<<

Sulphur0815 J. Zool., Lond. (1997) 243, 623-636

Survival and behaviour of castrated Soay (Ovis aries) in a island population on , St. Kilda,

P. A. JEWELL Department of Zoology, University of Cumbridge, Downing Street, CB2 3EJ

(Accepted 19 March 1997)

(With 5 figures in the text)

The free-living population of Soay sheep on the island of Hirta, St. Kilda, in the Outer , has been intensively studied since 1959. The present study was initiated to throw light on the causes of the high mortality rate of adult rams in comparison to that of ewes. In 1978, 1979 and 1980, a total of 72 male lambs was castrated within a day or two of birth. The survival of these castrates has been much longer than that of the entire rams, marked as controls, and longer than that of ewes of the cohorts of the same years. The daily activity pattern of the castrates was similar to that of ewes rather than that of rams. In particular, during the rut the castrates spent most of the daylight hours grazing, in contrast to the rams who were continuously moving and involved in agonistic and sexual encounters. This study substantiates the earlier assertion that the costs in energy of reproduction are a major cause of mortality in temperate zone ungulates. The social organization of some castrates was similar to that of females in that they remained with the home-range group of ewes into which they were born, but other individuals resembled males in that these castrates clubbed together in their own groups. These patterns of behaviour persisted throughout life.

Introduction Many species of ungulates exhibit marked sexual dimorphism, and there is a higher mortality rate among males than among females: there is intense competition between males to become successful sires (Jewell, 1976). In these ungulates, the ratio of males to females at birth is usually close to 1:l (Parkes, 1926) but, in the adult breeding population, females may greatly outnumber males. The feral Soay sheep that occupy the island of Hirta, St. Kilda, the most westerly of the in Scotland, show this condition to a marked degree. This population, showing cyclic fluctuations around 1200 head, has been studied continuously since 1959 (Jewell, Milner & Boyd, 1974; T. H. Clutton-Brock et al., 1991, 1992; Jewell, 1995), and it is recorded that mortality among rams is very high with a heavy incidence in ram lambs and yearling rams: the ratio of adult rams to ewes varies from I :3 to 1:12 through the phases of the cycle (see Fig. 2). Most sheep die in late winter when all suffer from inanition but it is presumed that the events of the rut, occurring in November, further reduce the chances of males surviving over winter. In the rut, not only are males in physical competition with each other for access to females, but they spend much less time feeding (Grubb & Jewell, 1974). In Soay rams, as the mating season approaches, testosterone levels rise and there is a marked increase in aggressive behaviour (Grubb & Jewell, 1973; Lincoln & Davidson, 1977). It is well known that castration reduces libido and aggressiveness (Mattner, 1980; Wodzicka-Tomaszewska, Kilgour & Ryan, 1981), and that castrated rams (also commonly called wethers) are placid animals, although sexual drive can be restored in them by the injection of testosterone (Fulkerson, Adams & Gherardi, 1981). In an attempt to reveal some of the sex-specific factors involved in the differential mortality of males, a number of male lambs on Hirta was castrated in the years 1978, 1979 and 1980. It 623 0 1997 The Zoological Society of London

Sulphur0815 Sulphur0816 624 P. A. JEWELL was hypothesized that the castrates would not attempt to participate in the rut and that their survival rate would be much better than that of ‘entire’ (uncastrated) rams. One of these castrates of 1980 is still to be seen alive today, nearly 17 years later (spring 1997)*, but all the entire rams of the same cohort are long since dead. It seems an appropriate time, therefore, to record the results: a preliminary note on the greatly enhanced survival of castrates has been published (Jewell, 1986). Intriguingly, there was an early observation on the survival of castrates on Hirta: when Soay sheep were introduced to Hirta from the adjacent island of Soay, in 1932, the consignment of 107 sheep included 22 wethers, these being male lambs of the year that had been castrated. Sixteen years later, James Fisher visited Hirta (Fisher, 1948) and afterwards remarked to Morton Boyd (Boyd, 1953) that he had seen some castrates still alive, recognizable by their ‘knob horns’. There is a very extensive literature describing the behaviour of sheep (recently reviewed by Lynch, Hinch & Adams, 1992), mostly relating to husbanded animals. While much quantitative ethology has been performed on ewes and lambs, there is little specifically about wethers. Much research has been carried out on the relationship between ewes and lambs, and on the sexual behaviour of rams (Lindsay & Pearce, 1984), but in commercial flocks the age structure and sex ratio are strikingly dissimilar to those in wild populations. The feral Soay sheep have afforded an opportunity to study behaviour in free-living conditions and with a full complement of rams (Grubb & Jewell, 1966, 1973; Jewell, Milner & Boyd, 1974; Jewell, 1995), and it is in this context that the behaviour of the castrates is particularly interesting. I know of no other study of a large mammal in which a group of castrates has been liberated into a free-living, unmanaged population in this way.

Methods

Soay sheep Soay sheep are a relic population of primitive domestic sheep similar to those that were widespread in Britain in the Bronze Age (Ryder, 1983). In their feral form they survived only on the small island of Soay in the St. Kilda group, but Soays were being brought to parks in mainland Britain at the beginning of the century (Elwes, 1913). After the evacuation of people from the nearby larger island of Hirta in 1930, a founding population of Soay sheep was established there. Over the years their numbers fluctuated between 600 and 1800 on the whole island, but the study was carried out on the sub-population that occupied the Village Glen and the nearby promontory of Ruaival (see Fig. l), numbering 300 to 500 animals. Soay sheep are polymorphic for fleece colour and horns (Boyd et al., 1964). The 2 basic fleece colours are light fawn (L) and dark brown (D) and, as with species of wild sheep, the fleece is usually cast completely every summer. The rams usually carry heavy horns in a close spiral, although occasionally crumpled horns are to be seen, and, very rarely, polled. Almost half the ewes have horns that are finer and more upright than the rams horns and almost half are completely polled (without horns); the remainder have small knobs of horn, or very poorly developed horns, called scurs. A description of home-range organization among Soay sheep is given in Jewell (1966) and Grubb & Jewell (1966), and the social behaviour of Soay sheep is described in detail in Jewell, Milner & Boyd (1974).

Castration In 1978, in a pilot study, 14 male lambs were castrated within 1 or 2 days of birth using the usual method for husbanded sheep of applying a rubber ring round the scrotum (‘Elastrator’, Elastrator Co., Blenheim,

*This castrate died on, or a day or two before, 23 March 1997. He was born on 22 April 1980 and so was one month short of 17 years old.

Sulphur0816 Sulphur0817 SURVIVAL OF CASTRATED FERAL SHEEP 625

Mulloch Mor Old village at the centre of the 'Onachai\ Village Glen study area \ /

FIG.1. Block diagram of Hirta and Dun giving an oblique aerial view from the south. A scale is provided by the island of Dun which is 1425 m long (after Macgregor, 1960). The dashed line encloses the Village Glen study area.

New Zealand). Fourteen other male lambs, caught concurrently, were tagged but not castrated to serve as controls (entire rams) see Table I. These lambs were caught over what proved to be the peak of the lambing period between 14 April and 22 April (median date April 19/20, see Jewell, 1989). An additional 16 male lambs (not shown in Table I) and 54 ewe lambs were tagged. No problems having arisen in the pilot study, a second visit was made in 1979 . This proved to be a crash year for the sheep population, however (see Fig. 2), and there had been heavy mortality in February and March. The surviving ewes were in poor condition, the lambs were light in weight, and many died before the end of the lambing period. No attempt was made, therefore, to castrate a large sample and only 8 castrates and 8 controls (3 of which died neonatally) were tagged, together with 33 other ram lambs and 44 ewe lambs. The following year, 1980, was one of recovery for the sheep (see Fig. 2). No dead sheep were found in the village area. A hundred male lambs were caught of which 50 were castrated and 50 became controls; 83 ewe lambs were caught . The male lambs were assigned to be castrated at random and the colour and horn types into which they grew had to be recorded later when they became mature. Forty-two castrates were seen in July 1983 and categorized in this way (see Results).

Tagging Lambs were given small Dalton plastic ear-tags each with a unique number and of a different colour for each year. The size of the numbers on these tags is too small to be read at a distance, even with field glasses, but small tags were used for aesthetic reasons as many National Trust parties visit the island.

The island of Hirta The island (Fig. 1) lies at 57"49' N lat., 08" 34' W long., and is 637 ha in extent (Macgregor, 1960). It is treeless but provides productive grazing and abundant shelter for the sheep. The sheep feed on the old pastures in the

Sulphur0817 Sulphur0818 Sulphur0818

0 01 € P P PI L1 LS SP I 0 00 0 0 I I& P P s 9 ZI 2s I P 11 s1 9z 82 I€ 8E ZP €P L9

0 01 I z z Z 6 I1 EE €8 0 00 0 0 0 I I€ P P s 9 11 0s 1 EZ ZE €€ 0s 0861 anla -I4 0 00 1 I 1 PS 01 21 OP 2 0 00 0 0 0 0 00 00 0 0 0 I 1 e 0 00 0 I P 9 L d d 0 00 1 I I I 1 Pt ZZ 0 00 0 0 0 0 00 00 0 0 0 00 1 0 00 0 I P 9 9 01 8L61 Yu!d

~ ___~ __ L6 96 S6 P6 €6 26 16 06 68 88 L8 98 58 P8 €8 28 18 08 Sulphur0819 SURVIVAL OF CASTRATED FERAL SHEEP 627

Total sheep Total adults (M+F) 1600 n 1800 F Lambs 1400 El Ewes Rams 1200 t 1000

800

600

400

200

n" ~ ~~ 52 53 5455 56 5753 59 6061 62 63 6465 66 6760a 7071 72 73 74 75767770 798001 8283 84 8586 07 88 89% 91 92 93 94 95

FIG.2. Year-to-year fluctuations in numbers of Soay sheep on Hirta, St. Kilda, taken from the summer census in May or June from 1952 to 1996. In some years a breakdown into the several categories of sheep was not achieved, and some counts were made by inexperienced volunteers. deserted village, on the heather-dominated hill slopes, and on cliff ledges and gulleys. A standard annual census of the sheep on the whole island was, and continues to be, made every year in early summer (Fig. 2) but intensive study is limited to the village area, Fig. I (T. H. Clutton-Brock et a/., 1991). Detailed accounts of the environment and the general methods of study are given in Boyd et al. (1964). and Jewell, Milner & Boyd (1974) where a large-scale fold-out vegetation map is also to be found.

Visits Following the last year in which castrations were carried out (1 980), I visited the island in October-November 1981, July-August 1983, October-November 1984, April 1985, August 1987, and July 1990. Other people from the Soay Research Project and the Large Animal Research Group at Cambridge have been on the island in every year from 1985 to the present.

Recording behuviour Focal unirnul follows were carried out on selected castrates and entire rams. When possible, 2 observers worked together, one following a castrate and the other following a control ram; activity was recorded every 5 minutes. Activity surveys: to compare the activity of rams, ewes, and castrates, all visible animals were recorded as grazing, resting, or engaged in some other activity (standing, walking, sexual, or agonistic encounters) in surveys through the village at half-hour intervals through the daylight hours. Continuous watches of castrates and rams provided details of behaviour; when a castrate who was a member of a group was selected, information on the behaviour of the whole group could be obtained as they remained together.

Results

Physical types The range of colours seen in the fleeces of castrates was the same as in other Soays, but in many

Sulphur0819 Sulphur0820 628 P. A. JEWELL castrates, as they grew older, the fleece was not fully cast and old matted fleece was retained. Many of the castrates, despite being male, failed to grow horns and remained polled (P); others grew small crumpled horns (Crump), rather larger than the scurs of ewes; most of the castrates, however, grew horns (H) that were similar to the horns of ewes (see photograph in J. Clutton-Brock et al., 1990). In a check of 42 castrates seen in July 1983, the categories recorded were: L, 11; D,3 1 and of these LH,6; LCrump,3; LP,2; DH,19; DCrump,3; DP,9. Castrates are much larger than ewes, and are taller and less stocky than rams and have longer limbs (for skeletal measurements see J. Clutton-Brock et al., 1990).

Social organization The small numbers of castrates and controls of the 1978 cohort of lambs were observed in April 1979, one year after birth. Their numbers had been reduced by the heavy mortality in the whole population in the winter of 1978/79 (see Fig. 2) and only six castrates and five controls out of the original 14 in each category were seen. These young animals, just becoming yearlings, were still in their maternal home-range groups and no unusual behaviour was recorded. Observations between 29 October and 9 November 1981 found many of the castrates still with females, presumably adhering to those groups into which they had been born. Typical, and persistent, grazing groups comprised one or two castrates, three or four adult ewes and the several lambs of these ewes. This behaviour of the castrates was not different from that of some young rams of the same age and many female groups had one-and-a-half-year-old rams among them: some groups had both castrates and entire young rams. The behaviour of the few older castrates (surviving from the 1978 and 1979 cohorts) was of great interest in being of two different kinds, some showing persistent association with particular female groups whilst others had associated together in ‘castrates only’ groups, none having joined existing ram groups. This behaviour was in contrast to that of rams of the same age many of whom had joined groups of older rams. Three such groups of castrates were regularly seen in the meadows of the village, one group of six individuals and two of four. (A few of the younger castrates of 1980 also adhered to these groups.) Observations in July-August 1983 revealed several stable groups of castrates as a conspicuous feature of the village sheep. One group of eight individuals that moved in the centre of the village and on the slopes of Oiseval in the east comprised two five-year-olds, one four-year-old, and five three- year-olds. This group was seen on almost every day of the visit and they were always together; they moved freely through the home ranges of ewes and the central village rams and no unusual interactions were observed. Another group of four moved at the east end of the village also using Oiseval, and a group of seven moved between the west meadows and the hill slopes to the west of the village. Yet another five castrates occupied a home range at the west end but they were often accompanied by ewes and a ram who behaved as an adherent to the group. Similarly, five castrates on Ruaival, further west, associated in a group that included ewes and their lambs. Notwithstanding the formation of these distinctive home-range groups, there were several castrates who did not associate with others of their kind but who remained as members of ewe home-range groups. The castrates moved freely through the home ranges of ewes and the rams that lived in the central village and no interactions were observed. A visit to Hirta between 23 October and 21 November 1984 covered the first half of the rut (as judged from the median date of lambing, 23 April 1985, see Jewell, 1989). The castrates were still associated in the conspicuous all-castrate home-range groups. All eight members of the central village group (all could be recognized individually) had survived and were still together. They had recruited another 1980 castrate making a group of nine. A second group of eight that ranged over the mid-village comprised one from 1978, one from 1979, and six from 1980. A group of seven castrates (two born in

Sulphur0820 Sulphur0821 SURVIVAL OF CASTRATED FERAL SHEEP 629 1978, one in 1979, four in 1980) was frequently seen on the slopes above the west end of the village. A large number of castrates moved between St. Brianan’s Fanks (fanks are old stone-walled exclosures formerly used for the growing of crops) further to the west and the promontory of Ruaival (see Fig. 1). They were seen most often as two distinct groups of four (one of 1989, three of 1980) and of six (one of 1978, five of 1980). Other castrates using the west side of the village appeared not to be adherents to stable peer groups but associated with ewe groups and occasionally with young rams. The alternative type of social organization also persisted. Certain castrates remained with ewe home-range groups and, in at least one instance, the castrate’s dam was known individually and she was a stable element of the group. Four castrates were known to have spent their entire lives in this way. The persistence of stable home-range groups, of castrates only, also remained a feature of this class of sheep throughout their lives. As castrates died, and the original groups became smaller, so an amalgamation of the survivors was noted, as in groups of entire rams.

Survival The persistence of the castrates in the population is a striking phenomenon (Table I). Their survival rate and longevity are evidently greater than that of entire rams. Although only one castrate is still alive today (spring 1997), 26 castrates were still alive in April 1991 at a time when all the rams from the same cohorts were dead, Quantifying the sequence of mortality has not been simple because of the impossibility of making a complete census during a short visit to the island. The sheep move widely and are often out of sight behind rocks or walls, or are out on the cliff ledges. Moreover, 10 lambs had originally been caught on Ruaival outside the area subsequently used for intensive studies. In addition, some wethers and entire rams born in the western meadows of the village may have taken up residence beyond the study area. Not all dead animals are found, but regular intensive searches at the end of March ensures that most deaths, which occur in late winter, are enumerated. Very few adult animals die between April and January. In any survey of the living animals, only a proportion of the tagged ones are seen. Although some wethers lost ear-tags, their distinctive appearance ensured that, with persistent counting, the total number of survivors (from the three cohorts combined) could be recorded (Table I). These total numbers are dominated by the large cohort of 1980, so it has proved practical to take the numbers present in that year as the starting point for the graph (Fig. 3) comparing the relative survival of rams, castrates and ewes.

Longevity The longest-living ram amongst the controls was 10 years old. The longest living ewes in the cohorts under consideration here were one from 1978 who died at 15 in 1993, one from 1979 who died at 14 in 1993, and one from 1980 who died at 14 in 1994. In July 1983, I saw a 13-year-old ewe who was successfully rearing twins. The longest-lived castrate is the one seen alive in March this year (1 997) who was born in 1980 and so is over 16 years old. It will be recalled (see Introduction) that James Fisher in his visit to Hirta in 1948 saw some castrates that were by then 16 years old.

Reproductive behaviour The visit to Hirta in October-November 1981 coincided with the early phase of the rut.

Sulphur0821 Sulphur0822 630 P. A. JEWELL

\ \ \

’\.I \

1980 1982 1984 1986 1988 1990 1992 1994 1996 Year

FIG.3. The survival of castrates (-O-) compared with that of entire rams (-m-) and ewes (-0-) standardized as a percentage of the starting numbers (given in Table I).

Occasionally, rams threshed clumps of vegetation and showed low-level aggression among themselves. Rather perfunctory investigation of ewes by rams and ram lambs was seen with frequent flehmen, but with no attempts to mount. No investigation of ewes by castrates was seen. The similarity of appearance of castrates and ewes was notable. There was some low-key interaction between rams and castrates: rams occasionally sniffed castrates and showed flehmen and head-twist towards them. Usually they ignored each other, although the castrates sometimes contested rams with head-butting intention movements. Ram lambs provoked reaction from castrates by approaching them and sniffing as if the castrates were ewes. The castrates either walked away or rebuffed the ram lambs by butting them. This could develop into chasing, reminiscent of the play of lambs. Sometimes mature rams also appeared to mistake castrates for ewes. A six-year-old ram made a low-stretch approach to the rump of a three-and-a-half- year-old castrate. The castrate stood still briefly and the ram then sniffed the ground just behind the place where the castrate had been standing and exhibited flehmen. This was the typical reaction shown to ewes who would normally urinate on the ground behind themselves following such an approach (see description and photographs in Jewel1 & Grubb, 1974). The visit in October-November 1984 covered the first half of the rut and at this time the castrates were between four-and-a-half and six-and-a-half years old, ages at which rams are in their prime. Courtship of ewes by rams and mating activity increased in intensity throughout the period the first mounting was seen on 2 November, a few mountings were seen daily after this and from 16 November onwards seven or eight mountings, between different ram and ewe pairs, were seen each day. Consortships were numerous, the rams making frequent testing of the urine of the ewes. In most respects, the activities of the castrates could not have been more different. Indeed, the group of nine castrates that moved in the central to east side of the village was seen out on the cliffs of Oiseval on 6 November and was not seen in the village again during the remainder of our visit. These castrates had distanced themselves from the main mating arena of the village. The castrates that

Sulphur0822 Sulphur0823 SURVIVAL OF CASTRATED FERAL SHEEP 63 I remained in the village seemed relatively indifferent to the mating activity that was going on around them. Castrates continued lying, cudding, or grazing despite agonistic encounters between rams, or mating activity, nearby; such circumstances usually provoked entire rams to participate, or join in the chasing of ewes. Nevertheless, a few encounters involving castrates were noted. A castrate did nudge a ram, and some castrates on seeing young rams blocking nearby ran over and broke up the confrontation. Two castrates engaged in a head to head confrontation before lying down. Only once was a castrate seen to investigate a ewe and to nudge her from both sides and rise to mount. Many of the encounters were provoked by the activity of entire males. A large ram (6 years old) sniffed castrates and performed kicks and twists towards them; on this occasion the castrates simply moved on. Occasionally castrates did urinate before moving off and then the ram would sniff the urine patch and make flehmen. Ram lambs were persistent in approaching castrates and ‘annoying’ them; these young males were butted and shown off by the castrates. The castrates were larger (at least taller) than most rams and so could hold their own and carry on with their own preferred activities.

Activity patterns The proportions of sheep engaged in different activities over the daylight hours is set out in Fig. 4. Activity on five selected days in 1984 is presented extending from the end of October (pre-rut) to 20 November (near the peak of the rut). It will be seen that the ewes, as expected, spent most of the time grazing, and the castrates presented a similar pattern: throughout November the wethers spent between 60 and 70% of the daylight hours grazing. The rams, in contrast, show a progressive reduction in the time spent grazing and, by mid November, this proportion had been reduced to about 25% of the daylight hours (a result similar to that recorded in November 1964, see Grubb & Jewell, 1974, fig. 7.4). Continuous records of the activity of selected individuals illustrate these activity patterns in more detail (Fig. 5). The distribution of bouts of grazing through the daylight hours demonstrate how intensively a ewe grazed (Fig. 5a). The castrate of Fig. 5b only differed in spending more time resting. The continuation of grazing through the afternoon and on into darkness characterizes the ewes and wethers. The record of a mature ram, however (Fig. 5c) shows how little time is spent grazing, most time being spent in moving or in social and sexual encounters. A castrate (Fig. 5d), recorded concurrently with the ram, was relaxed at midday, even apparently being asleep with its head on the ground, and showed the characteristic period of continous grazing as darkness supervened.

Discussion This study was initiated to throw light on the marked difference in survival rate of rams and ewes in this feral population. The population is limited by food and experiences intermittent ‘crashes’ in numbers followed by recovery over a few years (Boyd, 1974); analysis reveals an exaggerated cycle of ‘boom and bust’ (T. H. Clutton-Brock et al., 1991), but ewes survive better and live longer than rams despite the demands of pregnancy and lactation. It was postulated in earlier research (Boyd & Jewell, 1974) that a key factor bringing about this difference was the feeding behaviour of the two sexes during the rut in late autumn. Support came from the observation that rams spend much less time grazing than ewes during the whole period from late October to early December (Grubb & Jewell, 1974). The hypothesis proposed at the outset of the present research was that castrated males should not participate in the rut and should not reduce the time spent feeding. This is exactly what has been found. The feeding pattern of castrates was similar to

Sulphur0823 Sulphur0824 632 P. A. JEWELL

Standing, walking, antagonistic and sexual interactions

n = 30.7 34.8 47.1 52.8 42.5

100

a, m0 a, 50 i? aa,

n= 9.2 7.6 7.4 9.2 11.2 28 2 8 13 20 1984 Oct. NOV. NOV. Nov. Nov. FIG.4. The proportionate activity of sheep through the daylight hours, based on half-hourly scans, on selected days leading up to the peak of the rut (22 November) in 1984. n = the mean numbers of individuals seen at each half-hour scan.

that of ewes and their survival was even better. Stephenson (1994) has been able to support this interpretation by showing that ram lambs ‘hormonally castrated’ by the injection of a progestagen (which only lasts for the duration of the rut) spent more time feeding and survived better than intact ram lambs. The effects of overpopulation are so severe that the prospects of survival of animals born into successive cohorts in pre-crash, crash, and post-crash years are markedly different (T. H. Clutton-Brock et al., 1991, 1996). The castrates have survived four crash years and so have experienced several kinds of cohort- and age-related privations. This is a particularly revealing outcome of castration that permanently deprives males of the hazards and misfortunes of masculinity. A notable feature of male life histories on Hirta is the precocity of rams. Ram lambs, at seven

Sulphur0824 Sulphur0825 SURVIVAL OF CASTRATED FERAL SHEEP 633

Time of day 09.00h 10.00h 11.00h 12.00h 13.00h 14.00h 15.00h 16.00h I I I I I I I 1984 16 Nov.

16 Nov.

(c) Ram IIW IW I t.l 19 Nov.

19 Nov.

Grazing Standing, walking, Laying head 0antagonistic sexual down

FIG.5. Continuous activity records of sheep through the daylight hours in the first part of the rut in 1984. months of age, join in the rut, and are even more persistent in their endeavours to mate as yearlings at 19 months old. I had thought (Boyd & Jewell, 1974) that the experience gained in this way, and possibly the improvement of their position in the hierarchy of young rams, conferred reproductive success that compensated for heavy mortality. But a further factor was identified when it became clear that the mortality of mature rams, in the winter that terminates in a crash, can be so severe that the young rams form the majority of the male population in the next rut (Jewell, 1989). The direct reproductive success that the young rams gain in this way has been established (Bancroft, 1993; Stephenson & Bancroft, 1995). Notwithstanding this benefit, it may be noted that the survival rate of castrates as lambs and yearlings was high, demonstrating the benefit of not participating in the rut at that age. Physical castration alters many aspects of the anatomy and physiology of the male mammal affecting, amongst other things, the immune system, parasite load, and fat storage (Parkes, 1966; Glucksmann, 1978). (I observed castrated Soays, kept near Cambridge, to have considerable fat reserves in the omentum, and in commercial breeds of sheep, wethers are known to be relatively fatter than entire rams [Ritar et al., 1988; Thompson & Butterfield, 19881. Conspicuous deposits of fat, however, may give a false impression of adequate reserves since it has been shown that even the abundant fat reserves of reindeer suffice to meet only a small part of energy requirements in winter [Tyler, 1987a, h].)It must be noted that through most months of the year the survival rate of entire males and castrates is similarly good, and that deaths are concentrated in late winter in the months of February and March, when inanition takes its toll. Castration averts this disaster by promoting a continuous intake of energy through the late autumn and early winter. This results from the elimination of libido (Wodzicka-Tomaszewska et al., 198 1) and reduced aggressiveness of wethers (Mattner, 1980) and is not due to a difference in appetite because it has been shown in research on housed sheep (Gordon, 1964; Brown et al., 1979) that wethers exhibit a reduction of voluntary food intake in winter in the same way as rams.

Sulphur0825 Sulphur0826 634 P. A. JEWELL The longevity of the Soay castrates, at nearly 17 years, has been greatly superior to that of entire rams at 10 years (Grubb, 1974, also recorded a few rams surviving to this age). There is little published work that permits comparison with other mammals. Parkes (1966), quoting several authors, concluded that “in animals castration does not seem to have a beneficial effect on longevity”. Data from humans are conflicting: Hamilton & Mestler (1969) showed that eunuchs living in an institution for the mentally retarded had better survival than entire men, but Darwin, in 1856, noted: “Yet Eunuchs nor cut Stallions nor nuns are longer lived” (Barrett et al., 1987), and Nieschlag, Nieschlag & Behre (1993), in an analysis of historical documents on the longevity of male singers, found that the mean life span of castrati was not significantly different from normal singers. In cats (Hamilton, Hamilton & Mestler, 1969) and in rats (Talbert & Hamilton, 1965), castration does prolong life, and in observations on bank voles (Cleithrionomys glareolus) living in a natural outdoor enclosure it was shown that castrated males survived better than entire males (Gipps & Jewell, 1979). The present observations comprise the first experiments in which a substantial number of castrates has been released into a free-living population. Short (1979), working on the island of Rum, observed two castrated male red deer calves (castrated within a week of birth) and these animals remained closely attached to their mothers, within their respective groups of hinds, for the rest of their lives. An unexpected and interesting outcome of the present study is that the wethers displayed two distinct types of social cohesion. Some individuals stayed with females, known to be, or presumed to be, their mothers, in their natal home-range groups, and behaved socially very like females throughout their lives. Other castrates associated together in their own permanent castrate groups and in this respect were like normal males. These groups were apparently given cohesion by mutual recognition and acceptance. In physical appearance the castrates looked like ewes, even to the extent of showing similar horn types ranging from polled, through scurred (or crumple-horned), to homed. This effect of castration has long been known (Seligmann, 1906; Marshall & Hammond, 1914), but it is also certain that the odour of the castrates was changed (Izard, 1983). The different physical responses to castration have a genetic basis and it can be asserted that the different behavioural types, revealed by castration, have a genetic basis too (McGill, 1978). These novel observations would not have been revealed if adequately large numbers of lambs had not been castrated. In their review of the ethology of free-ranging domestic animals, Arnold & Dudzinski (1978) make only passing reference to wethers, and they have not been the subject of any previously published studies of behaviour. Yet wethers have been of very great importance in animal husbandry from ancient times (Trow-Smith, 1957; Killen, 1964), through the Middle Ages (Lloyd, 1977-78; Ryder, 1983) and on into the present century (Fraser & Stamp, 1957). Even today, on the island of in , where the sheep are traditionally confined to the foreshore (Hall, 1975) and eat only seaweed, I have observed that a high proportion of the sheep are wethers and serve as a ‘walking larder’ (J. Clutton-Brock, 1989). The data presented here provide a new dimension to the many factors considered by Boyd & Jewell (1974) for the management of the Soay flock on Hirta.

I thank the National Trust for Scotland and Scottish Natural Heritage for permission to work on St. Kilda, and the St. Kilda Detachment of the Royal Artillery and the Royal Corps of Transport for logistic support. 1 am grateful to Howard Payton for his assistance in the field, to Professor Colin Finn MRCVS and Drs Robert Gibson and Philip Armitage for assistance with the castrations, and to Professor T. H. Clutton-Brock and members of the Large Animal Research Group at Cambridge for much help in collating information. Dr J. Morton Boyd, Professor Stephen Hall and Professor Roger Short kindly made valuable comments on a preliminary draft. I am grateful to the National Geographic Society for financial support.

Sulphur0826 Sulphur0827 SURVIVAL OF CASTRATED FERAL SHEEP 635 REFERENCES Arnold, G. W. & Dudzinski, M. L. (1978). Ethohgy oJfree-ranging domestic animals. Amsterdam: Elsevier. Bancroft, D. (1993). Genetic variation andjtness in Soay sheep PhD dissert., University of Cambridge. Barrett, P. H., Gautrey, P. J., Herbert, S., Kohn, D. & Smith, S. (1987). Charles Darwin’.~notebooks, 1836-1844: 171. British Museum Natural History: Cambridge University Press. Boyd, J. M. (1953). The sheep population of Hirta, St. Kilda, 1952. Scott. Nut. 65: 25-28. Boyd, J. M. (1974). Introduction. In I.slandsurvivor.s: the ecology ofthe Souv ,sheep ofst. Kilda: 1-7. Jewell, P. A,, Milner, C. & Boyd, J. M. (Edsj. London: Athlone Press. Boyd, J. M., Doney, J. M., Gunn, R. G. & Jewell, P. A. (1964). The Soay sheep of the island of Hirta, St Kilda. A study of a feral population. Proc. zool. Soc. Lond. 142: 129-163. Boyd, J. M. & Jewell, P. A. (1974). The Soay sheep and their environment: a synthesis. In Island survivor,^; the ecology ofthe Soay sheep ofst. Kilda: 360-373. Jewell, P. A,, Milner, C. & Boyd, J. M. (Eds). London: Athlone Press. Brown, W. B., Forbes, J. M., Goodall, E. D., Kay, R. N. B. & Simpson, A. M. (1979). Effects of photoperiod on food intake, sexual condition and hormone concentrations in stags and rams. J. Physiol. (Cumb.)296: 58P-59P. Clutton-Brock, J. (1989) (Ed.). The walking larder. London: Unwin Hyman. Clutton-Brock, J., Dennis-Bryan, K., Armitage, P. L. & Jewell, P. A. (1990). Osteology of the Soay sheep. Bull. Br. Mus. (Nut. Hist.) 56: 1-56, Clutton-Brock, T. H., Price, 0. F., Albon, S. D. & Jewell, P. A. (1991). Persistent instability and population regulation in Soay sheep. J. Aninz. Ecol. 60: 593-608. Clutton-Brock, T. H., Price, 0. F., Albon, S. D. & Jewell, P. A. (1992). Early development and population fluctuations in Soay sheep. J. Anim. Ecol. 61: 381-396. Clutton-Brock, T. H., Stevenson, I. R., Marrow, P., MacColl, A. D., Houston, A. I. & McNamara, .I.M. (1996). Population fluctuations, reproductive costs and life-history tactics in female Soay sheep. J. Anim. Ecol. 65: 675-689. Elwes, H. J. (1913). Guide to the primitive breeds of sheep and their crosses. Edinburgh: R. B. Clark Ltd. Fisher, J. (1948). St. Kilda: a natural experiment. In The New Naturalist: a journal ofBritish natural history: 9 1-109. Fisher, J. (Ed.). London: Collins. Fraser, A. & Stamp, J. T. (1957). Sheep husbandry and diseases. 5th edn. London: Crosby Lockwood. Fulkerson, W. J., Adams, N. R. & Gherardi, P. B. (1981). Ability of castrate male sheep treated with testosterone to induce and detect oestrus in ewes. Appl. Anim. Ethol. 7: 57-66. Gipps, J. H. W. & Jewell, P. A. (1979). Maintaining populations of bank voles, Ckthrionornys glareoh, in large outdoor enclosures, and measuring the response of population variables to the castration of males. J. Anim. Ecol. 48: 535-555. Glucksmann, A. ( 1978). Sex determinution and sexual dimorphism in mammals. London: Wykeham Publications. Gordon, J. G. (1964). Effect of time of year on the roughage intake of housed sheep. Nature (Lond.) 204: 798-799. Grubb, P. (1974). Population dynamics of the Soay sheep. In Island survivors: the ecology of the Soay sheep ofst. Kilda: 242- 272. Jewell, P. A., Milner, C. & Boyd, J. M. (Eds). London: Athlone Press. Grubb, P. & Jewell, P. A. (I 966). Social grouping and home range in feral Soay sheep. Svmp. zool. Soc. Lond. No. 18: 179-2 10. Grubb, P. & Jewell, P. A. (1973). The rut and the occurrence of oestrus in the Soay sheep on St. Kilda. J. Reprod. Fertil. (Suppl.) 19: 49 1-502. Grubb, P. & Jewell, P. A. (1 974). Movement, daily activity and home range of Soay sheep. In Island suvvivor,s: the ecology of the Soay sheep ofst. Kilda: 160-194. Jewell, P. A., Milner, C. & Boyd, J. M. (Eds). London: Athlone Press. Hall, S. J. G. (1975). Some recent observations on Orkney sheep. Mammal Rev. 5(2): 59-64. Hamilton, J. B., Hamilton, R. S. & Mestler, G. E. (1969). Duration of life and causes of death in domestic cats: influence of sex, gonadectomy and inbreeding. J. Gerontol. 24: 427-437. Hamilton, J. B. & Mestler, G. E. (1969). Mortality and survival: comparison of eunuchs with intact men and women in a mentally retarded population. J. Gerontol. 24: 395-41 1. Izard, M. K. (1983). Pheromones and reproduction in domestic animals. In Pheromones and reproduction in mammal^: 253- 285. Vandenbergh, J. G. (Ed.). New York: Academic Press. Jewell, P. A. (1966). The concept of home range in mammals. Syinp zoo/. Soc. Lond. No. 18: 85-109. Jewell, P. A. (1976). Selection for reproductive success. In The evolution of reproduction: 71 -109. Austin, C. R. & Short, R. V. (Eds). Cambridge: Cambridge University Presa. Jewell, P. A. (1986). Survival in a feral population of primitive sheep on St Kilda, Outer Hebrides, Scotland. Natl Geogr. Rex 2: 401 -406. Jewell, P. A. (1989). Factors that affect fertility in a feral population of sheep. Zoo/. J. Liizn. Soc. 95: 163-174. Jewell, P. A. (1995). Soay sheep. In St Kilda the continuing story qf the islands: 73-93. Buchanan, M. (Ed.). Edinburgh: HMSO.

Sulphur0827 Sulphur0828 636 P. A. JEWELL Jewell, P. A. & Grubb, P. (1974). The breeding cycle, the onset of oestrus and conception in Soay sheep. In Island survivors: the ecology of the Souy sheep of St. Kilda: 224-241. Jewell, P. A,, Milner, C. & Boyd, J. M. (Eds). London: Athlone Press. Jewell, P. A,, Milner, C. & Boyd, J. M. (1974) (Eds). Island survivors: the ecology of the Soay sheep of St. Kilda. London: Athlone Press. Killen, J. T. (1964). The wool industry of Crete in the Late Bronze Age. Annu. Br. Sch. Athens 59: 1-15. Knight, T. W. & Lynch, P. R. (1980). Source of ram pheromones that stimulate ovulation in the ewe. Aninz. Reprod. Sci. 3: 133- 136. Lincoln, G. A. & Davidson, W. (1977). The relationship between sexual and aggressive behaviour, and pituitary and testicular activity during the seasonal sexual cycles of rams, and the influence of photoperiod. J. Reprod. Fertil. 49: 267-276. Lindsay, D. R. & Peace, D. T. (1984) (Eds). Reproduction in sheep. Cambridge: Cambridge University Press. Lloyd, T. H. (1977-78). Husbandry practices and disease in Medieval sheep flocks. Vet. Hist. 10: 3-14. Lynch, J. J., Hinch, G. N. & Adams, D. B. (1992). The behaviour nf sheep. CAB International & CSIRO Australia. Macgregor, D. R. (1960). The island of St. Kilda. Scott. Stud. 4: 1-48. Marshall, F. H. A. & Hammond, J. (1914). On the effects of complete and incomplete castration upon horn growth in sheep. J. Physiol. (Camb.)48: 171-176. McGill, T. E. (1978). Genetic factors influencing the actions of hormones on sexual behaviour. In Biological determinants (f sexual behaviour: 7-28. Hutchison, J. B. (Ed.). Chichester: John Wiley. Mattner, P. E. (1980). Gonadal hormone control of male sexual behaviour. In Behaviour in relation to reproduction, management and welfare of farm animals: 11-18. Wodzicka-Tomaszewska, M., Edey, T. N. & Lynch, J. J. (Eds). Rev. Rural Sci. No. 4: University of New England. Nieschlag, E., Nieschlag, S. & Behre, H. M. (1993). Lifespan and testosterone. Nature (Lond.) 366: 215. Parkes, A. S. (1 926). The mammalian sex ratio. Biol. Rev. 2: 1-5 1. Parkes, A. S. (1966). The internal secretions of the testis. In Marshall’s physiology of reproduction 111: 412-569. London: Longmans. Ritar, A. J., O’May, P. J., Williams, P. M., Gilbert, K. D. & Bond, E. M. (1988). Effects of castration and age on carcass composition of male lambs. Proc. Aust. Soc. Anim. Prod. 17: 457. Ryder, M. L. (1983). Sheep & Man. London: Duckworth. Seligmann, C. G. (1 906). Exhibition of the skull of a domestic sheep that had been castrated when young. Proc. zool. Soc. Lond. 1906: 903. Short, R. V. (1979). Sexual behaviour in red deer. In Animal reproduction: 365-372. Hawk, H. W. (Ed.). Beltsville Symp. Agric. Res. No. 3. Montclair: Allanheld, Osmun & Co. Short, R. V. (1985). Species differences in reproductive mechanisms. In Reproducrion in mammals 4. Reproductivefitness: 24- 61. 2nd edn. Austin, C. R. & Short, R. V. (Eds). Cambridge: Cambridge University Press. Stephenson, I. R. (1994). Male-biased mortality in Soccy sheep. PhD dissert., University of Cambridge. Stephenson, I. R. & Bancroft, D. R. (1995). Fluctuating trade-offs favour precocial maturity in male Soay sheep. Proc. R. SOC. Lond. B Biol. Sci. 262: 267-275. Talbert, G. B. & Hamilton, J. B. (1965). Duration of life in Lewis strain of rats after gonadectomy at birth and at other ages. J. Gerontol. 20: 489-491. Thompson, J. M. & Butterfield, R. M. (1988). Changes in body composition relative to weight and maturity of Australian rams and wethers. 4. Adipocyte volume and number in dissected fat partitions. Anim. Prod. 46: 387-393. Trow-Smith, R. (1957). A history of British husbandry to 1700. London: Routledge & Kegan Paul. Tyler, N. J. C. (1987~).Natural limitation ofthe abundance of the high arctic Svalbard reindeer. PhD thesis, University of Cambridge. Tyler, N. J. C. (1987b). Body composition and energy balance of pregnant and non-pregnant Svalbard reindeer during winter. Symp. zool. Soc. Lond. No. 57: 203-229. Wodzicka-Toniaszewska, M., Kilgour, R. & Ryan, M. (1981). ‘Libido’ in the larger farm animals: a review. Appl. Anim. Behuv. Sci. 7: 203-222.

Sulphur0828