University of Groningen
Costs of migration
Schmidt-Wellenburg, Carola Andrea
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to
cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2007
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Schmidt-Wellenburg, C. A. (2007). Costs of migration: Short- and long-term consequences of avian
endurance flight s.n.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.
Download date: 15-06-2017
CHAPTER 2
Energy expenditure and wing beat frequency in
relation to body mass in free flying Barn Swallows
(Hirundo rustica)
Carola A. Schmidt-Wellenburg, Herbert Biebach,
Serge Daan, and G. Henk Visser
Journal of Comparative Physiology B: Biochemical, Systemic, and
Environmental Physiology (in press), DOI 10.1007/s00360-006-0132-5
ABSTRACT
CHAPTER 2
Many bird species steeply increase their body mass prior to migration. These
fuel stores are necessary for long flights and to overcome ecological barriers.
The elevated body mass is generally thought to cause higher flight costs. The
relationship between mass and costs has been investigated mostly by
interspecific comparison and by aerodynamic modelling. Here, we directly
measured the energy expenditure of Barn Swallows (Hirundo rustica) flying
unrestrained and repeatedly for several hours in a wind tunnel with natural
variations in body mass. Energy expenditure during flight (ef, in W) was found to
increase with body mass (m, in g) following the equation ef = 0.38 * m0.58. The
scaling exponent (0.58) is smaller than assumed in aerodynamic calculations
and than observed in most interspecific allometric comparisons. Wing beat
frequency (WBF, in Hz) also scales with body mass (WBF= 2.4 * m0.38), but at a
smaller exponent. Hence there is no linear relationship between ef and WBF. We
propose that spontaneous changes in body mass during endurance flights are
accompanied by physiological changes (such as enhanced oxygen and nutrient
supply of the muscles) that are not taken into consideration in standard
aerodynamic calculations, and also do not appear in interspecific comparison.
20
INTRODUCTION
The energetic costs of transport have been approached both theoretically, from
aerodynamic theory, and empirically, from interspecific allometry. Aerodynamic
predictions are rather equivocal. The basic assumptions (Norberg 1996, Rayner
1990) lead to a prediction of scaling of the energetic costs during flight with a
mass exponent of 7/6 (= 1.17). If morphometric data are taken into account the
exponents derived from modelling of mechanical power increase to 1.59 (Rayner
1990). In Pennycuick's model "flight" (Version 1.10) an exponent of 1 is used.
Allometric studies based on direct measurements of energy expenditure in
different species also yield a wide range of exponents: from 0.74 (Butler and
Bishop 2000) to 1.36 (Rayner 1990). These interspecific equations compare
birds with different morphology. Some studies take morphology into account by
including parameters such as wing length, wing area and aspect ratio in the
models fitted. In such studies the mass exponents tend to be larger (0.87 - 1.93,
Rayner 1990) than when only body mass is included. Even these calculations
are not satisfactory for considerations at the intraspecific or individual level. In
the intraspecific comparison there is little variation in morphology such as wing
shape, but individual physiology may well change with spontaneous changes in
mass, for instance on an annual basis. In the Red Knot (Calidris canutus) during
the migration seasons major changes were reported for pectoral muscle mass,
size of the stomach and intestines, and fat stores probably all having an impact
on the basal metabolic rate (Piersma et al. 1996, Biebach and Bauchinger
2003). In the interspecific comparison, both physiology and morphology
represent adaptive states coevolved with body mass. Hence inter- and
FLIGHT COSTS AT DIFFERENT BODY MASSES
Birds can accumulate internal energy stores to cover periods of food shortage.
While fat is the main substrate (Blem 1976, 1980), protein stores can also be
increased (Lindström et al. 2000). The optimal amount of energy stored depends
on the balance between benefits and costs. Energy resources accumulated
increase the probability of surviving during cold exposure or food shortage. This
also holds for migratory flights, when birds have to cross ecological barriers such
as the Mediterranean Sea or the Sahara Desert. Energy stores also relieve them
temporarily from the need to forage such that they can allocate more time to
other activities. On the cost side, increased mass reduces manoeuvrability and
take-off ability and increases predation risk (Lindström and Alerstam 1992, Lind
et al. 1999, Nudds and Bryant 2002). There is only a small metabolic cost of
maintenance of fat (Schmidt-Nielsen 1997), but carrying an increased mass
during flight in general may well raise the energetic costs of locomotion in a
significant way. Little direct information on these costs is available.
21
intraspecific scaling may yield different dependencies of energy turnover on
body mass.
Two empirical studies have addressed the effect of body mass on energy
expenditure during sustained flights at the intraspecific level. These were done
on Red Knots, with body mass ranging from 100 to 190 g (Kvist et al. 2001) and
Rose Coloured Starlings (Sturnus roseus, 56 - 87 g, Engel et al. 2006). Both
studies observed lower additional costs at a higher body mass than predicted by
aerodynamic theory (exponent 1.2 - 1.6) or interspecific allometry (exponent
0.7 - 1.4). Energy expenditure during flight scaled with body mass to the power
of 0.35 (Kvist et al. 2001) or 0.55 (Engel et al. 2006).
It would be of interest to know how this intraspecific scaling exponent itself
varies with the size of the species. The interspecific allometric scaling exponent
for mass-specific metabolic rates is negative. Thus, such rates tend to be higher
in smaller species, and lower in larger birds. It is possible that the cost of
transport of extra energy stores varies in an adaptive manner between birds of
different size. It is therefore of interest to further explore intraspecific relations
between body mass and flight costs. We measured the energetic costs of
changes in body mass in a smaller sized species, the Barn Swallow. The
spontaneous accumulation of energy stores during the migratory period resulted
in changes of body mass of up to 35% in these swallows. Individual birds were
repeatedly flown in the wind tunnel at different body mass exploiting these
natural variations, and their cost of flight was measured with the doubly labelled
water method.
MATERIALS AND METHODS
CHAPTER 2
Birds
22
We used the Barn Swallow, a long distance migrant, as a model species. Nine
birds were taken from nests at about day 7 of age and hand raised. The chicks
were fed a diet of heart, curd, crickets, and bee larvae, supplemented with
vitamins and minerals. We raised the birds in a large aviary so that they could
practise flight immediately after fledging. During the experiment, the birds were
kept in an aviary (l x w x h ca. 2 m x 5 m x 2 m), sufficiently spacious to allow
flight. They received standard food ad libitum (insects, heart, curd, rusk, and
egg, supplemented with minerals and vitamins) and had unlimited access to
fresh water. Day length followed the course of the swallows' natural year. From
the autumn equinox onwards, day length was set to LD 12.8:11.2 h. This reflects
the light conditions (12.8 h from dawn civil twilight till dusk civil twilight) of a
southward migration towards the equatorial region. We used Osram Biolux®
fluorescent lights simulating the spectral composition of natural sunlight. Body
mass varied naturally due to premigratory hypertrophy, followed by gradual
weight loss. No artificial mass was added, and the birds were not food restricted
to manipulate weight. Intraindividual variations (maximum-minimum) in body
mass ranged from 2 to 36% of the minimum body mass, similar to observations
in the field (Pilastro and Magnani 1997, Rubolini et al. 2002). In the middle of the
experimental phase, wing length was measured with a ruler to the nearest
0.5 mm (mean wing length was 11.9 cm, SD between individuals 0.3 cm).
The experiments were performed in the wind tunnel of the Max Planck Institute
for Ornithology in Seewiesen, Germany (Figure 1.3), situated at 688 m above
sea level (see Engel 2006). The technical specification of the closed-circuit
tunnel is nearly identical with the one in Lund, Sweden (Pennycuick et al. 1997).
The flight section is 2 m long with an octogonal cross section of 1.2 m high and
1.2 m wide. It consists of transparent acrylic plates and glass to allow
observation of the birds during flight. The air speed distribution is homogeneous
over the entire flight section (in cross-section as well as longitudinally) and
turbulence in the flight section is negligible (0.04% at 10 m s-1; Engel 2005).
Downstream from the flight section there is a gap of 0.5 m giving access from
outside. The birds enter the tunnel through it, after which it is closed with netting.
The wind tunnel is secured with mist nets upstream of the flight section and
2.3 m downstream of the gap.
True air speed (and hence flight speed) was set at 10.3 m s-1 (SD over all 21
flights 0.04 m s-1). We chose this speed on the basis of observations of the birds'
flight performance. In a pilot study, we recorded the wing beat frequencies
(WBF) of one bird flying over a range of air speeds of 8.4 to 10.9 m s-1. WBF
followed a U-shaped curve, from which - analogously to the 'maximum range
speed' of the power curve (see Pennycuick et al. 1996, Bruderer et al. 2001) we calculated 10.4 m s-1 as the speed where number of wing beats per distance
covered is minimized. According to 'minimum power speed', the speed with least
wing beats per time was calculated as 9.6 m s-1 (Figure 2.1). During the
experiments, mean air pressure was 932.9 mbar (SD 9.2 mbar) and relative
humidity 65.7% (SD 11.5%), and air temperature was held constant at 16.4°C
(SD over 21 all flights was 0.8°C).
FLIGHT COSTS AT DIFFERENT BODY MASSES
Wind tunnel and flight speed
23
7.6
wing beat frequency [Hz]
7.4
7.2
7.0
6.8
6.6
6.4
1
2
6.2
8.0
8.5
9.0
9.5
10.0
10.5
11.0
11.5
-1
flight speed [m s ]
Figure 2.1
Wing beat frequencies (WBF, in Hz) of one Barn Swallow flying over a range of speeds
(average ± SE). From the fitted curve (WBF = 0.4*V² - 7.6*V + 43.0; with V as flight
-1
-1
speed, in m s ) we calculated 'minimum WBF speed' as 9.6 m s (1) and 'minimum WBF
-1
per distance-speed' as 10.4 m s (2).
Experimental protocol
CHAPTER 2
Prior to the experimental flights, the birds were trained to fly in the wind tunnel
for prolonged periods of time (2 - 6 h). All birds flew during one autumn migration
period (end of September until end of November). Some birds flew first with low
and later with high body mass, others vice versa.
24
The birds flew in pairs for 3.6 - 6.4 h (mean 5.4 h, SD between 21 flights 1.0 h),
covering on average 200 km. When birds tried to land, we chased them by
waving our hand at them. However, if birds interrupted the flight by landing or
tried to land repeatedly, they were allowed to rest for at least 1 h before
continuing the flight. Such flights are referred to as "interrupted", whereas
continuous flights are referred to as "non-stop". During resting periods of
interrupted flights birds were put in a box (ca. 0.2 m x 0.2 m x 0.2 m) to avoid
high locomotory activity. During the flights,
the time, standing next to the wind tunnel.
before and after the flight to the nearest
expenditure and wing beat frequency, we
mass during the experimental flight (m).
the observer was with the birds all
Body mass was taken immediately
0.01 g. In the analysis of energy
used the calculated average body
Energy expenditure was measured with the doubly labelled water method
(DLW). The evening before the experiment, food was removed from the aviaries
to enhance the accuracy of the DLW measurement. The next morning the birds
were injected intraperitoneally with about 0.11 g of a DLW mixture (enriched in
18
O by 59.9 atom percent, and in 2H by 36.7 atom percent). The injected dose
was weighed on an analytical balance (Sartorius BP 1215) to the nearest
0.1 mg. After the injection, the bird was placed in a dark box without access to
water and food to let the DLW mixture equilibrate with the bird's water pool. One
hour later, the bird was weighed and the "initial blood sample" of about 60 µl was
taken from the jugular vein. The jugular vein was chosen to avoid any
impairment of flight performance as might arise by any accidental haemorrhage
at the brachial vein. After the flight, usually 6.6 h after the initial blood sample,
the bird was weighed again, and another blood sample was taken as described
before: the "final sample". To calculate total body water at the end of a flight and
thus increase the accuracy of the DLW method, some birds were reinjected with
DLW afterwards, followed by another blood sample after one hour of
equilibration in a dark box. All blood samples were divided over four capillaries,
sealed immediately, and stored at 5°C for isotope analysis (see below). From
five birds blood samples were taken prior to the injection of isotopes to
determine the average background levels for 2H and 18O during our study, which
were about 0.0145% for 2H and 0.200% for 18O. In view of the high difference
between natural background levels and injected dose, we refrained from taking
more background measurements (Nagy 1980).
The isotope analyses were performed in triplicate or quadruplicate at the Centre
for Isotope Research according to the method described by Visser et al. (2000).
Briefly, for each sample 2H/1H and 18O/16O isotope ratios were determined with
the CO2 equilibration method and the uranium reduction methods, respectively
(Speakman 1997). The coefficients of variation for 18O and 2H enrichments
relative to the background levels were 0.97% and 0.76%, respectively. Rates of
CO2 production were calculated as described by Engel et al. (2006). As a last
FLIGHT COSTS AT DIFFERENT BODY MASSES
Energy expenditure during flight (ef)
25
step, these values were converted to energy expenditure using a conversion
factor of 27.8 kJ l-1 (Gessaman and Nagy 1988a). On average the turnover rate
of 18O was 2.96 (SD 0.277) times higher than that for 2H. As a consequence, the
DLW method as employed in this study appeared to be rather insensitive to
analytical errors. For example, if we changed one of the measured isotope
values by 1%, the calculated level of energy expenditure changed by 3.5% on
average.
The DLW method integrates the energy expenditure over the whole
measurement period between the "initial" and the "final" samples. This includes
both the energy expenditure during flight (ef) and the energy metabolised during
times in which the birds did not fly (er). In the birds with interrupted flights the
interruptions considerably affected the mean energy metabolism measured. As a
sensitivity analysis, we calculated the range of ef estimates using two
assumptions of energy expenditure during non-flight periods: (1) Neglecting the
energy expenditure during the resting time. These calculations are indicated as
efmax [W]; (2) Assuming the same energy levels during flight and non-flight.
Calculated flight costs are referred to as efmin [W]. efmax and efmin are the most
extreme estimates of the actual flight costs, with efmax providing the upper and
efmin the lower boundary. As true flight costs lie between these boundaries, we
estimated er using the DLW sessions with longer non-flight periods. The
estimated flight costs are indicated as efS (in W).
Wing beat frequency (WBF)
CHAPTER 2
Wing beat frequency (WBF, in Hz) was visually evaluated from digital video
records (Canon XL 1) of the experimental flights: we counted the wing beats
from the video tape every half hour for about 10 min (with a resolution 25
pictures s-1). We selected sequences of at least 15 wing beats of the birds
staying stationary, and evaluated 20 such sequences during each interval. The
Barn Swallows rarely flew with regular wing flapping. Mostly they showed
bounding flight with short "upstroke pauses" (Liechti and Bruderer 2002,
Bruderer et al. 2001, Pennycuick et al. 2000). These pauses were too brief to
quantify their duration. Therefore, WBF refers to the "effective wing frequency",
including sequences of bounding flight. As WBF did not change during the
flights, mean WBF per flight was used for further analysis.
26
Statistical analysis
The statistical analysis was performed in SPSS 13.0. Energy expenditure during
flight (ef) and wing beat frequency (WBF) were analysed in separate models with
the emphasis on unravelling patterns related to body mass. As the calculations
of ef in interrupted flights were very sensitive to the assumption on metabolism
during times the birds did not fly, we restricted the analysis of ef to non-stop
flights. We used mixed models with Restricted Maximum Likelihood for both
parameters. Parameters that did not significantly contribute to the explained
variance were excluded from the models in stepwise fashion. During the
elimination procedure, the AIC value was used as a criterion for the
improvement of the model. For the analysis of the effects on log10(ef) we initially
included individual as random and log10(m), log10(WBF), log10(WBF)*log(m), and
wing length as fixed factors. The final model for log10(ef) in non-stop flights is
based on individual and log10(m). For log10(WBF), we first included individual as
random and log10(m), flight performance (interrupted or non-stop), flight
performance*log10(m), and wing length as fixed factors. The final model for
log10(WBF) is based on individual and log10(m).
RESULTS
We measured energy expenditure in 21 flights. 13 sessions (in six individuals)
had non-stop flights lasting on average 92% of the session, i.e. of the total time
between "initial" and "final" blood sample. There were eight sessions (in seven
individuals), where flights were interrupted. These flights lasted on average for
66% and at most for 77% of the experimental time. We do not yet understand
why flights became interrupted. Two birds flew exclusively non-stop (two and
three flights), three individuals performed interrupted flights only (four flights). In
four birds, which flew both non-stop and interrupted, the interrupted flight was
always at the lightest body mass measured (eight non-stop and four interrupted
flights).
As non-flight time in the interrupted flights was on average 34% and
assumptions on metabolism during non-flight strongly affect the calculated ef, we
do not think that these data allow us to carefully measure the cost of flight in
these flights. Hence we restrict the analysis of flight cost to the 13 non-stop
flights. Even in these sessions, we need to allow for a different metabolic rate
during the non-flying time than during flight. This metabolic rate during rest,
FLIGHT COSTS AT DIFFERENT BODY MASSES
Energy expenditure during flight (ef)
27
indicated as er [W], is unknown, but it must be somewhere between two extreme
boundaries: a minimum of 0 and a maximum equal to the metabolic rate during
flight (ef, in W). We first derive flight costs from the non-stop flights for these two
extreme assumptions, and then we use the interrupted flights to obtain a more
realistic estimate in between these boundaries. Body mass in non-stop flights
ranged from 14.5 to 22.2 g, with an average of 18.5 g (SD 2.3 g). Birds flew on
average for 6.0 h (SD 0.5 h).
Upper and lower boundaries to flight costs
Neglecting the resting time, i.e. setting the metabolism er during non-flight at 0
W, leads to an average efmax of 2.28 W (range 2.06 - 2.79 W, SD 0.23 W).
efmax [W] increased with body mass m [g] following the equation (Table 2.1)
log10(efmax)= -0.292 + 0.511 * log10(m).
(equation 2.1)
Minimal flight costs can be computed under the assumption that resting is
equally costly as flight. (er = ef). This yields for the non-stop flights an average
efmin of 2.09 W (range 1.68 - 2.46 W, SD 0.20 W). The relationship between
efmin [W] and body mass m [g] is described as (Table 2.1)
log10(efmin) = -0.440 + 0.594 * log10(m).
(equation 2.2)
The upper and lower boundaries are plotted on double logarithmic scale as a
function of body mass in Figure 2.2.
These extreme assumptions provide us with a well-defined range in which the
true flight costs must lie. That this range is narrow is simply due to the fact that
the analysis is restricted to those sessions where there was little rest, so that
assumptions concerning er have not too much impact. Yet in view of the
unrealistic nature of these extreme assumptions, it is desirable to have a more
realistic estimate.
CHAPTER 2
Estimate of flight costs
28
We cannot apply basal metabolic rates such as provided by Gavrilov (1.41
kJ d-1 g-1) and Kespaik (1.51 kJ d-1 g-1, both cited in Gavrilov and Dolnik 1985) to
estimate er, as these are standardly derived during night. A better estimate of er
in the experimental situation can actually be derived from the eight sessions with
interrupted flights. We have used the DLW measurements in these sessions as
follows. We first assumed that flight costs in these sessions followed equation
2.2, i.e. at the minimum boundary. Subtracting the energy thus calculated for
flight from the total energy turnover in each session yielded a figure Er [Joule] for
the energy spent during rest time [seconds], which could be translated into Watt.
We then used the average mass-specific er (in W g-1) of these interrupted flights
to estimate Er in the 13 non-stop flights again and obtain a new ef value. As
before, we analysed the relation of ef and body mass for the non-stop flights in a
mixed model. With the new relation of ef and body mass, we again estimated ef
and consequently er in interrupted flights. We continued the iterative procedure
until the approximation was stable. This results in average efS of 2.13 W in the
non-stop flights (range 1.74 - 2.55 W, SD 0.21 W). efS [W] correlates with body
mass m [g] as
log10(efS) = -0.416 + 0.580 * log10(m).
(equation 2.3)
Table 2.1
Effects of log10(m) on log10(ef) (N= 13 in six individuals, flight time >80%). log10(WBF) did
not significantly affect log10(ef) and was excluded from the models.
Estimated
SE
effect
df
t
F
95% CI
p
intercept
log10(m)
-0.292
0.511
0.20
0.16
10.68
10.06
-1.47
3.28
2.16
10.73
-0.73 - 0.17
0.17 - 0.86
<0.05
intercept
log10(m)
-0.440
0.594
0.22
0.17
8.94
8.81
-1.99
3.43
3.96
111.77
-0.94 - 0.06
0.20 - 0.99
<0.05
intercept
log10(m)
-0.416
0.580
0.22
0.17
8.61
8.48
-1.91
3.40
3.64
11.53
-0.91 - 0.08
0.19 - 0.97
<0.01
log10(efmax)
log10(efmin)
log10(efS)
FLIGHT COSTS AT DIFFERENT BODY MASSES
For statistical details see Table 2.1. The scaling of the estimate of efS with body
mass is displayed in Figure 2.3. Irrespective of the different assumptions on er,
the slope of the relation ef and m is very robust and in all cases relatively low
(between 0.51 and 0.59).
29
3.2
energy expenditure during flight [W]
3
2.8
2.6
2.4
2.2
2
1.8
1.6
1.4
14
16
18
20
22
24
body mass [g]
Figure 2.2
Upper and lower boundary estimates of Energy expenditure during flight (ef in W) in
relation to body mass (in g), plotted on a double-logarithmic scale (N= 13 flights in six
individuals). The maximum and minimum estimates for ef are indicated by black symbols
and a solid line (efmax) and grey symbols and a dashed line (efmin), respectively. Different
estimates for single flights are connected by lines. The grey shaded area between the
two regressions indicates the range of the most extreme predictions of ef.
CHAPTER 2
Figure 2.3 (opposite page)
30
Energy expenditure during flight, ef in W, plotted in relation to body mass (in g) on a
double logarithmic scale. The regression (solid black line) refers to efS (dots). The grey
shaded area is the area between the upper and lower boundary (grey solid and dashed
lines, respectively) taken from Figure 2.2.
3.2
energy expenditure during flight [W]
3
2.8
2.6
2.4
2.2
2
1.8
1.6
1.4
14
16
18
20
22
24
body mass [g]
Figure 2.3
Mean wing beat frequency of all 21 flights was 7.00 Hz (range 5.67 - 8.18 Hz,
SD 0.62 Hz among 21 flights in nine individuals). Average body mass was
17. 8 g (range 14.5 - 22.2 g, SD 2.25 g). WBF did not change during flight (i.e.
we found no differences between the beginning of the flight, after 30 min, after
150 min, and as an average over the whole flight time). WBF was positively
correlated with body mass (linear mixed model for log10(WBF) in 21 flights (nine
individuals), with individual as random and log10(m) as fixed factor; F17.0= 9.3,
t= 3.05, p< 0.01 for log10(m), 95% CI of the slope was 0.12 - 0.64). This
relationship can be described as (Figure 2.4)
log10(WBF)= 0.381 + 0.376 * log10(m).
(equation 2.4)
WBF did not explain ef and was not correlated with either of the estimates of ef
(Spearman's rho with efmax= 0.082, p> 0.05, and with efmin= 0.253, p> 0.05,
N= 13).
FLIGHT COSTS AT DIFFERENT BODY MASSES
Wing beat frequency (WBF)
31
wing beat frequency [Hz]
8
7.5
7
6.5
6
5.5
14
16
18
20
22
24
body mass [g]
Figure 2.4
WBF in relation to body mass on a double logarithmic scale (N= 21 flight in nine
individuals). Filled circles refer to non-stop and open squares to interrupted flights, the
common regression (equation 2.4) is indicated with a solid line.
DISCUSSION
CHAPTER 2
Energy expenditure (ef)
32
Five previous measurements of energy expenditure during flight in Barn
Swallows in the field have been published. All five were taken during the
breeding season and calculated flight costs via time-budgets or mass-loss:
according to Hails' (1979) calculations, a Barn Swallow of 18.5 g would fly at
1.27 W, Lyuleeva's (1970) measurements would result in 1.39 W, and Turner's
would range from 1.57 (Turner 1982a, b,) to 1.90 W (1983). Based on
Westerterp and Bryant's measurements (1984) we would expect flight costs of
1.51 W in a bird of 18.5 g. The average efS of 2.13 W observed in our study thus
is about 10% - 70% higher than the flight costs of Barn Swallows estimated in
the field. Even our lowest estimates of flight costs (efmin, 2.09 W) are 11 - 66%
higher. Generally, wind tunnel measurements seem to yield higher ef than
measurements during free flight in the field (Masman and Klaassen 1987,
Rayner 1994).
True flight costs in the interrupted flights are difficult to assess, as they are
sensitive to assumptions on energy expenditure during non-flight, er. Estimates
of efmax yielded higher flight costs in interrupted than in non-stop flights,
especially at a lower body mass. Flight costs might well have been higher in
these flights: We chased the birds that landed repeatedly. This may have
increased their energy expenditure since it is known that take-off flights as well
as the starting phase of a flight are more expensive than steady state flights
(Nudds and Bryant 2000). Calculating ef assuming the same energetic costs
during flight and rest resulted in slightly lower efmin in the interrupted than in nonstop flights. We cannot distinguish whether in interrupted flights (1) flight costs
were higher than in non-stop flights, (2) energetic costs during non-flight were
higher than the often assumed resting metabolism during daytime of 1.25 x
basal metabolic rate, or whether (3) a combination of both effects occurred.
The sensitivity analysis showed, that estimates of ef in non-stop flights were very
robust with regard to different assumptions on er due to the fact that the birds
flew during nearly all of the DLW session. On average, the estimates for the
upper and lower boundary differed by only 0.19 W, i.e. efmax was on average 9%
higher than efmin. The following discussion on the effects of changes in body
mass on ef is based on our best estimate of ef.
Aerodynamic theories
Theoretical aerodynamic literature (Table 2.2.a) reports mass exponents for the
energy expenditure of avian flight which range from 1.16 to 1.59 (Pennycuick
1975, 1978, Rayner 1990, Norberg 1996). We found an exponent of 0.58
(95% CI 0.19 - 0.97). This was significantly lower than the lowest theoretical
value. The applicability of aerodynamic theories is difficult. They are based on
FLIGHT COSTS AT DIFFERENT BODY MASSES
Assuming the same flight costs for interrupted and non-stop flights, we obtained
an improved estimate for er. Mass-specific er in these sessions was on average
0.09 W g-1. That is 320% more than the 0.02 W g-1 calculated from Gavrilov's
and Kespaik's measurements (Gavrilov and Dolnik 1985) as 1.25 x basal
metabolic rate (the factor translating BMR into the active phase of the day,
Aschoff and Pohl 1970). This difference may appear to be high, but our birds
were neither kept in the dark nor at thermoneutrality during non-flight. Average
flight costs (efS) measured in this study equal 6.8 x estimated BMR or 1.3 x
estimated er.
33
fixed wing theory and better applicable to airplanes than to birds (Videler 2005).
Flight costs predicted from theory are often lower than directly measured. That is
not only a problem of assumptions on conversion efficiency, which are
necessary to translate (predicted) mechanical power into chemical power, ef.
Also other assumptions in the theory may not have robust foundations: In a
study by Pennycuick et al. (2000) on a Barn Swallow flying in a wind tunnel,
predicted flight costs were 50% lower than the costs calculated from flight
behaviour. The authors fitted the prediction by empirically changing the
assumptions for body drag and profile power ratio, but without aerodynamic
explanation.
Interspecific comparisons
In interspecific allometries mass exponents range from 0.67 (Videler 2005) to
1.93 (Rayner 1990). Wind tunnel studies in general appear to yield small
exponents, around 0.7 - 0.9, whereas allometric comparisons that include other
studies yield exponents of 0.8 - 1.4 (Table 2.2.b). The exponent of 0.58
measured in this study lies below this range, but is not significantly different from
Videler's value (0.67). The scaling exponent measured in our study has a wide
confidence interval due to the unavoidably small range of body mass in Barn
Swallows, and therefore it overlaps with some of the interspecifically observed
scaling exponents.
In migratory flights, the question of costs for transporting fuel cannot be
answered by interspecific allometric comparison. These regressions are based
on diverse species, with not only differences in size, but also in morphology,
ecology and evolutionary background. In addition, an increase in body mass
within individuals is qualitatively distinct from differences in body mass between
species. Within individuals, mainly fat is deposited, which is metabolically rather
inert and does not increase maintenance costs and basal metabolic rate as
other tissues would. Among species, larger birds are not necessarily fatter than
CHAPTER 2
Table 2.2 (opposite page)
34
Regressions for energy expenditure during flight in relation to body mass. The first
column gives the exponent of the correlation of ef and body mass. In the second column
the different equations are listed. Pout refers to mechanical power output, ef to total power
consumed during flight, M (m) is body mass in kg (g), b wing span in m, and S wing area
in m². Comments and study species are followed by the sources quoted in the last
column.
Table 2.2
Exponent
Equation
Comment / Species
Reference
a) Aerodynamic measurements and predictions
1.05
1.16
1.16
7/6
Pout = 12.9M1.05
Pout = 15.0M1.16
Pout = 14.95M1.16
Pout ~ M7/6
1.5
1.19
1.56
1.59
1.59
Pout ~ M1.5
Pout = 10.9M1.19
Pout = 24.0M1.56b-1.79S0.31
Pout = 27.21M1.59b-1.82S0.275
Pout = 27.21M1.59
at minimum power speed
at maximum range speed
at maximimum range speed
Theoretical prediction
Norberg 1990
Norberg 1996
Rayner 1990
Rayner 1990, Norberg
1996
Isometric wing morphology Pennycuick 1975, 1978
at minimum power speed
Norberg 1996
at minimum power speed
Norberg 1996
at maximum range speed
Rayner 1990
at maximum range speed; Rayner 1990
for individuals
0.67
ef ~ 60M0.67
maximum ef
Videler 2005
0.68
0.74
0.76
ef = 44.54M0.682
ef = 60.5M0.735
ef = 0.305m0.756
non-passerines
wind tunnel
Rayner 1990
Norberg 1996
non-wind tunnel
0.76
0.79
ef = 58.8M0.76
ef = 0.471m0.786
wind tunnel
wind tunnel
0.79
0.81
0.81
0.83
0.85
0.87
0.87
ef = 66.97M0.79
ef = 57.3M0.813
ef = 66.17M0.814
ef =1875.76M0.834b-1.690S1.732
ef = 49.4M0.851
ef =75.87M0.866b-0.175S0.279
ef = 69.5M0.87
all studies
Masman and Klaassen
1987
Butler and Bishop 2000
Masman and Klaassen
1987
Rayner 1990
Norberg 1996
wind tunnel
non-passerines
Rayner 1990
Rayner 1990
DLW; time-energy budget
wind tunnel
Norberg 1996
Rayner 1990a
wind tunnel
1.01
ef = 17.36M1.013b-4.236S1.926
1.36
1.37
1.51
1.93
ef = 542.73M1.355
ef = 51.5M1.37b-1.60
ef = 98.39M1.505b-2.539S0.236
ef = 39.84M1.93b-1.690S-0.553
Butler
and
Bishop
2000a
Masman and Klaassen
1987
Rayner 1990
all studies
passerines
all studies
Norberg 1996
Rayner 1990
passerine
Rayner 1990
c) Direct measurements: intraspecific
a
0.35
ef = cm0.35
cmin = 2.24, cmax = 2.57
non-passerine/
Calidris canutus
Kvist et al (2001)
0.55
0.58
ef = 0.74m0.55
ef = 0.38m0.58
passerine/ Sturnus roseus
passerine/ Hirundo rustica
Engel et al (2006)
this study
after Masman and Klaassen (1987)
FLIGHT COSTS AT DIFFERENT BODY MASSES
b) Direct measurements: interspecific
35
smaller ones. Instead they carry larger quantities for instance of muscle tissue,
which is metabolically more active than fat. Videler (2005) showed that the
dimensionless cost of transport (amount of work per unit weight and distance
covered) is not independent of body mass, but decreases with size. He
concluded that small and large birds are not scale models of each other.
Interspecific comparisons of ef are commonly based on different methods of
measurement. In wind tunnel studies, ef has often been measured by mask
respirometry, which adds additional mass and changes aerodynamics.
Calculating energy consumption via mass loss is prone to error when ambient
temperature is not defined. Many studies are necessarily restricted to short
flights, which have to be seen in a different ecological and physiological context.
Energy expenditure at the beginning of a flight, and thus during short flights, is
higher than in a later phase (Nudds and Bryant 2000). Furthermore, birds may
well adjust physiological traits to time. Flight costs during summer can therefore
be different from flight costs during long distance migration. Carrying food during
short foraging flights is a different task than flying for hundreds of kilometres
without rest and refuelling. These situations may well be associated with
different physiological adaptations.
Intraspecific scaling
CHAPTER 2
All three intraspecific studies on the effect of body mass on ef used DLW to
measure energy expenditure. The fraction of time in flight was similar in both
cases to our study (96% in Red Knots, Kvist et al. 2001; 94% in Rose Coloured
Starlings, Engel et al. 2006). er was corrected by 1.5 x basal metabolic rate in
Red Knots, and 1.25 x basal metabolic rate in Rose Coloured Starlings.
36
Both studies found mass exponents (Red Knot 0.35; Rose Coloured Starling
0.55) that were similar or slightly smaller than those observed in the Barn
Swallow (0.58). We may thus safely conclude that the dependence of flight costs
on spontaneous changes in body mass is much less steep than in the
comparison between species. This also holds for a small species like the Barn
Swallow. Kvist et al. (2001) proposed a change in flight muscle efficiency in
heavier birds allowing them to fly at lower costs than expected when fuelling up
for migration. A similar mechanism may be applicable to Barn Swallows.
However, it is not a priori clear why flight muscle efficiency should not be
maximized by birds with a lower mass.
Wing beat frequency
Implications
We showed that ef of Barn Swallows flying for long periods of time increases
less with body mass than aerodynamic calculations and interspecific
comparisons suggest. Benefits of a shallow increase of flight costs are obvious.
During migration, birds can reach farther regions and they are less dependent
on stopover sites (Weber et al. 1994). Heavy birds could reach their destination
faster as they might increase their flight speed as predicted from aerodynamic
theory. The consequences of the intraindividual scaling become conspicuous,
when we calculate the flight costs of a Barn Swallow of 18.5 g that increases its
mass by 1 g: according to our measurements, energy consumption is 3.2%
higher in the heavier bird, whereas the general aerodynamic prediction (with an
exponent of 7/6, see Rayner 1990) predicts an increase of 6.6%. With the
passerine exponent (1.355, Rayner 1990), the increase would be 7.7%, twice
FLIGHT COSTS AT DIFFERENT BODY MASSES
The average wing beat frequency of 7.0 Hz observed lies well within the range
of observations on Barn Swallows flying in wind tunnels or in the field. In a wind
tunnel study, Pennycuick et al. (2000) observed 6.8 and 6.9 Hz in birds flying at
10 and 11 m s-1. Nudds et al. (2004) reported 7.0 Hz in a 19 g bird. Bruderer et
al. (2001) measured frequencies of 2.5 - 8.5 Hz (average 6.9 Hz) in birds flying
over a range of speeds in a wind tunnel. WBF during level flight was 7.2 Hz in
the wind tunnel, compared to 5.4 Hz during migration in the field (Liechti and
Bruderer 2002). Another field study recorded 8.2 Hz during coursing (Warrick
1998). The observed WBF is 20 - 30% lower than aerodynamic predictions for
birds of that size, which range from 8.4 Hz (Pennycuick 1990), and 9.1 Hz
(Pennycuick 1996) to 9.8 Hz (Pennycuick 2001) or even 12.9 Hz (Norberg
1990). This discrepancy may be attributed to the fact that Barn Swallows are
aerial feeders with a special wing shape and aspect ratio, which differ in flight
performance from other species. The observed exponent of the allometric
relationship between WBF and body mass was 0.36 (95% CI 0.12 - 0.64). This
range includes the value of 0.5 suggested by Pennycuick (1996) for the
intraindividual scaling exponent. At the individual level, morphological
parameters like wingspan and wing area are constant, and they are similarly
scaled within a species. We therefore suggest that Pennycuick's proposition for
the scaling of WBF with body mass holds both at the individual and at the
intraspecific level.
37
the increase observed. Predictions for migratory strategies or flight range have
thus to be applied carefully.
Why flight costs increase less with body mass than expected from aerodynamic
theory and from interspecific comparisons is not fully understood. Fat load in
migrants may cause only a small increase in energy consumption during flight,
as these species deposit more fuel in the hind part of the body which then is
compensated for by lift power of the body and tail (Dolnik 1995). It is possible
that migratory birds undergo a training effect accompanied by physiological
changes, resulting in different muscle fibre composition or an improved supply
with oxygen and nutrients. Also maintenance costs and basal metabolic rate are
lowered during migration, as most organs are reduced in size (Piersma et al.
1996, Biebach and Bauchinger 2003). This may reduce flight costs at least
during long migratory flights. Not only physiological parameters but also flight
behaviour can be flexible: stroke amplitude, the extent to which the wings are
stretched and flexed, duration of upstroke pauses or wing beat frequency are
adaptable and are probably adjusted to different requirements. That becomes
obvious in the fact that ef cannot be predicted from WBF. WBF scales with body
mass with a lower exponent than does ef, indicating higher energy expenditure
per wing beat at an increased body mass. But as mentioned above, other
parameters of flight behaviour, which have not been measured, are flexible as
well. Mechanisms of these changes in flight behaviour and physiology are not
yet understood and require further investigation.
ACKNOWLEDGEMENTS
CHAPTER 2
We remember with gratitude the late Prof. Dr. E. Gwinner, whose inspiration and
support let to this project. We thank Ulf Bauchinger, Brigitte Biebach, Sophia
Engel, and Andrea Wittenzellner for their support especially of a temporarily
one-handed student. Berthe Verstappen determined the isotope enrichments at
the Centre for Isotope Research. David Rummel from the Institute of Statistics of
the Ludwig-Maximilians-University Munich provided helpful feedback on
statistics. The experiment was conducted in accordance with the German
legislation on the protection of animals. We thank two anonymous reviewers for
comments on the manuscript.
38
Some wetland birds (…) feed on grain, roots, seeds, worms, crickets, beetles, and other
insects. (…) Other birds feed on fish, worms and snakes, lizards, frogs, mice, and the
like, which are poisonous for most animals and for humans.
(fol. 20 v, 21 r)