The Condor93:943-95 I 0 The CooperOrnithological Society 1991

BROOD SIZE, FOOD PROVISIONING AND CHICK GROWTH IN THE PIGEON COLUMBA l

S. K. EMMS* AND N.A.M. VERBEEK Department of BiologicalSciences, Simon Fraser University,Burnaby BC V5A lS6, Canada

Abstract. We studied the effects of brood size and chick age on food provisioning rates to Pigeon Guillemot Cepphuscolumba nests, and differencesin growth parametersbetween chicks in broods of one and broods of two. Provisioning rates increased with chick age in the first part of the nestling period, and then declined. This decline was steeper in 1985, when overall provisioning rates were higher. Per-chick provisioning rates were lower to broods of two, resulting in one of the chicks having a lower growth rate, and possibly a lower fledgingmass, than its sibling. Growth parametersof the fastergrowing chick in broods of two did not differ significantlyfrom those of solo chicks. The decline in provisioning rate late in the nestling period is probably not due to reduced food availability. It may be due to reducedenergetic demands of the chicks or to a deliberate strategyon the parents’ behalf to induce fledging. At present we lack the information to distinguish between these two hypotheses. Key words: Pigeon ;Alcidae; food provisioning:brood size; chick growth.

INTRODUCTION tend to have smaller clutches (Preston 1968, Pe- Most members of the family (Alcidae) lay a tersen 198 1, Emms and Morgan 1989, Ewins clutch of one egg. The generally accepted expla- 1989), though this is partly explained by young nation for this is that parents cannot successfully tending to breed later (Asbirk 1979). Birds rear larger broods becauseof limitations on their laying clutches of two normally have a propor- ability to provide food for chicks (Lack 1968, tionately higher reproductive output than those Birkhead and Harris 1985). This explanation is laying clutches of one, mainly becauseeggs from supported by twinning experiments, which have one egg clutches have a lower hatching success normally shown that birds with artificial broods (Petersen 198 1, Emms and Morgan 1989, Ewins of two have been unable to rear both chicks suc- 1989), but sometimes because fledging success cessfully (Birkhead and Harris 1985 and refer- per chick also is lower (Cairns 198 1). Food de- ences therein). Cepphus guillemots typically lay livery rates to broods of two are normally higher a clutch of two eggs,and their ability to rear two than to broods of one, but per-chick rates are chicks is probably due to their feeding on inshore often no different (Cairns 198 1) or lower in broods benthic fishes rather than the offshore pelagic of two (Koelink 1972, Asbirk 1979, Petersen prey exploited by most other alcids (Bradstreet 1981). and Brown 1985). Because their foraging trips Food provisioning rates are also influenced by are shorter, they can deliver food to their nests chick age. Previous studies have found a variety at a higher rate, and consequently rear more of patterns, and it is important to distinguish chicks. between those measuring delivery rates (number Although the modal clutch size of Cepphus of fish per unit time) and those estimating pro- guillemots is two, a significant proportion (5-28 visioningrates (biomassor energyper unit time). percent) of breedersin most populations studied Drent (1965), Bianki (1967) and Koelink (1972) lay only one egg (Birkhead and Harris 1985). found that delivery rates increased with chick These are often young, inexperienced birds (As- age, and Petersen (198 1) found that rates in- birk 1979). Birds breeding later in the seasonalso creased until chicks were 25-30 days old and then declined markedly. Asbirk (1979) and Cairns (1987) found no effect of chick age on delivery I Received6 March 1991. Final acceptance2 July rates, though mean fish size increased with chick 1991. * Presentaddress: Department of Ecologyand Evo- age, so that older chicks were provisioned at a lutionaryBiology, Princeton University, Princeton, NJ higher rate. Cairns’s (1987) calculations also sug- 08544, U.S.A. gested a slight decline in provisioning rates for

19431 944 S. K. EMMS AND N.A.M. VERBEEK

chicks older than 25 days. Of these studies, only were classified as blennies (Blennioidea), scul- Cairns’s (1987) data were analyzed statistically, pins (Cottidae), lingcod Ophiodonelongatus, flat- and none controlled for other factors that can fish (Bothidae, Pleuronectidae), sandlance Am- affect delivery rates, such as time of day (e.g. modyteshexapterus, rockfish Sebastes spp., shiner Drent 1965, Slater and Slater 1972, Petersen perch Cymatogasteraggregata, herring (Clupei- 198 1, Cairns 1981, 1987), or tidal state (Belo- dae), and unidentified. We chose starting times pol’skii 1957, Preston 1968). so that the tidal height halfway through the pe- The aims of our study were (1) to investigate riod was either 0.75 m, 2.30 m, or 3.85 m. These how brood size and chick age influence food pro- heights corresponded to low-, mid-, and high- visioning rates in the Pigeon Guillemot Cepphus periods within the daily tidal cycle. columbawhen confounding variables are con- Mass-length regressions for locally caught trolled for statistically; (2) to assesswhy provi- blennies, , flatfish, and shiner perch were sioning rates may decreasein the latter part of used to calculate provisioning rates to individual the nestling period; and (3) to determine the con- nestsin grams wet mass per hour. Lingcod mass- sequencesof lower per-chick provisioning rates es were calculated from the mass-length regres- for the growth of individual chickswithin broods sion for sculpins, since these fish were similar in of two. shape and the massesof three lingcod found in guillemot nests closely fitted the predicted re- METHODS gression for sculpins. Data were excluded from The study was carried out in 1984 and 1985 on these calculations if parents delivered fish for the 200 pairs of Pigeon Guillemots breeding on which we did not have mass-length regressions, Mitlenatch Island (40”57’N, 125”OO’W), a 35.5 or if any fish were unidentified. ha rocky island in the Strait of Georgia, British We fitted logisticequations to the growth curves Columbia, Canada. We systematically searched of individual chicksand usedthe asymptotic mass for nestsearly in the seasonand visited occupied parameter A as a measure of fledging mass (Rick- nests every l-5 days throughout the breeding lefs 1967). Masses were log-transformed before period. We weighed chicks every 3-5 days in analysis. Because nestling guillemots typically 1984anddailyin 1985usingO-lOOgandO-500 reached their maximum mass near fledging, A g Pesola scales.In 1985, when nestswere checked was a reasonable estimate of fledging mass. The daily, chicks from two-egg clutches typically mass on the last day in the nest would have been hatched on the same day (n = 8) or on successive a poor measure because masses varied consid- days (n = 16). In one casethe inter-hatch interval erably (up to 30-40 g) from day to day during was two days. When chicks hatched on different the final part of the nestling period. We calcu- days, the first chick normally had gained mass lated growth rates in g/day by fitting linear re- by the time the second chick hatched and was gressions to the linear phase of growth, when easily distinguishable from its younger sibling. chicks were 5-20 days old. This is preferable to When chicks hatched on the same day, a size using the growth curve constant K of the logistic hierarchy normally developed within a few days. equation, as recommended by Ricklefs (1967), For convenience, we will refer to the heavier becauseK measuresthe speedof approach to the chick in a brood of two as the o( chick and the asymptote and can give misleading results if lighter chick as the /3 chick. comparisons are made between chicks reaching We observed food delivery rates to nests in different asymptotes(Husselll972, Gaston 1985). several nest groups. Observations were made Statistical analyseswere done using the BMDP from a blind, or from locations distant (100-200 statistical package (Dixon 1985). Proportional m) or concealed from the nest group using bin- data were arcsin square root transformed where oculars and 20-40 power telescopes.During each appropriate. 2 hr observation period we recorded the number of food deliveries to each nest, the type of fish RESULTS delivered, and its size, estimated as a multiple of the guillemot’s bill length to the nearest quar- FISH TYPES AND SIZES ter bill length. While this method may inaccu- The majority of the fish delivered in both years rately estimate absolutesize, it should not result were either blennies or sculpins. Small numbers in systematicbiases within our own data set. Fish of lingcod, flatfish, sandlance, rockfish, shiner PIGEON GUILLEMOT CHICK PROVISIONING 945

TABLE 1. Frequenciesand percentagesof fish types delivered to nests.

1984 1985 Fish type Number Percentage Number Percentage Blennies (Blennioidea) 306 43.7 119 30.8 Sculpins (Cottidae) 171 24.4 10.5 27.2 Lingcod Ophiodonelongates 48 6.9 53 13.7 Flatfish (Both&e, Pleuronectidae) 55 7.9 10 2.6 Sandlance 5 0.7 0 0.0 Ammodyteshexapterus Rockfish 22 3.1 24 6.2 Sebastesspp. Shiner Perch 1 0.1 17 4.4 Cymatogasteraggregata Herring (Clupeidae) 13 1.9 0 0.0 Unidentified 80 11.4 58 15.0

perch, and herring were also delivered (Table 1). A. Blennies The proportions of fish types delivered varied significantly among nests in both years (1984: x2 h 14- = 62.9, P = 0.0001, df = 18, n = 10 nests; 1985: E E 3 * 7: x2 = 90.8, P = 0.0001, df = 14, n = 8 nests. Fish 13- I .c 70 72 types were categorizedas blennies, sculpins, and 28 “others” and nests were included in the analysis F 2-‘ i only if expectedfrequencies for all fish categories ! 4 ll- 47 were greater than five). The proportions of fish 21 types delivered also varied significantly between brood sizes.In 1984, broods of two received more 10 ! I I I I I 1 blennies and fewer sculpins than expected (x2 = 0 1 2 3 4 5 6 10.0, df = 2, P < 0.01, IZ = 363 fish). In 1985, broods of two received more sculpins and fewer ‘others’ than expected (x2 = 14.6, df = 2, P c 0.001, n = 276 fish). To what extent these dif- ferences were due to an effect of brood size per 12 - B. Sculpins se, rather than to the variation among nests de- 7 scribed above is unclear. Eli- Both blenny and sizes differed signif- v. 14 49 icantly among chick age categories.For blennies, 52 c lo- 1 there was a marked increase in fish size in the 4g I 32 f third week of the nestling period (Fig. 1A). For P f sculpins the pattern was less obvious, but size : I 3 9

---t 81 FIGURE 1. The effect of chick age on the sizes of 0 1 2 3 4 5 6 (A) blennies and (B) sculpinsdelivered to nests.Blenny sizes differed significantly among chick age categories (ANOVA: F= 10.34, df= 5,310, P < 0.0001, Brown- Chick Age (Weeks) Forsythe test not assuming equal variances among 1967) and Bonferroni probabilities for multiple com- groups[Brown and Forsythe 19741).Means for the two parisons(see Neter et al. 1985). Sculpin sizes also dif- youngestage categories were significantlydifferent from fered significantly among chick age categories(ANO- all others (l-2: P < 0.01, l-3: P < 0.01, l-4: P < VA: F = 2.99, df = 5, 198, P = 0.048, Brown-Forsythe 0.001, l-5: P < 0.05, 2-3: P < 0.05, 24: P < 0.05, test). Only the means between age categories2 and 5 2-5: P < 0.00 1; all other tests n.s.; all P-values are for were significantlydifferent (P < 0.05). Sample sizesare separate variance t-tests (see Snedecor and Cochran given above data points. 946 S. K. EMMS AND N.A.M. VERBEEK

12 n 1984 A appeared to increase from the second week of 10 T ta 198.5 the nestling period onwards (Fig. 1B).

FOOD PROVISIONING RATES In 1984 chick age, time of day, and tidal state all had significant effects on provisioning rates. Square terms for chick age and time of day were 4 8 12 16 20 Nest Group Size also significant, indicating that the effectsof these variables were curvilinear (Table 2). Provision- ing rates increased during the first half of the nestling period, and then decreased.Rates were I21 R higher in the morning and evening hours, and IO were lower at high (Table 2, Fig. 2). In 1985 8 only chick age and its square had significant ef- 6 fects. Again, provisioning rates increased in the 4 first half ofthe nestling period and then decreased 2 (Table 2, Fig. 2). In neither year did nest group 0 size or brood size have significant effectson pro- LOW Mid High Tide visioning rates, though Figure 2d suggeststhat rates were higher to broods of two in 1985, and the failure to find a significant difference in this year may have been a problem of statistical pow- = 14 er. Overall, mean provisioning rate was lower in g 12 1984 (4.76 + 0.33 SE) than in 1985 (9.36 + 0.79 c 'O SE). Since the energy densities of the main fish 2 8 6 typesare similar (Dunn 1975, Cairns 1987), these z 4 patterns are unlikely to be affected by any sys- tematic changesin the proportions of fish types delivered. E 6 10 14 18 0 Time Date was not included in these analyses be- causeit was highly correlated with chick age. To determine whether the decreasein provisioning rates in the secondhalf of the nestling period was 14 1 due to a decreasein food availability late in the 12 season, rather than to an effect of chick age per 10 se, we analyzed the data for the first and second 8 6 halves of the nestling period separately. Because 4 of the reduced sample sizesand the large amount 2 of error variance in the data set, the overall re- 0 gressionswere significant only in the first half of 1 2 Brood Size 1984 and the second half of 1985. In 1984, time and its square and chick age and its square both

c E 2o1 T FIGURE 2. The effect of (A) nest groupsize, (B) tidal 16 state, (C) time of day, (D) brood size, and (E) chick age 12 on food provisioningrates (g/hr) to nests.Overall means 8 and standard errors are plotted. 1984: n = 2.51; 1985: n = 176. Since the effectsof the other variables are not 4 controlled for in any of the figures, these histograms 0 provide no statistical information and are presented 2 10 18 26 34 42 simply as a visual aid. See Table 2 for statistical anal- Chick Age ysis. PIGEON GUILLEMOT CHICK PROVISIONING 941

TABLE 2. Multiple regressionof factors affectingfood provisioning rates (g/hr) to nests. Squareterms for time of day and chick age were used to look for curvilinear effects of these variables. In broods of two chick age was arbitrarily defined as the age of the older chick. 1984: n = 251, R* = 0.099, F-ratio for regression ANO- VA = 3.815, P -c 0.0001; 1985: n = 176, R* = 0.112, F = 3.041, P = 0.0049.

1984 1985 P Slope SE t (2-h) SE I (2-L)

Intercept 1.135 - 13.795 Nest group size 0.020 0.078 0.259 0.796 0.354 0.738 0.480 0.632 Tidal state - 1.553 0.556 -2.795 0.006 2.040 1.256 1.625 0.106 Time of day -0.424 0.183 -2.315 0.022 0.761 0.530 1.446 0.150 Time of day2 0.016 0.006 2.416 0.014 -0.026 0.017 -1.465 0.145 Brood size 0.359 0.756 0.415 0.635 3.038 1.790 1.698 0.091 Chick age 0.664 0.186 3.578 0.0005 1.061 0.487 2.179 0.031 Chick arae2 -0.012 0.004 -3.023 0.003 -0.025 0.010 -2.675 0.008 still had significant effectson provisioning rates; pairwise differences were significant (Fig. 4B). tidal state no longer did (Table 3). This suggests Growth rates and asymptotic masseswere pos- that chick age itself, rather than date, may be the itively correlated (Y = 0.32, df = 46, P = 0.025). significant variable. In 1985, although the overall regressionwas significant (Rz= 0.17,F = 2.746, DISCUSSION df = 7, 92, P = 0.012) none of the coefficients FISH TYPES for individual variables were. In both years the majority of identified fish de- CHICK GROWTH RATES livered were either blennies or sculpins, as was found by Drent (1965) and Koelink (1972). The Growth parameters did not differ significantly observed variation among nests in the propor- between years, so data were pooled for further tions of different fish types, also documented in analysis. Growth rates differed significantly other studies (e.g., Drent 1965, Koelink 1972, among the three chick types (solo chicks and (Y Kuletz 1983), suggestseither that individual birds and p chicks in broods of two). This was due to specialize on different prey types, or that they solo chicks and cychicks both growing faster than have preferred foraging areas that differ in the fi chicks. (Figs. 3 and 4A). Asymptotic masses relative abundances of prey types. Kuletz (1983) also differed among chick types, although no found evidence for both possibilities.

FOOD PROVISIONING RATES 5oo . m The most consistent factor affecting provisioning . o ALPHA 5; 400- . BETA rates was chick age. In both years, rates increased E r! cn 300- V TABLE 3. Multiple regressionof factorsaffecting food provisioning rates (g/hr) to nestsin the first half of the 1984 breeding season. IZ= 130, R 2 = 0.186, F-ratio 3 for regressionANOVA = 3.969, P = 0.0006. 5v, *O”-loo- ?# ol . I ’ 1 . 1 1 ’ Variable Slope SE t (2-f&) 0 10 20 30 40 Intercept 0.517 Chick Age (days) Nest erou~ size -0.157 0.108 -1.449 0.150 Tidal-state -1.304 0.800 -1.631 0.105 FIGURE 3. Growth curves for solo chicksand 01and Time of day -0.617 0.221 -2.791 0.006 fl chicks in broods of two. Mean weights for all chicks Time of day2 0.024 0.009 2.820 0.006 in each category at each age are plotted. For clarity, Brood size 0.745 0.916 0.813 0.418 standard errors are not provided. Sample sizes range Chick age 0.930 0.326 2.853 0.005 from 1O-l 2 for the youngestchicks to 2-3 for the old- Chick age* -0.020 0.008 -2.233 0.027 est. 948 S. K. EMMS ANLJN.A.M. VERBEEK

7 12 17 22 27 32 37 42

Solo Alpha Beta Chick Age (days) FIGURE 5. Predictedmean food provisioningrates as a function of chick age,based on fitted partial re- gressioncoefficients from Table 2. For all independent variablesother than chickage, overall mean values for eachyear wereused in the regressionequation. 6.2 B 1 21 tlenatch, found that the biomass of fish in the a 6.1 - 10 rocky nearshorehabitats where guillemots forage f 14 peaked between July and September and did not z i 6.0 - decline until October, well after guillemot chicks have fledged. g -I 5.9- A secondpossibility is that the energy require- ments of chicks decreaseas growth slows down 5.81 when they approach adult size. In some Solo Alpha Beta the decreasein energy requirements due to slower growth is compensated for by an increase in re- Chick Type quirements for maintenance, so that total re- FIGURE 4. The effectof chick type on (A) growth quirements actually increase throughout the rate in g/day and (B) fledgingmass, calculated as the nestling period, or decrease only very slightly asymptoticmass parameter A of the logisticgrowth towards the end (Ricklefs 1979, 1987; Monte- curve(see Methods). Growth rates differed significantly vecchi et al. 1984; Klaassen et al. 1989). How- amongchick types (ANOVA: F = 14.1, df = 2,49, P < 0.0001, Brown-Forsythetest), due to p chicksgrow- ever, for fast-growing species the decline after ing moreslowly than the othertypes (solo-ol: ns., solo- the peak may be more marked (Roby 199 1). Cep- @:P < 0.01, (~$3:P < 0.001, all P-valuesare for sep- phus guillemots are relatively fast growing com- arate variance t-testsand Bonferroniprobabilities). pared to other alcids (Gaston 1985), so there Similarly,asymptotic mass differed among chick types could be a relatively sharp decline in require- (ANOVA: F = 4.25, df = 2, 50, P = 0.037. Brown- Forsythetest), although no individual t-testswere sig- ments near fledging. To try to assesswhether the nificant. Samplesizes are given abovedata points. decline in provisioning rates can be explained by a decline in energy requirements, we calculated predicted mean provisioning rates as a function during the first half of the nestling period and of chick age based on the regressioncoefficients then declined. What causes this decline? One of Table 2 (Fig. 5). In 1984 the decline in pro- possibility is that prey availability decreaseslate visioning rate late in the nestling period was small, in the season.Two lines of evidence suggestthat and can probably be explained by reduced met- this explanation is probably incorrect. First, pro- abolic demands. In 1985 the decline was much visioning rates to chicks fledging in the first half larger, from 12 g/hr on day 2 1 to 6.5 g/hr on day of the 1984 breeding seasonstill decreasedin the 35, the mean fledging age. It would be surprising latter part of the nestling period (Table 3), sug- if such a large decrease,reflecting a reduction in gesting that chick age has an effect at least partly provisioning rate of about 45 percent, resulted independent of date. Second, Moulton (1977), solely from a decrease in metabolic demands. working approximately 350 km southeastof Mi- However, since overall provisioning rates were PIGEON GUILLEMOT CHICK PROVISIONING 949 much higher in 1985, chicks probably matured young. Despite Asbirk (1979), it appears that faster, and so required less food later in the nest- pairs with broods of two, or artificial broods of ling period. three, do not normally supply food at optimal A third possibility is that parents deliberately rates for both their chicks. This could be due to reduced delivery rates to induce fledging at an parents being unable to provide extra food be- optimal time for themselves (Ydenberg 1989). cause of foraging limitations, or to a trade-off Since foraging trips are energetically expensive between present and future reproduction (Wil- (Drent and Daan 1980, Roby and Ricklefs 1986) liams 1966, Charnov and Krebs 1974). Although and may incur an increased risk of predation we have no direct evidence for the latter hy- (Harris 1980, Watanuki 1986), reduced delivery pothesis, two pieces of evidence suggestthat it rates probably are beneficial to parents. If chick is worth considering. First, provisioning rates de- survivorship is only weakly dependent on fledg- clined rapidly in the latter part of the nestling ing mass above some minimum threshold mass, period in 1985. If parents had provided food to a decline in provisioning rate late in the nestling broods of two at the maximum observed rate period may only slightly reduce the fitness of throughout the nestling period this should have current offspring, but considerably reduce the re- allowed the /I chick to reach a higher fledging productive coststo the parents. To test this hy- mass, even if it grew more slowly. We argued pothesis we need to know how survival rates of earlier that the observed changesin provisioning both chicks and adults are affectedby provision- rate with chick age were probably not a conse- ing rates. Such data are not yet available for Cep- quence of changes in food availability. Second, phus. Asbirk (1979) found that birds laying two eggs and rearing one or two young to fledging had a CHICK GROWTH RATES lower survival rate than those laying two eggs Provisioning rates did not increase significantly but rearing no young. The changein survival rate with brood size in either year. Consequently, from 92 percent to 83 percent reduced the ex- provisioning rates per chick were lower to broods pectation of further life from 11.5 years to 4.8 of two, resulting in the @ chick having a lower years. Sample sizes were small, so the difference growth rate (Figs. 3, 4A). Neither Drent (1965) in rates was not significant, but note that even nor Koelink (1972) mention an effect of brood small decreasesin survival rate greatly reduce size on provisioning rates, but several studies of the expected number of future breeding oppor- Black Guillemots C. grylle showed that delivery tunities. rates increased with brood size (Asbirk 1979, The longer nestling periods and lower fledging Cairns 1981, Petersen 1981). massesof p chicks could reduce their probability In this study, the cost of lower per-chick pro- of overwinter survival. However, while some visioning rates to broods of two clearly fell main- studiesof seabirdshave shown that early fledging ly on one of the chicks (Figs. 3 and 4). Growth and higher fledging massesincrease survivorship rates of solo and a chicks did not differ from one (Perrins et al. 1973, Iarvis 1974), studies of sev- another, but both grew significantly faster than eral alcid specieshave failed to find suchan effect /I chicks. A similar pattern was found for as- (Lloyd 1976; Hedgren 1979, 1981; Harris and ymptotic masses, though the effect was less Rothery 1985). As far as we are aware, no data marked. Figure 3 shows that p chicks did not are available for Cepphusguillemots chicks. compensate for their lower growth rates by re- maining in the nest longer and eventually fledg- ACKNOWLEDGMENTS ing at the same mass as their older sibs. Rather, they fledged at lighter massesand took longer to This work was funded by NSERC grant No. A0239 to NAMV and by the following scholarshipsto SKE: do so: p chicks normally fledged 4-8 days after The William and Ada Isabelle Steele Memorial Grad- their sibs. uate Scholarship (1984-1985), the Anne VallCe Eco- Artificially increasing the brood size to three logical Fund Scholarshio (1986) and an SFU Presi- has normally resulted in chicks fledging at lower dent’s Research Stipend-( 1986).‘ We thank the British Columbia Parks Branch for permission to work on masses(Koelink 1972, Petersen 198 l), but As- Mitlenatch, Anne and Art Lechasseurfor invaluable birk (1979) found that fledging masseswere ac- logisticalsupport, John Wieczorek, Rod Malcolm, and tually higher in five caseswhere pairs raised three Joel Hagan for help with fieldwork, Brad Thomas for 950 S. K. EMMS AND N.A.M. VERBEEK statistical assistance,and Bruce Lyon, David Cairns, GASTON, A. J. 1985. Developmnt of the young in the Daniel Roby, and an anonymous reviewer for sug- Atlantic Alcidae, p. 3 19-354. In D. N. Nettleship gesting improvements to an earlier version of the and T. R. Birkhead [eds.], The Atlantic Alcidae: manuscript. the evolution, distribution and biology of the inhabiting the Atlantic Ocean and adjacent water areas. Academic Press, London. LITERATURE CITED HARRIS, M. P. 1980. Breedingperformance of puffins ASBIRK,S. 1979. The adaptive significance of the Fraterculaarctica in relation to nestdensity, laying reproductive pattern in the Cep- date and year. Ibis 122:193-209. _ . - phus grylle. Vidensk. Medd. Dan. Naturhist. Fo- HARRIS,M. P., AND P. ROTHERY. 1985. The post- ren. 141:29-80. fledging survival of young puffins Fratercula arc- BELOPOL’SKII,L. 0. 1957. Ecologyof seacolony birds tica in relation to hatching date and growth. Ibis of the Barents Sea. Israel Program for Scientific 127~243-250. Translations. Jerusalem (Translated from Rus- HEDGREN,S. 1979. Seasonal variation in fledging sian, 1961). weight of guillemots, aalge. Ibis 121:356- BIANKI. V. V. 1967. . shorebirds and alcids of 361. Kandalaksha Bay. Israel Program for Scientific HEDGREN,S. 1981. The effect of fledging weight and Translations, Jerusalem (Translated from Rus- time offledging on survival ofguillemot Uria aalge sian, 1977). chicks. Omis Stand. 12:5l-54. BIRKHEAD,T. R., ANDM. P. HARRIS. 1985. Ecological HUSSELL,D.J.T. 1972. Factors regulatingclutch size adaptations for breeding in the Atlantic Alcidae, in arctic passerines.Ecol. Monogr. 42:3 17-364. p. 205-232. In D. N. Nettleship and T. R. Birk- JARVIS,M.J.F. 1974. The ecological significanceof head leds.1.The Atlantic Alcidae: the evolution. chick size in the South African Gannet Sula ca- distribution and biology ofthe auks inhabiting the pensis(Lichtenstein). J. Anim. Ecol. 43:1-17. Atlantic Ocean and adjacentwater areas.Academ- KLAASSEN, M., C. BECH, D. MASMAN, AND G. ic Press, London. SLAGSVOLD.1989. Growth and energeticsofArc- BRADSTREET,M.S.W., AND R.G.B. BROWN. 1985. tic Tern chicks(Sternaparadisaea). Auk 106:240- Feeding ecology of the Atlantic Alcidae, p. 264- 248. 3 18. In D. N. Nettleship and T. R. Birkhead [eds.], KOELINK,A. F. 1972. Bioenergeticsof growth in the The Atlantic Alcidae: the evolution, distribution Pigeon Guillemot, Cepphuscolumba. MSthesis, and biology of the auks inhabiting the Atlantic Univ. British Columbia, Vancouver, British Co- Ocean and adjacent water areas. Academic Press, lumbia. London. KULETZ,K. J. 1983. Mechanisms and consequences BROWN,M. B., AND A. B. FORSYTHE.1974. The small of foraging behavior in a population of breeding sample behavior of some statisticswhich test the Piaeon Guillemots. M.S.thesis. Univ. . equality of several means. Technometrics 16:129- IGine, CA. 132. LACK, D. 1968. Ecological adaptations for breeding CAIRNS,D. 1981. Breeding,feeding, and chick growth in birds. Methuen, London. of the Black Guillemot (Cemhus nrvlle) in south- LLOYD,C. S. 1976. The breedingbiology and survival em Quebec. Can. Field-Nat: 95:3‘i 2-3 18. ofthe Alca torda L. D. Phil. thesis,Univ. CAIRNS.D. K. 1987. The ecoloav-. and eneraetics of Oxford, Oxford, England. chick provisioning by Black Guillemots. Condor MONTEVECCHI,W. A., R. E. RICKLEFS,I. R. KIRKHAM, 89~627-635. AND D. GABALDON. 1984. Growth energeticsof CHARNOV,E. L., AND J. R. KREBS. 1974. On clutch nestling Northern Gannets (Hula bassanus).Auk size and fitness. Ibis 116:217-219. 101:334-341. DIXON, W. J. (ED.). 1985. BMDP statistical software MOULTON,L. L. 1977. An ecologicalanalysis of fishes manual: 1985 printing. Univ. Calif. Press, Berke- inhabiting the rocky nearshoreregions of northern ley, CA. , . Ph.D.thesis, Univ. DREN~, R. H. 1965. Breeding biology of the Pigeon Washington, Seattle, WA. Guillemot Ceoohuscolumba. Ardea 53:99-160. NETER.J.. W. WASSERMAN.AND M. H. KUTNER. 1985. DRENT,R. H., AND-S.DAAN. 1980. The prudent par- Applied linear statistical models. 2nd ed. ent: energeticadjustments in avian breeding. Ar- PERRINS,C. M., M. P. HARRIS,AND C. K. BRITTON. dea 68~225-252. 1973. Survival ofManx ShearwatersPufinuspuf- DUNN, E. H. 1975. Caloric intake of nestling Double- finus. Ibis 115:535-548. crested Cormorants. Auk 92:553-565. _ PETERSEN,A. 1981. Breeding biology and feeding EMMS,S. K., AND K. MORGAN. 1989. The breeding ecologyof Black Guillemots. D. Phil. thesis,Univ. biology and distribution of the Pigeon Guillemot Oxford, Oxford, England. (Cepphuscolurnba) in the Strait of Georgia,p. IOO- PRESTON,W. C. 1968. Breeding ecology and social 107. In K. Vermeer and R. W. Butler, [eds.], The behavior of the Black Guillemot Cepphusgrylle. ecology and status of marine and shoreline birds Ph.D.thesis, Univ. Michigan, Ann Arbor, MI. in the Strait of Georgia, British Columbia. Spec. RICKLEFS,R. E. 1967. A graphical method of fitting Publ. Can. Wild. Serv., Ottawa. equations to growth curves. Ecology 48:978-983. Ewms, P. J. 1989. The breeding biology of Black RICKLEFS,R. E. 1979. Adaptation, constraint, and Guillemots CepphusgrylZe in Shetland. Ibis 13 1: compromisein avian postnataldevelopment. Biol. 507-520. Rev. 54~269-290. PIGEON GUILLEMOT CHICK PROVISIONING 951

R~CKLEFS,R. E. 1987. Response of adult Leach’s SNEDECOR,G. W., AND COCHRAN,W. G. 1967. Sta- Storm-Petrels to increased food demand at the tistical methods. 6th ed. Iowa State Univ. Press, nest. Auk 104:750-756. Ames, IA. ROBY, D. D. 1991. Diet and postnatal energeticsin WATANUKI,Y. 1986. Moonlight avoidance behav- convergenttaxa of plankton-feedingseabirds. Auk iour in Leach’s Storm-Petrels as a defense against 108:131-146. Slaty-backedGulls. Auk 103:14-22. ROBY,D. D., AND R. E. RICKLEFS. 1986. Energy ex- WILLIAMS,G. C. 1966. Natural selection,the cost of penditure in adult least auklets and diving petrels reproduction,and a refinement of Lack’s principle. during the chick-rearingperiod. Physiol. Zool. 59: Am. Nat. 100:687-690. 661-678. YDENBERG.R. C. 1989. Growth-mortalitv trade-offs SLATER,P.J.B., AND E. P. SLATER. 1972. Behaviour and the evolution of juvenile life histories in the of the tystie duringfeeding of the young. Study Alcidae. Ecology 70: 1494-1506. 19:105-113.