CENSUSING HOLE-NESTING BY VISUAL COUNTS

By DAVID CAIRNS

INTRODUCTION Recentattempts to censusbreeding seabirdsover large geographical areas, such as eastern Canada (Brown et el., 1975) and Britain and Ireland (Cramp et el., 1974), have focusedattention on the problem of obtaining maximum censusaccuracy from surveyteams whose time and manpower are often limited. Ideally, direct nest countsshould be made to indicate best the number of breeding pairs (Nettleship, 1976), but counts of at or near colonies continue to be used because of re- strictionsof time or becausethe nestsare hidden or inaccessible(e.g., Harris, 1976). The numberof birdsat a colonydepends on daily and seasonalcycles, as well as on random and weather-induced fluctuations. Several Euro- pean workers,including D¾ck and Meltofte (1975) and Birkhead (1978), have derived correctionfactors to convertcounts of ledge-nestingCom- mon Murres (Uric aalge)made at specifictimes of the day and the season into estimatesof the breeding population. Lloyd (1975) and Harris (1976) used seriesof repeated countsof Common Murre and (Alcatorda) colonies to calculatethe expectedcount error, but were un- able to determine the number of nestersbecause of the inaccessibility of the colonies. Hole-nestingseabirds such as Black ( g•ylle), Atlan- tic Puffins (Fraterculaarcrice), and most cannot be counted on the nest, and little information is available on the relation between the number of birds found near the colonyand the true breeding popula- tion. In this paper I present data on the daily and seasonalcolony at- tendancepatterns of guillemots,puffins, and Razorbills,and from these data determine optimum count times, correction factors, and expected errors for the estimationof breeding populations from visual counts.

METHODS This study wasconducted on St. Mary's Island (50ø18'N, 59ø39'W) on the north shore of the Gulf of St. Lawrence, and at Cap d'Espoir (48ø25'N,64ø19•W) on Qu6bec'sGasp6 Peninsula. On St. Mary's,puffins and Razorbillswere counted on land, and guillemotswere counted on land and water, from a low hill overlookinga 0.6-km-wide cove. I did not attempt sea counts for puffins and Razorbillsbecause their distri- butions on the water tended to be continuous from near shore to the limit of vision,and any countswould probablyhave dependedmore on visibilityand wave height than on the actual number of birds present. Guillemotswere censusedat Cap d'Espoirfrom three cliff-top vantage points. At one of these (Cap d'Espoir South) the curvature of the cliff

Bird-Banding,Autumn 1979, 50(4): 358-364 Vol.50, No. 4 CertsusingHole-nesting Auks [359

BIRDSON WATERLAND •• 40:•0

,,,60 0 .J

co 40 >4o

u_ 20 o cr 0

• 20 20

w • o o 27i iI •0i 20i .•0i Ioi 2or 3oI • 27 I •0I 20I .•0I •0I 20I 30I ?I May June July AugustMay June July August FIGURE1. Seasonalchanges in Black ,, and Razorbillattendance at the colony. (a) guillemot numbersat St. Mary's, from countsbetween 0500 and 0700. (b) guillemotnumbers at Cap d'EspoirSouth, from countsbetween 0500 and 0700. (c) guillemotnumbers at Cap d'EspoirNorth, from countsbetween 0500 and 0700. (d) puffin numbersat St. Mary's, from countsbetween 0500 and 1300. (e) Razorbill numbersat St. Mary's, from countsbetween 0500 and 1900.

permitted countsof birds on both land and water, but at the other two only birds on the water could be seen; these latter counts have been summed as Cap d'Espoir North. The number of nests on St. Mary's was determined by thorough ground searchesrepeated every one or two weeksthroughout the breed- ing season.At Cap d'Espoir South I locatednests by observingarrivals and departuresof adults, thus graduallyassembling a completemap of nest sites. In all study areas the birds were counted hourly from 0400 to 2000 on 10 censusdays. The censusesran from 28 May to 7 August 1977, on St. Mary's,and from 27 May to 29 July 1978, at Cap d'Espoir.All times are Atlantic Standard Time.

RESULTS Thirty-one pairsof Black Guillemots,18 pairsof Atlantic Puffins,and 22 pairs of Razorbillsnested in the St. Mary's studyarea. The guillemot colonyat Cap d'Espoir South numbered 17 pairs. Numerous guillemotswere visiblein the three count areason all cen- sus days, but substantialattendance on St. Mary's was limited to the I, 360] D. Cairns -Banding Autumn 1979

BIRDSON WATER BIRDS ON LAND 40 LU • (c) • 2o

• 2o • o I I I I 20 (d) m4ol z•2o g

0 0400 0800 1200 1600 2000 20 • ,:•..•, ,(b) 0400 0800 1200 1600 2000 Tl'm o T•ne FIGURE2. Daily attendancepatterns of BlackGuillemots, Atlantic Puffins,and Razorbills. Data from 10 censusdays 28 May to 7 August for St. Mary's, and from 10 census days 27 May to 29 July for Cap d'Espoir. (a) guillemots,St. Mary's. (b) guillemots, Cap d'EspoirSouth. (c) guillemots,Cap d'EspoirNorth. (d) puffins, St. Mary's.(e) Razorbills,St. Mary's. periods of 25 June-19 July for puffins, and 11 June-19 July for Ra- zorbills(Fig. 1). Attendancefor all speciesand count areastended to be highestin the morning, although guillemotsand puffins at St. Mary's alsoshowed distinct evening peaks(Fig. 2). Guillemot attendanceat Cap d'Espoir North and South decreasedthroughout the day. Razorbill numbers on St. Mary's also decreased,but in a much lessregular fash- ion. In order to determine the time periodsduring which countswere the leastsubject to variation,I groupedhourly countsinto three-hour blocks and computed their meansand coefficientsof variation (Table 1). counts in all three study areas were least variable between 0500 and 0700. Puffin counts were most stable between 0500 and 1300, but the variabilityof Razorbillcounts fluctuated irregularly throughout the day. The correctionfactors (k) neededto estimatethe number of breeding pairs from visual countsconducted at the optimum times listed above are presentedin Table 2. In the caseof St. Mary's guillemots,the ap- plication of a correctionfactor to the mean of severalearly morning countswould yield a relativelyaccurate estimate of the breedingpop- ulation. The expectederrors of the Cap d'Espoircounts are higher than the errors from the St. Mary'scounts (71% and 82% vs. 42% for single counts), but reasonableaccuracy could still be achieved if the counts were repeated a number of times. Vol.50, •o. 4 CertsusingHole-nesting Auks [361 362] D. Cairns Bird-Banding Autumn 1979

TABLE 2. Correctionfactors and estimatedpercent errors for the predictionof numberof breeding pairs from visualcounts.

Estimated error of the mean Cot- (95%confidence limits) rection Species Time factor (k) • 1 count 5 counts l0 counts

Black Guillemot St. Mary's 0500-0700 0.67 42% 19% 13% Cap d'EspoirS. 0500-0700 0.95 71% 32% 22øI Cap d'Espoir N. 0500-0700 -- 82% 37% 26% Atlantic Puffin 0500-1300 0.66 136% 61% 43% Razorbill 0500-1900 1.53 186% 83% 59%

lk= number of breedingpairs mean visual count

The expectederrors of populationestimates of puffinsand Razorbills are much higher than thoseof guillemots:even after 10 countsthe er- ror levelsremain at 43% for puffins and 59% for Razorbills(Table 2).

DISCUSSION A strong morning attendance peak is a common feature of Black Guillemot colonies,having been reported from the Gulf of Finland (Koskimies, 1949; Bergman, 1971) and Shetland (Slater and Slater, 1972),as well as from BritishColumbia for the PigeonGuillemot (C. columba)(Drent, 1965). Guillemotnumbers were highestin the early morningat both St. Mary'sand Cap d'Espoir,but the St. Mary'sbirds alsoshowed a lesserevening attendance peak (Fig. 2). Maximumnum- berswere recorded at Cap d'Espoirat 0400 and thesecounts probably includedsome nonbreeding individuals, because the totalcount at Cap d'EspoirSouth exceeded the populationof the colonyon severalocca- sions. The differencebetween the correctionfactors for the St. Mary'sand Cap d'Espoir South colonies(0 67 vs. 0.95) suggeststhat there is no universallyvalid ratio of the number of guillemotsvisible near the col- ony to the breeding population.Visual countsmay therefore be more usefulas indicesof year-to-yearpopulation changes than as indicators of true colonysize. Count variationin British Razorbilland murre col- oniesseems to be greateramong colonies than amongyears within the samecolony (Harris, 1976),but informationon year-to-yearchanges in BlackGuillemot attendance patterns is lacking. The numbersof seabirdsat a colonycan be alteredby severalunpre- dictablephenomena, including visits of birds from other colonies,visits of nonbreedingbirds, and weatherand factors.My dataare insuf- ficient to computecompensating factors for the effectsof weatherand Vol.50, No. 4 CertsusingHole-nesting Auks [363 tide on guillemot attendance,but probably rough seassubstantially re- duce water counts becausethe birds are harder to see among waves. Tidal amplitude is weak (lessthan 1.5 m) at both St. Mary's and Cap d'Espoir,but in areaswith a large tidal range guillemot attendancevaries closelywith tide level, and the daily cycleis maskedby a tidal one (Pres- ton, 1968; Cairns, 1978). In some colonieshuman or other disturbance may discouragebirds from appearing near the colony. This may have been the caseat Cap d'Espoir South, where a large colonyof Herring (Larusargentatus) occupied the talus heap in which the guillemots nested,and the guillemotswere seldomseen on land. Puffins and Razorbills are much more variable in their attendance than are guillemots,and their counts are of little use as indicatorsof true population levels.The much greater stabilityof British Razorbill countsas reported by Lloyd (1975; estimatederror as low as 17% for single counts) was apparently due to the fact that many ledge-nesting birds were visibleon their nest sites,a situation that occursinfrequently in Gulf of St. Lawrence colonies. Fortunately, both puffins and Razorbillscan usually be censusedby direct nest counts.In those colonieswhere nest counts are precluded by inaccessibleterrain or becausethe birds nestin deep or hidden crev- ices, censusingof these specieswill continue to pose problems. Black Guillemotnests are nearly alwaystoo well concealedto be locatedduring seabirdcensuses, but, as the data in this paper suggest,properly con- ducted visualcounts can provide a useful populationindex.

SUMMARY Data from daily and seasonaloccurrences of hole-nestingBlack Guil- lemots,Atlantic Puffins, and Razorbillsare used to determine optimum count times, correctionfactors, and expectederrors for the estimation of breeding populationsfrom visual counts.The relative stabilityof early morning guillemotcounts in three studyareas suggests that such countscan provide reasonablyprecise indices of colony size if they are repeated several times. However, differences in attendance patterns amongcolonies will precludethe direct estimationof true breedingpop- ulation from visualcounts. The attendanceof puffins and Razorbillsat the colonyfluctuated widely, and visualcounts of thesespecies are of limited use as populationindicators.

ACKNOWLEDGMENTS I axn indebted to G. Foreman and L. Chislett, lightkeeperson St. Mary's Island, for their many and varied services,and to D. Berryman for field assistanceat Cap d'Espoir.I thank R. Boulangerof the Cana- dian Ministry of Transport for providing accommodationon St. Mary's, and R. Bissonof the Canadian Wildlife Servicefor logisticalhelp at Cap d'Espoir. Fieldwork on St. Mary's was supported by grants from the National ResearchCouncil of Canada to J. Badard, and by a special program of the Canadian Wildlife Service for university research.J. 364] D. Cairns Bird-Banding Autumn 1979

B6dard, W. Cairns, and D. Nettleshipimproved the manuscriptwith their comments.

LITERATURE CITED BERtMAN,G. 1971. GryllteistenCepphus grille in einem Randgebiet:Nahrung, Brutre- suitat, Tagesrhythmusund Ansiedlung. CommentationesBiol., 42: 1-26. BIRKHEAD,T. R. 1978. Attendancepatterns of guillemotsUria aalgeat breedingcolonies on Skomer Island. Ibis, 120: 219-229. BROWN, R. G. B., D. N. NETTLESHIP, P. GERMAIN, C. E. TULL, AND T. DAVIS. 1975. Atlas of Eastern Canadian . Ottawa, Canadian Wildlife Service. CAIRNS,D. 1978. Someaspects of the biologyof the BlackGuillemot (Cepphus grille) in the estuary and the Gulf of St. Lawrence. M.Sc. thesis,Universit• Laval, Quebec, Qua. C•aP, S., W. R. P. BOURNE, AND D. SAUNDERS. 1974. The Seabirds of Britain and Ireland. London, Collins. DRENT,R. H. 1965. Breedingbiology of the PigeonGuillemot, Cepphuscolumba. Ardea, 53: 99-160. D¾cI•,J., ANDH. ME•.TO•TE.1975. The GuillemotUria aalgepopulation of the Faeroes, 1972. Dansk orn. Foren. Tidsskr., 69: 55-64. HAaR•S, M.P. 1976. The seabirds of Shetland in 1974. Scot.Birds, 9: 37-68. KOSKIMIES,J. 1949. Somemethodological notes concerning the waterfowlcensus in the archipelago.Papers Game Research, Helsinki, 3, 18 p. L•.oYD,C. 1975. Timing and frequencyof censuscounts of cliff-nestingauks. Brit. Birds, 68:507-513. NETTLESHIP,D. g. 1976. Censustechniques for seabirdsof arctic and eastern Canada. CanadianWildlife Service Occ. Pap. No. 25, 31 p. PRESTON,W. C. 1968. Breeding ecology and social behavior of the Black Guillemot, cepphusgrille. Ph.D. Dissertation,University of Michigan. S•.ATER,P. J. B., ANDE. P. SLA•rER.1972. Behaviourof the tystieduring feeding of the young. Bird Study,19: 105-113. D•partementde biologie, Universit• Laval, Quebec, Quebec, Canada G1K 7P4. (Presentaddress: Research Unit, CanadianWildlife Service, Bedford Instituteof Oceanography,P.O. Box 1006, Dartmouth,Nova Scotia,Canada B2Y 4A2.). Received14 October 1978, accepted29 May 1979.

Recentissues of Bird-Bandingand especiallyVolume 50 havebenefitted from the expert adviceof numerousreferees who have freely donatedtheir time and energies.The Editor acknowledgeswith much gratitude the counselfrom John C. Aldrich, Dean Amadon, Keith A. Arnold, Oliver L. Austin, Jr., W. Wilson Baker, Keith L. Bildstein,Hans Blok- poel, Jerram Brown, Joanna Burger, Roger Clapp, Mary H. Clench, Alexander Cruz, Ralph W. Dexter, William Drury, R. Michael Erwin, J. Bruce Falls, Warren Flock, Sidney A. Gauthreaux,William G. George,Thomas Grubb, J. w. Hardy, Brian A. Harrington, Helen Hays, Carl Helms, CharlesJ. Henny, RichardHolmes, Henry S. Horn, CharlesE. Huntington,Jerome A. Jackson,Douglas Johnson, Ned K. Johnson,Richard Kaminsky, John H. Kennard, Kathy Klimkiewicz,James Kushlan,Charles F. Leck, J. P. Ludwig, Wallace MacBriar, Richard Marsh, Martin McNicholl, Brooke Meanley, L. Richard Me- waldt, Douglas Mock, Helmut Mueller, Bertram G. Murray, Stephen A. Nesbitt, Ray- mond J. O'Connor, John C. Ogden, Janet c. Ollason, Frank A. Pitelka, Sergej Post- upalsky, Eloise F. Potter, Frank W. Preston,C. John Ralph, Gilbert S. Raynor, Merle E. Richmond, William Sladen, Paul A. Stewart, Robert W. Storer, Milton W. Weller, and John C. Wingfield.