Food and Feeding Ecology of Puffins

Bird Study ISSN: 0006-3657 (Print) 1944-6705 (Online) Journal homepage: https://www.tandfonline.com/loi/tbis20 Food and Feeding Ecology of Puffins Peter Corkhill To cite this article: Peter Corkhill (1973) Food and Feeding Ecology of Puffins, Bird Study, 20:3, 207-220, DOI: 10.1080/00063657309476382 To link to this article: https://doi.org/10.1080/00063657309476382 Published online: 23 Jun 2009. Submit your article to this journal Article views: 714 Citing articles: 21 View citing articles Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalinformation?journalcode=tbis20

Food and Feeding Ecology of Puffins by Peter Corkhill This study shows that although annual fluctuations occur in the main prey types of Skomer Puffins, food availability does not appear to limit breeding success. Most birds found food close to the colony, showing peaks of activity early in the day and again in late afternoon; and two out of four experimental pairs were able to rear 'twins', though the growth-rates of these were less than for single chicks. Some interesting observations are given on the feeding of the chick and on kleptoparasitism of adults by Jackdaws and gulls. WITH THE EXCEPTION of general observations by Lockley (1934, 1953), Harris (1970) and Pearson (1968), no detailed study of the diet of Puffins Fratercula arctica in British waters exists. Myrberget (1962) gives detailed information on the food and feeding habits of Puffins in northern Norway, and a recent study by Nettleship (1972) provides information from the east coast of Newfoundland. This paper reports on a study made at Skomer, a rocky cliff-bound island of 292 ha. (722 acres) situated off the southwest tip of Wales. Although quantitative data are lacking, there is evidence in the form of old accounts and photographs that the Puffin population has declined appreciably in the last century. This work was undertaken as part of a broader study to investigate reasons for the decline, and discover if possible the main factors limiting Puffin populations in the Irish Sea. Observations covered the breeding seasons of 1969 and 1970, with some preliminary work in 1968. METHODS Food samples were collected from adult Puffins returning to feed their chicks, either by startling the adults as they came to land and causing them to drop their fish, or by catching in mist-nets adults carrying fish. The latter method proved the more satisfactory. After each sample was obtained the surrounding area was thoroughly searched to ensure that no fish were overlooked in the vegetation. Puffins caught carrying fish were ringed to ensure that the same individuals were not being repeatedly recaptured; and bill measurements were taken to evaluate the role of the sexes in feeding the young (Corkhill 1972). Each food load was examined to determine the prey species, total weight, and the length of each undamaged food item (measured from tip of snout to base of tail). Samples 207

BIRD STUDY were collected systematically throughout the chick-rearing period (8 June to 25 July). Puffins were seen carrying fish before and after these dates in both seasons, but the numbers involved were too few to make the collection of food samples practical. Most of the samples were collected during the late afternoon fishing period. There was no evidence that the prey species composition varied throughout the day. RESULTS The Food The two most important fish species brought to the colony (based on frequency of occurrence) were sand-eels Ammodytes marinus and sprats Clupea sprattus; together these comprised 95.6% of the total of 1,387 individual food items examined (Table I). Small numbers of the greater sand-eel A. lanceolatus also occurred and there was a species of Ammodytes which could not be positively identified due to the immaturity of the specimens. Very small herrings Clupea harengus made 3.26% of the total and usually appeared at the beginning of the season when other food was perhaps less plentiful. TABLE L SPECIES OF FISH BROUGHT TO PUFFIN CHICKS ON SKOMER ISLAND DURING 1969 and 1970 Species Number % of Total Ammodytes marinus 744 53.6 Ammodytes lanceolatus 3 0.2 Ammodytes sp. 10 0.7 Clupea sprattus 582 42.0 Clu pea harengus 45 3.3 Gadus spp. 3 0.2 TABLE II. RELATIVE IMPORTANCE OF THE MAJOR FOOD GROUPS IN TWO SEASONS, DIVIDED (FOR CONVENIENCE) INTO FOUR FEEDING PERIODS A (10/6-20/6) B (21/6-1/7) C (2/7-12/7) D (13/7-23/7) Total 1969 Ammodytes 111 (82.8%) 83 (38.8%) 10 (4.6%) 3 (1.7%) 207 (28%) Clupea 21 (15.7%) 131 (61.2%) 204 (94.8%) 173 (98.3%) 529 (71.5%) Others 2 (1.5%) 1 (0.6%) 3 (0.5%) Total 134 214 215 176 739 1970 Ammodytes 142 (92.7%) 98 (85.9%) 179 (87.3%) 132 (74.6%) 551 (84.9%) Clupea 10 (7.3%) 16 (14.1%) 26 (12.7%) 45 (25.4%) 97 (15.1%) Total 152 114 205 117 648 Table II compares the relative abundance of sand-eels and sprats brought to the colony in 1969 and 1970. (For convenience the chickrearing period in both seasons was divided into four eleven-day subsections.) During 1969 sprats were the more important and comprised 71.5% of the total (based on frequency of occurrence), while in 1970 sandeels predominated and accounted for almost 85%. In both seasons sandeels were more important at the beginning of the chick-rearing period, after which in 1969 there was an almost total change to sprats. During 208

FOOD AND FEEDING ECOLOGY OF PUFFINS feeding period D-1970 sprats again increased in importance and, after the collection of food samples had ceased, observations showed that the remaining Puffin chicks were fed exclusively on them. It is not clear whether sprats were less abundant until later in the season in 1970, or if sand-eels were preferred. In both seasons the total mean size of Ammodytes prey items was similar (1969, mean 53.1, range 33-65 mm.; 1970, mean 57.2, range 33-67 mm.) although during 1970 sprats tended to be slightly larger (1969, mean 37.4, range 24-43 mm.; 1970, mean 46.0, range 26-65 mm.). No tendency could be detected for Puffins to select larger fish for older chicks; in fact, the mean length of all prey species combined declined as the season progressed, due to the increasing frequency of sprats. Both sand-eels and sprats were smaller than those listed by Harris (1970) for Puffins breeding in Pembrokeshire, the Fame Islands, and Norway. 10-1969 1970 8-7- 6-5 A Figure 1. Mean load weights in four feeding periods showing standard errors. Weight of Loads Throughout the chick-rearing periods 137 loads of fish were collected from Puffins returning to the colony (75 in 1970 and 62 in 1969). Figure 1 compares the mean weight of loads in 1969 and 1970, with each season divided into four eleven-day periods as in Table II. In 1969 a trend 209

BIRD STUDY emerged for loads to increase in weight as the season progressed. In 1970, however, this trend was not apparent because Puffins were catching heavy loads very early in the season, possibly because food was more abundant than usual at that time of year. Loads were significantly heavier during 1970 (mean load weight 8.42±0.33 gm. S.E.) than in 1969 (6.30±0.28 gm. S.E.). In feeding period B-1970 a sharp drop occurred in the weight of the loads collected. In attempting to explain this apparent decline in the fishing efficiency of the Puffins, it is perhaps important that almost 75% of the samples obtained in this period were collected on two days when wind speeds approached gale force. 16-14- 12-10- 8-6 A Figure 2. Number of fish per load, showing mean load size in four feeding periods, and standard errors. Number of Fish per Load Figure 2 compares the average number of fish per load in the two seasons, divided into four feeding periods as in Figure 1. A basic difference between the seasons emerges, for during 1970 the average number of fish per load varied little throughout, whereas in 1969 the number of fish per load increased gradually from 7.8±0.8 S.E. to 14.7±0.9 S.E. Loads contained significantly fewer fish during 1970 with a seasonal average of 8.3±0.4 S.E. compared with 11.9±0.6 S.E. fish per load in 1969. The smallest load encountered in the two seasons contained two fish and the largest 30 fish : the latter comprised very small sprats and herrings with a total weight of only 3.55 gm. 210

FOOD AND FEEDING ECOLOGY OF PUFFINS Feeding Rhythms In a general examination of the Puffins' feeding habits Lockley (1934) states : 'In the first few weeks the young bird is fed with small sand-eels and the minute freshly hatched fry of fishes which swim close inshore in June and July. These providential shoals are brought up against the island shores with every flood tide, and it is noticeable that the adult Puffins take advantage of this, fishing over the flood and resting over the ebb.' Lockley examined the daily rhythm of Puffins bringing fish to an estimated 530 burrows on Skokholm, and found that most Puffins fed their young early in the morning and again in the afternoon. He did not attempt to relate this to tidal rise and fall. Observations by other workers (Myrberget 1962, Roberts et al. 1963) tend to confirm this general bimodal feeding pattern, but have not taken into consideration possible tidal effects. The feeding pattern described by Roberts et al. indicates a fishing peak at 0900 hours on 23 July 1963, which is considerably later in the morning than peaks demonstrated in earlier examples; however, no observations were available before 0600 hours. In order to examine the possibility of a tidal effect on the daily feeding rhythm, a study area containing an estimated 85 Puffin chicks was kept under continuous observation from 0330 to 2200 hours on 25 June 1969, when most Puffins were feeding young. To avoid a chance relationship between peak feeding time and high water, observations were repeated on 20 July when the tidal rhythm was reversed. The rhythm of feeding intensity on both occasions (Figures 3 and 4) agreed with that found by Lockley et al. (1947) and Myrberget (1962). A sharp peak of activity occurred early in the morning, followed by a lesser and more protracted peak in the late afternoon. Figure 3 shows that the afternoon fishing peak of 25 June roughly coincided with high water; that this is a chance effect is clear when the situation in Figure 4 is taken into consideration, since the afternoon fishing peak then coincided with low water. On neither occasion did the more intense early morning fishing peak coincide with high water, which suggests that most of the birds engage in fishing as soon as it is light. It is interesting to note the similarity of fishing rhythms between the Skomer Puffins and those examined by Myrberget (1962) at latitude 62 20'N. in northern Norway where, close to the Arctic Circle, more hours of daylight when food could be obtained are available during the summer months. Number of Feeds per Chick In 1953 Lockley estimated that on average Puffin chicks were fed less than twice a day, using data from the total number of feeds delivered during one day to a study area containing an estimated population of 534 pairs. Myrberget (1962), from direct observations on 40 burrows, found Puffin chicks were fed between one and five times a day, 2.5 times on average. On 26 June 1969 I maintained a continuous watch on ten marked burrows, known to contain single chicks, throughout daylight from 0330 to 2200 hours, noting the time of each food delivery. On 20 July observations 211

BIRD STUDY were repeated, but two extra burrows were included in the sample. Table III shows that the average number of feeds delivered to each chick was 5.3 on 26 June and 8.3 on 20 July, with an overall average of 6.9 feeds per chick per day. cc 75 ± *4 ul.7:( 50., x x....hw co,, I u_ N. O 25 / I ui.- < o- J 90 LW 06.00 ogoo 1200. 15.00 18 00 21.00 TIME (HOURS) I I 0 Figure 3. Rhythm of fish landings and relation to tide (dotted line) on 25 June 1969. 75 I 50 LL u_ O 25 u) 9 0 06.00 0900 12.00 15.00 18, 00 21.00 TIME (HOURS) Figure 4. Rhythm of fish landings and relation to tide (dotted line) on 20 July 1969. TABLE III. NUMBER OF FEEDS PER DAY RECEIVED BY 10 YOUNG PUFFINS ON 26 JUNE AND 12 YOUNG PUFFINS ON 20 JULY. (EACH BURROW CONTAINED ONE CHICK.) Date Number of Mean meals! Std. dev. Range chicks day 26 June 1969 10 5.3 2.05 2-8 20 July 1969 12 8.3 2.17 4-12 212

FOOD AND FEEDING ECOLOGY OF PUFFINS The Role of the Sexes Although it has long been known that the Puffin chick is fed by both parents (Lockley 1934) there has been no detailed attempt to evaluate the share of the sexes during the feeding period, largely because of the difficulty of sexing Puffins at this time of year. Information on this was collected in two ways, by direct observation of colour-ringed birds of known sex, and by sexing a sample of Puffins caught when carrying fish, using the method of bill measurement described by Corkhill (1972). During watches on 26 June and 20 July 52 feeds were recorded by colour-ringed birds of known sex. Four pairs were involved and on 58% of the occasions the female was responsible. From a total of 53 Puffins captured when carrying fish it was possible to sex 49 according to the size of their bills (the other four birds having bills of intermediate proportions); of this sample 28 (61%) were females. The results confirm that both sexes are actively involved in feeding the chick. Feeding Behaviour Because of the difficulty of observing events which normally take place underground in the nest chamber, the method used by Puffins to feed small chicks has not previously been described. Myrberget (1962) noted that small heaps of fishes are often found on the floor of the nest chamber, and that young Puffins peck at small objects which are lighter in colour than the floor of the chamber. He further stated that the brightly coloured bills of the parents aroused no curiosity in newly hatched chicks, and he described a food-begging posture. After the 1969 breeding season I constructed an underground hide with a glass panel next to a nest chamber where a pair of Puffins had successfully reared their chick. The pair returned in 1970 and I was able to observe a chick being fed on three occasions before it was four days old. The adult returned to the nest carrying fish and stood in the centre of the chamber with the fish dangling from its bill, uttering a soft repeated clicking call which I had not previously heard. The chick came out of the darkness at the back of the chamber and started to peck at the dangling fish, taking them directly from the parent's bill and swallowing them whole. After three or four fish had been taken in this manner the adult dropped the rest on to the nest floor, whence they were retrieved and eaten by the chick. The chick squeaked excitedly throughout but was not seen to flap its wings. On no occasion did I see the chick peck at its parent's coloured bill. When chicks are older and more agile the amount of time spent underground by parents feeding them decreases considerably, and may be as short as two or three seconds on each occasion (Myrberget 1962 and personal observation). It is probably that the more experienced chicks take all their food from the floor of the burrow, and in many cases the food is dropped just inside the entrance hole. Despite the underground nest and the large coloured bill, the method used by Puffins to feed their chicks is very similar to that employed by 213

BIRD STUDY Razorbills Alca torda. In this species also young chicks are fed initially from the parent's bill, pecking at the silvery fish; but as soon as the chick shows interest and starts to feed, the rest of the catch is laid on the ground. Guillemots Uria aalge, however, normally bring a single large fish to their young and lay the fish on the ground so that it may be picked up by the chick and swallowed head first (personal observation). Feeding Range Information on the feeding range of Skomer Puffins was obtained by timing the fishing journeys of colour-marked individuals (see Pearson 1968), and by direct observations at sea of Puffins carrying fish. In order to reduce error due to Puffins stopping at sea to rest or preen, fishing trips were only timed during the morning and late afternoon peak periods when feeding activity is concentrated. As in Pearson's study no time was allowed in the calculations for catching the fish because of the difficulty of obtaining data on this aspect. The fishing range therefore represents the maximum distance a Puffin could have travelled in the time available. TABLE IV. THEORETICAL MAXIMUM FEEDING RANGE OF SKOMER PUFFINS. (ESTIMATED FLIGHT SPEED OF PUFFINS CARRYING FISH=48 KM. OR 30 M.P.H.) 25 June 1969 20 July 1969 Total Number of Fishing Trips 20 42 26 Mean Duration (minutes) 105 78.6 85.5 Maximum distance which could have been flown from the nest (miles) 48 km. (30) 31.4 km. (19.6) 32.4 km. (20.3) In Table IV the theoretical maximum feeding range of Skomer Puffins, calculated from a total of 62 timed trips, was 36.8 km. (20.3 miles). This is considerably less than the 187 km. (85.7 miles) calculated by Pearson (1968) from a total of nine feeding flights for Puffins on the Fame Islands. From general observations made from the land and from boats travelling at known speed, the flight speed of Puffins carrying fish was estimated at 48 km. or 30 miles per hour. Even on the basis of a flight speed of 80 km. or 50 miles per hour, as postulated by Meinertzhagen (in Pearson 1968), the theoretical maximum feeding range increases only to 56 km. (35 miles). On 17 July 1970 a line transect was run from a boat travelling between Skomer and Grassholm, an island lying 13 km. to the west, and the positions of Puffins carrying fish were noted during the afternoon fishing period. The procedure was repeated on the return journey an hour later. Although single Puffins and occasionally small groups were encountered over the whole distance, an estimated 85% of all Puffins were concentrated within 3 km. of Skomer. No Puffins carrying fish were seen outside this limit, which indicates that on this occasion the Puffins were fishing very close to the colony. Further support is provided by the fact that the samples collected from Puffins returning to the colony in 1970 often contained live prey. Harris (1970) also noted that fish presented to young auks on Skomer and Skokholm were sometimes still alive. 214

FOOD AND FEEDING ECOLOGY OF PUFFINS KLEPTOPARASITISM On Skomer adult Puffins carrying fish for their young are subject to attacks from Herring Gulls Larus argentatus, Lesser Black-backed Gulls L. fuscus and Jackdaws Corvus monedula which attempt to steal their fish. During continuous watches on 25 June and 20 July 1969, when records were kept of the total number of food deliveries to a study area containing an estimated 85 Puffin chicks, the number of Puffins attacked by predators was also recorded together with data on the species and success of the predator involved. TABLE V. KLEPTOPARASITISM RECORDED FROM 0430 TO 2200 HOURS (BST) ON A STUDY AREA CONTAINING AN ESTIMATED 85 PUFFIN CHICKS, 12 PAIRS OF JACKDAWS AND PART OF THE FEEDING TERRITORIES OF TWO HERRING AND ONE LESSER BLACK-BACKED GULLS 25 June 1969 20 July 1969 Total Total number of food loads brought to colony 584 709 1,293 % attacked by Gulls 1.16 1.12 1.16 % of successful Gull attacks 28.5 25.0 26.6 % attacked by Jackdaws 15.24 nil 6.88 % of successful Jackdaw attacks 6.7 nil 6.7 % of total number of food loads lost to predators of both species 0.86 0.14 0.46 Table V shows that on 25 June 15.24% of all Puffins carrying fish were attacked by Jackdaws, and a further 1.16% were attacked by both species of gull. On 20 July, although the level of gull attack had not changed, no incidents involving Jackdaws were noticed. On 25 June it was estimated that 12 pairs of Jackdaws were breeding in the study area and were at that time engaged in feeding large young. In the interval preceding the second watch newly fledged Jackdaws were seen, but by 20 July all juveniles and the majority of adults had left the island. The low level of gull attacks is thought to be unrepresentative of the rest of the Skomer colony, because only a small section of the feeding territories of the gulls overlapped the study area. This was situated within 50 m. of human habitation, which probably inhibited gull activity. The success rate of the Jackdaws in obtaining fish was relatively low, being only 6.7% of the total number of attacks made. This contrasts sharply with the mean success rate for both gull species, which was 26.6%. This disparity in the success rates on 25 June may be partly due to the difference in body size between Jackdaws and gulls, but the hunting methods employed were also distinct. All the observed gull attacks commenced from an airborne patrol immediately above the main concentration of Puffin burrows, the gull swooping to intercept an incoming Puffin!n. the air. Puffins robbed by this method either dropped their fish at the initial encounter or after a short chase down to the sea. Jackdaw attacks were made by a bird standing in the colony, often partially hidden in the vegetation, attempting to intercept an incoming Puffin as it landed close by. Puffins did not usually drop their fish unless the Jackdaw managed to make 215

BIRD STUDY physical contact, hanging on to the tail or wing. Puffins often escaped from Jackdaw encounters either by bouncing back into the air and flying away, or by rushing into their burrows, where the Jackdaws would not follow. On other occasions gulls have also been observed hunting from a station on the colony, the chosen method of attack perhaps depending on wind speed and direction. ATTACKED BY JACKDAWS TOTAL DELIVERIES Figure 5. Frequency of kleptoparasitism by Jackdaws on 25 June 1969. The histogram in Figure 5 demonstrates the daily rhythm of Puffin food deliveries and the relative pressure throughout the day. The hatched areas represent the percentages of incoming Puffins that were attacked, and changes in the predation pressure are apparent. It is highly advantageous for a Puffin to provision its chick during the early morning feeding peak. Jackdaws were most active between 0700 and 1100 hours when up to 57% of Puffins carrying fish were attacked. The low level of Jackdaw activity before 0600 hours suggests that the birds were then exploiting another food source away from the colony. Considering the food requirements of Jackdaw chicks and the low success rate of the Jackdaws robbing Puffins, it is probable that stolen fish does not constitute a major part of their diet. 216

FOOD AND FEEDING ECOLOGY OF PUFFINS The exceptions to this general conclusion are certain individual Jackdaws whose expertise in obtaining fish is considerably above average. One such bird, recognised on the study area on 25 June from its regular hunting station outside its breeding burrow was responsible for 60% of the successful attacks by this species. Despite the small number of fish loads actually lost to Jackdaws (0.85% of the total loads delivered), it is thought that Jackdaw attack may influence the provisioning performance of certain Puffins nesting away from the edge on the flat cliff top. These birds are more at risk to Jackdaw attack as they usually alight first on the cliff edge and run the remaining distance to the burrow entrance. On Skomer Great Black-backed Gulls Larus marinus, Carrion Crows Corvus corone and Razorbills have also been observed robbing Puffins, but no predation by these species occurred during watches on the study area. REARING TWINS To investigate the Puffins' ability to rear more than one chick, four pairs were each provided with an extra chick in 1970. In each case very young chicks, chosen to match the weight of the original as closely as possible, were introduced. Two pairs successfully reared twins, but the other two reared only one chick each, losing the second early in the experiment. One chick lost weight rapidly and died a few days after its introduction, while the other introduced chick disappeared from the burrow from an unknown cause. In the case of both successful pairs, one chick gained a weight advantage over its sibling quite early in the experiment, and this difference slowly increased with time. In one set of twins the original chick was the heavier at fledging, while in the second set the introduced chick had the weight advantage. Figure 6 compares the growth curves of the two sets of successful twins with a normal growth curve calculated from almost daily weights of ten single chicks in the same study area. After the eleventh day of life the twins were significantly lighter than single chicks and their growth curves were generally depressed from then onwards. The mean weight of the twins at fledging was 267.2±17.7 gm. S.D. compared with a mean weight of 284±21.5 gm. S.D. for single chicks. However, the twins took longer to reach their fledging weight; the mean length of their fledging period was 45±1.0 days S.D. compared to 37.3-1.9 days S.D. for single chicks. In both cases the lighter member had a fledging period two days longer than its sibling. DISCUSSION Differences between the two breeding seasons under consideration in respect of the weight of fish loads, the number of fish they contained, and the frequency of the main prey species, suggest that Puffins have considerable flexibility in their hunting methods. They are therefore able to adapt to changes in the availability and distribution of their food supply. 217

BIRD STUDY 218

FOOD AND FEEDING ECOLOGY OF PUFFINS Figures 1 and 2 demonstrate a basic difference in the load sizes between the two seasons. During 1969 loads were significantly lighter than in 1970 but contained a greater number of food items due to a much higher proportion of the relatively larger sand-eels in 1970. The tendency for sandeels, although the more important prey species at the beginning of the season, to be later replaced by sprats, is thought to reflect a change in the availability of fish around Skomer. Myrberget (1962) also found that in northern Norway sand-eels and Pollachius were more frequently caught early in the season, whereas clupeids dominated later in the summer. It is possible that in some seasons an unusual abundance of fish fry is available and that the Skomer Puffins then select large clupeids as noted by Harris (1970), thus increasing their fishing efficiency. Comparison with data in Nettleship (1972) reveals that Puffins on Great Island, Newfoundland, brought heavier loads containing fewer and larger fish than did those on Skomer, and that individual chicks were fed fewer times per day. Nettleship (pers. comm.) also thought that Puffins on Great Island were obtaining fish at a greater distance from the colony than the range suggested by the data for Skomer Puffins. Lockley (1953) records that the fish food of Puffins at Skomer and Skokholm is largely small fry of Gadus and Clu pea with some sand-eels; however, Gadus species were only very rarely taken by Puffins in 1969 and 1970. Pollack Gadus pollachius and Coal Fish G. virens are both very common around Skomer (personal observation). It is possible that these species are less available to Puffins because of their more benthic habits and therefore remain unexploited except in times of food shortage. Razorbills and Guillemots also breed in large numbers on Skomer. Harris (1970) suggests that competition for food between the different auk species is avoided at times of food shortage by their preference for different prey sizes. During feeding period B-1970, 20 food items examined from Razorbills on Skomer comprised 16 Ammodytes marinus (mean length 64.2 mm.) and four Clupea sprattus (mean length 54.2 mm.). In the case of both prey species, Razorbills caught fish that were slightly larger than those caught by Puffins during the same period. The sample size of Razorbill food items is unfortunately too small for a more detailed comparison. However, it is not clear why, if larger fish were available to Razorbills, the Puffins were catching even smaller fish than those examined at other Puffin colonies (e.g. Pearson 1968, Harris 1970, Myrberget 1962), unless there was a much greater availability of smaller fishes. Twinning experiments of two types undertaken by Nettleship (1972) on Great Island showed conclusively that Puffins were unable to rear more than one chick at this colony. That two pairs of Puffins did succeed on Skomer indicates a greater availability of food in 1970. It is thought that the weight difference between siblings was exaggerated by aggressive behaviour between the chicks, as the lightest and weakest individual was constantly discovered in a less favourable position in the burrow. Plumb (1965) also noticed aggressive behaviour between Razorbill twins on Skokholm. 219

BIRD STUDY It is concluded that because of the success of the twinning experiment and the apparently contracted fishing range of Skomer Puffins, the availability of natural food to the adults was not a limiting factor in this colony in 1970. ACKNOWLEDGEMENTS I should like to thank the West Wales Naturalists' Trust for allowing me to carry out this work while employed as warden on Skomer. I am grateful to Dr C. M. Perrins, Dr D. N. Nettleship, J. L. F. Parslow and J. W. F. Davis for criticism and advice; to C. T. Macer for identifying fish samples', and Dr D. L. Urry for preparing the figures. SUMMARY The food brought to Puffin chicks on Skomer Island, Pembrokeshire, was examined in 1969 and 1970. The most important food species were sand-eels Ammodytes marinus and sprats Clupea sprattus. The weight of fish loads and the number and species of the fish they contained were found to vary in the two seasons, probably reflecting a change in the species available to Puffins. Daily feeding rhythms are described, most fish being brought to the colony in the early morning and late afternoon. No connection was found with tidal effects. Puffin chicks were fed on average 6.9 times per day, more frequently than in other studies elsewhere, and both sexes were active in feeding the young. The feeding behaviour of adult Puffins with very small young was watched from a specially constructed hide. Skomer Puffins were found to feed quite close to the colony and to have a theoretical maximum feeding range of 32 km. (20 miles). Kleptoparasitism by Jackdaws and gulls was examined in a study area estimated to contain 85 Puffin chicks. Twinning experiments in 1970 showed that some Puffins had the ability to rear more than one chick, suggesting that in this season the food supply was not critical in limiting breeding success. REFERENCES CORKHILL, P. 1972. Measurements of Puffins as criteria of sex and age, with special reference to sexual differences. Bird Study, 19: 193-201. CORKHILL, P. 1971. Factors affecting auk attendance in the pre-egg stage. Nature in Wales, 12:258-262. DICKINSON, H. 1958. Puffins and burrows. Skokholm Bird Ohs. Rep., 27-34. HARRIS, M. P. 1970. Differences in the diet of British auks. Ibis, 112:540-541. LocKLEy, R. M. 1934. On the breeding habits of the Puffin; with special reference to its incubation and fledging periods. Brit. Birds, 27:214-223. LOCKLEY, R. M. 1953. Puffins. London. MYRBERGET, S. 1962. Undersokelser over forplantningsbiologien til lunde (Fratercula arctica). Egg, raging og unger. Papers of the Norwegian State Game Research Inst., 2, no. 11. NErrLEsHip, D. N. (in press). Breeding success of the Common Puffin (Fratercula arctica L.) on different habitats at Great Island, Newfoundland. Ecological Monog. PEARSON, T. II. 1968. The feeding biology of sea-bird species breeding on the Fame Islands, Northumberland. I. Anim. Ecol., 37 : 521-552. PLUMB, NA,. 3. 1965. Observations on the breeding biology of the Razorbill. Brit. Birds, 58: 449-456. ROBERTS, C.. K. BROADHURST, and P. GARRETT. 1963. Observations on the movements of breeding Puffins. Skokholm Bird Ohs. Rep., 17-24. Tuoc, L. M. 1960. The Murres. Canadian Wildlife Series 1. Ottawa. P. Corkhill, The Nature Conservancy, 2 Dent Bank, Middleton-in-Teesdale, Co. Durham. 220