Foraging Radii and Energetics of Least Auklets (Aethia pusilla) Breeding on Three Bering Sea Islands

We studied the relationship between the foraging radius and energy economy of least auklets (Aethia pusilla) breeding in colonies on three islands in the Bering Sea (St. Lawrence, St. Matthew, and St. George Islands). The distance to which auklets commuted on foraging trips varied by more than an order of magnitude (5-56 km), but mean field metabolic rate (FMR) did not vary significantly among birds from the three islands. These observations indicate that allocation to various compartments of time and energy budgets is flexible and suggest that least auklets may have a preferred level of daily energy expenditure that is similar across colonies. We modeled the partitioning of energy to various activities and hypothesize that the added cost of commuting incurred by auklets from St. Lawrence Island (foraging radius, 56 km) was offset by reduced energy costs while foraging at sea. Data on bird diets and prey abundances indicated that auklets from St. Lawrence Island fed on larger, more energy-rich copepods than did auklets from St. Matthew Island (foraging radius, 5 km) but that depth-averaged prey density did not differ significantly between the birds' principal foraging areas. However, previous studies have indicated that zooplankton abundance is vertically compressed into near-surface layers in stratified waters off St. Lawrence Island, which suggests that variation in foraging eficiency may have resulted in part from differences in the accessibility of food rather than its absolute abundance in the water column. Our findings demonstrate why long-distance commuting between colony and feeding areas need not significantly affect total daily adult energy expenditures, even in heavily wing-loaded species such as auklets. Interpretation of the energetic ecology of auklets and other seabirds can strongly benefit from information regarding foraging distributions, diets, and prey accessibility.


Introduction
Breeding seabirds commute to sea on foraging trips that vary greatly in distance and duration. Foraging radii observed in different species span more than four orders of magnitude, from <200 m to >2,000 km (Brown 1976;Kuletz 1983 Weimerskirch et al. 1993;Jouventin et al. 1994). In some species, individuals feed at widely varying distances on different days and even on consecutive foraging bouts (see, e.g., Anderson and Ricklefs 1987; Wanless, Harris, and Morris 1990). Whereas theoretical models can be used to predict the maximal foraging radius of a species (Andersson 1978; Pennycuick et al. 1984;Flint 1991), they provide little insight into the ecological causes or energetic consequences of intraspecific variation in commute distance. Because of the high cost of flight, these consequences are expected to be substantial, yet few empirical studies have attempted to examine the relationships between foraging radius and energy expenditure.
We studied the energetic ecology of least auklets (Aethiapusilla) breeding on different islands in the Bering Sea. These smallest members of the alcid family are planktivorous divers that specialize on copepods during the summer breeding season (B~dard 1969; Springer and Roseneau 1985; ). Auklets have a specialized flight morphology associated with their subsurface foraging mode and use muscle-powered (nongliding) flight exclusively while commuting between colonies and pelagic foraging sites. Their short wings, which are adapted for underwater propulsion, result in high wing-disc loading and poor fuel economy in flight (Pennycuick 1987). Although long-range foraging by auklets is therefore unexpected, radically different spatial distributions of foraging birds have been observed around different colonies. In the southern and central Bering Sea, auklets typically forage just offshore island colonies (Hunt et al. 1978). In contrast, birds breeding on islands in the northern Bering Sea often forage at sites more than 50 km offshore Hunt, Harrison, and Cooney 1990). (fig. 1). Our objectives here are (1) to compare field metabolic rates (FMRs) of birds with commute distances varying by more than an order of magnitude, (2) to deduce patterns of energy use and acquisition by integrating FMR data with observed foraging radii and other behavioral information in a simple model, and (3) to use insights derived from the model to explore the ecological impacts of physical and biological differences among the marine environments of the three islands.

Field Procedures
Rates of energy utilization were measured in adult least auklets by means of the doubly labeled water method (Lifson and McClintock 1966;Nagy 1983). At both St. Lawrence and St. Matthew Island colonies, adults were captured by stretching mist nets flat over regions of scree under which nests were located. Birds were trapped in the nets as they emerged from the nests to depart for sea or as they attempted to deliver meals to their chicks after Forty birds with chicks were given intrapectoral (muscle) injections of 0.3144 mL doubly labeled water (3HH180). Isotope enrichments in the injection solution were 0.33 mCi 3H per milliliter and 97 atom % 18O. Injected birds were held for 1 h in individual cloth bags to allow the injected isotope to equilibrate with the body water. After 1 h, two 70-L.tL blood samples were taken by puncturing the brachial vein and filling heparinized capillary tubes. The capillary tubes were sealed with Critocaps and stored in a shaded area (40C) for up to 2 d.
After initial blood samples were taken, birds were released near the point of capture. In all but one case, the injected birds flew immediately out to sea. We checked nets every 30-60 min throughout the day. When an injected bird was found again in the net, it was removed and reweighed, and a second blood sample was taken and processed as above. The bird was again released near the point of capture. In several cases, injected individuals were captured a third time, and a third blood sample was taken.
Adults (injected and noninjected) often regurgitated food loads from their sublingual pouch upon hitting the net. We routinely scraped these regurgitations from the rocks under the nets. In addition, food samples were obtained from auklets collected at sea at locations where large numbers of birds were observed. All food samples were preserved individually in 85% ethanol for later identification.
To determine the foraging distributions of least auklets at sea, we conducted counts from the bridge of the research vessel Alpha Helix while the ship was underway on transects radiating from each island. All birds observed on the water within a 300-m arc from directly ahead of the ship to 900 off the beam were counted, and data were entered directly into a handheld microcomputer.
Zooplankton abundance and community structure were determined by conducting vertical tows of a 1-m diameter, 505-jm mesh plankton net when the boat stopped at sampling stations. The net was towed from a position 3-4 m above sea bottom to the surface, with mean tow depths of 32 m (St. Matthew Island) and 29 m (St. Lawrence Island).

Analysis of Samples and Calculations
Blood samples were analyzed in the laboratory of K. A. Nagy at the Laboratory of Biomedical and Environmental Sciences, UCLA. Water was removed from the blood samples with a vacuum microdistillation technique. Tritium concentration in each water sample was measured in duplicate by adding 10 gL of the distillate to 7 mL of scintillant and counting 3H activity on a liquid scintillation counter until the coefficient of variation was 0.7%. Oxygen-18 concentrations were measured with the proton activation method (Wood et al. 1975;Nagy 1983) and subsequent counting of 18F with a gamma counter. Oxygen-18 samples were counted in triplicate for each blood sample until coefficients of variation were <1.0% for each group of triplicates. Tritium and 180 levels were corrected for background by measuring 3H activity and the 1"O concentration in blood samples taken from six uninjected birds (four from St. Lawrence Island and two from St. Matthew Island). The means of these background values were subtracted from the isotope levels measured in each of the samples from injected birds before proceeding with calculations of CO2 production.
Total body water volume was determined from the 180 levels in initial blood samples via the isotope dilution space method (Nagy 1983). Rates of CO2 production were calculated from initial and final isotope levels and water fraction via equation (2) in Roby and Ricklefs (1986). Field metabolic rate was calculated from CO2 production by assuming an energy equivalent of 26.8 J (mL CO2)-1 (Roby and Ricklefs 1986). For birds that were recaptured twice (n = 3), we calculated an overall time-averaged FMR using the data from the two successive measurement intervals.
Because auklets were recaptured opportunistically, the measurement pe- Food samples were identified and quantified by microscopic examination in the laboratory. The relative importance of different prey species was assessed by counting the total number of prey items in each food sample and computing the percentage of each prey type represented in each sample. Our identifications were completed prior to the description of Neocalanus flemingeri, which is very similar to the better-known Neocalanusplumchrus (Miller 1988;Miller and Clemons 1988). Our taxon N. plumchrus therefore includes an unknown proportion of N. flemingeri.

Statistical Analyses
Bird survey data from the radial transects were reduced by constructing cumulative frequency distributions (CFDs) of the distance of foraging auklets from each island. We used pairwise two-sample Kolmogorov-Smirnov tests (Sokal and Rohlf 1981) to look for differences in the at-sea distribution of foraging birds around the three colonies. The null hypothesis for these tests was that the maximum difference between the CFDs of birds from any two islands was no greater than might be expected by chance alone. In a strict sense, the assumptions underlying this test (or any other statistical test) are not rigorously met because transect observations do not provide random samples of birds. Venrick (1986) conducted Monte Carlo studies and showed that when individuals from a population are not sampled randomly, the resulting distribution of the Kolmogorov-Smirnov statistic is dependent on the spatial heterogeneity of the population. Specifically, Type I error rates increase with increasing spatial heterogeneity. In our study, the potential for Type I error is reduced because we apply the Kolmogorov-Smirnov test to CFDs assembled from multiple radial transects from each island (i.e., determinations of foraging distributions of birds are based on extensive sampling effort in both dimensions of two-dimensional horizontal space). Because we conducted nonorthogonal comparisons among the islands, we employed Bonferroni adjustments of significance levels to protect against Type I error from multiple tests.
Copepod densities around St. Matthew and St. Lawrence Islands were compared by means of a t-test. Food samples and regurgitations were analyzed with Mann-Whitney U-tests because these data were not normally distributed. The FMR data were analyzed with a parametric ANOVA. Because of the relatively small sample sizes generated by the doubly labeled water study, we chose a significance level of P = 0.1 to reject the null hypothesis of homogeneity of mean FMRs among the three islands. We estimated the power of this statistical test by generating 100 samples of normally distributed data with means, variances, and sample sizes as in our study and then count-ing the number of times (out of 100) that the null hypothesis would be correctly rejected. In contrast, C marshallae was absent from food samples from St. Lawrence Island, where Neocalanus plumchrus and Neocalanus cristatus were im- P < 0.0001). Reasons for this difference are not known with certainty, but the stomach samples generally contained few identifiable prey items, whereas regurgitations contained up to several thousand. Furthermore, because of their small size and high surface-to-volume ratios, copepods are probably subject to more rapid digestion than are any larger prey items that may occasionally be taken, thereby deflating the apparent importance of copepods in food samples obtained from stomachs of birds at sea. In contrast, food loads in sublingual pouches, which were being carried to the chicks, were generally regurgitated intact and did not suffer from this potential source of bias. For these reasons, we feel that the data from regurgitations reflect the typical composition of auklet diets more accurately and precisely than do the data from stomach samples. Vertical plankton tows conducted within 10 km of St. Matthew Island indicated that densities of C. marshallae averaged 50.9 m-3 (+ 12.0 SE, n = 12 tows). Offshore St. Lawrence Island, densities of the two larger copepods varied with distance from the island (fig. 3) We measured the sizes of the three principal copepod prey species represented in food samples from least auklets. Mean lengths of C. marshallae, N. plumchrus, and N. cristatus were 3.9 mm, 4.8 mm, and 9.5 mm (n = 50 each), with no overlap.

Field Metabolism
Field metabolic rates of birds from the three colonies are shown in table 3. Variances were homogeneous among the islands (Bartlett's test, X2 = 2.2, df = 2, P> 0.1). A parametric ANOVA did not detect a significant difference among birds from the three colonies (F2,31 = 0.25, P> 0.75). However, the power of this test, estimated via simulation, was only 32%. In other words, we had only a 32% chance of rejecting the null hypothesis if it was in fact false, given the sample sizes and variances observed. The difference between mean FMRs from any two islands would have had to exceed approximately 40-50 kJ d-1 in order for us to have been able to judge these means to be significantly different using Fisher's least significant difference (LSD) procedure. (table 3), but rates of body mass change did not differ among the three islands (ANOVA, F2,32 = 2.05, P> 0.1). Zero fell within 95% confidence intervals constructed around mean rates of mass change for each island, which suggests that auklets were in a state of energy balance during the course of the study.

Energy Economy ofBreeding Auklets
We can use data from this study and information gleaned from the literature to reconstruct time and energy budgets of least auklets rearing chicks at the three island colonies. Birds breeding on all three islands spend most of the day at sea but spend the night inactive at the colony (B~dard 1969; Roby and Brink 1986; B. S. Obst, R. W. Russell, G. L. Hunt, Jr., Z. A. Eppley, and N. M. Harrison, personal observation). We assume that energetic costs during periods of nocturnal inactivity are close to BMR. We neglect possible thermoregulatory costs because there is no information on the microclimate in the nest crevices; however, we suspect that ambient temperature in a crevice may remain within the auklets' thermoneutral zone.
During the day, the birds forage at sea and only return to the colony briefly to drop off chick meals. An auklet's daily time budget is therefore divided principally between nocturnal inactivity, commuting between the colony  and foraging sites, resting on the sea surface, and diving for food. We denote these respective time allocations by Tn, T,, Tr, and Td (h d-') and the associated energetic costs by En, Et, Er, and Ed (kJ h-1). Field metabolic rate can then be expressed as FMR = EnTn +EtTt + ErTr t+ EdTd. (1) As explained earlier, we estimated Tn to be 7 (2)

We took Vand Et to correspond to the maximum-range speed (Vmr) predicted from Pennycuick's (1975, 1989) flight mechanics theory. Both Vm1, and E, are dependent on R and C because (1) the length of a commute affects body mass, which in turn affects flight speed and flight cost, and (2) the number of commutes determines the payload mass that must be transported during each return trip, which also affects flight speed and cost (Pennycuick 1975, 1989). We calculated Vm, and Et with the software in Pennycuick (1989). Wingspan, body mass, and payload mass are important determinants of flight performance. The mean wingspan of 15 auklets measured on St. George Island was 0.34 m (Roby and Ricklefs 1986). Mean body masses from the three islands were assumed (table 3). The payload mass transported during each return trip is a function of daily food consumption by chicks and the number of trips undertaken to meet this demand.
It is not known whether chick provisioning rates varied among colonies, but we assume here that they did not. Food requirements of auklet chicks on St. George Island averaged 13.9 g per parent per day (Roby and Brink 1986). We assume that the digestion of meals consumed by adult auklets occurs while the birds are at sea. Payload mass is therefore calculated as 13.9 g divided by the number of return trips made by each parent per day. The predicted flight speeds and costs that we calculated using Program 1 in Pennycuick (1989)   because of the nearly identical foraging distributions of auklets from the two islands ( fig. 2). Our analyses indicated that it would be impossible for auklets from St. Lawrence Island to have completed more than two daily foraging trips to the median foraging distance, given the observed energy expenditures (table 3)

Islands may have made only two trips (since the mean chick-meal delivery rate on St. George Island was 2.6 d-'; Roby and Brink 1986), the reconstructed budgets for these islands are relatively insensitive to the number of trips assumed because of the small flight distances involved.
Finally, it is of interest to assess the foraging performance of auklets on each of the islands. We will characterize foraging performance in terms of the proportion of time at sea spent underwater and the rate of energy intake while underwater. Both of these values are dependent on Td. To estimate Td, information on Er and Ed is required. Unfortunately, no information is available on these energetic costs in small alcids. In a laboratory study of larger alcids (murres, Uria spp., body mass ca. 1 kg), absorptive resting metabolic rates on water in the thermoneutral zone exceeded BMRs by about 50% (Croll and McLaren 1993). Water temperatures in our study areas were below the lower critical temperature for the much smaller least auklets (10"-150C; Roby and Ricklefs 1986), so thermoregulatory contributions to Er were probably significant. We assumed a value of 2 X BMR for Er.

The cost of underwater locomotion is also poorly understood. Laboratory studies of penguins (Butler and Woakes 1984; Baudinette and Gill 1985; Culik and Wilson 1991; Kooyman et al. 1992), murres (Croll and McLaren 1993), and ducks (Woakes and Butler 1983;
Stephenson 1994) have yielded widely varying measurements in the range 1-5 X BMR. In contrast, field studies of free-living penguins have suggested higher figures (5-10 X BMR) that seem generally comparable to the cost of flapping flight (Nagy, Siegfried, and Wilson 1984;Gales and Green 1990;Chappell et al. 1993). Because of the wide disparity among results from these different studies, we did not attempt to estimate Ed and instead performed a sensitivity analysis to exarline the dependence of estimated measures of foraging performance on the value of Ed assumed. After substituting Tr = 24 -Tt -Td into equation (1) and then rearranging, we have

Recalling our assumptions about the values of En (= BMR), Er (= 2 X BMR), Et (table 4), and time budgets (fig
where Yd denotes the gross rate of energy acquisition while actively diving for food (kJ h-1). In equation (4), the expression on the left represents daily gross energy acquisition, and the expression on the right represents daily gross energy demands. We assume that these two values are equal because our data did not indicate any significant trends in body mass change. Further support for the assumption of energy balance is provided in a recent study by Jones (1994), who found that breeding least auklets undergo an abrupt, programmed loss of mass at the time when the chicks hatch but that body mass then remains stable during the chick-rearing period.

Rates of energy intake while underwater can now be calculated for auklets foraging around each of the three colonies by means of equation (4), the FMR data (table 3), and the estimates of Td Obtained from equation (3).
Estimates of Yd were highly sensitive to the cost of underwater locomotion assumed (fig. 5, bottom). However, this analysis suggests that rate of energy intake by St. Lawrence Island birds was approximately four times the rate realized by birds from the other two islands regardless of the actual cost of underwater locomotion (fig. 5, bottom).

Discussion
Despite a greater than 10-fold difference in mean foraging radius, mean energy expenditures differed by less than 6% among auklets from the three colonies. It should be noted, however, that although a statistical test did not indicate a significant difference, the Type II error rate was high because of our small sample sizes. A simulation showed that even if the null hypothesis was false, we would fail to reject it 68% of the time, given our sample sizes and variances. Nevertheless, we would argue that the observed differences were ecologically unimportant, even if we had been able to declare them statistically significant. If total energy expenditures were directly related to the proportion of time allocated to energetically costly flight, as might be expected, then FMRs of auklets from St. Lawrence Island should have been much higher than those of birds from the other two islands. This was not the case, and, in fact, mean FMR tended to be highest in auklets breeding on St. George Island, where the birds foraged just offshore. Despite their much greater allocation of time to energetically costly commuting flight, mean daily energy expenditures of St. Lawrence Island auklets were actually lower than those of St. George Island birds, and only 5% higher than those of St. Matthew Island birds. These results suggest that even in heavily wing-loaded seabirds such as auklets, long-distance commuting between colony and feeding areas need not significantly change total adult energy expenditures.
Although we did not study the compensatory mechanism directly, our model indicates that the added cost of commuting for auklets from St. Lawrence Island was probably offset by a reduction in total underwater foraging costs resulting from greater foraging efficiency. The difference between estimated rates of energy intake by auklets from St. Lawrence Island and auklets from the other islands was approximately fourfold. What factors might have been responsible for such a large difference in foraging efficiency?
Differences in rates of energy intake while underwater could be due to differences in prey abundance, the energy content of prey, and/or the distribution of prey in the water column. Our plankton studies did not indicate a significant difference in overall prey abundance between St. Matthew and St. Lawrence Islands, so this explanation seems unlikely. However, the taxonomic composition of the copepod fauna differed radically between the islands. Large, energy-rich Neocalanus copepods were absent from the middle shelf domain waters near St. Matthew Island. Collections at sea and regurgitations from netted birds showed that the dominant prey of St. Matthew Island auklets was C. marshallae, which, at about 4 mm, is smaller than the typical prey of auklets from St. Lawrence Island (Harrison 1987;). The difference in energy content between C. marshallae and Neocalanus spp. is even greater than suggested by the differences in length because the former species is a slighter copepod that stores less lipid (S. Smith, personal communication in Hunt and ). Thus, although the numerical density of copepods did not differ significantly between the islands, differences in the energy content of individual prey items almost certainly resulted in differences in overall energy density in the water column. Higher rates of energy intake by St. Lawrence Island auklets therefore probably resulted partly from their ability to exploit more energy-rich prey.
Although mean copepod densities did not differ between the auklets' foraging areas around St. Lawrence and St. Matthew Islands, the availability of a food resource to a consumer depends on both the absolute abundance of the food in the environment and its accessibility. Food accessibility can be defined as the relative difficulty of reaching a food item (Moermond 1990), and it is very likely that differences between the marine environments of the different islands have important impacts on food accessibility.
At about 80 g, least auklets have an estimated maximum diving depth of 15 m (calculated with the regression equation for alcids in Burger [1991]). The depth-averaged levels of copepod density reported here are based on net tows conducted from a depth of about 30 m, so only about half of the water column sampled by our plankton nets was accessible to the auklets. Although vertical heterogeneity in prey density was not quantified by our plankton tows, previous studies in the region demonstrated an important source of such heterogeneity. North of St. Lawrence Island, the Anadyr current forms a bottom layer, intruding eastward under Bering Shelf water. Density differences between these two water masses are large, and a strong, stable, and shallow pycnocline is often present offshore. Least auklets from St. Lawrence Island did not feed near shore, where waters were vertically mixed, but were instead consistently concentrated over the stably stratified water offshore . Copepods and other zooplankton are known to accumulate at interfaces between water layers (Harder 1968; Barroclough, Lebrasseur, and Kennedy 1969), and Hunt et al. (1990; see their figs. 6, 7, 9, 12, 15) presented acoustic evidence of dense plankton layers at and above the pycnocline (ca. 10-15 m depth) off St. Lawrence Island. They hypothesized that auklets preferentially foraged over the offshore, stratified water because prey availability was enhanced there. Foraging over a shallow, stable pycnocline should result in higher food accessibility and lower search and pursuit costs because (1) the presence of near-surface food layers is spatially predictable, (2) diving depth required to reach these shallow food layers is reduced, and (3) prey is concentrated in these layers (i.e., the mean distance between prey items is smaller).
In contrast to St. Lawrence Island, St. Matthew Island is located in the middle domain of the Bering Sea Shelf (Springer and Roseneau 1985). The only area where we encountered large numbers of foraging auklets was in Sarichef Strait off the northwest end of the island, where copepods may have been concentrated to some extent by a tidal rip over a shallow sill. We found no evidence of auklets foraging over the near-surface scattering layers offshore St. Matthew Island, which were dominated by jellyfish (K. O. Coyle et al., unpublished data). We therefore conclude that differences in both the energy density of prey and the accessibility of prey in the waters surrounding St. Lawrence and St. Matthew Islands are likely sources of variation in the foraging efficiency of least auklets.
Although our assumptions and predictions seem plausible in light of available information, several caveats should be noted. First, we have assumed that when departing from their colonies, auklets follow straight flight paths to foraging sites and that after arriving at these sites, the birds do not again engage in flight until the return trip is initiated, when they return directly to the colony. However, the pelagic distributions of auklets show intriguing patterns of spatial autocorrelation across a wide range of scales, which may derive from multiscale searching behavior at sea (Russell et al. 1992). If auklets employ such multiscale foraging strategies, then our estimates of time and energy allocations to flight are probably conservatively biased. Second, we used a theoretical model (Pennycuick 1975(Pennycuick , 1989) to estimate flight speeds and costs. If the theory is wrong, or if auklets fly at speeds other than V,-during the chick-rearing period (cf. Norberg 1981; McLaughlin and Montgomerie 1985), then our calculations are obviously inaccurate. Third, we had no information on the energetic cost of resting on the water, and we assumed that this cost was 2 X BMR on the basis of studies of other, larger species. Finally, we assumed that the amount of food delivered to chicks did not vary among islands. Recent studies of other alcid species lend some support to this assumption. In particular, Burger and Piatt (1990) found that rates at which common murres fed their chicks did not vary significantly among years, even though the abundance of capelin (the murres' principal prey) varied up to 10-fold within and between years.
The similarity of mean energy expenditures among least auklets in such diverse ecological situations is striking and raises the question of whether observed FMRs were at or near some biologically important benchmark level. Drent and Daan (1980) postulated the existence of a "general energetic upper level for sustained work" of 4.0 X BMR in parent birds feeding their young. Peterson, Nagy, and Diamond (1990) termed the ratio of sustained FMR to BMR "sustained metabolic scope." Compilations of published data on breeding seabirds have shown that although sustained metabolic scope does indeed tend to cluster around 3 or 4, it nevertheless varies considerably (1.6-6.6) among different species (Birt-Friesen et al. 1989; Mehlum, Gabrielsen, and Nagy 1993). Although the 4.0 X BMR hypothesis can now be rejected as a general rule (Peterson et al. 1990;Koteja 1991), few workers have considered the alternative possibility that breeding birds may have a species-specific preferred level of energy expenditure that is relatively invariant, even across an array of widely varying environments (but see Montevecchi, Birt-Friesen, and Cairns 1992).
In least auklets, the mean sustained metabolic scope of birds breeding on different islands in the Bering Sea showed little variation, ranging from 2.94 on St. Matthew Island to 3.11 on St. George Island. Despite this ho-mogeneity of mean daily energetic expenditures, allocations to various compartments of time and energy budgets (e.g., commuting vs. foraging) appeared to be highly flexible and responsive to variation in the marine environment. These observations suggest that least auklets may have a preferred level of daily energy expenditure of about 3 X BMR, although the high levels of variation in FMR indicate that this level is below any upper ceiling on FMR that may exist for the species.
Conclusions concerning the existence of a characteristic species-specific preferred level of energy expenditure will have to await a more robust field sampling program. Our FMR sample sizes were very low, as is often the case in doubly labeled water studies. Furthermore, logistical constraints prohibited at-sea observations of the behavior of the specific birds for which FMR was measured, and some of our conclusions are necessarily contingent upon the assumption that the study birds employed "typical" auklet behavior (i.e., that they foraged at the midpoint of the distribution of birds away from the colony). It is important to stress that while mean FMRs showed remarkably little variation among colonies, measured FMRs were quite variable within colonies, which suggests that our assumptions concerning typical auklet behavior were probably oversimplified. Obtaining simultaneous FMR measurements and activity budgets for individual seabirds remains a difficult challenge for the future--although not an insurmountable one, in light of the recent successes with satellite telemetry (Weimerskirch et al. 1993;Jouventin et al. 1994).
Our findings indicate that interpretation of the energetic ecology of seabirds can benefit significantly from information regarding foraging distributions, diet, and prey availability. However, because investment in chicks is a potential energy "sink" that was not accounted for in our studies, we are unable to assess the selective impact of intercolony variation in foraging radius. The effects of variable commute distances on chick provisioning rates and other aspects of auklet reproductive performance such as growth rates and fledging success are presently unknown and should be explored in future studies.