Heart Rates and Abdominal Temperatures of Free-Ranging South Georgian Shags, Phalacrocorax Georgianus

The South Georgian shag (Phalacrocorax georgianus Lönnberg) is a foot-propelled pursuit diver (Ashmole, 1971) that feeds on benthic-dwelling fish. It is capable of diving to considerable depths (maximum 116 m; Croxall et al. 1991) and for long durations (maximum 5.2 min; Wanless et al. 1992), yet retains the ability to fly. The birds exhibit distinct patterns in their diving behaviour and can be categorized as being short, shallow divers (diving for ⩽120 s and to <20 m), long, deep divers (diving for >120 s and to >35 m) or mixed divers (Croxall et al. 1991; Wanless et al. 1992). These two studies have estimated that over 50 % of all dives made by the South Georgian shag are for longer than their estimated aerobic dive limit (ADL; Kooyman et al. 1983). Thus, the inference is that the birds must obtain a proportion of their energy requirements via anaerobic metabolism. This would be consistent with the theory of Ydenberg and Clark (1989) that a pursuit-diving bird will resort to anaerobic metabolism to increase dive duration and hence foraging time. However, it must be remembered that the calculations of ADL are based on a number of estimates (e.g. the diving metabolic rate) that have rarely been tested and have certainly never been tested in the South Georgian shag.

Studies investigating the effect of forced diving on the physiology of cormorants, P. auritus, in the laboratory (Mangalam and Jones, 1984; Jones and Larigakis, 1988) found that submergence caused a profound bradycardia. This cardiac response indicates that the cormorant undergoes extensive peripheral vasoconstriction and switches to anaerobic metabolism in the underperfused tissues (Irving, 1939). If wild birds also resort to anaerobic metabolism, as is indicated when they exceed their estimated ADL, then they too may show this cardiac response.

However, ducks rarely exhibit a bradycardia when diving voluntarily (Butler and Woakes, 1979; Furilla and Jones, 1987) and the blood supply to the active muscles is maintained at the same level as when exercising in air (Bevan and Butler, 1992b), indicating that they are metabolizing aerobically.

Bradycardia and, presumably, its associated peripheral vasoconstriction occur in voluntarily diving ducks only in response to external factors such as access to the surface being temporarily denied (Stephenson et al. 1986) or when being chased.

Penguins are extremely well adapted to their aquatic existence, e.g. they have reduced pneumatization of the bones to reduce buoyancy and increased subcutaneous fat deposition for insulation. However, these adaptations also mean that they have lost their flight capabilities. The South Georgian shag, as well as having a considerable diving ability that is surpassed only by that of the larger penguin species, retains its ability to fly, which is a highly aerobic activity. This dual lifestyle is further complicated because the birds use their wing muscles for flight and their leg muscles for underwater propulsion and, therefore, both sets of muscles must be well developed. They also do not appear to possess a substantial fat layer to aid insulation (R. M. Bevan, personal observation), as in the penguins, and therefore must use other thermoregulatory mechanisms to control body temperature.

How are these birds capable of such an amazing diving ability? The purpose of the present study was to record heart rate and abdominal temperature continuously in free-ranging South Georgian shags. These measurements were then related to the behaviour of the bird in the field, with particular attention being given to the physiological responses to diving and flying.

The study was undertaken at the British Antarctic Survey (BAS) base on Bird Island, South Georgia (54 °S, 38 °W) during the 1992–1993 austral summer. Four South Georgian shags (SGS1–4) were used in the present study, two females and two males (mean mass 2.39±0.09 kg; S.E.M.) from different nests, which were brooding/rearing 1–3 chicks.

Heart rates and abdominal temperatures of the free-ranging birds were determined using heart rate data-loggers (HRDL; Woakes et al. 1995). The HRDLs weighed 19 g or approximately 0.8 % of body mass. They were programmed to record the number of heart beats in every 15 s period and to take a reading of abdominal temperature every 60 s (for bird SGS4, abdominal temperature was recorded every 30 s). Programming was achieved via a purpose-built interface connected to an A/D converter (PCLL-711, Advantech Co. Ltd) housed in a lap-top computer (316LT, Dell). After programming, the HRDL was encapsulated in wax and then in silicone rubber to ensure biocompatibility. The entire HRDL was sterilised in alcohol before implantation.

The birds were caught at the nest site and transported back to the operating facilities (approximately 1 km distance) where they were anaesthetised with a halothane-enriched air:O₂ mixture (induction 1–1.25 %, maintenance 0.75–1.25 %). The HRDL was implanted as described by Bevan et al. (1995a), the time of implantation being noted for synchronisation with a time–depth recorder (see below). A long-acting antibiotic (LA Terramycin) was injected intramuscularly into the pectoralis muscle, and the bird was transported back to the colony. It took approximately 2 h to transport the birds, implant them and return them to the colony. The study site was monitored for several hours each day, and the times of arrival and departure and periods of flying of the implanted birds were recorded.

The HRDLs had been tested previously in two other species, the gentoo penguin Pygoscelis papua and the Antarctic fur seal Arctocephalus gazella. In the gentoo penguin, a heart rate radio transmitter (Woakes and Butler, 1975) was attached to the HRDL so that both devices used the same electrocardiogram (ECG) electrodes. The dual device was encapsulated in wax and silicone rubber and was implanted into the abdominal cavity (Bevan et al. 1995a). The bird was then occasionally walked on a treadmill over a period of several days (Bevan et al. 1995b). At random intervals, the ECG was recorded from the radiotransmitter for periods of at least 2 min. The HRDL transmits a click when it recognises the ECG and so the audio signal from a receiver was recorded at the same time. There was no difference between the timing of the ECG events from the two devices. A subsequent analysis showed that the HRDL accurately counted the number of heart beats in each time period. The same was also found to be true for the tests performed with the Antarctic fur seal.

The birds were recaptured 3–4 days after the implantation procedure and a time–depth recorder (TDR; mk V, Wildlife Computers, Inc.) was attached. The interval between the implantation procedure and TDR attachment was designed to allow the birds to recover. Previous work using HRDLs has shown that 1–2 days is sufficient. The TDR weighed 50 g and had a front cross-sectional area of 5.7 cm². The mass of the TDR was, on average, 2.1 % of the body mass of the birds on capture. It was attached to the feathers on the dorsal body surface between the wings and anterior to the base of the tail using quick-drying epoxy resin (RS Components). Epoxy resin was also applied to the front of the TDR to produce a more hydrodynamic profile. The TDR was programmed to sample depth every 1 s (Boyd, 1993) and to record the periods for which the device was dry. By positioning the TDR such that the conductivity switch was distal to the head, the switch was often submerged when the bird was on the water suface and was never dry for more than a few seconds (R. M. Bevan, K. R. Reid and I. L. Boyd, personal observations). It was therefore possible to determine when the birds landed on and left the water and hence to infer when they were flying or resting ashore.

At the end of the monitoring period (see Table 1), the birds were recaptured and transported back to the holding facilities where the oxygen consumption of three of the birds was measured in air and when the birds were on water. Bevan et al. (1994, 1995b) provide full details of the respirometry equipment used. It consisted of a Perspex respirometer with a monitored flow of air passing through it. The oxygen and carbon dioxide levels in the air entering and leaving the box were monitored using a paramagnetic O₂ analyser (Servomex 570A) and an infrared CO₂ analyser (Servomex 1410), respectively. For measurements when the bird was dry, the respirometer was placed on an inactive treadmill, whereas for measurements with the bird resting on the water, the respirometer was held above a water channel and the edges of the box were submerged to maintain a seal. The effects of resting on water on the energy expenditure, heart rate and abdominal temperature could then be determined.

Finally, the data-loggers were removed using the same operating procedure as for implantation, and the data were extracted for later analysis. The birds were then taken back to the colony and released. The colony was not monitored after the study in order to reduce any further disturbance. However, previously implanted birds which had the devices removed earlier in the study were seen feeding their chicks.

A dive was deemed to occur when the depth was greater than 2 m. The parameters calculated from the depth data (see Fig. 1) were time of dive, dive duration, maximum depth, duration of the post-dive surface interval, descent rate, ascent rate, transit time (time spent descending and ascending), time spent foraging (bottom time), dive cycle duration (a dive cycle is composed of the dive plus the subsequent surface interval) and dive bout duration (see below for definition of a bout).

Descent rate, ascent rate, descent time, ascent time and bottom time were calculated from the depth data using a modified version of the iterative regression described by Yeager and Ultsch (1989) and Nickerson et al. (1989). Their technique applies a two-segment analysis to a data set and thus calculates two ‘best-fit’ regressions and the point where the two lines intersect. In applying the technique to the diving data, each dive was split into two parts. For the first half of a dive, the slopes of the two regression lines calculated from the data correspond to the rate of descent (line dr, Fig. 1) and to the rate of change of depth while at the bottom (line br1, Fig. 1). As the abscissa is time, the point of intersection is equivalent to the time that the descent phase (dt) finishes and foraging time begins. For the second half of the dive, the slopes of the regression lines correspond to the rate of change of depth while at the bottom (line br2, Fig. 1) and the rate of ascent (line ar, Fig. 1). The point of intersection is the time at which foraging ends and ascent (at) begins. The time spent in the foraging area (bottom time) is the difference between the time at which the bird finished its descent and that at which it started to ascend (bt). The minimum duration of a dive for this analysis to be carried out was 13 s, because this was the time required to fulfil the requirements for the sample size of each regression.

The diving behaviour of the birds was split into discrete bouts. To determine the bout-ending criterion (BEC) of each individual, a log-survivor curve of the surface interval data was plotted (Gentry and Kooyman, 1986). The BECs for birds 1, 2 and 3 were surface intervals >255 s, >358 s and >322 s, respectively. Bird 4 showed two clear inflections in the log-survivor curve at 122 s and 856 s. Visual inspection of the data suggested that these BECs were correct (see Croxall et al. 1991) and that the larger BEC of bird 4 occurred when dives were for longer durations (>180 s).

A number of different heart rate parameters were calculated using the depth data as a guide (see Fig. 1). These were pre-dive heart rate, mean heart rate during diving, post-dive heart rate, mean heart rate while at the surface between dives, minimum heart rate during a dive, heart rate over each dive cycle and mean heart rate over each dive bout. As the sampling rates of the HRDL and TDR were different, it was not possible to synchronise the two data sets exactly. In addition, the HRDL stored the number of beats over 15 s periods. Therefore, the first heart rate measurement taken after the start of the dive (Fig. 1) may include a proportion of the heart beats which occurred during the pre-dive period and a proportion that occurred during the dive. Likewise, the heart rate measurement taken at the end of the dive may include a proportion of heart beats which occurred during the dive and a proportion occurring during the surface interval. To circumvent this problem, heart rates were analysed during diving only if the dive duration was greater than 59 s. The first heart rate measurement during diving to be analysed was the first taken when a full 15 s (the sampling frequency of the HRDL) of heart rate counting during a dive had elapsed. Similarly, the last analysed heart rate of a dive was taken before surfacing. Consequently, if a dive lasted 60 s and started 5 s into the HRDL sampling regime, three heart rate measurements were used: the first covering the period 10–24 s, the second covering the period 25–39 s and the final measurement covering the period 40–54 s. The mean of these gives the mean heart rate during the dive. A similar analysis was also applied to the surface intervals (Fig. 1). Through this mode of analysis, it was hoped to eliminate the effects of surfacing and diving when substantial changes in heart rate may occur (Butler and Woakes, 1979).

The temporal changes in heart rate with dive duration were analysed without the above restrictions. The heart rate value that occurred when diving started was used as time 0; thereafter, the heart rate values were plotted with respect to this timing.

The minimum heart rate during a dive was the lowest value recorded during each submergence. The mean heart rate over the dive cycle was calculated from the first analysed diving heart rate up to, but excluding, the first heart rate measurement of the next dive (dive cycle heart rate was only calculated if the dive occurred within a diving bout). Pre-and post-dive heart rates were those measured during the 15 s sampling periods immediately prior to and immediately after a dive.

The HRDL was temperature-calibrated after removal from the bird by immersing it in a water bath. The water temperature was varied between 18 °C and 42 °C and measured with a precision thermometer (±0.02 °C). After calibration, the HRDL was plunged into cold water (15.0±0.5 °C) to determine the response time. The mean time for a 90 % response was 225±15 s (N=4). However, since the HRDLs responded to any temperature change within 60 s, the temperature data were adjusted to take this time lag into account.

Statistical testing was performed using the SYSTAT (Systat Inc.) software. Mean values are given ± S.E.M. The 95 % level was taken as statistically significant. Regression analysis was performed using a least-squares regression (Zar, 1984). Differences between two means were tested with a paired t-test (Zar, 1984). Differences between more than two means were tested by analysis of variance (ANOVA). If a significant difference was found to exist between more than two means, post-hoc tests were performed to determine where these occurred (Zar, 184).

The mean duration of heart rate monitoring of the four birds was 138.6±5.8 h (Table 1), while the mean duration of depth monitoring was 39.2±8.33 h. All but one of the birds used in the present study gained or maintained body mass during this period (Table 1). There was no apparent difference in the behaviour of the experimental birds from the rest of the population, and foraging behaviour was very similar to that of unencumbered birds (Wanless et al. 1992). Fig. 2 shows representative traces obtained from the HRDL and TDR fom two different birds.

Table 2 summarizes the diving data analysis. In total, 900 dives deeper than 3 m were recorded. SGS4 was the only bird of the present group to engage in deep (>60 m) or long (>180 s) diving activity (Croxall et al. 1991; Wanless et al. 1992). The dives from this bird were therefore also split into short (<180 s) and long (>180 s) dives for further analysis. The maximum depth attained by the individual birds ranged from 35 to 101 m. The depths and durations of dives made by individual birds were not normally distributed but were negatively skewed. The mean median depth was 22±5 m and the mean median dive duration was 85±19 s (range 50–119 s). For SGS4, the median dive duration for dives lasting more than 180 s was 244 s.

Diving activity only occurred between 04:27 h and 20:11 h local time, i.e. during the hours of daylight and, although it could be split into bouts (see Table 3), these bouts were not confined to specific times of the day. There was no correlation between dive depth or duration and time of day.

Each bird showed a highly significant positive correlation between dive duration and maximum depth (P<0.001, mean r=0.98±0.01), but, although dive duration was significantly related to the duration of post-dive intervals shorter than 900 s (P<0.001), there was a much weaker correlation (r=0.61±0.11). There was also a significant, positive correlation between the time spent at the bottom and dive duration (P<0.001, r=0.97±0.02). When the time spent at the bottom was expressed as a percentage of dive duration, this relationship weakened considerably (r=0.55±0.06) (Fig. 3). In fact, for SGS4, the percentage of the dive spent at the bottom was negatively correlated with dive duration. This was due to a reduction in the proportion of time spent at the bottom in dives lasting more than 180 s (50 %) compared with dives lasting less than 180 s (70 %) (Table 2; Fig. 3). Consequently, the proportion of time spent in transit to the feeding area must be increased in dives of over 180 s.

The duration of dive bouts varied between birds (range 43.2–101.6 min, mean 68.3±12.6 min), as did the number of dives per bout (range 10.7–34.6, mean 21.2±5.3) (Table 3). The number of dive bouts per day varied between three and six, with a mean of 4.6±0.3.

A total of 133,023 measurements of heart rate were obtained from the four birds over a mean recording period of 138.6±5.8 h. For three of the four birds, heart rate was bimodally distributed, but for one (SGS2) there was a trimodal distribution (Fig. 4). The lower peak frequencies occurred at 93±5 beats min⁻¹ and the upper peak at 307±10 beats min⁻¹. The middle peak of SGS2 occurred between 152 and 159 beats min⁻¹. The mean maximum recorded heart rate was 511±48 beats min⁻¹. However, 99 % of the heart rates recorded were less than 360±19 beats min⁻¹. There was a daily pattern to heart rate, common to all birds, with high values occurring during the hours of daylight and low values during the night (Fig. 5). Resting heart rates were obtained when the birds were at the nest and during the hours of darkness and ranged between 85 beats min⁻¹ and 143 beats min⁻¹ for individual birds, with a mean of 104.0±13.1 beats min⁻¹ (Table 4).

The high heart rates which occurred during the day coincided with the diving activity (Fig. 5). For most birds, this was shown by an increase in heart rate between 04:00 h and 05:00 h local time. SGS4 did not dive until after 09:00 h local time, but again this was reflected by an concomitant elevation in heart rate (Fig. 5). Pre-and post-dive heart rates did not differ significantly (Fisher LSD test, P=0.370). The pre-dive heart rate was also not significantly different from the mean heart rate measured during the surface interval (P=0.212), whereas the post-dive heart rate was found to be significantly higher (P=0.039; Table 4). The minimum heart rate recorded during a dive (64.8±5.8 beats min⁻¹) was significantly lower than both the mean heart rate measured during the whole period of submersion (P=0.013) and the mean resting heart rate (P=0.012). There was no significant difference between mean heart rate during diving and that at rest (P=0.979).

There was only a very weak (not significant) correlation between the pre-(r=0.06, P=0.266) and post-dive (r=−0.02, P=0.336) heart rates and the depth of the succeeding or preceding dive. This was also true when the two variables were correlated with dive duration (r=0.10, P=0.358 and r=−0.02, P=0.456). The minimum heart rate recorded during diving was, however, significantly negatively correlated with both dive depth (r=0.53, P<0.001) and dive duration (r=0.57, P<0.001, Fig. 6). The mean heart rates recorded during a dive also fell as the dive progressed (Fig. 7). In all birds, the mean heart rate at, or near, the end of the dive was below the resting level for that bird.

Periods of flights of varying durations were recorded for all of the birds. Most flights were short and consisted only of leaving the colony and landing on the sea a few hundred metres away, or flying from the sea to a rocky outcrop. The duration of these flights was usually less than 30 s. For heart rate analysis, only flights longer than 120 s were used. These flights took place when the birds foraged at a greater distance from the colony. The absolute duration of these flights could not be ascertained as the birds flew out of sight of the observer, but the mean duration for which it was possible to follow the birds was 221±29 s. Mean heart rate over this period was 309.5±18.0 beats min⁻¹, which was 3.05 times the resting level (Table 5).

Resting abdominal temperature of the South Georgian shag, measured when the birds were in the colony overnight, was 39.1±0.2 °C (Table 5). It was significantly higher (paired t-test, t=−5.222, P=0.014) after flying for more than 120 s, when it had risen to 40.1±0.3 °C. At the start of a diving bout, abdominal temperature was again marginally above the resting level at 40.0±0.5 °C (Table 3), but this was not a significant increase. Abdominal temperature fell exponentially during a diving bout (Fig. 8) and by the end of a bout had fallen by, on average, 4.9±1.6 °C (range −2.4 to −9.4 °C) (Table 3). The mean minimum abdominal temperature was 31.6±2.1 °C. There was a significant difference between the body temperature at rest, during flight and at the start and end of a diving bout (ANOVA, P=0.008). The temperature at the end of a diving bout (35.1±1.7 °C) was significantly lower than that at rest and that at the start of a diving bout (Tukey LSD, P=0.044 and P=0.014, respectively).

The rate of oxygen consumption of three South Georgian shags in air was 17.33±0.54 ml min⁻¹ kg⁻¹, rising by 58 % when the birds were placed on water at 6.9 °C (Table 6). This was similar to the percentage increase in heart rate, which rose from 162±27 beats min⁻¹ in air to 238±17 beats min⁻¹ when the birds were placed on the water, an increase of 47 %. There was no change in body temperature as a result of being placed on the water. Both the heart rate and body temperature of the birds in air were elevated above the resting level of the birds when free-ranging in their natural environment (Tables 5, 6).

The physiological measurements presented in this paper are some of the first to be recorded from a free-ranging bird that uses foot propulsion to dive. However, even though South Georgian shags use their feet to propel themselves through the water, they exhibit a diving ability that is comparable to that of the penguins (Croxall et al. 1991).

In general, the diving behaviour of the birds in the present study agrees with that reported by Croxall et al. (1991) and Wanless et al. (1992). The diving behaviour of three individuals (SGS1–3) classifies them as shallow or intermediate divers. None of these three birds showed the remarkable diving performance shown by SGS4 or reported previously (Croxall et al. 1991; Wanless et al. 1992). SGS4 performed one bout consisting of 31 dives that were all deeper than 90 m and longer than 240 s. There were no other indications that this bird was remarkable in any way, and it is possible that long deep dives would have been recorded from the other birds had they been been monitored for longer. In the future, it would be interesting to investigate whether the different diving behaviours described by Wanless et al. (1992) are confined to individuals, or whether each bird exploits all available strategies depending on the situation in which it finds itself.

The mean descent and ascent rates of all the birds (Table 2) were of a similar magnitude to those found by Croxall et al. (1991) but faster than those reported by Wilson and Wilson (1988). They also covered a much smaller range and they certainly did not reach the highest values of 5.0 m s⁻¹ found by Croxall et al. (1991) or the 4.0 m s⁻¹ estimated by Wilson and Wilson (1988). In the present study, a sampling rate of 1 s was chosen for the depth measurements, which provides a much more precise analysis than from sampling every 15 s (Boyd, 1993). It is likely, given the very constant rates of descent and ascent recorded in the present study, that the high rates found by others are incorrect. Indeed, these high rates represent values of approximately 10 body lengths s⁻¹, comparable with the sustained speed attainable by salmonid fish. Interestingly, when comparing the long and short dives of SGS4, both the ascent and descent rates were higher for the long dives than for the short ones. This contrasts with the theory of Thompson et al. (1993) who predicted that, in seals at least, swim speed should decrease as depth of dive increases.

The mean transit times reported here are comparable with those of birds classified as intermediate divers (Croxall et al. 1991). However, the time spent at the bottom, equivalent to the foraging time, is more similar to the bottom time reported for the deep dives in the birds studied by Croxall et al. (1991). Wilson and Wilson (1988) reported a significant relationship between estimated bottom time and dive depth in two species of cormorant. This was also found in the present study, although not by Croxall et al. (1991). In the present study, both travel time and time spent at the bottom increased with dive duration (see Fig. 3), while the percentage of the dive time spent at the bottom increased at a much lower rate. For SGS4, the relationship was actually reversed, with the proportion of the dive spent at the bottom decreasing for longer dives. This is in agreement with the model proposed by Houston and Carbonne (1992) which predicts that the time spent in the foraging area should first increase and then decrease as travel time increases.

There was a distinct modality to the heart rate frequency distributions. For SGS2, there were three distinct peaks, whereas the others showed only two peaks (Fig. 4). The heart rates of the birds during diving were distinctly bimodal (see Fig. 2), a feature that is commonly seen in birds diving voluntarily in the laboratory environment (e.g. Butler and Woakes, 1979; Furilla and Jones, 1987; Stephenson et al. 1992) and in free-ranging gentoo penguins (R. M. Bevan, P. J. Butler, A. J. Woakes and J. P. Croxall, unpublished data). The lower peak in the heart rate frequency distributions contains both diving and resting heart rates. The middle peak shown by SGS2 was due to this bird having a very high mean resting heart rate. There are few other data with which to compare our results. Although Kooyman et al. (1992) recorded heart rate in unrestrained emperor penguins, Aptenodytes forsteri, their birds were constrained to dive from a hole cut in the ice and to which they had to return. The recording devices used could only measure heart rates up to 120 beats min⁻¹, which these birds have been shown to exceed (Kooyman et al. 1992; Kooyman and Pongannis, 1994). Other studies measuring heart rate in free-ranging diving birds and mammals have shown a similar bimodal pattern to that reported here (Fedak et al. 1986; Thomson and Fedak 1993; R. M. Bevan, P. J. Butler, A. J. Woakes and J. P. Croxall, unpublished data).

The pre-and post-dive heart rates were measured immediately before and after a dive took place. However, because of the different sampling rates of the depth and heart rate data-loggers, it is likely that these partly included diving heart rates. The pre-and post-dive heart rates presented here are therefore probably lower than if they had been measured on a beat-to-beat basis (see Butler and Woakes, 1979; Furilla and Jones, 1987).

Tufted ducks, Aythya fuligula, in the laboratory environment show a pre-dive tachycardia. Upon submersion, heart rate initially decreases to sub-resting levels but during the remainder of the dive rises above the resting level. This is also true of cardiac output, blood flow to the exercising muscles and metabolic rate (Woakes and Butler, 1983; Bevan and Butler, 1992b). Only when the birds extend their dive duration towards their calculated ADL or are temporarily prevented from surfacing does their heart rate fall below resting levels (Stephenson et al. 1986; Bevan et al. 1992). The heart rate data for the free-ranging South Georgian shag (Fig. 7) show certain similarities to those from tufted ducks performing ‘extended’ dives (dives in which the birds have to swim long horizontal distances under water to obtain their food; Stephenson et al. 1986). SGS1–3 show an initial reduction in heart rate followed by a slow rise, then, as dive duration increases, there is a further reduction. This temporal pattern is similar to that seen in tufted ducks.

The mean heart rate during diving was not significantly different from resting levels but was lower than levels for birds resting on the water channel (Table 4) and is probably lower than the heart rate while exercising at the surface (see Woakes and Butler, 1983). However, heart rates measured during the inter-dive period were similar to the elevated levels measured during flight. As a consequence, the mean heart rate measured over the dive cycle was significantly higher than the resting level, suggesting that these birds have an increased metabolic rate over the complete dive cycle. This feature was also found in the tufted duck (Bevan et al. 1992). The tufted duck has to work hard to overcome buoyancy in order to remain submerged, and it is therefore not surprising that these birds show an elevated dive cycle metabolic rate. Their cardiac output is, however, lower during diving than when working at a comparable rate at the surface. This reduction in cardiac output suggests that there is a lower blood flow to some tissues and organs during diving, although not to the working muscles (Bevan and Butler, 1992b).

The South Georgian shag is probably less buoyant than the tufted duck. The relative work required by the bird to remain submerged will therefore also be lower. At depth, buoyancy will be further reduced owing to the compression of the air-filled spaces between the feathers, which will again reduce the energy required to remain submerged. As the birds are active hunters, they will presumably be searching when at the seabed, which will require work to be performed by the active leg muscles. Any reduction in buoyancy will, therefore, reduce the total energy requirement and hence save the oxygen reserves.

Although the overall metabolic rate during a dive cycle is elevated, this does not preclude a reduction in metabolic rate during the portion of the dive when the birds are submerged. Indeed, the mean minimum heart rate of the South Georgian shags during diving was lower than the resting rate, which can be interpreted as a true ‘diving bradycardia’, which is indicative of a switch to anaerobic metabolism during submergence (Irving, 1939). Further evidence implying the use of anaerobic metabolism during diving is that the minimum heart rates recorded during diving in these free-ranging birds are similar to those found in forcibly submerged cormorants (Jones and Larigakis, 1988), a procedure that invokes severe vasoconstriction and anaerobiosis (Irving, 1939; Scholander, 1940; Butler and Jones, 1982). A gradual switch to anaerobic metabolism is also indicated by the progressive reduction in heart rate as a dive progresses.

That the estimated ADL is frequently exceeded by the South Georgian shag is well documented. However, the calculations of the ADL may be flawed. The variables used to calculate the ADL are (1) the usable oxygen stores in the body and (2) the rate at which these stores are used up. Neither of these variables has been measured in the South Georgian shag, but both can be estimated using data from other species. The ADL of 120–150 s proposed by Croxall et al. (1991) used oxygen storage data obtained from penguins; unfortunately, they do not state how the rate of oxygen usage was calculated. Wanless et al. (1992) predicted an ADL of 174 s assuming the metabolic cost of diving to be only 1.5 times the standard metabolic rate (SMR), but suggested that the diving metabolic rate would be between 2 and 10 times SMR, which could shorten the ADL considerably (from 130 to 26 s). At a diving metabolic rate of 2.5 times the resting rate (the rate obtained from diving tufted ducks; Bevan et al. 1992), the ADL in the South Georgian shag would be less than 100 s and over 50 % of the dives made by the birds in the present study would have exceeded it.

At the point in the dive at which the ADL is reached, if the assumptions regarding the calculation of ADL are correct, then the animal has utilized all of the available oxygen. Even tissues such as the brain would then have to obtain their energy from a route other than oxidative metabolism. One possible solution is that, rather than making a dramatic switch to anaerobic metabolism at the ADL, the birds gradually increase the proportion of the energy requirement that is supplied by anaerobiosis throughout the dive. The body thus gradually conserves the remaining oxygen stores for the tissues that are oxygen-dependent. Evidence supporting this possibility could be inferred from the gradual reduction in heart rate that occurs as dive length increases.

Another possible solution is to decrease the overall metabolic rate as the dive progresses (Bevan et al. 1992). This could be achieved in a number of ways. First, as a shag dives deeper, the air spaces within its body and feathers will gradually be compressed, and its buoyancy and hence the amount of work required to dive will then be reduced (Stephenson, 1994). Second, we reported a reduction in abdominal temperature with dive duration. If changes in abdominal temperature reflect temperature changes occurring in other tissues, then the metabolic rate of these tissues will also fall, thus reducing the overall metabolic rate (see also Hill et al. 1987, for Weddell seals, Leptonoychotes weddellii). For example, if a tissue has a Q₁₀ of 2, then a 10 °C change in temperature will cause a 50 % reduction in the metabolic rate of that tissue, a reduction in overall metabolic rate and thus an increase in the ADL. The most probable situation is that a combination of the mechanisms mentioned above is used and that the energy cost of diving is lower than the previously assumed values. Like Weddell seals, therefore, these birds may only infrequently exceed their true ADL.

The cause of the reduction in abdominal temperature during diving in free-ranging birds can only be speculated upon, but it is likely to be a combination of ingesting cold prey items (Grémillet and Plös, 1994) and the physical act of diving in cold water. The latter is particularly likely because of the high conductance and heat capacity of water, coupled with the reduced insulation of the body caused by compression of the feathers during a dive (Stephenson, 1994). Other studies have shown that diving animals lose heat during submergence (MacArthur, 1984; Hill et al. 1987; Bevan and Butler, 1992a; R. M. Bevan, P. J. Butler, A. J. Woakes, I. L. Boyd and J. P. Croxall, unpublished data) although the temperature of the active muscles remains high through the generation of heat (Ponganis et al. 1993).

South Georgian shags have to elevate their metabolic rate to maintain their abdominal temperature when resting on water in a respirometer. This elevated metabolic rate is coupled with an increase in heart rate. High heart rates were also recorded at the end of every dive bout while the free-ranging South Georgian shags were on the water surface. These periods last for approximately 10 min, following which the abdominal temperature starts to rise. This elevated heart rate, which probably reflects an increase in metabolic rate, could be due to a number of factors: (1) the act of resting on cold water (Jenssen et al. 1989; Bevan and Butler, 1992a), (2) the activity asssociated with preening after a dive bout (the extra work generating heat), (3) the removal of anaerobic metabolites, and (4) specific dynamic action. Specific dynamic action, which is a byproduct of assimilating ingested food, could be an energetically inexpensive way of regenerating the heat lost through diving. It is clear that more research is needed on the effects of food intake on the thermoregulation of diving animals.

The fact that the South Georgian shag uses different groups of muscles for its two locomotory modes (swimming/diving and flying) may also help explain the apparent paradox between the estimated ADL and the recorded dive durations. The flight muscles are inactive during diving and, consequently, their demands on the oxygen stores will be restricted to, at most, a basal maintenance level which may be reduced even further if the temperature of these muscles is allowed to fall. Although not measured in this study, the flight muscle mass of the Phalacrocoridae amounts to approximately 15–20 % of the total body mass (Hartman, 1961). If the metabolic rate of this tissue is reduced, it would represent a considerable reduction in the overall metabolic rate. The actual tissues that are using the oxygen stores during a dive may, therefore, be only a relatively small fraction of the body mass and, even though these tissues are working at an elevated rate (Bevan and Butler, 1992b), the available oxygen stores may be sufficient for most dives to remain predominantly aerobic.

The recorded heart rates of South Georgian shags during flight were much lower than would predicted from recent allometric equations (486 beats min⁻¹; Bishop and Butler, 1995). This may represent an adaptation of the cardiovascular system of the South Georgian shag to the dual roles of flying and diving through an increase in the cardiac stroke volume or an increased capacity of the tissues to extract oxygen from the blood.

This study shows how the physiology and behavioural ecology of the South Georgian shag are inextricably linked. It also raises some interesting questions about the control of the physiological processes employed by this free-ranging diving bird, particularly those mechanisms concerning thermoregulation and metabolism during submergence. It is likely that a number of mechanisms are employed by the South Georgian shag to increase the time that they are able to remain submerged and therefore the time they have available for foraging. These mechanisms may also be used by other diving animals and may therefore assist in our interpretation of the physiology and behavioural ecology of diving birds and mammals in general.

The authors would like to thank everyone at the British Antarctic Survey base at Bird Island for their help. This work was supported by NERC grant GR3/7508.