Alasdair I. Houston, School of Biological Sciences, University of Bristol, Woodland Road, Bristol BS8 1UG, UK. Tel.: 0117 9287481; fax: 0117 3317985; e-mail: A.I.Houston@Bristol.ac.uk
The extant auks show three strategies of chick rearing – precocial (chicks leave the nest site when a few days old), intermediate (young raised to a mass of around 20% of adult mass) and semi-precocial (young raised to a mass of around 65% of adult mass). It is not known which strategy the extinct Great Auk used. In this paper, we investigate this issue by a novel combination of a time and energy budget model and phylogenetic comparison. The first approach indicates that for reasonable estimates of the equation parameters, the Great Auk could have followed an intermediate strategy. For a limited range of parameters, the Great Auk could have followed the semi-precocial strategy. Phylogenetic comparison shows that it is unlikely that the Great Auk followed a precocial strategy. The results suggest that the Great Auk followed an intermediate strategy as does its presumed closest extant relative the Razorbill.
Including the extinct Great Auk (Pinguinus impennis), there are 23 species of birds in the family Alcidae (Gaston & Jones, 1998). A striking feature of this family is that it exhibits three patterns of nestling development (Sealy, 1973; Gaston, 1985a; Gaston & Jones, 1998). Most auks, including puffins (Fratercula spp.), rear their chicks to at least 65% of adult weight, and the young, independent birds leave the colony capable of flight. This pattern is known as the semi-precocial strategy. At the other extreme is the precocial strategy, in which the young are not fed at the nest. A few days after hatching, when able to swim but unable to fly, the chicks go to sea with their parents. Four species of murrelet adopt this strategy (Gaston, 1992; Gaston & Jones, 1998). The Common Murre (Uria aalge), the Thick-billed Murre (Uria lomvia) and the Razorbill (Alca torda) adopt what is known as the intermediate strategy. After around 20 days at the colony, the chicks, often weighing less than 25% of the full adult mass, go to sea with the male parent who feeds them for several weeks (Sealy, 1973; Gaston, 1985a; Gaston & Jones, 1998). (We follow the terminology used by Gaston & Jones (1998); for a detailed review of terminology, see Gaskell (2004).)
The largest of the recent Alcidae, the flightless Great Auk was driven to extinction in the 19th century before a detailed study of its behaviour or ecology was made (Bengtson, 1984). In this paper, we investigate which pattern of chick rearing was adopted by the Great Auk by using two theoretical approaches. One involves a time and energy budget, based on that constructed by Houston et al. (1996), to determine limits on fledging mass. The other is a phylogenetic comparison of the Great Auk within the Alcidae.
The Great Auk is thought to have weighed in the region of 5000 grams (Bèdard, 1969; Livezey, 1988; Fuller, 1999) and stood some 60–70 cm in height (Bengtson, 1984). (For geographical variation in size see Burness & Montevecchi (1992).) The Great Auk was likely to have been an efficient diver, able to capture prey at considerable depths (Gaskell, 2000). This view is supported by allometric analysis (Halsey et al., 2006). The wings of the Great Auk were shorter compared to body size than its flying alcid relatives (Livezey, 1988) and were used for ‘subaqueous flight’ (Bengtson, 1984). The flightlessness of the Great Auk seems to have been an extreme specialization for pursuit diving, convergent with that of Antarctic penguins, the Spheniscidae.
The Great Auk’s inability to fly presumably imposed restrictions on its possible behaviour, leading to speculation about the method used to care for its young. It has been assumed (e.g. Strauch, 1985) that the Great Auk used the intermediate strategy because this is the strategy used by its presumed closest living evolutionary relative, the Razorbill. Prince & Harris (1988) estimated that if the Great Auk followed the intermediate strategy, then a parent would have to make 4–5 trips a day if the load delivered was 75 g per trip. They suggest that this schedule might be difficult for a flightless bird to achieve and argued that it would have been more efficient for the Great Auk to have delivered one large meal a day by regurgitating food. Prince and Harris point out that penguins regurgitate food from their stomach and suggest that the large load that can be carried in this way makes up for a reduction in nutritional value because of partial digestion. Hobson & Montevecchi (1991) also argue that the Great Auk fed its chick by regurgitation. Based on an early account of the Great Auk’s time ashore, Fisher & Lockley (1954) and Bengtson (1984) suggest that the chick was fed at the nest site for a period as short as 9 days. Birkhead (1993) goes further and argues that the Great Auk followed a precocial strategy. He bases this on the fact that there are few records of chicks observed at colonies, together with a relatively short breeding season. Birkhead shows that unlike the precocial murrelets, the size of the Great Auk’s egg was not large for its body size. Thus, egg size does not suggest that the precocial strategy was followed, but Birkhead points out that the egg size seems large enough to make precocial development possible. He also notes that the Great Auk’s inability to fly would have limited its ability to bring food to its chick. In their chapter on development of the auks, Gaston & Jones (1998) suggest that the Great Auk could not have adopted the semi-precocial strategy unless it fed its chick by regurgitation. They conclude (p 100) that ‘We cannot rule out the possibility that the young of the Great Auk were completely precocial and were never fed on land.’Gaskell (2004) reviews the data and argues that the Great Auk followed a precocial strategy. We now provide the first formal evaluation of how the Great Auk raised its young. Our first approach uses a time and energy budget to investigate the size of chick that the Great Auk could have raised. Our second approach involves phylogenetic trees and ancestral character state reconstruction using parsimony.
The time and energy budget model
Houston et al. (1996) construct a model in which an explicit link is made between the energy delivered to the chick and the time allocation and energy expenditure of the parent. The budget is constructed for a parent rearing a single chick, and data from the Thick-billed Murre (Uria lomvia) is used to investigate the constraints acting on the parent. We now apply this approach to the Great Auk. The 24-h time budget of the parent is divided between five activities (all measured in hours):
1Ta: Time spent engaged in activities such as brooding or resting during the hours of darkness
2Tr: Time spent resting during the daylight hours
3Tc: Time spent collecting food for the chick
4Ts: Time spent collecting food for itself
5Tw: Time spent swimming between the breeding site and the foraging area
Each of these activities has a corresponding metabolic rate: ma, mr, mc, ms and mw, respectively (kJ/h). It is assumed that the metabolic rates for the activities of foraging and swimming are the same; i.e. mc = ms = mwm, which allows some simplification.
The parent’s contribution to the chick’s daily energy requirement is defined as Ec. Because both parents are assumed to care, Ec is half the chick’s requirement. The average assimilated energy content of a meal is Em. The mean number of foraging trips per day by a parent is then
The time that the bird spends swimming is
where R is the foraging range (kilometres) and v is the swimming speed (kilometres/h). It is assumed that the parent can forage for itself while collecting food for the chick and so no extra trips are necessary.
The time that the parent spends collecting food for the chick is
where γ is the rate of energy gain.
The 24-h time budget must satisfy:
The parent’s expenditure, E, is given by
The requirement that the parent should balance its energy expenditure gives
Estimation of parameter values
The body mass of the great auk is likely to have been about 5000 g (Bèdard, 1969; Livezey, 1988). There are various equations for estimating the basal metabolic rate (BMR) from body mass. Bryant & Furness (1995) measured the BMR of northern seabirds and obtained the equation
where M is body mass in g and BMRD is BMR in kJ/day.
For a mass of 5000 g, this predicts that BMR is just under 70 kJ/h. Bryant and Furness point out that their equation gives a higher value of BMR than the nonpasserine equations of Lasiewski & Dawson (1967) and Aschoff & Pohl (1970) or the equation for seabirds obtained by Ellis (1984). The lowest value is given by the resting phase nonpasserine equation of Aschoff & Pohl (1970), which predicts a BMR of about 35 kJ/h. We take a value of 70 kJ/h as our baseline value, but we also explore the implications of BMR being 35 kJ/h. Gaston (1985b) takes the time spent in other activities (Ta) for chick-rearing Thick-billed Murres to be 12 h per day. Ta values of around 12 h are mostly considered in this study.
The Adèlie penguin (Pygoscelis adeliae) is of similar height and weight to the Great Auk and is known to swim at a speed of 2.2 m/s (Culik et al., 1994). This corresponds to 7.92 km/h so a value of 8 km/h seems a reasonable baseline value.
Chappell et al. (1993) give an average value of 8.2 times the BMR for the metabolic rate of swimming, diving and bottom time in Adèlie Penguins. In their discussion, they point out that this value is higher than other estimates of the metabolic rate of swimming in penguins. They go on to suggest that a value of six times BMR may be more reasonable. Culik et al. (1994) argue that even this is too high and present evidence for a value of 2.9 to 4.3 times BMR. We take four times BMR as our baseline value, which means that m =280 kJ/h.
The resting metabolic rate mr can be estimated from data for ‘at nest’ metabolism across seabirds in general; Birt-Friesen et al. (1989), eqn (4). This gives a value of 92.3 kJ/h. It is difficult to estimate ma. Gaston (1985b) assumes a value of 1.9 BMR for Thick-billed Murres. In the case of the Great Auk, this implies that ma = 133 kJ/h.
Prince & Harris (1988) point out that the large auks carry loads that are 1.1–1.4% of their body mass. They suggest that if the Great Auk carried food in its bill, the load would have been around 1.5% its own body mass: i.e. 75 g. The energetic content of the fish taken by puffins may be as low as 4.05 kJ/g (Whiting) and as high as 10.9 kJ/g (Sprats) (Harris & Hislop, 1978). Brekke & Gabrielsen (1994) give values for the energetic content of Arctic Cod range of 4.9 kJ/g and 7.4 kJ/g. Assimilation efficiency is about 75% (Davis et al., 1989; Brekke & Gabrielsen, 1994). These values suggest that the energetic content of the load brought by the parent Great Auk (Em) could be as low as 228 kJ or as high as 613 kJ. We take 400 kJ as the baseline value, 600 kJ as a high value and 200 kJ as a low value. To investigate the consequences of the parent feeding the chick by regurgitating food, we also consider a much larger value of Em. Data on the load brought back to the young by Adèlie penguins (Trivelpiece et al., 1987) suggests that values in the region of 1800 kJ are possible in this case.
Average dive duration for Adèlies is reported as 73 s. This figure has to be doubled to give the dive plus surface recovery time (Chappell et al., 1993). Clowater & Burger (1994) reported that Pigeon Guillemots (Cepphus columba) are successful in one of every 10 dives, and assuming this to be a reasonable estimate for the Great Auk, this would mean 24 min to collect 1 fish (73 × 10 × 2 (s)) – approximately two fish per hour. For a 38-g fish, this suggests a rate of energy collection (γ) in the region of 600 kJ h−1. This line of argument involves many assumptions, so we will consider a wide range of values for γ.
Estimates of the amount of energy required by a chick that fledges at a given mass were obtained using the relationship between the peak daily metabolized energy, D (measured in kJ), of a chick and the chick’s mass at fledging, x (g) given by Weathers (1992):
This gives the total peak energy delivered by both the parent birds, i.e. each parent would supply half of this amount. For a 5000-g Great Auk raising a chick to 20% of its own mass (a reasonable value for the intermediate strategy – see Gaston, 1985a), Ec would be 556 kJ. Following a semi-precocial strategy, even to a relatively small fledging mass of 65% of the adult mass, would require Ec to be 1397 kJ.
Drent & Daan (1980) suggested that the upper limit to daily energy expenditure (K) is four BMR. The baseline value of BMR gives a daily value of 1680 kJ, giving a value of 6720 kJ for K. Kirkwood (1983) suggests that the maximum energy assimilated in a day is given by 1713 M0.72 where M is mass measured in kilograms. For a 5-kg Great Auk, this gives 5458 kJ.
Olson et al. (1979) argue that the prey of the Great Auks breeding on Funk Island, Newfoundland was available within a 2-km radius of the Newfoundland coast. This would mean a foraging range of about 65 km (Bradstreet & Brown, 1985). This is in line with Davoren et al. (2003) who report that some Common Murres breeding on Funk Island feed at about 45 km from the colony and 30 km from shore, whereas others feed at about 60 km from the colony and 10 km from shore. Despite their doubts about the argument presented by Olson et al., Bradstreet and Brown suggest that a radius of 65 km is possible. They also draw attention to accounts that suggest a radius of 15–20 km. Various values of R are considered in this study. The parameter values for the Great Auk are summarized in Table 1.
The energy that a parent delivers to its chick may be limited by a time constraint or an energy constraint. The time constraint is given when the parent has no time to rest. This can be found by setting Tr = 0 in eqns (4) and (5). From these equations plus eqn (6),
Eqns (1) and (3) can then be used to express Tw and Tc in terms of Ec. It follows that
The energy constraint can be found by setting E = K in eqns (5) and (6) and using eqn (4) to eliminate Tr. This gives
For any given set of parameters, the constraint that limits the energy delivered to the chick is the one that predicts the lower value of Ec. An example is given in Fig. 1. In this figure, Ec is plotted against γ for R =5 km, with all other parameters at their baseline value.
Because of the uncertainty associated with the parameters γ and R, we plot the maximum energy delivered to the chick as a function of one of these parameters. Figure 1 shows the energy delivered to the chick as a function of γ if the parent is constrained by energy (filled symbols) or by time (open symbols). There are two cases; in one, Em = 400 kJ, in the other (corresponding to feeding by regurgitation) Em = 1800 kJ. In each case, the time constraint line lies below the energy constraint line. This means that the parent is constrained by time. (In subsequent figures, we only show the constraint line that limits the parent.) We note that when Em = 400 kJ, the parent is still time constrained if K =5000 kJ, which is below the Kirkwood limit. The horizontal dotted lines show the value of Ec necessary to achieve a fledging mass of 65% adult mass (semi-precocial) and 20% adult mass (intermediate). It can be seen that γ would need to be about 540 kJ/h for the intermediate strategy to be possible and in the region of 830 kJ/h to achieve semi-precocial fledging mass. Figure 2 shows Ec as a function of R when γ = 800 kJ/h. The figure shows that the foraging radius R can have a strong effect on the amount of energy brought to the chick. In the baseline case (Em = 400 kJ), the intermediate strategy is not possible if R is greater than 15 km (Fig. 2a). When the chick is fed by regurgitation (Fig. 2b), the intermediate strategy is possible when R is 50 km. Once again, the parent is constrained by time in all cases.
Given that the parent is time constrained, we can use eqn (10) to obtain an equation for a critical value of R as a function of γ if a particular value of Ec is to be achieved. We can then use this equation to find combinations of R and γ that enable the parent to raise a chick to a given fledging mass. For example, if the chick is to follow the intermediate strategy and fledge at a mass of 1 kg, then Ec must be 556 kJ. The line labelled (i) in Fig. 3 gives the critical value of R as a function of γ in this case. All combinations of R and γ below the line enable the parent to bring at least 556 kJ a day to the chick and hence mean that the intermediate strategy can be adopted. Along the line, Ec = 556 kJ, and so, for a given value of γ if R is above the line, then Ec is < 556 kJ. The other lines in the figure are for the semi-precocial strategy under different assumptions about Em. Even when Em = 200 kJ, the semi-precocial strategy is possible if the foraging radius is very low. If the chick was fed by regurgitation (case iv), then the semi-precocial strategy could have been adopted for fairly large values of R.
The sensitivity of our conclusions is analysed in Table 2. For each parameter that influences the maximum energy that can be delivered to the chick in a day, the table gives the effect on this energy of a 10% increase and a 10% decrease in the parameter. We note that the parameters K and mr do not appear in the time constraint equation (eqn 10). Because the parent is limited by time in all the cases in the table, these parameters have no effect on the energy that can be delivered and so have not been included in the table. In case (a), R =5 km and Ec is about 1327 kJ. Using the function proposed by Weathers (1992), this converts to a fledging mass of approximately 3040 g. This is about 60% of adult mass, which is slightly below the lowest percentage seen in semi-precocial species (Sealy, 1973; Gaston, 1985a). In case (b), R =20 km and Ec is about 422 kJ, which corresponds to a fledging mass of about 700 g. This is about 14% of adult mass, and so the intermediate strategy might just be possible (Gaston, 1985a).
Table 2. Sensitivity analysis. For each parameter, the table gives the energy that can be delivered to the chick (kJ) if the parameter’s value is increased by 10% or decreased by 10%. In all cases, the parent is constrained by time. The time constraint also applies when mr is perturbed from its value of 90 kJ/h and K is perturbed from its value of 6720 kJ and so these parameters have no effect.
a. R =5, Ec = 1326.8 kJ, mean number of trips per day = 3.3
b. R =20 km, Ec = 422.2 kJ, mean number of trips per day = 1.1
It is likely that the time constraint limits the parent in all the cases that we have considered so far because, compared to a bird that flies, the Great Auk’s rate of energy consumption while travelling is low. The effect of the rate of energy consumption m is explored in Fig. 4. Four cases are illustrated; the baseline case (m =280 kJ/h), one case in which m is lower (140 kJ/h) and two cases in which it is higher (420 kJ/h and 560 kJ/h). When m =560 kJ/h, the parent is constrained by energy expenditure rather than time. This value corresponds to 8BMR, as suggested by Chappell et al. (1993). This high rate makes life difficult for the parent. It needs a gain rate γ of about 700 kJ/h to break even and a value of about 1060 kJ/h to adopt the intermediate strategy. (If Em = 600 kJ, the intermediate strategy can be adopted if γ is above about 980 kJ/h.)
Our analysis suggests that the Great Auk could have adopted the intermediate strategy under a fairly wide range of parameters, and even a semi-precocial fledging mass of 65% of adult mass could have been achieved if the foraging radius R was low or the chick was fed by regurgitation (see Fig. 3). If the low value of BMR is adopted, then conditions are easier for the parent, and a given energy to the chick can be delivered at higher values of R (see Fig. 5).
Birkhead (1993) suggested that the constraints imposed on the Great Auk by being unable to fly were such that a precocial strategy was the most likely pattern to have been used. Only two genera in the alcids use this strategy, Endomychura (the Xantus’ and Craveri’s Murrelets) and Synthliboramphus (The Ancient and Japanese Murrelets). The likelihood of the Great Auk having a precocial post-hatching development pattern is discussed in the next section.
Phylogenetic analyses of both morphology (Strauch, 1985) and, more recently, vintage DNA (Moum et al., 2002) support a sister species relationship between the Great Auk and the Razorbill but differ in whether their sister group is the Little Auk or the Murres (Uria). A supertree analysis (Thomas et al., 2004) that provides a synthesis of published phylogenies resolves this disagreement in favour of the morphological tree (Fig. 6), but this result is surprising given the strong support provided by the molecular data for the alternative. Furthermore, the morphological result is weakened by Strauch’s (1985) use of post-hatching development pattern (semi-precocial, precocial or intermediate) as a character in his analysis, for which he assumed that the Great Auk followed the intermediate strategy. Thus, we investigated the implications of both alternative phylogenetic placements of the Great Auk and Razorbill for interpreting the chick-rearing strategy of the Great Auk. Parsimonious interpretations of character evolution were examined using MacClade (Maddison & Maddison, 1992).
It is stated (e.g. Grande & Bemis, 1998) that the treatment of missing entries in parsimony analysis involves parsimoniously attributing a known character state to the taxon with the missing entry, thereby giving a reconstruction of the unknown feature. This is not strictly correct. The determination of tree length requires a reconstruction of character states only of the nodes in the tree. It does not require state attributions for the missing entries of terminal taxa, which are ignored (algorithmically they are assigned a state set that includes all the character states, D. L. Swofford, pers. comm.).
Conceptually, reconstruction of the state of the node at which a terminal taxon with a missing entry joins the tree (representing a hypothetical ancestor of the terminal taxon) implies that the unknown state of the terminal taxon is either the same as that of its ancestral node (a compositional inference) or is some other unique condition that is derived from the state of the ancestral node (a relational inference). The tree itself is equally consistent with either of these interpretations, and preference for the first requires an additional application of parsimony beyond that needed to determine phylogenetic relationships. In the present case, the seemingly innocuous assumptions that the Great Auk had a chick-rearing strategy and that this strategy is either precocial, semi-precocial or intermediate, justify compositional inferences.
Parsimonious optimization of the chick-rearing strategy character on the supertree (Fig. 6) unambiguously indicates that the Great Auk followed an intermediate strategy. With the alternative hypothesis (inset Fig. 6), the semi-precocial or intermediate reconstructions are equally parsimonious, with unambiguous support only for the conclusion that the Great Auk did not follow a precocial chick-rearing strategy.
The phylogenetic analyses reveal that the Great Auk is most parsimoniously interpreted as having an intermediate or semi-precocial (molecular tree) or an intermediate (morphological tree and supertree) chick-rearing strategy. The results unambiguously suggest that the chick-rearing strategy of the Great Auk was not precocial and they favour the inference that it was intermediate.
Birkhead (1993) suggested that the possibility that the Great Auk was precocial should not be dismissed on evolutionary grounds, because this would be to disregard the unique way in which natural selection has acted on this bird, rendering it flightless. It should be noted, however, that Livezey (1988) and Bengtson (1984) argue that being flightless is an extreme adaptation in the Great Auk for more effective pursuit diving and less extreme adaptations achieving the same purpose are common throughout the auks. Increased body mass and wing loading are seen in the evolution of the alcids, the extremes of which characterize the Great Auk whose very high wing loading of 22 g cm−2 made it unable to fly (Livezey, 1988).
The energy that could be delivered to a chick has been investigated by the use of a time and energy budget model. The fact that the parent is limited by time for most parameter values simplifies the analysis because K and mr have no effect. We have found that the Great Auk should have had little difficulty in following an intermediate strategy provided that the foraging radius R was not too large. In contrast, there is a limited range of parameters within which it is possibility that the Great Auk followed a semi-precocial strategy. This strategy is possible if the Great Auk regurgitated food to its chick. The Little Auk (Alle alle) feeds its chick by regurgitating food brought back in a throat pouch (Bradstreet & Brown, 1985; Pedersen & Falk, 2001). Gaston & Jones report that anatomical evidence for this has not been found. Another possibility is that, like penguins, the Great Auk regurgitated food from its stomach (Prince & Harris, 1988). If the chick was not fed by regurgitation, it seems unlikely that the semi-precocial strategy could have been adopted if the foraging radius was greater than 15 km, and even a foraging radius of 10 km would have required a high gain rate. Houston et al. (1996) showed that, for the Thick-billed Murre, foraging within a smaller than usually observed radius could allow them to adopt the semi-precocial strategy rather than the intermediate one. So the possibility that the Great Auk could have reared its chicks to 65% of its own mass does not mean that the chicks did not follow an intermediate strategy. It can be seen that following an intermediate strategy would certainly have been more energetically ‘comfortable’ for the Great Auk and this is made even more likely by its phylogenetic position within the Alcidae.
We thank Michael Jennions and an anonymous referee for comments on a previous version of this ms.