British

Vol. 57 No. 2

FEBRUARY 1964

Observations on the St. Kilda By Estlin Waters

(Plate 8)

INTRODUCTION ST. KILDA'S DISTINCTIVE race of the Wren troglodytes hirtensis (plate 8b) has long excited the interest of naturalists. It was Martin (1698) who first mentioned on this lonely archipelago almost forty miles beyond the Outer . The many subsequent accounts have been summarised by Armstrong (1953) and Williamson (1958). I was Medical Officer on St. Kilda between May 1961 and September 1962, and, except during short absences from 20th October to 1st December 1961 and from 20th February to 19th March 1962, I was able to make observations on the Wrens there at all times of the year. I was confined to the main island, , and as with all previous observers my studies were chiefly concentrated on the population inhabiting the Village Glen. TOTAL POPULATION OF HIRTA Williamson (1958) made a thorough study of past observations on the numbers of St. Kilda Wrens and showed that the only reliable method of estimating the population is by a dawn chorus census before the onset of nesting. Using this method in 1957, Williamson located 116 pairs on Hirta between 21st and 31st May. I repeated the census between 27th May and 4th June 1962, counting the singing Wrens between 03.45 and 05.30 hours* on all days when the weather was suitable. Winds were the main difficulty experienced, since accuracy tended to fall off rapidly as they increased, and counting was abandoned if they were above Beaufort force 4. To reduce further the effects of wind, the coastline was not always covered systematically on successive days, the route being varied to enable counting to be done on the •All times given in this paper are GMT.

49 BRITISH BIRDS

FIG. I. Plan of Village area, Hirta, showing the approximate boundaries of the territories of St. Kilda Wrens Troglodytes troglodytes hirtensis in 1962. No Wrens were seen at the western end of the Village street in territory no. 2 {cf. fig. 2) (drawn by Robert Giilmof) sheltered side of the island. My total for 1962 was 92 singing males on Hirta and this probably represents a real though slight decrease in the population from the figure of 116 obtained by Williamson in 1957. The general pattern of distribution was similar to that in 1957, with the main concentrations occurring on Cam Mor, the Mullach Bi coast and Oiseval. Bleak stretches, such as Na h'Eagan and most of the island away from the cliffs, were again sparsely populated. The survival rates of island Wrens are likely to depend on climate. Williamson (1958) suggested that the St. Kilda Wren was well able to survive severe storms but might suffer more in a cold spell. Winters on St. Kilda are extremely stormy and the island has been described as one of the windiest places in the British Isles (Green 1959). The winter of 1961-62 was no exception there and gusts of over 130 m.p.h. were recorded on an anemograph (Waters 1962). That the St. Kilda Wren survives in such conditions is due to its extremely close associa­ tion at that time of year with shelter among rocks, as I have described elsewhere (Waters 1962). Storms are so frequent that it exists for most of the winter sheltering under and between rocks and crags; it does not even have to leave its rocky labyrinth to obtain food. It therefore seems unlikely that more frequent or more severe storms would greatly effect the population in winter.

5° OBSERVATIONS ON THE ST. KILDA WREN

FIG. 2. Plan of Village area, Hirta, showing the approximate boundaries of the territories of St. Kilda Wrens Troglodytes troglodytes hirtensis in 1957 {reproduced from Williamson 19)8) Winters on St. Kilda are usually a few degrees warmer than on in the (Green 1959). Severe cold is uncommon; thus during the winter of 1961-62 air frost was recorded on only eight days, the minimum temperature recorded was 260 F. and the daily maximum was always above 32°F. Temperatures lower than this, if they do occur, are extremely uncommon. Snow probably lies on the ground for up to 14 days each winter, and did so in 1961-62, but unless it is particularly deep, or accompanied by hard frost, it probably affects the St. Kilda Wren less than the European Wren T. t. troglodytes on the mainland, where the species has to forage more in the open. The troglodyte existence of the St. Kilda Wren is a good protection against both wind and snow.

TERRITORY The size of the territory of the St. Kilda Wren varies considerably in different areas. The population is densest on Cam Mor, where in 100,000 square yards Williamson estimated nine or ten pairs in 1957 and I estimated eight pairs in 1962. It is less dense in Village Glen, where my territory observations were made. The extent of neutral zones in Village Glen has aroused some lively controversy in letters in this journal (Armstrong 1959, Williamson 1959). It was therefore 5i BRITISH BIRDS with particular interest that I studied this area in 1962. The boun­ daries of territories, plotted from the positions of Wrens singing from March to June, are shown in fig. 1. This is compared with William­ son's results in 1957, reproduced as fig. 2. The size of the territories on the whole did not appear to have in­ creased since 1957, despite there being one pair less in 1962. William­ son (1958), after a review of past literature, said that it was clear that the situation of some territories remained much the same from year to year. This is strikingly demonstrated by figs. 1 and 2. As will be shown below, a similar constancy from year to year appears to apply also to the site chosen for the nest. It is unlikely that many of the 1957 Wrens were alive in 1962 as, according to Armstrong (1955), the greatest age to which a Wren has been known to live is only five years. There is a large amount of neutral ground in Village Glen which is not defended by song; nor is it much used for obtaining food for the nestlings. After establishing the extent of the territories by observing singing Wrens, I confined my studies mainly to territories nos. 4, 5 and 6. Except for a small area of about 200 square yards among the boulders at the base of Oiseval and immediately above the enclosures at the south-east corner of territory no. 5, all Wrens seen during the breeding season were within the limits defined by the positions of singing males. It was only after the young had flown, when family parties moved indiscriminately around the Village Glen, that these territories broke down. Then family parties were seen unmolested in the neutral areas and in the territories of other pairs. I do not know at what stage the territories are again of importance, but it is certainly well before the following spring. In January and early February, when there was no song, a definite and consistent boundary had been established between the Wrens which later held the territories plotted as nos. 5 and 6. This remained constant at least from January until July. During January and February I could not drive a Wren along the stone walls from one territory into another. The Wren would become agitated as it neared the dividing line, start a noisy rattle, and finally break back past me, flying in an arc to the cover of the wall behind me. This flight in itself was unusual in the winter when the attachment of Wrens to rocky cover is so extremely strong. In the winter, as in the spring, territories nos. 5 and 6 were separated by a neutral zone about 75 feet across at the narrowest point. Armstrong (1955) suggested that a reduction in song implied a reduc­ tion in territorialism. It would appear, however, that with the St. Kilda Wren territorialism occurs even during the complete absence of song. Even inside the territories shown in fig. 1 there were large areas of

52 OBSERVATIONS ON THE ST. KILDA WREN unsuitable habitat rarely or never visited by the Wrens. They were seldom seen on the arable grassland, for example. Harrisson and Buchan (1934) estimated that 85% of the food for the nestlings came from 1% of the area of the territory. This was certainly true in 1961 and 1962 in any short period of time (up to a couple of days), but the Wrens would then discover another area with a rewarding food supply and make repeated visits there instead. It was frequent, or even usual, for the male and female of any pair to work different areas, but these were not constant from day to day. (The sexes could often be differentiated in the field as one or the other was frequently a partic­ ularly dark or light , though true dimorphism does not occur as there are many intermediates; in territory no. 3, where both birds were of similar coloration, the female was caught and ringed.) During the winter months the Wrens of Village Glen were even more restricted to certain parts of their territories, mainly along the dry-stone walls or, in the cases of those in territories nos. 1 and 7, among the boulders of the storm-beach. Even on the few calm days during the winter, they were rarely seen more than a few inches from the shelter of stones. While this protection may be essential for the survival of such weak fliers on such a stormy island, it severely res­ tricts their foraging areas. During the winter they were seldom observed foraging around the cleits in territories nos. 3, 4 and 5 (plate 8a) except the ones built into the main stone walls. The other cleits are often separated by fifty feet or more of open grassland, a distance apparently sufficient to discourage Wrens from flying regularly from one to another. These Village Wrens had territories of over 10,000 square yards, yet in winter each was almost confined to a length of less than a quarter of a mile of dry-stone wall, with its incorporated cleits.

SONG Conflicting opinions about the loudness of the songs of St. Kilda Wrens compared with those of mainland Wrens have been given in the past. So much depends on the acoustics of the surroundings, however, that further subjective impressions are of little value. Nevertheless, the song of the St. Kilda Wren is distinctly different from that of the mainland race: it is sweeter, has more trills, and lacks the 'throaty' quality of T. t. troglodytes. The seasonal incidence of song, as observed in 1961 and 1962, is given in fig. 3. This differs from that of the mainland race, which sings all the year round (Witherby et al. 1938), but is similar to that of the Wren T. t. ^etlandicus(V'enables and Venables 195 5). My only December record involved a single song phrase, lasting about three or four seconds, on the afternoon of the 12th. No further song was 53 BRITISH BIRDS

F: G . 3. The song output of the St. Kilda Wren Troglodytes troglodytes hirtensis, based on observations made chiefly between 08.00 and 10.00 hours in the Village area, Hirta. The figures indicate the number of twelve-minute periods per hour in which song was heard, calculated as weekly averages. The question marks show periods when no observing was done {drawn by Robert Gillmor) noted until 20th February when short bursts were heard at 07.30 hours and 10.30 hours. A complete record was kept of Wren song heard in each 12-minute period from 19th March to 14th June 1962, as well as occasionally at other times of the year. Fig. 3 is based on these observations. The amount of song at the end of March and during the first two weeks in April showed daily fluctuations and none at all was heard on 22nd, 25th and 26th March, nor from 30th March to 4th April. This may be related to the exceptionally cold spell about this period, as shown in fig. 4. From 5 th April song was heard daily until the third week in July, reaching a maximum during the last two weeks of May and the first week of June. The dawn censuses in both 1957 and 1962 were carried out during this peak period. From the end of the first week in April to the end of June there were well marked dawn and less prominent evening choruses. Occasional songs were heard during every hour of the day, with the fewest between 12.00 and 15.00. Wrens wTith nestlings up to a week old continued to join in the dawn chorus, but were then quiet for the remainder of the day. The evening chorus had ceased to exist by the end of June, when only an occasional song was heard after 11.00 hours. By early July the dawn chorus had become desultory and there was then less than a tenth of the amount of song heard on mornings in late May. By the middle of July there was no true chorus at dawn, but song was now spasmodic at all times of the day, even if rather less frequent in the afternoon; this pattern continued at least until well into October. My data show considerable differences from those of Harrisson and Buchan (1936) in late July and early August of 1931. During this period in 1961 and 1962 there was no evening chorus, nor did I note a second peak of song in mid-August. Harrisson and Buchan correlated the resurgence of singing which they recorded then with the young leaving the nest. In 1961 and 1962 all the young in Village Glen, 54 OBSERVATIONS ON THE ST. KILDA WREN

FIG. 4. The relationship between song output of St. Kilda Wrens Troglodytes troglodytes hirtensis and temperature in the Village area, Hirta, between 25 th March and 23rd April 1962. As in fig. 3, the output is given as the number of twelve- minute periods per hour in which song was heard between 08.00 and 10.00 hours. The temperatures shown are the daily maximum and minimum screen temperatures. After 23rd April both temperature and song remained fairly constant {drawn by Robert Gillmor) where most of my song observations were made, had left the nest by the last week of July. Even then, however, there had not been an appreciable increase in song, the males still being fully occupied in carrying food for the young. This may indicate a poorer food supply than in 1931 and is perhaps evidence in support of Williamson's (1958) view that the Village area is becoming less suitable as a Wren habitat. My observations elsewhere on Hirta, although not extensive enough for detailed comparison, suggest that song may have been more frequent than in Village Glen during July and August. The dawn chorus used to begin well before local sunrise (calculated from the- Astronomical Ephemeris H.M.S.O. for a position 58CN, 8°3o'W). The first song of 1962 was 24 minutes before sunrise on 20th February. Table 1 gives the times of the first song compared with local sunrise on several other dates. St. Kilda Wrens were never heard to sing after sunset. The time 55 BRITISH BIRDS Table I. Time of first song of St. Kilda Wren Troglodytes troglodytes birtensis compared with time of local sunrise (all GMT) MAY JUNE JULY 13th 26th 27th 2nd 10th 12th 15th 23rd 5th

First song 03.05 02.45 02.50 02.40 02.20 02.25 02.25 02.35 02.45 Local sunrise 04.16 03.49 03.49 03.37 03.32 03.32 03.31 03.32 03.41 Minutes before 71 64 59 57 72 67 66 57 56 of the last song varied only slightly during the main song period from about 18.35 hours in the third week of March (when it was only 14 minutes before local sunset) to 19.55 hours on 4th May (31 minutes before local sunset). It then became somewhat earlier and the latest song in June was at 19.42 hours on the 10th (112 minutes before local sunset). The usual territorial song of the St. Kilda Wren lasts from five to seven seconds, but is shorter than this early and late in the season. A longer and more subdued sub-song was heard on several occasions between 28th April and 10th June, when uninterrupted phrases of about 20 seconds frequently followed one another with only a second's pause in between; the longest such phrase I timed lasted 31 seconds. These longer bursts were probably courtship songs, although a second Wren was not always apparent. On 28th April, when no mate was present, one Wren uttered two long sub-songs each of over 20 seconds, then gave a double chirp note and immediately moved along the storm- beach collecting nesting material. Armstrong (1955) has recorded that European Wrens occasionally sing courtship songs 'absent- mindedly' while engaged in building; from this observation it appears that St. Kilda Wrens may do the same. In the normal singing position the tail of the St. Kilda Wren is not raised, though it is cocked when the bird is excited or anxious. Only rarely did I hear counter-singing between two Wrens in the Village Glen, doubtless due to the small number now inhabiting this area; counter-singing was frequent on Cam Mor. Singing in flight was very common and was observed on most days in May and early June. The open terrain of St. Kilda makes this habit easily observed. Arm­ strong (1955) regarded flight-song as playing a more important part in the insular races than it does among European Wrens. Feeding and singing often alternated, especially about the middle of the morning in May. I watched one Wren thus occupied on 20th May 1962: after several songs from a stance on the top of a dry-stone wall he stopped, darted five feet up into the air, caught an insect and flew on to another stone further along the wall, where he continued to sing. This was,

56 OBSERVATIONS ON THE ST. KILDA WREN incidentally, the only time I saw a Wren taking prey in flycatcher fashion. It is known that Wrens sing from progressively higher perches as the season advances. That this is particularly marked in the St. Kilda race is perhaps surprising in view of its general habitat. The Wrens of the Village Glen use man-made objects to obtain more prominent perches. Early in the year the song is delivered from walls and stones on the sides of cleits. By the end of April the highest stones on a length of wall or on the top of a cleit are selected as song- perches. A Wren on the storm-beach sang from the top of a rusty 40-gallon oil-drum which gave it a commanding view over the boulders. Early in May the Wren from territory no. 5 started singing from the top of the chimney-pot on the Factor's House, and on 3rd June sang from an electric cable twenty feet above the ground. Strong winds at this time had little or no effect on the amount of song, but did drive the Wrens from the higher perches. By July the higher perches were no longer used.

BREEDING The greater spread of nesting that occurs in the northern insular Wrens applies to the St. Kilda race, as has been shown by Williamson (1958). In 1962 none of the six pairs within the perimeter wall of Village Glen was double-brooded, but there was a considerable spread of nesting. I was able to record the dates when five of these broods left the nest: they were 24th June, 26th June, 5th July, 21st July and 22nd July. In none of these territories was the loss of a first brood suspected. Thus the spread of nesting in the Village area was four weeks. I found a nest with young on Cam Mor on 3rd June 1962 and these young had left by the 18th June, at least six days earlier than any in the Village area. My 1961 observations were less complete, but no second broods were suspected in the Village area where all the young had fledged by the end of July. Nest-building was observed between 19th April and mid-June; an isolated instance of nest-building in January 1959 has been published by Williamson and Boyd (i960). The nests were constructed mainly of dried grass with occasional green or dried , bracken and mud. As has been noted previously, the junction of roof and side wall inside a cleit is a frequent nest-site in the Village area. During late May and early Tune 1962 I carefully inspected all the cleits enclosed by the perimeter wall and east of Amhuinn Mhor, and also the group of cleits around Tobar Childa and the mediaeval village at the foot of Conachair. In this area there were 150 cleits that appeared suitable for Wrens' nests. Thirteen Wrens' nests were found inside these cleits, and at least seven were obviously recently built. Two remained

57 BRITISH BIRDS in sites where I had found them in 1961, while some others were probably two years old. In addition to these thirteen nests, there were the remnants of several much older ones. Only three of the seven or more recent nests were subsequently lined and used for breeding. The lining was of feathers, but some nests had wool as well (the island's are sloughing their fleeces at this season). All thirteen nests were built at the highest convenient point inside the cleit. Their height, which thus depended on the height of the cleit, varied from 3 feet 6 inches to 6 feet 8 inches above the ground. All except one were situated in the half of the cleit nearest the doorway, sometimes immediately above the lintel stone over it. In the case of the only nest found well away from the doorway, the cleit was an especially well-lit one with several large gaps in the side walls. The siting of the nest near the doorway is probably related to the better illumination there, for, as described below, a Wren foraging late in the evening may have difficulty in entering a nest in the darkness of a cleit. The doorway is, however, little used as an entrance by the Wrens; they usually enter and leave through a series of chinks in the sides and far end. Twelve of the nests had the large side-entrances typical of this race (plate 8c). The remaining one was the only unroofed St. Kilda Wren's nest I ever saw. Three recent nests were found in the Village area in sites other than in cleits, but only one of them was actually laid in. This, and the fact that only three of the seven or eight recent nests in cleits were used for breeding, suggests a rather larger number of cock's nests than was indicated for the St. Kilda Wren by Armstrong (1953, 1955)- In 1962 in the Village area there was an average of over one and a half cock's nests to every nest used. In territory no. 6 three cock's nests were completed before the pair finally built in a fourth site. The typical race in Britain, however, averages many more cock's nests than this. The nest-sites used for breeding by the six pairs within the perimeter wall in 1962 showed a remarkable similarity to those recorded by Williamson (1958) in 1957. The Wrens of territories nos. 4 and 5 nested in cleits. The pair in no. 3 reared their young in a little pocket under the collapsed turf roof of an old black-house (this nest was pointed out to me by Dr. J. Morton Boyd who had found others in the same building in several previous years). The Manse bird built a cock's nest in the turf covering the roof of a cleit but later nested in the erosion scarp, as did the pair in territory no. 7. While the nests inside cleits had nothing to guard their entrances, those built outside these structures in the Village area and on the erosion scarp had their entrances sheltered and camouflaged by tufts of vegetation or turf. There was no evidence of polygamy among the Wrens in the Village area in 1962, nor did I suspect it during 1961. There was, however, 58 OBSERVATIONS ON THE ST. KILDA WREN an apparently unattached male in 1962. This Wren was colour-ringed when caught in a mist-net on the storm-beach on 28th April. His song output became gradually less during May while that of the other male in the territory increased. Both patrolled territory no. 7 and, although the colour-ringed bird usually kept more to the western end, there was a large area of overlap. By the end of May the colour- ringed bird was no longer to be seen on the beach. I saw him again early on the morning of 3rd June when he flew on to a cleit immedi­ ately after the male of territory no. 4 had alighted there. The occupier of the territory sang loudly twice and then the colour-ringed bird flew off in the direction of the Dry Burn without making a noise of any sort. During the time both were on the cleit they were on opposite sides and not in each other's view. My interpretation of this event was that the unattached male saw the other fly on to the cleit and, unsure of its sex, followed it. The loud 'territorial' song told him that the other was a male in possession of the territory, whereupon he left without a challenge. I never saw this colour-ringed Wren again.

Incubation There are no previous data on the lengths of the incubation and fledging periods of the St. Kilda Wren, except that Williamson (1958) recorded that the combined period at one nest was 40 days. Table 2 summarises my observations at three nests in 1962.

Table 2. Incubation and fledging periods at three nests of the St. Kilda Wren Troglodytes troglodytes hirtensis in 1962 Clutch Eggs Incubation Young Fledging Territory completed hatched (days) left nest (days)

Tobar Childa 21 st May 10th June 20 26th June 16 No. 3 19th June 5 th July 16 No. 4 19th May 7th June 19 + 24th June 17 or earlier

In all three cases the completed clutch was of six eggs. At the Tobar Childa site only one nestling eventually flew and two eggs remained in the nest for at least two days after it was deserted; the fate of the other three eggs is not known as disturbance was kept to an absolute minimum (the nests were visited once or twice each day and inspected while the adult Wrens were away foraging). The eggs in territory no. 4 were cold when felt on the first three days and, although the clutch had been completed on or before 19th May, they were probably not brooded regularly until the 22nd. The observations on incubation given in table 3 are the first for the St. Kilda Wren. All incubation was by the female as is the case with 59 BRITISH BIRDS Table 3. Incubation rhythm at two nests of St. Kilda Wren Troglodytes troglodytes birtensis in Village area on various days in June 1962 Average Average Screen Days duration duration Time temperature before of brooding of absences (GMT) (°F) hatching (minutes) (minutes)

10.00-11.00 5i° 1 15.6 5.0 12.00-13.00 5i° 1 7.6 3-9 13.00-14.00 5*° 1 7.0 3-5 14.00-15.00 55° 5 9-5 5.0 15.00-16.00 55° 5 9.0 6.8 17.00-18.00 Si° 1 11.6 6.0 17.00-18.00 54° 0 12.0 5-5 18.00-19.00 50° 1 13-3 6-3 18.00-19.00 5i° 5 Remained on nest 18.00-19.00 5°° 2 Remained on nest 19.00-20.00 50° I 13.6 5-7 20.00-21.00 53° 0 12.4 3.0 the other races of this species. The male patrolled the territory, sang and occasionally visited the area of the nest which, however, he never entered. The longest brooding session I observed was 25 minutes, the shortest six minutes. The extremes of foraging times were twelve minutes and one minute. At the time of the last observation in table 3 at least one, but not all, of the eggs had hatched. At this nest the parent continued to make sorties, returning from the last one at 21.27 hours—only about eight minutes before local sunset. On this return she appeared to have some difficulty in finding the entrance to the nest in the fading light inside the cleit. I do not think she was bringing food. These observations may not be typical because they occurred close to hatching, but they do suggest that incubating St. Kilda Wrens sometimes have a longer active day than other races (see Armstrong 195 5). This could be related to a poorer food supply.

Nestlings Except in the case of the Tobar Childa nest all the eggs of a clutch hatched within a period of twenty-four hours. All egg-shells were removed and carried well away from the nest within the following twenty-four hours. The visits of the male became more frequent about the time of hatching, and from the first few days he took a roughly equal share in the bringing of food at all three nests watched. This was so even at the Tobar Childa site where only one nestling was fledged. During the first few days brooding continued both by day and by night. On one occasion both adults were together in the nest for seven minutes; the young were then three days old. 60 OBSERVATIONS ON THE ST. KILDA WREN Observations on the feeding of nestling St. Kilda Wrens have been made by other observers. Harrisson and Buchan (1936) recorded 10 to 13 visits per hour with both parents active. Bagenal (1958) observed a much higher rate with 20 to 30 feeds per hour for the greater part of the day. Bagenal remarked that his results were more comparable to those obtained for other races and particularly the mainland form. My own observations at the nest in territory no. 4, though covering only three hours, are given in detail in table 4 as they support the findings of Harrisson and Buchan both in the rate of feed­ ing and in the periods of 'rest' away from the nest which Bagenal found extremely rare. Such intermissions are not prominent in mainland Wrens feeding young, but have been observed in the Hebridean Wren T. t. hebridensis(Armstrong 1955) and the Shetland Wren (Yeates 1948).

Table 4. Number of visits to nest in territory no. 4 of St. Kilda Wrens Troglodytes troglodytes hirtensis on 9th June 1962 when the six young were two days old Visits; per hour Longest interval Times both Time (GMT) 6" 9 6*+? 0" adults together

13.00-14.00 7 5 12 13 mins 14 mins 2 14.00-15.00 3 4 7 27 mins 11 mins 3 18.00-19.00 6 4 10 26 mins 8 mins 3

I noticed that visits became more frequent as the young grew older, though I have no hourly figures for comparison. It may be relevant that Bagenal's twenty-five hours of observation were all within six days of the young leaving the nest. The observations in table 4 and those of Harrisson and Buchan concerned broods where the young were rather smaller. Whitehouse and Armstrong (1953) have shown that the number of visits per day greatly increases during the fledging period in the case of the mainland Wren; from the above data this seems also to apply to the St. Kilda Wren. Faecal sacs were removed regularly by both sexes and carried for distances of up to 100 feet from the nest where they were deposited on selected sites which used to contain a score or more of the droppings. The pair in territory no. 5 used a metal pipe and a wooden post as depositing points. Young St. Kilda Wrens are quiet for at least the first two days. The eyes of one were open on the fifth day, when there was consider­ able growth of the body and wing feathers. Armstrong (1955) has recorded that the. eyes of T. A troglodytes begin to open about the seventh day, but earlier in other species of wrens. Even when their eyes are open, nestling St. Kilda Wrens still react vigorously to audi- 61 BRITISH BIRDS tory stimuli. Adults returning with food approach through the chinks in the lower part of the cleit wall. Once inside they gain height with two or three diagonal flights across the cleit, at times almost hovering while flying in this confined space. The young in the nest must hear the loud whirring of the parent's wings because they begin calling and begging well before the adult arrives at the nest entrance.

Young out of the nest The six young Wrens in territory no. 4 left the nest during the after­ noon of 24th June 1962. As I entered the cleit at 13.00 hours one of them was perched at the entrance of the nest and immediately flew out past me. The others were calling from within the nest and after watching for a few minutes I left the cleit. When I returned an hour and a half later all had flown. During the remainder of the afternoon and evening these young Wrens flew distances of up to a hundred feet at a time. They tended to fly higher than the adults and would perch on more prominent objects in the open, often on top of the turf roof of a cleit. The wind was north-westerly with occasional gusts reaching about 25 m.p.h. To judge from their behaviour a gale at this time might cause a high mortality among newly fledged Wrens. The young uttered a high-pitched squeaking note and tended to keep together. They showed no fear of man and repeatedly perched on human heads and shoulders despite anxiety notes from both parents. Five of these six from territory no. 4 were caught by hand between 15.00 hours and 19.00 hours on the day they left the nest. They weighed 9.5, 10.o, 10.0, 10.5 and 11.0 gm. (average 10.2 gm.). None showed any traces of down. The weight on leaving the nest is probably very variable. The only one of the brood from territory no. 3 to be caught weighed 12.5 gm. on the evening of the day it left the nest. The fledging period had been one day shorter and this bird still showed traces of down on the nape. The single youngster raised from the Tobar Childa territory weighed 14.5 gm. three days before leaving the nest. In contrast, the adults may lose weight during this busy season. The female from territory no. 3 was caught in a mist- nest on nth June 1962 and again on the 26th and, during these eight days of incubation and seven of brooding and feeding the young, she had decreased in weight from 14.5 gm. to 12.5 gm. (both weighings being carried out between 19.00 and 20.00 hours). Towards dusk the young Wrens huddled together in sheltered spots and later roosted in a cleit. The following day they were much shyer and could not be caught by hand. They kept in one or two parties and were fed by both parents. There were still three young together three weeks later and during this time they had often wandered outside the territory boundaries. 62 OBSERVATIONS ON THE ST. KILDA WREN Discussion The incubation period of the St. Kilda Wren at 19 to 20 days is three or four days longer than that of the mainland Wren in Britain (Arm­ strong 1955), and five or six days longer than that of the Shetland Wren (Armstrong 1952). Bagenal (1958) has suggested that a longer incubation period in the St. Kilda Wren may be due to its larger egg. However, the egg of the Shetland Wren is only very slightly smaller than that of the St. Kilda Wren, Witherby et al. (1938) giving the average measurements as 18.58 x 13.58 mm. and 18.55 x 13.93 mm. respectively. Yet the Shetland race appears to have an even shorter incubation period than the mainland Wren whose eggs average only 16.66 x 12.77 mm. From my observations on the nest in territory no, 4 the completed clutch may not be brooded for three days. If this is usual, it would account for the extended incubation period. The fledging period of 16 or 17 days is one or two days longer than the average of 14.9 days for the mainland race in Britain (Armstrong 1955); the only comparable figures for the Shetland Wren (16 days) and the Hebridean Wren (20 days) are based on single nests in each case (Armstrong 1955). The St. Kilda observations strengthen Armstrong's suggestion that longer fledging periods occur in the Wrens of northerly regions. This could be related to a poorer food supply in the 'bleak' habitats, but it is interesting to note that the fledging period of the single youngster from the Tobar Childa territory was not appreciably shorter than those of the other broods observed. Perhaps this difference in fledging periods, originally determined environmentally by the poorer food supply, has become genetically fixed.

ACKNOWLEDGEMENTS During my stay on St. Kilda I received help in various ways from the Nature Conservancy and, in particular, from Dr. J. Morton Boyd. I am grateful to Kenneth Williamson and my father, Dr. E. T. Waters, for reading this paper in manuscript and offering many helpful sug­ gestions, and to Robert Gillmor for preparing figs. 1, 3 and 4 for publication.

SUMMARY A dawn census of singing males showed the population of the St. Kilda Wren Troglodytes troglodytes birtensis on Hirta to be 92 pairs in the spring of 1962 compared with 116 pairs in the spring of 1957. The territories of the Wrens in the Village area were determined and the results compared with previous observations; terri­ torial activity during the winter was also studied. Large tracts of neutral ground in the Village area were not visited by Wrens until after the young had fledged. Observations on song are summarised; the seasonal incidence was found to be similar to that of the Shetland Wren T. t. ^etlandicus. The Village Wrens were single-brooded in both 1961 and 1962; the spread of

63 BRITISH BIRDS nesting in the Village area was four weeks in 1962. The incubation period was 20 days at one nest and at least 19 at another; in the latter case brooding was irregular for at least the first three days. The fledging period at three nests was 16 or 17 days. Observations are given on incubation rhythm, on the number of visits to feed nestlings, and on the behaviour and weights of young leaving the nest.

REFERENCES ARMSTRONG, E. A. (1952): 'The behaviour and breeding biology of the Shetland Wren'. Ibis, 94: 220-242. CI95 3): 'The history, behavior, and breeding biology of the St. Kilda Wren'. Auk, 70: 127-150. :—(1955): The Wren. London. (1959): 'The behaviour and breeding environment of the St. Kilda Wren'. Brit. Birds, 52: 136-138. BAGENAL, T. B. (1958): 'The feeding of nestling St. Kilda Wrens'. Bird Study, 5: 83-87. GREEN, F. H. W. (1959): 'The weather of St. Kilda, winter 1957-1958'. Weather, 14: 19-21. HARRISSON, T. H., and BUCHAN, J. N. S. (1934): 'A field study of the St. Kilda Wren, Troglodytes troglodytes hirtensis, with especial reference to its numbers, territory and food habits'. /. Anim. EeoL, 3: 133-145. ' ("93Q: 'Further notes on a field study of the St. Kilda Wren (troglodytes, troglodytes hirtensis Seeb.), with especial reference to its nest habits and song'. Scot. Nat., 1936: 9-21. MARTIN, M. (1698): A Late Voyage to St. Kilda, etc. London. VENABLES, L. S. V., and VENABLES, U. M. (1955): Birds and of Shetland. Edinburgh and London. WATERS, W. E. (1962): 'The birds of St. Kilda—winter 1961-62'. Scot. Birds, 2: 227-233. WHITEHOUSE, H. L. K., and ARMSTRONG, E. A. (1953): 'Rhythms in the breeding behaviour of the European Wren'. Behaviour, 5: 261-288. WILLIAMSON, K. (1958): 'Population and breeding environment of the St. Kilda and Wrens'. Brit. Birds, 51: 369-393. (1959): 'The behaviour and breeding environment of the St. Kilda Wren'. Brit. Birds, 52: 138-140. and BOYD, J. M. (i960): St. Kilda Summer. London. WITHERBY, H. F., et al. (1938): The Handbook of British Birds. London, vol. 2. YEATES, G. K. (1948): Bird Haunts in Northern Britain. London.

64 PLATE 8. St. Kilda Wrens Troglodytes t. hirtensis and their habitat. Above, part of territories nos. 3, 4 and 5 in Village Glen (page 50), showing the cleits (dry-stone storage chambers) in which they nest. Below, note the pronounced bars on the flanks and the large-entranced nest (pages 49-64) {photos: Estlin Waters)