We studied Laysan and black-footed albatrosses at Tern Island (23.87°N, 166.28°W), French Frigate Shoals, Northwest Hawaiian Islands during the 2002–2003, 2004–2005, and 2005–2006 breeding seasons, and Indian yellow-nosed albatrosses at Amsterdam Island (37.86°S, 77.52°E), Southern Indian Ocean, during the 2006-07 breeding season. We used satellite telemetry to determine at-sea locations of foraging albatrosses during late incubation and brooding to characterize differences in behavior between species and reproductive stages. During brooding, we also measured field metabolic rates of tracked albatrosses using the doubly labeled water technique (Lifson and McClintock, 1966; Nagy, 1980; Speakman, 1997).

Seventy-five adult albatrosses were equipped with satellite platform terminal transmitters (30 g Pico-100, Microwave Telemetry, Columbia, MD; and 35 g SPOT4, Wildlife Computers, Redmond, WA) during late incubation (within 2 weeks of hatch date; 10 Laysan, 11 black-footed, 11 yellow-nosed albatrosses) and brooding (15 Laysan, 13 black-footed, 15 yellow-nosed albatrosses). Satellite tags were attached to dorsal feathers with adhesive tape (tesa®, Hamburg, Germany), and satellite transmissions were downloaded via the Argos satellite system (Service Argos, Inc., Largo, MD). Individuals were also equipped with temperature recorders (10 g Lotek LTD 2400 and 1100, Lotek Wireless, St. John's, Newfoundland) attached to a plastic leg band so that temperature recordings (±0.05°C) could be used to characterize activity patterns while at sea (Wilson et al., 1995). Foraging activity is only presented for the brooding period, when high-resolution (12 s) temperature records were available. In all cases, total mass of deployed devices was <2% of bird body mass, which is under the recommended limit for albatross tracking studies (Phillips et al., 2003). Sex was determined from blood samples (Fridolfsson and Ellegren, 1999) for all individuals tracked. All protocols employed in this study were approved by the Institutional Animal Care and Use Committees, University of California Santa Cruz.

We delimited albatross foraging tracks based on visual observations of departure and arrival times from twice-daily nest checks during incubation, and hourly nest checks from dawn to dusk during brooding. To remove unlikely Argos locations, tracks were filtered using the Iknos Toolkit (Y. Tremblay, unpublished program) for Matlab (The MathWorks, Natick, MA), following Kappes et al. (2010). First, a speed filter of 80 km h⁻¹ was applied to transit rates between successive locations (following Hyrenbach et al., 2002; Suryan et al., 2006) to remove unrealistic flight speeds (Spear and Ainley, 1997). Next, the maximum change in azimuth was set to 170° to remove track spikes between successive locations that are likely to be erroneous (Keating, 1994; Freitas et al., 2008). Finally, to avoid errors in transit rate determination (Hays et al., 2001), the minimum time between successive fixes was set to 10 min.

We calculated maximum distance traveled from the colony using great-circle distances to account for the earth's curvature. We divided great-circle distances between off-colony Argos locations by the time between successive locations to calculate average transit rate. To characterize albatross foraging activity patterns, we determined the proportion of time spent in flight and the frequency of landings on the sea surface. Landing rates are indicative of feeding effort (Weimerskirch et al., 2000; Shaffer et al., 2001a) because albatrosses must land on the sea surface in order to consume prey (Conners et al., 2015). Previous research has demonstrated that take-offs and landings are the most energetically demanding activities albatrosses engage in at sea (Weimerskirch et al., 2000), and landing rates of wandering albatrosses (Diomedea exulans) are correlated with field metabolic rates (Shaffer et al., 2001a). We used an algorithm (Iknos toolkit for Matlab; Y. Tremblay, unpublished program; Kappes et al., 2015) to identify landings based on rapid changes in temperature, and stable periods associated with sitting on the sea surface (Wilson et al., 1995), for those individuals equipped with temperature records. We defined daylight hours based on civil twilight (sun no more than 6° below the horizon) using NOAA's solar calculator in the maptools package in R (Lewin-Koh and Bivand, 2010) and temporally-matching to tracking locations.

Doubly labeled water was used to determine field metabolic rates (FMR) of Laysan, black-footed, and yellow-nosed albatrosses at sea (FMR_at−sea) and at the nest (FMR_on−nest) during the brooding period (Lifson and McClintock, 1966; Nagy, 1980; Speakman, 1997). Fifteen birds of each species were captured at the nest immediately following a mate switch, and an initial blood sample (0.5–3.5 ml) was collected from a vein on the tarsus. Albatrosses were given an intraperitoneal injection of 1.6–1.9 ml sterile water containing 0.9% NaCl, and either 34.5 atom percentage oxygen-18 and 35.9 atom percentage deuterium (Laysan and black-footed albatrosses), or 29.8 atom percentage oxygen-18 and 5.0 Mbq g⁻¹ of tritiated water (yellow-nosed albatrosses). Mass of injectate (±0.01 g) was determined by weighing the syringe before and after injection using a portable field balance (Ohaus Corp., Pine Brook, NJ). Each bird was weighed to the nearest 50 g using either a spring-loaded Pesola (Pesola AG, Baar, Switzerland) or Salter scale (Salter Weightronix Ltd, West Bromwich, UK) and then placed in a box or holding pen; isotopes were allowed to equilibrate for ~90 min (Shaffer et al., 2001b) before a second blood sample was collected. Three Laysan, two black-footed, and three yellow-nosed albatrosses were held for 3 h to help ensure equilibration when there was evidence that injections may have been made into the gastrointestinal cavity or cutaneous fat. All individuals were equipped with satellite tags and/or temperature recorders and released at the nest. After completion of a foraging trip, each bird was recaptured and a third blood sample was collected, within 2–3 h of returning to the colony. Satellite tags and temperature recorders were then removed and final body mass was measured. To determine FMR_on−nest, three Laysan, four black-footed, and three yellow-nosed albatrosses were subsequently captured after 2 days at the nest and a fourth blood sample was collected from these individuals. In one instance, a Laysan albatross did not depart to sea after release, but instead switched with its mate again and remained on the nest; for this individual only FMR_on−nest was calculated. This double-switching behavior was also observed at Laysan and black-footed albatross nests checked daily to determine attendance patterns of control individuals.

Due to equipment limitations, one black-footed albatross was equipped with a temperature recorder only, two yellow-nosed albatrosses were equipped with satellite tags only, and one black-footed albatross was equipped with a GPS tag (TechnoSmart, Rome, Italy) rather than a satellite tag; in this case, the GPS record demonstrated that the individual stayed near Tern Island overnight and then returned to the nest (short departures were also observed in control pairs, with both members of the pair remaining at the nest for several days in some cases). As this was not representative of FMR_at−sea or FMR_on−nest, the metabolic rate measured for this individual is not included in the subsequent analyses.

All blood samples were collected with a syringe and 21–25 gauge needle, transferred to a vacutainer (B-D brand with spray-coated lithium heparin, Beckton-Dickinson, Franklin Lakes, NJ) and stored in a cooler with cold packs until centrifugation on the day of collection. Plasma was transferred to 2 ml cryogenic plastic screw cap vials (with silicon O-rings; Corning Inc., Corning, NY) and frozen until isotopic analyses were performed. Aliquots of water distilled from plasma samples (following Ortiz et al., 1978) were then used to determine specific activity of deuterium by laser-absorption spectroscopy (University of California Davis, Davis, CA) or tritium by scintillation spectrometry (LS 6500, Beckman Coulter Inc., Fullterton, CA) in triplicate; specific activity of oxygen-18 was determined by mass ratio spectrometry (Metabolic Solutions, Nashua, NH).

The initial dilution space of oxygen-18 was used to calculate the volume of initial total body water. To calculate final body water volume, body mass at recapture was multiplied by the initial fractional water content, which has been validated in Shaffer et al. (2006). CO₂ production was calculated using equation 2 in Nagy (1980); this equation assumes that body mass of the animal changes linearly with time. For two yellow-nosed albatrosses and one Laysan albatross, FMR was determined using the single sample method described in Speakman (1997), because initial total body water calculations suggested isotopes were not fully equilibrated when the post-equilibration blood sample was collected. FMR (mL g⁻¹ h⁻¹) was converted to kJ (and W) by applying a conversion factor of 24.7 kJ = 1 L CO₂ following (Pettit et al., 1988), based on chemical composition of the diet of Laysan albatrosses (Harrison et al., 1983); this conversion factor was assumed to approximate CO₂ yield from the diet of black-footed and yellow-nosed albatrosses. Mass-specific FMR (W kg⁻¹) was determined by dividing FMR by mean body mass so that energy expenditure is more directly comparable among species. Because FMR calculations included time spent at the nest before departure and after arrival at the nest, measured FMR was corrected based on visual observations of departure and arrival times at the nest. Following methods of Costa and Prince (1987), FMR_at−sea was calculated as:

We compiled average FMR and body mass values for Hawaiian and yellow-nosed albatrosses with values from other studies (male and female wandering albatrosses (Shaffer et al., 2001a), shy albatrosses (Thalassarche cauta; Green and Brothers, 1995, Abstract from First International Albatross and Petrel Conference, Hobart, Australia), gray-headed albatrosses (T. chrysostoma; Costa and Prince, 1987), black-browed albatrosses (Thalassarche melanophrys; Shaffer et al., 2004), and Laysan albatrosses during incubation (Pettit et al., 1988) to provide a mechanism for comparing energy expenditure among species (Shaffer, 2011). Given large differences in mass between wandering albatrosses and other studied albatross species, we tested for the effect of genus Diomedea in the relationship between log-transformed FMR and body mass values.

Statistical analyses were implemented in the program R (R Development Core Team, 2010). We used ANOVA and Tukey multiple comparison tests (Hothorn et al., 2008) to investigate differences in foraging behavior between species and reproductive stages, and species differences in FMR, total body water, body mass, and foraging activity; Bonferroni-corrected P-values are presented for multiple comparisons. Trip characteristics, landing rates, time in flight, total body water, and FMR for each species did not differ significantly between the sexes, therefore males and females were grouped for all analyses. To meet normality assumptions, percent time in flight was arcsine transformed and landing rate was log transformed prior to analysis. Simple linear regression was used to investigate relationships between foraging behavior and FMR_at−sea for each species, and to examine the allometric relationship between log-transformed FMR and body mass for albatrosses from this and other studies. P-values reported are two-tailed, and the significance level was set as P = 0.05. All averages are reported as Mean ± SD.

Hawaiian albatrosses traveled significantly farther during incubation compared to brooding (more than four times as far on average; Table 1; P < 0.001 for pair-wise tests), whereas maximum foraging range did not differ between incubation and brooding for yellow-nosed albatrosses (P = 0.08). All three albatross species took foraging trips of longer duration during incubation compared to brooding (P < 0.01 for all pair-wise tests): Hawaiian albatrosses took trips that were more than three times longer, whereas yellow-nosed albatrosses took trips that were just under two times longer on average (Table 1). During the incubation period, the majority of Hawaiian albatrosses traveled north of the Tern Island colony to cooler waters of the North Pacific Transition Zone; during brooding, their movements were restricted to warmer waters of the subtropical gyre (Figures 1, 2). Yellow-nosed albatrosses foraged in similar thermal environments during the incubation and brooding periods (Figure 2). Black-footed albatrosses traveled more rapidly during incubation compared to brooding (P = 0.02); transit rates did not differ between breeding stages for Laysan and yellow-nosed albatrosses (Table 1).

During the incubation period, Hawaiian albatrosses traveled farther (over 700 km on average) and more rapidly (9 km h⁻¹ on average) than yellow-nosed albatrosses (P < 0.02 for pair-wise tests), but spent a similar amount of time at sea (between 7 and 10 days on average). During brooding, all three species traveled similar distances (~400 km) and durations (3–4 days). Laysan albatrosses traveled more rapidly than yellow-nosed albatrosses during brooding (8 km h⁻¹ on average; P = 0.004), but this was the only species difference in transit rate observed. Sea surface temperature along foraging tracks was most similar among species during incubation; Hawaiian albatrosses used warmer waters than yellow-nosed albatrosses during brooding (Figure 2).

Percent time in flight and landing rates did not differ between reproductive stages for any of the three species (Table 2). During incubation, Hawaiian albatrosses spent more time in flight at night (70–80%) compared to yellow-nosed albatrosses (~50%; Table 2; P < 0.01 for pair-wise tests). During brooding, Hawaiian albatrosses spent more time in flight both during the day (~90%) and at night (80–90%) compared to yellow-nosed albatrosses (~70% during day; ~40% at night; Table 2; P < 0.01 for pair-wise tests). Yellow-nosed albatrosses had higher overall and daytime landing rates than black-footed albatrosses during incubation (P < 0.001), but there were no species differences in overall or daytime landing rates during brooding. All three species demonstrated diel patterns in foraging activity (Table 2).

Body mass differed between the three species studied; black-footed albatrosses were significantly heavier than Laysan albatrosses (P < 0.001), which were significantly heavier than yellow-nosed albatrosses (P = 0.002; Table 3). Total body water (%) did not differ significantly between Laysan and black-footed albatrosses, but each species of Hawaiian albatross had a lower percentage of total body water compared to yellow-nosed albatrosses (P < 0.001 for pair-wise tests). On average, albatrosses gained mass during the foraging trip but the change in mass did not differ between species and was highly variable among individuals (Table 3).

Mean absolute FMR_at−sea (W) was greatest for black-footed albatrosses (P = 0.005, Laysan pair-wise test; P < 0.001, yellow-nosed pair-wise test), and similar for Laysan and yellow-nosed albatrosses (Table 3). Mean absolute FMR_on−nest (W) did not differ between albatross species but sample sizes were low for this parameter (Table 3). Mass-specific FMR_at−sea (W kg⁻¹) and FMR_on−nest (W kg⁻¹) did not differ between species. The ratio of FMR_at−sea to FMR_on−nest was lowest for black-footed albatrosses (1.9), and similar for Laysan (2.6) and yellow-nosed albatrosses (2.4).

For each species, we investigated the relationship between at-sea behavior and mass-specific field metabolic rates, however, FMR_at−sea (W kg⁻¹) was not statistically related to foraging range, trip duration, transit rates, the percent time in flight, the number or frequency of landings, change in mass, or water influx rates.

We plotted the allometric relationship between FMR and body mass for albatrosses from this and other doubly labeled water studies to provide a mechanism for appropriately comparing energy expenditure among species (Figure 3; Shaffer, 2011). We found a significant effect of the genus Diomedea on the relationship between log-transformed FMR and body mass [β = 0.60, t₍₆₎ = 3.19, P = 0.02], therefore our discussion focuses on the allometric relationship of the smaller albatross species (solid line; Figure 3) rather than the line for all studied species (dashed line; Figure 3). We found that Hawaiian albatrosses during brooding fall below the regression line for smaller albatross species, whereas yellow-nosed albatrosses fall above this line (Figure 3).

During the incubation period, all three albatross species foraged within convergence zones between warm subtropical waters and cool subpolar waters (Figure 1) where productivity is regionally enhanced (Lutjeharms and Valentine, 1984; Barange et al., 1998; Read et al., 2000; Olson, 2001; Polovina et al., 2001) and albatross prey resources are aggregated (Harrison et al., 1983; Gong et al., 1993; Yatsu et al., 1993; Pearcy et al., 1996; Pinaud et al., 2005; Conners, 2015). To reach these habitats, Laysan and black-footed albatrosses traveled significantly farther than yellow-nosed albatrosses and spent more time at sea. During brooding, all species reduced time at sea and Hawaiian albatrosses retracted their foraging ranges to the warm, oligotrophic environment close to the breeding colony (Figures 1, 2), where prey abundance is likely lower (Ashmole, 1971; Ballance et al., 1997). Conversely, yellow-nosed albatrosses were able to forage in a similar thermal environment during both reproductive stages, despite restricted movements during brooding (Figures 1, 2).

Hawaiian albatrosses spent more time in flight than yellow-nosed albatrosses, especially at night and during the brooding period. This is in agreement with our hypothesis, as we expected that spending more time in flight would be suitable for foraging in a low-productivity environment with unpredictable prey resources (Weimerskirch et al., 2005). Reliance on flight to travel between dispersed prey patches is seen in other tropical seabirds where greater flight proficiency is associated with lower productivity habitats (Ballance et al., 1997). Compared to other albatross species that forage in more productive marine habitats such as subtropical and polar convergences, continental shelf-breaks and slopes, and coastal upwelling zones (Tickell, 2000; BirdLife International, 2004), yellow-nosed albatrosses spent a similar proportion of time in flight (44–69%; Weimerskirch and Guionnet, 2002; Phalan et al., 2007). Contrary to our hypothesis, we did not detect a consistent difference in overall landing rates between Hawaiian albatrosses and yellow-nosed albatrosses during brooding, despite contrasting foraging environments. Landing rates during brooding were similar between species during the day, but higher for Laysan albatrosses at night. Higher landing rates at night for Laysan albatrosses may be related to nocturnal feeding in this species (Harrison et al., 1983; Conners et al., 2015). The tendency of all species to forage during daylight hours is supported by previous research (Fernández and Anderson, 2000; Weimerskirch and Guionnet, 2002; Kappes et al., 2015).

Brooding Laysan and black-footed albatrosses did not differ in terms of body composition; however, both Hawaiian albatrosses demonstrated significantly lower total body water when compared to yellow-nosed albatrosses. Lower total body water suggests that Hawaiian albatrosses have comparatively greater lipid reserves (Reilly and Fedak, 1990; Ellis and Jehl, 1991; Groscolas et al., 1991), which may be related to foraging in a low-productivity environment. If foraging conditions are poor during brooding, adults may rely on lipid body stores obtained during the incubation period for self-maintenance and then allocate food resources acquired at sea during the brooding period to rapidly-growing chicks (Weimerskirch and Lys, 2000).

Mean absolute FMR_at−sea was greatest for black-footed albatrosses, and similar for Laysan and yellow-nosed albatrosses. Greater absolute FMR_at−sea in black-footed albatrosses can be explained by larger body size (Nagy, 2005) and higher wing loading in this species (Suryan et al., 2008). Contrary to our prediction, Laysan and black-footed albatrosses did not exhibit lower mass-specific FMR_at−sea compared to yellow-nosed albatrosses. We expected that Hawaiian albatrosses would minimize foraging costs by employing a comparatively economical foraging strategy in response to sparse, unpredictable local prey resources (Flint and Nagy, 1984; Weimerskirch et al., 2005). We therefore also combined our results with published research and examined residual variation in the allometric relationship between body mass and FMR_at−sea to further evaluate the comparative energy expenditure of Hawaiian and yellow-nosed albatrosses (discussed below).

We did not find significant relationships between FMR_at−sea and foraging range, trip duration, transit rates, the percent time in flight, the number or frequency of landings, change in mass, or water influx rates within each species. As predicted, Hawaiian albatrosses spent more time in flight than yellow-nosed albatrosses, however, we were not able to detect a relationship between time in flight and energetic costs within species. During brooding, overall landing rates did not differ between species and were highly variable. Previous research using the doubly labeled water method demonstrated a relationship between energetic costs and landing rates in wandering albatrosses (Shaffer et al., 2001a), therefore we expected that landing rates would be related to energy expenditure within species in this study. The lack of a relationship between landing rates and field metabolic rates may be explained by the relatively smaller size of Hawaiian and yellow-nosed albatrosses (2–4 kg) compared to the larger wandering albatross (8–10 kg; Tickell, 1968).

While field metabolic rates of black-footed and yellow-nosed albatrosses have not been studied during incubation, Pettit et al. (1988) measured FMR_at−sea and FMR_on−nest of Laysan albatrosses at Tern Island during the incubation period. Estimates of FMR_on−nest during incubation were similar to our estimates of FMR_on−nest during brooding, however, estimates of FMR_at−sea during incubation were higher than our estimates of FMR_at−sea during brooding for all three species (Pettit et al., 1988; Figure 3). This contrasts with a study of wandering albatrosses, where FMR_at−sea was higher during brooding, compared to the incubation period (Shaffer et al., 2003). Although estimated FMR_at−sea was greater during incubation for Laysan albatrosses, activity patterns during incubation and brooding were similar (this study; Kappes et al., 2015). This provides further evidence that activity patterns do not relate directly to energy expenditure in this species, contrary to findings for wandering albatrosses (Weimerskirch et al., 2000; Shaffer et al., 2001a). Higher FMR_at−sea during the incubation period could reflect effort directed at assimilating lipid stores while foraging at distant, preferred habitats (Welcker et al., 2009), so that adults are able to effectively provision young chicks while foraging in an oligotrophic marine habitat during brooding.

To further examine variation in energy expenditure among species, we evaluated the allometric relationship between FMR and body mass for smaller albatross species and examined where Hawaiian and yellow-nosed albatrosses fell in relation to the plotted regression line (Figure 3). We found that Hawaiian albatrosses during brooding fell below the regression line, whereas yellow-nosed albatrosses fell above this line (Figure 3). This indicates that Hawaiian albatrosses expend comparatively less energy at sea during brooding after accounting for species differences in mass, which may be related to foraging in an oligotrophic environment during this reproductive stage. It may also be related to the fact that Hawaiian albatrosses breed during boreal winter when winds are stronger, compared to conditions during the Indian Ocean austral summer when yellow-nosed albatrosses breed. Laysan albatrosses during incubation fall well above the regression line, indicating that this species expends comparatively more energy during this reproductive stage, which may be related to effort directed at assimilating lipid stores in distant, preferred habitats.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer GS and handling Editor declared their shared affiliation.