, the overall population of sheep has increased from around 1000 animals in the 1980s to a possible plateau of around 1700-1800 animals since 2000. A similar increase has been seen in Village Bay from around 250 animals in the 1980s to around 500 since 2000⁷. This has been attributed to an increase in warmer weather, with climate change resulting in an increase of biomass, particularly of plants preferred by sheep.

Change in population size is the balance between all animals born in a year and those that die. In any system all animals will eventually die and, in unmanaged populations in the absence of predators, sheep die from a combination of lack of condition on entering the winter (because they have failed to gain sufficient fat over the summer and autumn), winter weather, and any additional co-morbidities such as parasite load, teeth loss and other issues. Although absolute numbers of animals that have died can be derived from the population size, more informative data were provided for us by the Soay Sheep Project of actual mortalities in various age classes for the Village Bay sheep. These animals are ear tagged at birth, and estimated death dates can be calculated from finding carcases of known individuals or estimated winter of death in animals not seen again in subsequent annual censuses. These data demonstrate that animals are by far the most vulnerable in their first winter (with survival of 35.9% of juvenile females and 30.2% of males). Yearling sheep (entering their second winter) have the next lowest survival (79.4% and 66.7% for female and male sheep respectively). Thereafter, until reaching old age (7+ years), survival is high for adults (91.3% for ewes and 71.4% for rams). Longevity is as great as 15 years for females, and 10 years for males. Throughout, and in common with studies in both wild and domestic species, males have greater risk of mortality at every age than females. It is important to note that average survival rates mask complexities over different years as animals respond to environmental pressures. Thus, survival can vary markedly from year to year within an age cohort, which is only partly explained by population density (and hence competition between sheep for resources), but is also affected by winter weather, infection rates of parasitism, and the age and sex structure of the population⁵⁰. Data from the 1990s, which includes data from populations compared before and after a significant reduction in size²⁴, demonstrated high survival in all age classes before and after a crash, but lower survival, especially for juveniles, in a year when there is a large decrease in population numbers. In those years, bodyweight was related to survival (lighter animals were less likely to survive), but this relationship was not seen in other years. In addition, pregnant females were more likely to die in years with a large reduction in size, but breeding status did not affect survival in other years. However, since the last large reduction year in 2012, survival of adults and yearlings has remained high, with 100% survival in some years, whereas most mortality is seen in the juvenile animals, and the older members of the population.

4.4.2 Goats on Rum/Southern Uplands:

For the smaller number of goats per herd on Rum and in the Southern Uplands, individual animals were identifiable by coat colour and horn shape, and aged either by known kidding dates or counting horn rings⁴². Individual animals were considered to be alive if seen at least once on a weekly basis, and annual surveys are undertaken in April or May. Therefore, for some years there were more detailed records of survival of individual age and sex classes within some relatively short time periods than is possible with the larger and less accessible Soay sheep population (although for the latter there are greater continuous data over many years). In total, between 103 and 161 animals were recorded each year in five years of data on Rum. As is common in wild and domestic small ruminants, animals are most vulnerable in their first year of life, and kid mortality varied from 28.6% to 57.1% in different hefts. Female mortality rates also varied by heft from 11.9% to 37.5% per year. Male mortality was lower than that of females at early stages in the Rum population but then increased. However, in the Southern Uplands male kid mortality was greater than for females⁴⁶, as is more commonly seen in wild and domestic small ruminants. In addition, survival of adult goats in the Southern Uplands seemed to be higher than those on Rum, with survival rates of 90% for females and 73% for males. Male goats were much more likely to die in September to December than at other times of the year on Rum and at higher rates than females. In contrast, female mortality was highest in April and August and lower at other times of the year⁴². Longevity in the Rum population was 12 years for both males and females; for goats in the Southern Uplands females up to 10 years old have also been recorded⁴⁵. The lower overall survival of males leads to skews in population towards females, where more than three times as many females are recorded as present than males⁴⁶.

The detailed information on specific hefts on Rum has allowed investigation of the environmental and other factors that may influence survival. This suggested that it was winter birth season conditions (since most kids were born in January and February) that had the greatest impact on survival of kids, not density dependence or availability of caves for shelter. However, for females a combination of winter thermal conditions and shelter are important for survival. For this population mortality and growth rates did not seem to be density dependent and the author observed that competition for food between goats was very rarely observed. However, demonstration of density-dependent effects can be challenging, and the population may also be constrained by interspecific competition with red deer.

4.4.3 Causes of mortality:

For small ruminants the first year of life has the greatest risk of death and early life mortality is high. Estimated mortality of Soay sheep in Village Bay (from known birth dates and for lambs that were not present in the later census) suggests that average lamb survival to weaning was 73%, but with significant between-year variation (from 50% to 90% survival across years). For the goat data only 34% of kids were reported to survive to six months in the Southern Uplands⁴⁶, with 97% of deaths occurring in the first month of life in goats on Rum and inversely related to ambient temperature, suggesting death from a combination of hypothermia and starvation. Winter-born kids had a three-fold higher mortality than those born in spring. Young animals are also vulnerable following weaning, when they need to expend energy in forming social relationships and may participate in the rut in their first year of life. For goats in the Southern Uplands, Bullock⁴⁵ suggested that individuals losing more than 20% of their body weight over winter were likely to die the following spring.

Participation in the rut for male sheep and goats carries a significant energetic cost in fighting, risk of injury and a drastic reduction in time spent in foraging and feeding behaviour. This means that males can enter the winter period in lower body condition and are more susceptible to winter weather, parasite burdens and other issues that reduce survival. In both species males in their first year of life may participate in the rut, although this is more intense in older males. Soay sheep of both sexes are known to mature early⁵¹ and juveniles can participate in the rut, with around 10% of males producing offspring in their first year and 37% of females experiencing a pregnancy in their first year⁵². For goats, an estimate of 20% juvenile pregnancies is reported, although numbers of animals were low⁴⁶. Male juveniles are more successful in the rut when the population density is low and are more likely to survive the winter if they mate successfully, probably as an indirect impact of larger juveniles at the start of the rut being more successful. Much of the data on participation in the rut of juveniles is derived from success in leaving offspring and the impact of participating unsuccessfully in the rut on the survival of juvenile males is not clear. Juvenile males, as with all males, invest more time and energy in carrying out reproductive behaviours and less in feeding and energy conservation. Juveniles are also likely to lose more fights, sustain more injuries and be chased out of preferred locations than mature males. However, older males may start to accumulate injuries from fighting, experience loss of teeth that can limit their ability to utilise forage efficiently and develop chronic conditions, such as lameness. Anecdotal evidence by SSP suggests that they rarely see rut-related injuries, and tooth loss and lameness are very rarely seen in Soay sheep on Hirta.

Reproductive costs of females include the experience of harassment from rutting males, which can reduce their foraging behaviour and becoming pregnant, which increases their nutritional demands. Although juvenile female Soay sheep can become pregnant in their first year, this is population-density and body-mass (weight) dependent⁵². This means that juveniles are less likely to be pregnant when the population is higher and when their weight is lower. The population of sheep on Hirta appears to be maintaining a higher number than in the 1980s and 1990s, and animal body mass is decreasing, which may be contributors to the decrease in the frequency of juvenile pregnancies reported in Soay sheep in the Village Bay population since the early 1990s. For juvenile females pregnancy is associated with a substantial increase in the risk of mortality. However, as with males, heavier juvenile females are more likely to survive than lighter females⁵². In domestic sheep, older females may experience the cumulative impacts of successive energetically expensive pregnancies and lactations, and can also experience worn or loss of teeth and the accumulated effects of aging. Evidence from SSP on Soay sheep suggest that no links have yet been documented in mortality and past reproductive investment or tooth loss, but there may be changes in immune function and ranging behaviour in older animals.

Post-mortem examinations of carcases suggest many animals are emaciated at death and, in the absence of predation, may be most likely to die of starvation, probably secondary to other predisposing factors, such as health status. The impact of population size on likely survival is variable depending on species, location, age cohort and other factors. Winter weather is a significant contributor to mortality in both species, where those animals with the greatest reserves as they enter the winter are more likely to be able to utilise fat reserves when forage quality is poor and maintain body temperatures in poor weather, and may also be able to compete more effectively for food sources. In addition, these animals may be able to partition more energy to maintaining immune responses against parasites or other pathogens. For both Soay sheep on Hirta and goats on Rum cleitan and caves, which are used by the animals to provide shelter from the weather, were associated with dead remains of conspecifics, suggesting a similar withdrawal to a sheltered location at end of life or to resist the cold weather in both species.

4.5 Observations of lives of feral small ruminants

Most studies of feral small ruminants have focused on population dynamics and feeding patterns. Typically, these have focused attention on key events in animals’ lives that are associated with these life history traits (the rut, kidding or lambing and winter survival). This means there is less information about what could be considered the normal life of the animals outside these key events, such as social groupings, social interactions, use of home ranges and habitat features, and dominance or sociability (other than between males in the rut). Nevertheless, to understand the full welfare and adaptations of these animals to their naturalised environments this information is relevant. In this section we summarise some of the information that is available considering these elements of the animals’ lives.

As a highly social species, Soay sheep form distinct social groups, which can be advantageous in obtaining social information, finding food, avoiding predation (in situations where predation occurs), caring for young and coping with environmental challenge. As described above [Section 4.3], specific individuals can be a repository of information about the environment and act as important individuals in exploiting the environment. Female group sizes in Soay sheep were on average 4 individuals but ranged from 1-10 ewes⁵³. Home-range size, and local sheep population density, had positive impacts on ewe survival, especially in poor winter weather. In addition, ewes who had strong partnerships with other ewes also had increased survival. In wild sheep species, and feral and domestic sheep, withdrawal to more remote areas of the home range to give birth is commonly reported, with ewes then returning to the social group followed by their lambs. Observations of suckling behaviour in Soay sheep on Hirta suggest that home-range quality for ewes (amount of forage available) did not influence maternal care expressed towards their offspring⁵⁴, suggesting that ewes invested in their current offspring at the potential expense of future lambs. Although play in wild, feral and domestic lambs is commonly observed, including our own observations whilst visiting St Kilda, there do not appear to be systematic data collected on incidence of play behaviours in Soay lambs. Play has been considered an indicator of positive welfare since it is known to disappear when animals are in poor welfare or risky environments (as has been demonstrated in sheep) and is considered to be associated with immediate psychological benefits (such as enjoyment) and longer-term fitness and health benefits⁵⁵.

Southern Upland goats form social groups of 2-10 female goats, although groups of up to 22 animals were also recorded. Males were more commonly observed in groups of 2-4 animals. Group sizes were smallest in the spring and increased in the autumn⁴⁶. Females choose secluded and sheltered areas to give birth. Unlike sheep, kids are left ‘lying out’ by females going to feed even on the first day of life as part of the ‘hider’ pattern of maternal care in this species. Dams stayed within 100m of a kid and returned within 2 hours when it was newborn, but thereafter females could move up to 600 m away and leave their kids for up to 9 hours. Lying out decreased after the first week of life to 50% by the second week and 5% by the 4^th week. Weaning was completed by 5-6 months of kid age. Before this, kids were seen to be closely associated with the mother until the rut, where male kids were further from their mothers than females. Female kid separation from their mothers occurred later when the next kidding season occurred.

Rutting behaviour in both species is composed of aggressive encounters (threats and displays followed by escalation to fight or displacement of one male) and fights (where both males clash horns; in goats this may involve the standing on the hind legs to deliver a blow from above). Aggressive encounters in goats were reportedly short (less than 1 minute) and relate to appraisal of fighting ability of opponents. Fights ranged from 1 to 140 minutes in duration. Dominance was related to age in the first two years of life (two-year-old males were dominant over juveniles and yearlings), thereafter to weight and horn size. Male-male aggression was much higher in the rut than at other times for the year, which is related to presence of oestrous females⁴⁶.

4.6 Learnings from farmed small ruminant populations

In discussions with various stakeholders about the Soay sheep populations it was suggested that these animals could/should be managed in the same way as hill sheep populations. Therefore, it seemed pertinent to consider what this might entail and to evaluate comparable metrics. Although overall numbers of Soay sheep deaths have been highlighted to us, this is not a particularly useful statistic since in an unmanaged population all animals will eventually die, and some of these animals have died at a considerably older age than is typically seen in hill sheep farming. The reasons why animals may die and the suffering involved (see also the welfare section below) seem more important considerations. Therefore, we have considered how and when sheep die in managed situations, drawing parallels to Soay sheep where possible and it made sense to do so.

Lamb mortality and survival data have long been studied in sheep populations as these are both a welfare issue for managed sheep and a constraint on productivity. Across many studies and years lamb mortality (from birth to weaning) has remained unchanged at 15%, although individual farms or studies report lower (rarely below 10% however) or much higher (50% or more) mortality before weaning⁵⁶. However, this study considered lamb losses across many countries and farming systems. Survey investigations across farmed sheep in the Scottish Highlands and Islands report annual unexplained lamb losses (of healthy viable lambs, termed ‘black loss’) of 6.6% or 7.6% when considering the period from marking (typically 6 weeks of age) to weaning^57,58. Note that these losses are in addition to the known losses of lambs at birth (when 50% of all pre-weaning mortality occurs). An additional study of EID tagged lambs on 5 farms suggested a mean black loss of 17.4% (per farm range 10.5% to 25.8%)⁵⁷. Mean annual losses of unmanaged Soay lambs from birth to a comparable time for weaning in domestic lambs (April to September) was 27%.

The management of juvenile sheep on hill farms diverges sharply from the unmanaged situation. Typically, on farm male lambs will be castrated (currently by methods without pain relief⁵⁹), and all males and most females will be finished and slaughtered within the first year of life. Thus, we have no comparable data to compare to rates of juvenile survival in Soay sheep, since most juvenile lambs in managed systems do not live beyond the first year of life. Mating of ewe lambs (who lamb in the first year of life) commonly occurs in some management systems, but is not frequently seen in hill sheep management. Juvenile ewes have lower ovulation rates, lower fertilisation rates and higher embryo losses than young adult ewes. Juvenile ewe fertility is positively related to live weight, with heavier ewe lambs being more likely to conceive, although this was not related to lamb survival. In one study, fertility was around 30% for lighter ewe lambs and over 80% for heavier ewe lambs⁶¹. The lower end of this figure is comparable to juvenile Soay ewes, with 37% of ewes becoming pregnant in the first year of life.

In managed ewes, the prime age for breeding is between 2 and 6 years with less than 5% of ewes recorded as lambing at 7 years⁶². When leaving the breeding flock, approximately 7% of ewes are described as ‘missing – presumed dead’. Otherwise, UK ewes are culled for teeth problems (usually missing incisor teeth) (38.9%), ‘age’ (23.5%), body condition (6.8%) or reproductive disorders (5.4%). Data on survival of rams in breeding sheep farms are markedly absent from the scientific literature, although anecdotal evidence suggests relatively poor survival of rams compared to ewes. Research from Australia⁶³ reported three times greater annual losses in ram flocks (21%) compared to ewe flocks (7%), but otherwise we are unable to find any comparable data on ram survival. However, we note that most adult males are not present on farms and their management is considerably different from that of the ewe flocks.

Data on management of goat populations is hampered by fewer studies in a relevant location (most data were found on goats farmed in Africa and Asia), but kid mortality ranged from 11.5% to 37%⁵⁶. In extensively managed goat herds, postmortem examinations of dead kids suggest that mis-mothering, hypothermia, white muscle disease and enteritis are significant causes of kid mortality. A study of dairy goats in New Zealand suggested that an average of 10.4% of live-born female kids died before mating, although 20% mortality occurred on some farms⁶⁴. For kids that died in the first 14 days of life, septicaemia, starvation or dehydration were the main causes of mortality; for those dying at older ages gastrointestinal disorders, disbudding-related injuries (such as thermal damage, bacterial meningoencephalitis, and necrosis of the cerebrum) and septicaemia were the main causes of death⁶⁵.

4.7 Learnings from wild small ruminant populations

Behaviourally, wild and domestic sheep have been shown to be very similar, particularly in terms of reproductive responses, maternal care and survival^36,66. Therefore, wild sheep behaviour has been used as an example to aid understanding of the behaviour of domestic sheep, and to compare with survival characteristics. Wild and feral sheep (bighorn, Dall’s, mouflon, Soay) have all been shown to withdraw to more remote parts of their home range to lamb, selecting secluded areas with shelter, with dry and south-facing aspects preferred lambing habitat³⁶. Selection of birth sites is related to improved ability to protect lambs from cold and hypothermia and thus this is an important part of improving lamb survival.

Dall’s sheep, Ovis dalli, in Alaska show annual survival rates of 29% for lambs and 84% for adults. Lamb survival is positively influenced by body mass. Greatest lamb mortality is seen in the youngest lambs (at birth)⁶⁷. Juvenile survival in mountain bighorn sheep is very variable by year and significantly influenced by predation, e.g., 20% survival in a year with high coyote predation (typically lambs under 8 weeks old) but 71% in other years⁶⁸. Other studies suggest that ewe condition in late winter had a strong and positive impact on summer lamb survival in bighorn sheep, Ovis canadensis, with lambs from ewes in relatively good body condition having four times the survival of those in poor condition⁶⁹.

Mountain bighorn sheep have the highest risk of mortality in younger ewes and very high survival rates in females aged 2-7 years. Annual survival in wild sheep (mountain bighorn in Canada) has been reported as 89.6% in males and 93.9% in females (prime-aged sheep) and 77.7% and 85.9% in older animals⁷⁰. Yearling survival varies by year, but only survival of yearling females in one location was shown to be population-density dependent. In this population males were hunted for trophies, typically males over 5 years, and the natural mortality of younger males aged 3-5 years was increased, possibly because of participating in the rut at early ages.

A study of wild goats, Capra aegagrus, in Iran identified climatic variables as having significant impacts on population fluctuations, although anthropogenic factors (habitat degradation and encroachment, and hunting/poaching) were also relevant⁷¹.

Wild ungulates, in common with many mammalian species, adapt to seasonal variation in food supply by laying down fat stores when food is abundant, and adapting their activities and energetic requirements when food is scarce. Wild Iberian ibexes, Capra pyrenaica, in the Mediterranean, adjust their fat stores by increasing fat in warm seasons, and utilising stored resources in the cold period⁷². The timing and extent of these changes is age and sex dependent, but juvenile and yearling goats were more depleted than adults and so were more vulnerable to winter conditions, particularly snowfall, than older animals. Younger animals need to partition energy to both growth and maintenance in the summer, and so are less able to develop fat stores. These patterns are seen in wild and domestic ungulates in both natural, farmed and experimental studies.

4.7 Learnings from other feral or naturalised populations

4.7.1 Cattle on Swona: Although it was our intention to also consider the welfare of the feral cattle on the island of Swona as part of our report, there are few published data to support our investigations. Swona is a private island located in the Pentland Firth south of Orkney and inhabited by a single family until 1974. Thereafter the cattle remained on the island when the islanders departed and were fed until 1977, when several animals were removed for sale⁷³. The cattle are Aberdeen Angus and Shorthorn crosses and are one of the few feral cattle herds of Bos taurus. Numbers vary around 25 animals, although there was a reduction in animals in the harsh winter of 2017/18, and there seems to have been a subsequent increase in population size with 37 animals recorded in 2026. The cattle graze grass and forage on seaweed. As with the feral sheep and goat populations, cows have better survival than bulls. The cattle herd is described as grazing together as a single herd and the bulls do not hold territories, although they are recorded as occupying a different part of the island to the cows⁷⁴.

4.7.2 Chillingham cattle: Although known as ‘wild’ cattle, in fact these are feral cattle, descended from domestic cattle⁷⁵. They are maintained on a 134-ha site and receive some minimal management (culling on welfare grounds, hay feeding). The cattle have been isolated in the park since at least 1646 and are genetically distinct, although subject to intense inbreeding. In the past, these cattle have experienced heavy winter mortality and die off of young and senescent animals, but the numbers are now managed more proactively with nutrition and culling to prevent mortalities from natural causes. Current numbers are around 120 animals. Annual adult female (>3 years) survival is around 90% (yearly range: 88.7-93.2%) and male survival a little lower (81.4%-94.6%). There is no evidence of density-dependent mortality. Although die offs of over 30% have been observed in Chillingham cattle on two occasions, management policy is now to avoid these welfare impacts. It should be noted that the Chillingham cattle are readily accessible for management and cannot be considered as an isolated population from a management perspective, unlike island populations.

4.7.3 Sable Island horses: A population of around 500-700 feral horses exists on a sand island, Sable Island National Park Reserve, in Nova Scotia, Canada. These horses seem to have originated from a series of introductions in the second half of the 18^th century and seem to be most related to horses of Nordic origin. The population has been isolated from the addition of new horses since 1935 and protected from all human interference (both positive and negative) since 1960. There are no predators on the island, but the horses are subject to unpredictable and harsh conditions. Annual survival of 3-year-old mares was highest at 95.5% compared to 86.6% in adult mares, although adults had highest fecundity (61.6% pregnancy) compared to 3-year-olds (40.2%)⁷⁶. Two-year old mares had intermediate survival (88%) but were rarely pregnant. Survival of foals and yearlings (87.6% and 85.3% respectively) was higher than seen in young small ruminants. Population density and winter weather interact to predict adult female survival. Overall, and compared to domestic horses, the population was unusually young (39-50% of adults were less than 3 years old and only 15% of females and 11% of males were aged over 9 years). In addition, and contrary to most wild herbivores including other horse populations and feral sheep, the sex ratio was male-biased in adults, although equal numbers of males and females were present at younger ages⁷⁷, and foal survival was female biased. The authors suggest this may be related to more recent feralisation in this population compared to other feral populations and greater sexual dimorphism in other species, which can carry a cost for males.

Sable Island horses are unusual compared to other feral horse populations (e.g., feral horses in Grand Canyon, in Australia, Konik horses in Poland, etc.) as they may not receive any interventions as mandated by Canadian legislation, including reintroductions of other horses, feeding or veterinary care⁶. This was in response to public pressure against a plan to round up and export or cull them all in the 1950s.

4.7.4 Other feral horse populations: There are many feral horse populations world-wide, including populations in North America, Australia, Europe and South America. Many of these populations are relatively recent feralisation events, with populations being unmanaged sometimes for only 50-100 years in the USA, and more active management occurring in recent years⁷⁸. A meta-analysis of these populations suggests that horses generally have a negative impact on environment quality for most parameters, particularly at higher population-density levels, due to grazing activity⁷⁹. This often then leads to efforts to cull or eradicate the feral populations, which can cause high profile public debates. However, in other places, feral ponies of native breeds are considered to be valuable for conservation grazing as they are less selective than sheep. These ponies are variously described as ‘semi-wild’ (British native breeds⁸⁰) or ‘semi-feral’, such as primitive Konik ponies in Poland, or Przewalski’s wild horses, Equus przewalskii, where they are widely used for conservation and habitat management⁸¹. The welfare of free-ranging native ponies (Exmoor and Dartmoor breeds) on common land or conservation grazing suggested populations were in good health and showed moderate fearfulness responses to humans. These populations are relatively small and accessible, and this permitted welfare to be assessed on an individual animal basis⁸².

4.7.5 Arabian camels in Australia: Arabian camels were first brought to Australia in the 1800s, specifically to aid exploration and exploitation of the internal wilderness regions⁸³. With the advent of a road infrastructure and railways camels were no longer needed and were either shot or released, eventually making their way into the desert regions in central Australia. Camels have been feral in Australia for about 100 years, but were largely ignored until the 1980s when aerial surveys provided the first evidence of their numbers. Assessments of population size suggest at least one million feral Arabian camels living in an area of 3.3 million km² in 2008⁸⁴. Most literature on feral camels is negative, citing their destruction of the environment, competition with livestock and damage to infrastructure⁸⁵, and emphasising the need for culling and population control. However, Crowley⁸³ argues that camels thrive in the desert environment and, despite concerted efforts at eradication, continue to exploit the desert habitat, for which their adaptations are well suited. This suggests that the welfare of the camels may be good, even if their impact on Australian environments is not.

4.7.6 Rewilding studies: Rewilding studies differ from the other examples discussed, where animals became feral through accidental or historic releases, as the animals are deliberately released with the intention that they should live a natural and functionally-wild life. Examples include the release of Konik horses, British native ponies, water buffalo, Bubalus bubalis, cattle, bison, Bison spp, and small ruminants in various European locations (particularly the UK, Netherlands and Denmark^86,87). These projects are not without controversy, particularly around animal welfare issues and the differing views of wildness, suffering and natural living⁸⁸. The themes identified as issues with these projects are very similar to those relating to the Soay sheep living on St Kilda: concerns about whether welfare is primarily avoiding suffering or about living a life where animals can express natural behaviours; whether welfare should be assessed at the level of the individual or the population; whether supplementary feeding should be given (particularly if the animals are constrained within the landscape in some way); and whether severely sick or injured animals should be euthanised to avoid extreme suffering.

A particularly notorious example is the rewilding experiment at Oostvaarderplassen, where Konik horses, red deer, Cervus elaphus, and Heck cattle were introduced to a 22.5 square mile wetland area in the 1980s and 1990s. Whether these animals were ‘wild’ or not and human responsibilities towards the animals were contested from the start, and mortality in poor winters was a cause of great public concern⁸⁹. In particular, the rich grazing (on reclaimed polders) and mild summers allowed the population to expand rapidly to levels that exceeded the carrying capacity of the site during the winter. Failure to appreciate how predators affect population dynamics of herbivores was also a factor. Various committees considered how to manage the population, particularly providing shelter in the winter, providing greater access to more land for dispersal and euthanising suffering animals, but most were not implemented and following a large population crash during a very harsh winter in 2018, the experiment was abandoned. These animals are now managed in situ with feeding and culling of some animals to maintain the population at a fixed size.

Section summary

The review of different feral populations internationally suggests different values are placed on the animals in different situations: in some cases populations are considered to be invasive, destructive and to require eradication (camels, goats and some equine populations), where their welfare is almost never considered; or populations are considered to embody values of wildness, heritage and to require protection (Sable island horses, Chillingham cattle, conservation grazing). These align with the differing attitudes expressed towards the feral sheep and goat populations considered in our report. In all cases, animals have found themselves in their current locations through human interventions, and their welfare needs do not differ because of human values. The study of British breeds of pony in different environments is noteworthy as an example of attempts to directly measure animal welfare in different situations; data which are typically absent from most studies of feral populations.

4.8 Management options and constraints

In our discussions with various stakeholders for both naturalised sheep and goat populations a number of potential scenarios arose as putative means to manage the populations. Currently the Soay and Boreray sheep on St Kilda are treated as if they are a wild population of sheep and receive no active management. The population in Village Bay (about one third of the total population on Hirta) is caught once a year in late summer as a survey of numbers and the lambs there are caught and ear tagged at around 3 days of age for research purposes (individual animal identification is a Home Office requirement for experimental studies), but this information is not used for other purposes. The North Ronaldsay sheep are actively managed with gathers three times a year and the removal of some animals in the autumn with the goal of maintaining a stable population of sheep numbers. Management of the goat populations is sporadic and primarily through culling of animals, when these are perceived to be too many or damaging to the environment. Although not in Scotland, an alternative scenario of removing and rehoming whole hefts of animals was described for the primitive goats in the Cheviots. The various options for management of naturalised small ruminant populations form the basis of our welfare impact review (below), but will be outlined here. These are described in order of increasing control of the lives of the animals. Note not all these options need to be considered on their own, e.g., some could be combined and, in some cases, we envisage that they must be combined to be effective. These have been deliberately developed without any regulatory, moral or ownership considerations, but to allow the direct welfare situation for the animals to be evaluated. At this point we also do not consider the practical feasibility of implementing these management actions.

4.8.1 Unmanaged populations/hands-off approach

This is essentially the status quo for the Soay sheep on St Kilda and treats the animals as if they are wild. Thus, births, mating and deaths are unsupervised, animals make choices related to their social and physical environment, and no human intervention takes place to provide supplementary nutrition or health treatments. The sheep on North Ronaldsay also receive similar management approaches with the exception of shearing in the summer (note that Soay sheep are a wool-shedding breed which naturally loses its fleece in the summer), but some animals are removed from the population each year. These are described as being ‘selected’ in the autumn, but the criteria on which these are based are not given. However, the sheep on North Ronaldsay are confined to the margins of the islands and have limited environmental choices. They are also much more accessible than other populations as there is a permanent human presence on the island.

Although the management of some of the primitive goat herds may also resemble the ‘hands off’ approach taken with the Soay sheep, some of the goat herds in the Southern Uplands and Rum may be subjected to ad hoc culling.

4.8.2 Euthanasia of suffering animals

The focus of attention and concern has been the animals, primarily those in Village Bay on Hirta, which can be seen dying or in very poor welfare (typically starvation, but may also include injury) and understandable frustration that nothing appears to be done to manage suffering. However, these situations could occur in any of the naturalised small ruminant populations we consider and indeed for wild animals in general. One option we considered was that these animals could be identified and euthanised as swiftly as possible to limit the amount of suffering. This may only be possible in cases where the animals can be found and approached. However, use of technologies such as wildlife camera traps, mortality trackers and others may improve the identification of such animals. This approach directly targets the main concern for these animals: the suffering associated with dying of natural causes. It is possible that only a proportion of the sheep on Hirta and none of those on Soay or Boreray could benefit from this intervention, although it may be possible to implement a specific programme of searching for these animals at specific times of year when natural mortality is known to be greatest.

4.8.3 Support for the existing population

Here we considered that the existing populations of small ruminants could receive additional support that may improve the quality of their lives and could target the key areas of concern that were expressed to us by various stakeholders (deaths from starvation, high parasite burdens of some populations, early participation in the rut as a risk factor for poor winter survival in juveniles).

4.8.3.1 Provision of supplementary feeding

Additional nutrition to supplement that available in the environment is commonly provided to hill sheep populations over the winter and it may be possible to provide some additional nutritional support to some of the populations under consideration. This is limited by the need to prevent rats or other rodents gaining access to the islands of St Kilda as inadvertent “passengers” on delivery ships and consequent impacts on important bird species nesting there. However, use of feed blocks or bagged concentrated feed would avoid these issues and could be deployed in some areas. A relevant question here is whether the animals would recognise these as food items and that increased feeding will improve fecundity, the number of juveniles that participate in the rut and increase population density that may lead to higher levels of disease and mortality. Note this would be considered a continuing commitment that would need to be provided every year and would drive population growth and increased risk of overgrazing.

4.8.3.2 Provision of health treatments

Typically, farmed small ruminants are vaccinated against important diseases (such as clostridial disease, pasteurellosis, and toxoplasmosis), drenched with parasiticides and may also receive topical applications of treatments against ectoparasites. Of these, the most important for naturalised small ruminant populations is probably treatment against gastrointestinal parasites as they reportedly carry high worm burdens. Parasite resistance is traded off by ruminants against reproduction – females have finite resources, which are apportioned to current reproductive effort (successful rearing of a lamb) rather than future reproduction (as females who are rearing a lamb have higher spring faecal egg counts than those who do not⁹⁰). This phenomenon, known as the peri-parturient rise in parasitism, is also seen in domestic sheep. Higher levels of faecal egg counts in spring were associated with reduced winter survival in the following winter and suggests that parasitism is an important component of mortality in reproducing female Dall’s sheep. Observations on Soay ewes, given experimental treatments with anthelminthics, suggest ewes were in better condition in the spring and their lightest lambs had better survival. However, early growth and development of juveniles were not affected by maternal anthelminthic treatment⁹¹.

Health treatments could only be applied to those animals that can be caught, which is typically the Village Bay population of Soay sheep, the North Ronaldsay sheep and some of the primitive goat populations in the Southern Uplands. Note this would be considered a continuing commitment that would need to be provided every year.

4.8.3.3 Reproductive management

On farms male lambs and kids are often castrated within a few days of birth to allow reproductive management, particularly on extensive farms where separation of males and females can be challenging. This is usually to allow humans to select which animals will breed, but could also play a role in limiting population size and preventing early engagement in the rut by juvenile males (reproductive management of females is not commonly practised on farms except by separation from males). Experimental castration of some male Soay sheep on Hirta was carried out in the past and is associated with greater survival in castrated males compared to entire males⁹². Limitations would be whether castrating some male lambs would have any impact on the number of juvenile females that become pregnant. In addition, all methods of physical castration are permanent, cause pain to a greater or lesser degree⁵⁹ and would require some human-made decisions about which male animals should breed. Immunocastration of ram lambs is a possibility⁹³, which has the advantage of being painless, but is not permanent and does require more frequent handling to achieve effective sterility. Contraception to limit female reproductive activity, through oral, injectable or implanted devices, has been used in ungulates in zoos to limit breeding⁹⁴, but its application in wild animals for population management internationally has been very limited^95,96. Particular constraints include the need to inject animals, as no oral formulation exists, which may limit its application to the Village Bay population of sheep or other applications where animals can be trapped, identified and handled. Note this would be considered a continuing commitment that would need to be provided every year and may not be a practical long-term solution.

4.8.4 Population reduction

A number of stakeholders expressed concern that there were too many animals on the St Kilda archipelago, and some form of population reduction was needed to reduce suffering. However, we note that for the Soay sheep large winter reductions in population numbers occurred at much lower numbers than the current population (where population crashes have not been recorded since 2012), and not all mortality appears to be density dependent. For primitive goat populations, where the animals are typically in herds and hefts of much smaller numbers (a hundred or so as maximum) than the naturalised sheep populations, and for North Ronaldsay sheep, some population reduction is already occurring. We considered two different approaches to removal of a proportion of the population from the naturalised situation: culling or removing to a different location, and two different methods of removal (darting or trapping).

4.8.4.1 Annual culling of a proportion of the population

Here we consider that to maintain a defined population size, culling of some of the animals would need to take place on an annual or semi-annual basis given the prolificacy of small ruminants. This form of management is already implemented in some of the small ruminant populations we are considering. In discussions with those involved in management of naturalised small ruminant populations or published reports, it appears that this approach should consider social groups that are removed together (as described for the goats in the Cheviots). Additionally, removal of animals may still not achieve the desired number of animals (reports from both North Ronaldsay and Lundy where attempts are made to maintain a specific number of sheep, but typically do not achieve this). We did not consider here the ideal population size to be achieved for each population, given that this would require additional consideration not specific to welfare and is likely a human construct of what is desirable. We thus only considered the impacts on welfare of a decision to remove some animals, likely by stalking and shooting similar to red deer management. In addition, considerations of closed seasons, neonatal dependency periods and selection of suitable animals for removal would need to be considered in this scenario.

4.8.4.2 Annual trapping for removal of a proportion of the animals to be rehomed

Another form of population size reduction is the removal of animals to an alternative location, likely with additional management of the population in the new location. Animals can be trapped by various methods, where they can be caught, restrained, confined and transported to a new location. As with culling we consider that this would also need to take place on an annual or semi-annual basis to achieve a desired level of population reduction. Further, the length and type of journeys these animals will need to undertake to reach a new location, whether suitable habitat can be found for the animals to be removed, and the issues associated with re-domestication of the animal are relevant here. We note that some animals can become ‘trap-shy’ and it may become increasingly difficult to capture animals for removal over time. Animals may also be injured during trapping.

4.8.4.3 Annual sedative darting of animals for removal of a proportion of the population to be rehomed

This scenario is similar to that described with trapping, but may avoid some of the issues with animals becoming wary of trapping and may also reduce the stress associated with capture if animals are sedated. However, it is relevant to note that effective darting of small animals, such as Soay sheep or primitive goats, may require animals to be approached at quite close quarters to be effective, which may elicit a flight response and limit animals that can be caught by this method. This would be particularly difficult for Soay sheep outside Village Bay, which appear to have large flight distances. Sedation is also not without risk, potentially causing injuries or deaths.

As a variant on this approach, it was suggested to us that some of the animals could be sufficiently tame that they could be encouraged to follow feed either to a pen or onto a landing craft (e.g., for Soay sheep). For example, in well-handled farmed sheep it may be possible to encourage them to follow a feed bucket as an alternative to driving with a dog. Despite its initial attractiveness as a low-stress approach, we are unconvinced that these feral animals are tame (for example, the Soay sheep in Village Bay are habituated to human presence, but still exhibit a marked flight response when approached or trapped), and they are also completely unfamiliar with food other than forage, so unlikely to associate a bucket or bag with anything rewarding. Because we are dubious that this can be achieved, we did not consider this method further.

4.8.5 Managing the populations in situ

In this scenario the animals are treated as if they are farmed sheep or goats and managed in a similar way to a farmed population, although still maintained in the environment and landscapes in which they currently live. To achieve this, it is likely that the animals would need to be confined in some way, although hefting might be possible in unfenced areas, and many of the interventions described above (nutritional and health support, reproductive management, annual removal/killing of some animals) would need to be applied. It is likely that decisions relating to breeding management and selection of animals to be removed from the populations would reflect human decisions about desirable outcomes, which may not reflect those that the animals may have made themselves. Surplus animals would need to be removed from the population in some way: if killed for human consumption this would require transport to a suitable abattoir.

4.8.6 Removal of all animals from the current location

In this scenario we considered that, if the welfare outcomes for the animals were sufficiently poor (or other concerns are present such as need for eradication of the species in a particular location), then removal of the population in its entirety from the locality could be a relevant outcome. Under this condition we considered two different approaches: capture and removal of all animals to a managed site elsewhere and culling of all animals from their current location. This approach would, of course, represent a once-and-for-all solution, although practically may not be achieved in a single action.

4.8.6.1 Removal of the whole population to another site

In this scenario the animals would be caught and removed in the ways described under 4.8.4.2 and 4.8.4.3. This approach would emphasise that in the opinion of human managers the current location is either not capable of meeting the animals needs or is impeding other human needs or wants. It is likely that, for at least some of the larger populations under consideration, this may take several years to achieve. Similar issues as to the types and structures of journeys that the animals would need to undertake, the suitability of new habitats, and the adaptation of the animals to a new location would be relevant to this scenario.

4.8.6.2 Culling of the entire population

As with 4.8.6.1, this scenario would reflect concerns that the current environment is not suitable in some way, and it is no longer considered acceptable for the animals to remain. If new locations cannot be found or are considered too impractical, then the culling of all the animals remains as an alternative option. As discussed above, this may need to take place over several years for larger populations and to ensure that all animals are removed. Given the remoteness of some of these locations this may be hard to achieve and may require prolonged engagement of trained stalkers and shooters. Of additional note here is also the conservation status of these animals: Soay and North Ronaldsay sheep are considered priority breeds for conservation by the Rare Breeds Survival Trust, as is the ‘Old English’ goat breed.