Colony Attendance and Population Monitoring of Least and Crested Auklets on St

The Condor 9297406 0 The CooperOmitholopjd Society1990

COLONY ATTENDANCE AND POPULATION MONITORING OF LEAST AND CRESTED AUKLETS ON ST. LAWRENCE ISLAND, ALASKA ’

JOHN F. PLATT, BAY D. ROBERTS, AND SCOTT A. HATCH Alaska Fish and Wildlif ResearchCenter, U.S. Fish and Wildlife Service, 1011 E. Tudor Rd., Anchorage,AK 99503

Abstract. Diurnal and seasonalpatterns of attendanceof Least Auklets (Aethia pusilla) and CrestedAuklets (A. cristatella)were studiedin 1987 at breedingcolonies on St. Lawrence Island, Alaska. Numbers of aukletsattending eight 200-m2 plots on talus slopeswere counted throughout the day by observers on 11 occasionsthrough the breeding season.Numbers attending smaller plots were recorded on time-lapse film on 7 1 different days. Another 16 200-m2 plots were censusedfor auklets using surfacecounts. Within-day patterns of attendance were extremely variable over small and large temporal scales.Peaks of attendance occurredin late morning and late evening, with a 7- to 12-hr period of absencein the middle of the day. Attendance varied markedly between days, and numbers were negatively correlated with wind speed and the magnitude of tidal oscillations.Patterns of attendancealso varied with stage of breeding, and counts were least variable during incubation and early chick rearing. Whereas Least Auklet numbers peaked during prelaying, Crested Auklet numbers peaked during incubation. Counts indicated that auklets at Kongkok Bay have increasedabout twofold since studies in the mid-1960s. Recommendations are made for future monitoring of auklet populations. Key words: Least Auklet; Aethia pusilla; CrestedAuklet; Aethia cristatella;census; population monitoring;St. LawrenceIsland; time-lapsephotography.

INTRODUCTION of breeding and nonbreeding auklets to and from Least Auklets (Aethia pus&z) and Crested Auk- breeding sites and related attendance at the sur- lets (A. cristatella) are two of the most abundant face to activities of birds below the surface of seabirds in the North Pacific, but numbers are talus slopes.Despite these advances, no consen- difficult to censusand population estimates for sus has been reached on the best methods for most coloniesare still tentative @owlset al. 1978). monitoring auklet populations. Both species lay their eggs in natural crevices We studied diurnal and seasonal attendance among the rubble of boulder fields or glacially- patterns of Least and Crestedauklets on St. Law- formed talus slopes, and attendance at colonies renceIsland in summer 1987, usingsurface counts is extremely variable. Those factors make it dif- of auklets and time-lapse photography. The pur- ficult to estimate numbers of breeding birds or pose of the study was to identify sourcesof vari- monitor population trends. ation in attendance and refine methods for mon- BPdard (1969b) was the first to develop meth- itoring populations. We compare our data with ods for censusing auklets. He counted birds at- results of previous studies to assesschanges in tending 200-m* plots on talus slopesand extrap- the populations of auklets at St. Lawrence Island. olated to estimate total population sizes of METHODS colonies on St. Lawrence Island, Alaska. How- ever, Btdard did not quantify diurnal or seasonal We worked at Kongkok Bay (63”24’N, 17 l”49’W) variations in attendance. Byrd et al. (1983) and near the southwestcape (Fig. 1); in the area de- Roby and Brink (1986) demonstrated that the scribed in detail by Btdard (1969b). We estab- numbers of Least and Crested auklets in atten- lished 16 200-m* auklet study plots on: (1) the dance on slopes varied widely both daily and inland talus slopesof Owalit Mountain to study seasonally.They also monitored the movements breeding biology (Piatt et al., in press) and attendance patterns, and (2) the northern and southern slopesof Kongkok Basin for censusing ’ ’ Received 13 March 1989. Final acceptance2 Oc- (Fig. 1). tober 1989. We counted birds sitting on the surfaceof talus

[971 98 J. F. PIATT, B. D. ROBERTS AND S. A. HATCH

For comparing our surface counts with those of BCdard and Searing, we averaged the second, T. LAWRENCE third, and fourth highestcounts of auklets at each plot to calculate peak densities. Counts were conducted between 05:00-l I:00 Bering Daylight Time (BDT). We found two of the same plots used by Searing, and located other plots in the same vicinity (Lane Iyakitan, pers. observ.). Sur- face countsat Owalit were conductedduring June (prelaying) and at Kongkok in late July (chick rearing). We corrected for seasonalvariation by counting at Owalit and Kongkok simultaneously in July, and extrapolating from the June and July counts at Owalit. We extrapolated from surface oAuklet Plots counts to total populations by using our mea- sures of average auklet densities and Bedard’s estimates of total colony areas for Owalit and 07 1 Kongkok. BERING SEA RESULTS ATTENDANCE PATTERNS Diurnal patterns of auklet attendance were typ- FIGURE 1. Studyarea and locationof auklet plots on St. LawrenceIsland. Contourintervals in feet. ically erratic at small temporal (5 min) and spatial scales(Fig. 2). Least and Crested auklet numbers habitat (hereaftercalled “surface counts”) on eight on time-lapse plots were negatively correlated on of the Owalit study plots every 0.5 hr on 11 73% of 77 count-daysobserved, and 48% of those different days. We also placed g-mm time-lapse negative correlations were significant. At larger cameras on three Owalit plots with about 1O-m2 (30 min) time scales, Least and Crested auklet areasto take time-lapse photosevery 5 min. Birds numbers were positively and significantly cor- recorded by observers or time-lapse film repre- related each day on all eight Owalit plots (Fig. sent only that portion of the population present 3) and on 94% of time-lapse count-days when on top of the talus rocks (i.e., birds in crevices, the data were averaged over 0.5-hr time inter- flying about the colony, foraging, or loitering at vals. Stronglysignificant correlations (Least: mean sea near the colony are excluded). Auklet num- r = 0.67 f 0.096 SE; Crested: r = 0.62 f 0.23) bers were later counted from the films using a between time-lapse counts averaged over 0.5 hr time-lapse projector. Complete diurnal counts and observer counts on eight Owalit plots were were obtained for 7 1 different days from 30 May observed on all days except 23 August, when few to 26 August. Attendance patterns were analyzed auklets attended the colony (Fig. 3). Time-lapse with nested ANOVA and variance component and surface counts revealed that auklet atten- proceduresin SAS (1985). We used the 5% level dance increasedrapidly and plateaued during the of statistical significance and the term “signifi- morning hours, declined slowly in early after- cant” is usedhereafter in its statistical senseonly. noon until no birds were present for 7-l 2 hr, and BCdard (1969b) censusedauklets prior to egg peaked again at lower levels late in the evening. laying by counting numbers attending a number Analysis of the components of variation in of 200-m2 plots during the early morning, se- numbers of auklets on Owalit plots indicated that lecting the secondto fourth highest counts to get within-day fluctuations accounted for most of a mean density per plot, and extrapolating from the total explained variance in auklet attendance plots to the total area of habitat used by auklets (Least: 78%, Crested: 53%). After removing to get a population estimate. He adjusted surface within-day variation from the analysis, and using counts for the total area of plots covered with only data from 06:00-09:00 BDT (period used suitable substrate,and the depth of the talus rock for censusingby Bedard 1969b), between-day and layer. Searing (1977) used exactly the same pro- between-plot variance accounted for most of the cedure as BCdard (1969b). variation (Table 1). By grouping time-lapse data AUIUET ATTENDANCE PATTERNS 99

DIURNAL ATTENDANCE PATTERNS

CRESTED AUKLET 5 8

P bl 0500 0600 0700 0600 0900 1000 1100 1200 2400 0100 0200 TIME OF DAY (21 JUNE) PIGURE 2. Diurnal attendanceof Leastand Crestedauklets at plot 3, Owalit Mountain on 21 June 1987. Countsare for a 10-m>area every 5 min from time-lapsefilm. Note breakin time between12:10-23:55, when no aukletswere present.

into hours and months, we obtained an inte- dance of both specieswas negatively correlated gratedpicture ofauklet attendance (Fig. 4) which with wind speed, which accounts for many days was more accuratethan the observer counts (Fig. of low attendance, particularly in June (Figs. 3 3) becausebetween-day variance was smoothed and 5). Attendance also was negatively correlated out by the large sample size. In most respects, with the range of daily tidal oscillations. Corre- patternsrevealed by time-lapse and surfacecounts lations improved when data were grouped over were similar. Attendance shifted from early 7-day intervals. Crested Auklet attendance was morning to afternoon hours as the breeding sea- also positively correlated with daily maximum son progressed,and Crested Auklet attendance air temperature and rainfall, although those cor- peaked in July whereas Least Auklet attendance relations were probably an artifact of analysis was similar in June and July. due to a seasonal trend of increasing air tem- Integration oftime-lapse data within days (Fig. perature and rainfall. 5) revealed that Least Auklet attendance peaked POPULATION TRENDS during prelaying whereas Crested Auklet attendance slowly built up to a peak during incuba- At Kongkok Basin, we observed higher densities tion. Both speciesstopped attending the surface of Least Auklets, but similar densities of Crested of talus slopesafter mid-August, although chicks did not fledge until late August and early Sep- TABLE 1. Componentsof variationin attendanceof tember (Piatt et al., in press). Figure 5 also il- Leastand Crestedauklets at plotson Owalit Mountain, lustratesthe magnitude of daily variation in sur- St. LawrenceIsland. NestedANOVA on countscon- ductedbetween 06:00-09:00 BDT. faceattendance. Coefficients of variation in auklet

attendance over 2-week intervals indicated that 96variance explained attendance was least variable during late incu- SOllICe Least Auklet Crested Auklet bation or early chick rearing (Table 2). Plot 11.8 22.9 The influence of environmental conditions on Periodof summer 0.0 10.2 attendance were assessedby measuring correla- Day (within period) 45.7 13.4 tions between daily mean attendancevalues from Plot*period 0.0 time-lapse plots (excluding the fledging period) Plot*day 8.2 i.1 Within day (error) 34.3 and environmental variables (Table 3). Atten- 4317 100 J. F. PIATT, B. D. ROBERTSAND S. A. HATCH

Early egg-laying

FIGURE 3. Diurnal attendanceof Leastand Crestedauklets at eight 200-m* plots on Owalit Mountain at differentstages of the breedingseason.

Auklets in 1987 compared to Searing’s estimates respectively, whereas Crested Auklets increased in 1976 (Table 4). The mean densities of both 1.7- and 1.4-fold, respectively, over the 20-year speciesdecreased on the north side, but increased interval since Bedard’s investigation. Extrapo- on the south side of Kongkok. The density of lating to the total area surveyed by BCdard,total Least Auklets on Owalit did not change since auklet populations in Owalit (2 1,000 birds) and Searing’s investigation, but we noted a large in- Kongkok (730,000 birds) are now about twice crease in density of Crested Auklets. At both what they were during Bedard’s time. locations, Searing observed higher densities of Least and Crested auklets in 1976 than BCdard DISCUSSION had in 1964. Basedon Bedard’s (1969b, p. 39 1) original es- ATTENDANCE PATTERNS timates, the overall density of Least Auklets in- Breeding auklets arrive at Kongkok Bay in May creased1.6- and 2.7-fold at Owalit and Kongkok, and early June, and establish breeding territories AUKLET ATTENDANCE PATTERNS 101

AUKLET ATTENDANCE on talus slopes where they engage in courtship activities, defend against competitors, and select sites to lay their eggs in crevices under the talus rocks (Btdard 1969b, Sealy 1975). Subadults arrive during incubation and remain until fledging. They also loiter on talus slopes during the day, interacting with other auklets and prospecting for breeding sites (Btdard 1969b, Roby and Brink 1986). During prelaying and incubation periods, at INCUBATION least one adult stays in the breeding crevice during the day and frequently both adults occupy egg sites overnight (Sealy 1968, Roby and Brink 1986). Adults exchange incubation duties about every 24 hr (Sealy 1968, Roby and Brink 1986). During chick rearing, one adult auklet stayswith the chick for most of the day during the first 5- 10 days of life (Roby and Brink 1986; Piatt et CHICK-REARING al., in press), and thereafter chicks are increas- (AUGUST) ingly left unattended during the day while adults forage or sit on talus slopes. On the basis of those observations, one might expect the numbers of auklets on talus slopesto be highest during prelaying when neither mate is restricted to an egg site, and lowest during \,. U,‘,. . .,. .,.,. 6 b 10 12 12 lb 18 20’ 22 24 !! chick rearing when both parents are foraging for HOUR chick meals. This was true for Least Auklets, FIGURE 4. Diurnal attendanceof Least and Crested even though the arrival of subadults during in- auklets integrated over time during prelaying, incu- cubation may inflate slope attendance by lo- bation, and chick rearing. Each data point is the mean 50% (Bedard 1969b; Roby and Brink 1986; Ian (*SE bars) attendanceper hour per month calculated Jones, pers. comm.). Crested Auklet attendance, from time-lapse countson all plots conductedat 5-min intervals. however, was markedly lower during the prelaying period. At St. Lawrence Island, Crested Auk-

PRE-LAYING ,, INCUBATION ,,CHICK-REARING,

CRESTED AUKLET

FIGURE 5. Seasonalattendance of Least and Crested auklets as determined from time-lapse film. Each data point is the mean (+SE bars) attendanceper day on lo-m2 surfaceplots calculatedfrom time-lapse counts on all plots conductedat 5-min intervals. 102 J. F. PIATT, B. D. ROBERTS AND S. A. HATCH

TABLE 2. Variation in Least and Crested auklet attendanceduring different periods in their breeding cycle at St. Lawrence Island (counts of auklets at 5-min intervals on time-lapse film plots).

Least Auklet CrestedAuklet Period Date n R SE cv .z SE cv Prelaying 1-15 Jun 1,289 0.51 0.03 229 0.10 0.01 426 16-30 Jun 2,846 0.48 0.02 238 0.37 0.02 278 Incubation 1-15 Jul 3,139 0.49 0.02 216 0.76 0.03 224 16-31 Jul 2,095 0.67 0.03 188 1.58 0.06 175 Chick rearing l-l 5 Aug 2,171 0.49 0.02 199 1.43 0.05 166 16-3 1 Aug 1,470 0.05 0.01 549 0.07 0.01 526

let egg sites are deeper and less accessible to the day to chicks (Byrd et al. 1983, Roby and Brink surface than Least Auklet egg sites, and Crested 1986). Auklets might spend more time under the talus Crested Auklets may suppressthe numbers of during prelaying and therefore were underrep- Least Auklets attending shared talus slopes resented on surface counts compared to Least through aggressiveintimidation (Bedard 1969b, Auklets. Alternatively, Crested Auklets (ca. 270 Knudtson and Byrd 1982). Our data support this g) may have to forage longer than Least Auklets hypothesis becauseon a 5-min time scale, Least (ca. 85 g) becauseof their higher food demands, and Crestedauklet numbers on 1O-m2 time-lapse and attendance may not peak until zooplankton plots were negatively correlated about 75% of blooms result in densities of prey sufficient to the time. Attendance at plots was often synchro- support the extra cost of commuting from the nized; auklets of the same speciesoften arrived colony (BCdard 1969b; Piatt, in press).In support and departed togetherin groupsof 20-50, as they of this hypothesis, Byrd et al. (1983) noted a do while foraging at sea (Bedard 1969a). This nearly twofold increase in Crested Auklet activ- partially accounts for the erratic diurnal atten- ity (total movement in and out) between pre-egg dance data (Figs. 2, 3) and, in light of the neg- laying and late incubation at Buldir Island where- ative correlations in fine-scale attendance, fur- as Least Auklet activity remained similar over ther suggeststhat Least Auklets may conducttheir that time. activities at the surface in short bouts between Attendance by both speciesdiminished after periods of intense Crested Auklet activity. hatching of chicks and apparent attendance Diurnal patterns of attendance changed sea- stopped completely during late chick rearing. sonally for two reasons.First, seasonalvariations Most of the birds attending the surface at that in day length determine the timing of departure time may have been subadults (Dan Roby, pers. in morning and return in evening. Auklets ap- comm.). Although the attrition in numbers was parently do not forage at night (Bedard 1969a, partly a result of fledging chicks and adults leav- 1969b; Roby and Brink 1986). Similarly, day ing the colony, breeding birds stopped loitering length determines the total amount of time avail- on slopes at that time. The total movement of able for foraging, and therefore the duration of adults actually increasestwo- to fourfold during the midday break in attendance. This may ex- chick rearing as adults bring four to six meals a plain why the midday gap in attendance at St.

TABLE 3. Spearman’s rank correlationsbetween mean numbers of auklets attending time-lapse plots per day and environmental variables (ns = nonsignificant).

LeastAu&t Crested Auklet Variable n I P I P

Wind speed 55 -0.29 co.05 -0.31 co.05 Tide range 55 -0.07 ns -0.28 co.05

MaximumTide (7-day temperature means) 559 -0.690.26 co.05ns -0.540.48

TABLE 4. Mean densitiesof Leastand Crestedauklets on Owalit Mountainand in KongkokBasin, St. Lawrence Island. Countsfrom 1987 indicatedto be significantlyhigher ( ****P i 0.0001) or not significantlydifferent (ns) from countsin 1976 by Searing(1977). Differencestested using two-tailed f-tests. Counts in 1964 from B&lard(1969b).

1987 1976 1964 Location n x SE n R SE n .z LeastAuklet Owalit Mtn. 24 53 3.0 ns 24 64 4.9 ? 36 KongkokBasin 48 90 5.9 **** 48 56 5.0 39 34 CrestedAuklet

OwalitKongkok Mtn.Basin 4824 4232 4.46.0 ****ns 4824 4013 12.61.7 39? ::

Lawrence Island (7-12 hr) is greater than that shore fronts increase in intensity as tidal oscil- observed further south at the Pribilofs (5-9 hr; lations become larger, and this may serve to con- Roby and Brink 1986) or Buldir Island (5-8 hr; centrate zooplankton in predictable locations Byrd et al. 1983). Secondly, auklets must make (Alldredge and Hamner 1980, Brown 1980, Ha- more visits to the colony during chick rearing to ney 1988). We found that auklet attendance was deliver chick meals, and activity stretchesover negatively correlated with the range of tidal os- a longer portion of the day at that time. Although cillations when the data were grouped over 7-day a midday gap in attendance continues through intervals. This suggeststhat birds were foraging chick rearing, this is deceptive becauseadults still more during periods when tidal action was come and go at reduced levels, but they do not strongest,and spentmore time at the colony when loiter on slopesafter delivering food (Byrd et al. tidal movements were minimal and prey were 1983, Roby and Brink 1986). Like Byrd et al., presumably less available. Wind speed was also we found that Least Auklets usually returned negatively correlated with auklet attendance, about 1 hr earlier in the morning than Crested particularly during the prelaying period. This may Auklets. Some Least Auklets use prey patches have been because high winds interfered with closer to the colony (e.g., at sites of coastal up- activities at the colony so birds just stayed away, welling) than patchesexploited by Crested Auk- or high winds may interfere with foraging activ- lets (BCdard 1969a), and therefore may be able ities at sea, so birds are forced to spend more to get back to the colony more quickly after the time foraging. dawn exodus. Auklets from Kongkok Bay may regularly for- MONITORING AND CENSUSING age over 60 km from their colonies, and by late POPULATIONS chick rearing as far as 150 km from colonies BCdard’s (1969a, 1969b) ‘surface count’ tech- (Piatt et al. 1988; Hunt et al., in press). Some nique is relatively simple to apply and interpret, auklets must foragewithin about 30 km of Kong- large numbers may be counted by a single ob- kok to be able to return within an hour or so server, and there may be a direct relationship after the dawn exodus (assuming flight speedsof between birds at the surface and the number of ca. 60 km/hr). For auklets to make a return trip breeding pairs occupying the talus (Sealy 1975; to foraging areas 150 km away would require Ian Jones, pers. comm.), although the ratio re- about 6 hr (allowing 1 hr for feeding), which may mains to be determined. On the negative side, explain the total absenceof loitering adults dur- our study indicated that Crested Auklets were ing late chick rearing. At that time, delivery of least abundant before egg laying and counts of five chick meals per day (Roby and Brink 1986) both auklet specieswere most variable at that would require a commitment of about 30 hr/ time. Furthermore, the deletion of the highest day, or about 15 hr of a 17- to 18-hr day for each count to estimate mean densities seemsarbitrary adult. (Byrd et al. 1983). Tidal conditions also may influence atten- At Buldir Island, Byrd et al. (1983) estimated dance behavior via its effect on prey availability. auklet numbers by mark-recapture and net Tide rips near headlands and turbulence at off- movement techniques, and compared those with 104 J. F. PIATT, B. D. ROBERTS AND S. A. HATCH estimates from surface counts. Mark-recapture and convenient size) should be conducted over effortswere unsuccessfulbecause auklets became morning and early afternoon, and for statistical net-shy after initial capture and leg bands were purposes, the top 10 counts should be used to difficult to seeon birds walking or resting on talus get a mean value of attendance per day. Counts slopes.The net movement of unmarked birds to should be conductedover at least a 14day period and from study plots was used to estimate num- to get a mean value of attendance over a com- bers of auklets, and results suggestedthat surface plete tide cycle. Days with strong winds (e.g., counts underestimated numbers about lo-fold. >40 km/hr) should be avoided for censusing. Furthermore, results indicated that count vari- The number of plots counted will depend on ability was lowest during late incubation and ear- manpower available, but a reasonable number ly chick rearing. Theoretically, the net movement (e.g., 2 10) in representative habitat should be technique should provide reasonable estimates established and permanently marked for later of total birds using breeding habitat. However, comparisons. For monitoring purposes,it is suf- the technique is difficult to apply in areas of high ficient to apply these methods to sample plots auklet density, requires a substantial observer over succeedingyears. effort, yields highly variable counts, and is dif- Extrapolation to total numbers of breeding ficult to interpret without knowledge of breeding birds requires knowing the ratio between the phenology and the activities of adult birds (Byrd number of adult birds present on talus slopesand et al. 1983). the number of breeding sitesbelow the talus, and At the Pribilofs, Roby and Brink (1986) also measuring the total habitat used by auklets. We monitored the net movement of marked birds recommend the use of the net movement tech- over brief periods at a small colony. They con- nique for determining ratios of attending (adults cluded that the best time to censusbreeding adults and subadults) to breeding birds. Although dif- was in the evening during late chick rearing be- ficult to apply to large areas or in areas of high cause at that time, only adults with meal deliv- auklet density (Byrd et al. 1983) this technique eries were observed arriving at the colony. How- could be usedto establishratios in relatively small ever, our study indicated that attendance during areas (e.g., 10 to 20 mZ) of larger surface count late chick rearing was more variable than at any plots. Besides obtaining a correction factor for other time, and more work is required to deter- surfacecounts, subadult birds could be identified mine what proportion of breeding adults would on net movement watches, and that would help be counted on those evening arrivals. To extrap- in interpreting seasonaland annual variations in olate from net movements to total population attendance. size, this method requires that birds be captured For monitoring, we also recommend the use and banded, and only small numbers can be of 35mm time-lapse photography. Whereas the monitored by a single observer. One similar and 8-mm cameras we used were adequate to quan- promising approach is to count the numbers of tify some aspects of attendance behavior, and banded adults carrying food in gular pouches time-lapse counts were strongly correlated with throughout the chick-rearing period to estimate surface counts, the number of birds monitored the numbers of auklets per unit area that suc- was quite small. The 35mm camera could be cessfully rear chicks (Ian Jones, pers. comm.). used to monitor attendance in much larger plots Like the net movement technique, however, this (e.g., 50 to 200 m2) with permanent boundaries requires a considerable time investment and is highlighted by rope or flagging tape during the limited by the number of individual birds that study period. We recommend that photos be tak- can be monitored at one time. en about every 20 min through morning hours Basedon our experienceand the results of pre- of attendance over the same 2-week period that vious investigations, we recommend the follow- surface counts are conducted. Establishing one ing approach for censusingand monitoring auk- or more permanent time-lapse plots would pro- let populations. Censusing should be conducted vide a valuable secondarysource of information by using the surface-counttechnique over at least for monitoring populations because the time- a 2-week period during late incubation and early lapse technique provides a measure of atten- chick rearing (i.e., when Crested Auklet attendance that integratesfar more data over days and dance is maximal and variability in counts is weeks than can be obtained from surfacecounts. minimal). Counts on 200-m2 plots (a workable Furthermore, it can yield a single average value AUKLET ATTENDANCE PATTERNS 105

of attendance with known variance for compar- biles in the 1960s and the population of foxes ison between years, and the influence of diurnal has been reduced since Btdard’s time (Lane Iya- and daily variations in attendance is minimized kitan, pers. comm.). Perhaps more importantly, because data are integrated over the whole pe- the traditional Yup’ik method of capturing auk- riod. lets in nets on talus slopes,and removing chicks from crevices, had fallen into disfavor at Kong- POPULATION TRENDS kok even in Bedard’s time (Bedard 1969b). We If the density estimates of BCdard(1969b), Sear- found numerous stone walls and blinds at both ing (1977), and ourselves in 1987 are compara- Owalit and Kongkok that were used in the past ble, then it appears that over the period 1967- for hiding while netting birds, but thoseappeared 1987, Least Auklet densities increased dramat- not to have been used for many years, and we ically at Kongkok and slightly at Owalit, whereas observed no netting in 1987. Thus, the indirect Crested Auklet densities increased markedly at (disruption of adult attendance and chick feed- Owalit, but only slightly at Kongkok. Overall, ing) and direct (loss of breeding adults) effectsof the densitiesofboth speciesincreased about two- netting have been reduced or eliminated since fold since Bedard’s study was conducted. Our Btdard’s time. estimates for 1987 would be slightly higher if we Finally, the different patterns of population in- had made the corrections for percentageof plots creaseat Kongkok and Owalit exhibited by Least covered by suitable breeding substrate and the and Crestedauklets may result from competitive depth of talus, as described by BCdard (1969b). interactions between speciesand habitat differ- However, BCdard apparently did not measure ences between those sites. The habitat at Kong- talus depths at Kongkok (he made no mention kok is definitely favored by Least Auklets, es- of Owalit where the talus is much deeper), and pecially on the south side, because the talus is his corrections for percentagecover were mini- composed of a shallow layer of small boulders mal, so our estimates were probably conserva- (Bedard 1969b, Searing 1977). The reverse is tive. found on Owalit, where a deep layer of talus Although statistically significant, we are cau- composedof large boulders would be favored by tious in suggestingthat the changesobserved were Crested Auklets (Btdard 1969b, Searing 1977). biologically significant. We still know little about The fact that Crested Auklet densities have re- interannual variations in attendancepatterns, and mained the same despite a large increasein Least particularly about how subadult numbers con- Auklets at Kongkok, and that Least Auklet den- tribute to total variability. Attendance behavior sities declined slightly while Crested Auklets in- of subadults needsfurther study. Subadult plum- creasedat Owalit (1976-l 987) supportsthe idea agesare confusing, however, and we often found that asymmetric interference competition exists it difficult to distinguish subadults from adults. between speciesfor breeding habitat, with Crest- Plumages need to be better described before at- ed Auklets being superior in their favored habitat tendance patterns of different age classescan be because of their larger size and aggressivebe- studied with confidence. havior (BCdard 1969b, Byrd et al. 1983, this We can only speculate on the causes of the study). apparent increase in auklets at Kongkok and Owalit. If populations were limited by food, it is ACKNOWLEDGMENTS possible that zooplankton became increasingly We thank the Sivuqaqand SavoongaNative Corpo- more abundant or available to auklets over the rationsfor permissionto workon St. LawrenceIsland. 20-year period examined because of favorable In locatingstudy areas used by previousinvestigators oceanographicconditions or a reduction in fish and for assistancein the field, we are greatlyindebted speciesthat could compete with auklets for zoo- to Lane Iyakitan. We alsothank Wayne Lidster,John Wells. Chris Hanev. and Joel Hubbard for helo at plankton (Springer and Roseneau 1985). How- Kongkok.We aregiateful to Vem Byrd,William Dru- ever, a reduction in predation on auklets by arctic ry, Tony Gaston,Ian Jones,Karen Mance,Dan Roby, foxes (Alopex lugopus) and humans is the most and Alan Springerfor reviewingprevious drafts of the likely explanation for the apparent increase in manuscript.This study was funded partially by the Minerals ManagementService (MMS), U.S. Depart- auklet numbers. The trapping of foxes by native ment of the Interior (DOI), throughan intra-agency Yup’ik eskimos increasedon St. Lawrence Island agreement(No. 14-12-000 l-3039 1) with the Fish and with the arrival of modern traps and snowmo- Wildlife Service,U.S. DOI, aspart of the MMS Alaska 106 J. F. PIATT, B. D. ROBERTS AND S. A. HATCH

Environmental Studies program. Additional funding Common Murres and Atlantic Puffins to schools and logistic support was provided by the Alaska Fish of capelin. Stud. Avian Biol. and Wildlife Research Center, the Alaska Maritime PIATT, J.-F., S. A. HATCH, B. D. ROBERTS,W. W. National Wildlife Refuge, and Theodore Crossof New LIDSTER.J. L. WELLS.AND J. C. HANEY. 1988. York. Populations, productivity, and feeding habits of seabirds on St. Lawrence Island, Alaska. Alaska Fish and Wild]. Res. Center, Final Report for Min- LITERATURE CITED erals Management Service, OCS Study MMS-88- 0022, Anchorage, AK. ALLDREDGE,A. L., AND W. M. HAMNER. 1980. Re- PIATT,J. F., B. D. ROBERTS,W. L. LIDSTER,J. L. WELLS, curring aggregationof zooplankton by a tidal cur- AND S. A. HATCH. In press. Breeding biology of rent. Estuarine Coastal Mar. Sci. 10:31-37. Least and Crested auklets at St. Lawrence Island, BBDARD,J. 1969a. Feeding of the Least, Crested,and Alaska. Auk. Parakeet auklets around St. Lawrence Island, ROBY, D. D., AND K. L. BRINK. 1986. Breeding bi- Alaska. Can. J. Zool. 47:1025-1050. ology of Least Auklets on the Pribilof Islands, BBDARD,J. 1969b. The nestingof the Crested, Least, Alaska. Condor 88:336-346. and Parakeetauklets on St. LawrenceIsland, Alas- SAS. 1985. SAS user’s guide:statistics. SAS Institute, ka. Condor 71:386-398. Cary, NC. BROWN,R.G.B. 1980. Seabirds as marine animals, SEALY,S. G. 1968. A comparative study of breeding p. l-39. In J. Burger, B. L. Olla, and H. E. Winn ecologyand timing in plankton-feedingalcids (Cy- [eds.],Behavior of marine animals. Plenum Press, chlorrynchusand Aethia spp.) on St. Lawrence Is- New York. land, Alaska. M.Sc.thesis.Univ. of British Colum- BYRD,G. V., R. H. DAY, AND E. P. KNUDTSON. 1983. bia, Vancouver, B.C. Patterns of colony attendance and censusingof SWY, S. G. 1975. Influence of snow on egg-layingin auklets at Buldir Island, Alaska. Condor 85:274- auklets. Auk 92528-538. 280. SEARING,G. F. 1977. Some aspectsofthe ecologyof HANEY, J. C. 1988. Foraging by Northern Fulmars cliff-nesting seabirds at Kongkok Bay, St. Law- Fulmarus glacialis at a nearshore, anti-cyclonic rence Island, Alaska during 1976, p. 263-4 12. In tidal eddie in the Northern Bering Sea, Alaska. Environmental assessmentof the Alaskan Conti- Colonial Waterbirds 11:318-32 1. nental Shelf,Annual Reports,Vol. 5. BLM/NOAA, HUNT, G. L., N. M. HARRISON,AND T. COONEY. In OCSEAP, Boulder, CO. press. Foragingof Least Auklets: the influence of SOWLS,A. L., S. A. HATCH,AND C. J. LENSINK. 1978. hydrographicstructure and prey abundance.Stud. Catalog of Alaskan seabird colonies. U.S. Dept. Avian Biol. Interior, Fish and Wildl. Serv. FWSIOBS-78/78. KNUDTSON, E. R., AND G. V. BYRD. 1982. Breeding SPRINGER,A. M., AND D. G. ROSENEAU.1985. Co- biology of Crested, Least, and Whiskered auklets pepod-basedfood webs:auklets and oceanography on Buldir Island, Alaska. Condor 84: 197-202. in the Bering Sea. Mar. Ecol. Prog. Ser. 21:229- PIATT, J. F. In press. The aggregativeresponse of 237.