Hovering Energetics and Thermal Balance in Anna's Hummingbirds (Calypte anna)

Abstract

We studied the energetics of hover-feeding Anna's hummingbirds, using three different simultaneous techniques: heat loss as estimated via thermal imaging, metabolic rate as measured at a feeder mask using flow-through respirometry, and aerodynamic power estimated from wingbeat kinematic data. These three methods yielded comparable estimates of power output at ambient air temperatures ranging from 18 degrees to 26 degrees C, whereas heat imbalance at higher air temperatures (up to 34 degrees C) suggested loss by mechanisms other than convection and radiation from the body, such as evaporative cooling and enthalpy rise associated with exhaled air and excreted water and convective heat loss from the patagia. Hummingbirds increased wingbeat frequency and decreased stroke amplitude as air temperature increased, but overall muscle efficiency was found to be approximately constant over the experimental range of air temperatures.

Full text

Hovering Energetics and Thermal Balance in Anna’s Hummingbirds ( Calypte anna )
Author(s): DennisEvangelista, MaríaJoséFernández, MadalynS.Berns, AaronHoover, and
RobertDudley
Source:
Physiological and Biochemical Zoology,
Vol. 83, No. 3 (May/June 2010), pp. 406-413
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/10.1086/651460 .
Accessed: 11/12/2013 11:00
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
.
The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to
Physiological and Biochemical Zoology.
http://www.jstor.org
This content downloaded from 152.2.15.253 on Wed, 11 Dec 2013 11:00:44 AM
All use subject to JSTOR Terms and Conditions

406
Hovering Energetics and Thermal Balance in
Anna’s Hummingbirds (Calypte anna)
* Corresponding author; e-mail: devangel@berkeley.edu.
Physiological and Biochemical Zoology 83(3):406–413. 2010. 䉷2010 by The
University of Chicago. All rights reserved. 1522-2152/2010/8303-9091$15.00
DOI: 10.1086/651460
Dennis Evangelista
1,
*
Marı´a Jose´ Ferna´ndez
1
Madalyn S. Berns
2
Aaron Hoover
3
Robert Dudley
1,4
1
Department of Integrative Biology, University of California,
Berkeley, California 94720;
2
Department of Bioengineering,
University of California, Berkeley, California 94720;
3
Department of Mechanical Engineering, University of
California, Berkeley, California 94720;
4
Smithsonian Tropical
Research Institute, P.O. Box 2072, Balboa, Republic of
Panama
Accepted 1/20/2010; Electronically Published 3/26/2010
ABSTRACT
We studied the energetics of hover-feeding Anna’s humming-
birds, using three different simultaneous techniques: heat loss
as estimated via thermal imaging, metabolic rate as measured
at a feeder mask using flow-through respirometry, and aero-
dynamic power estimated from wingbeat kinematic data.These
three methods yielded comparable estimates of power output
at ambient air temperatures ranging from 18⬚to 26⬚C, whereas
heat imbalance at higher air temperatures (up to 34⬚C) sug-
gested loss by mechanisms other than convection and radiation
from the body, such as evaporative cooling and enthalpy rise
associated with exhaled air and excreted water and convective
heat loss from the patagia. Hummingbirds increased wingbeat
frequency and decreased stroke amplitude as air temperature
increased, but overall muscle efficiency was found to be ap-
proximately constant over the experimental range of air
temperatures.
Introduction
Among vertebrates, hummingbirds (Trochilidae, Apodiformes)
are among the smallest endotherms and exhibit extremely high
mass-specific metabolic rates (Weis-Fogh 1972; Suarez 1992).
In addition to their small sizes, hummingbirds are the only
birds capable of sustained hovering flight, an energetically de-
manding form of locomotion that is associated with high levels
of metabolic power input (Bartholomew and Lighton 1986;
Suarez et al. 1990; Suarez 1992) and mechanical power output
(Lasiewski 1963; Wolf and Hainsworth 1971; Epting 1980; Wells
1993; Chai and Dudley 1996). Because of their small size and
costly mode of locomotion, hummingbirds represent an im-
portant taxon with which to evaluate maintenance of endo-
thermic balance in the face of environmental challenge (Miller
1996).
Despite their high energetic cost of flight and exposure to
wide fluctuations in ambient air temperature, hummingbirds
maintain energy balance through varied behavioral and phys-
iological mechanisms such as entering nocturnal torpor and
altering foraging strategies according to environmental con-
ditions (e.g., Hixon and Carpenter 1988; Gass and Garrison
1999; Ferna´ndez et al. 2002). Hummingbirds also economize
by substituting heat generated during flight for that required
for thermoregulation (Berger and Hart 1972; Chai et al. 1998),
although the magnitude of this response may vary with body
size (see Welch and Suarez 2008). Variation in wingbeat ki-
nematics in response to variable air temperature may also alter
efficiency so as to advantageously augment metabolic heat pro-
duction (Chai et al. 1998; see also Zerba and Walsberg 1992).
Although hummingbird hovering energetics are now well
studied, the quantitative extent of heat dissipation has not been
evaluated simultaneously for comparison with estimates of met-
abolic and mechanical work. For starlings in forward flight, the
use of infrared thermography has enabled identification of dif-
ferent modes of heat loss as well as an independent estimate
of flight muscle efficiency (Ward et al. 1999). The specific mech-
anisms of heat retention used by hovering hummingbirds, pos-
sibly enabled by variable conductance and the use of thermal
windows, may be of broader ecological and evolutionary rel-
evance given the historical patterns of trochilid diversification
into colder montane habitats (Altshuler and Dudley 2002;
McGuire et al. 2007). To investigate these mechanisms, we an-
alyze simultaneous measurements of surface temperature, met-
abolic rate, and wingbeat kinematics during hovering flight of
Anna’s hummingbirds over a range of ambient air tempera-
tures. By evaluating temperature-dependent changes in kine-
matic, energetic, and efficiency variables, we seek to determine
whether elevated surface temperatures and heat dissipation are
limiting factors in hot air and whether excess heat loss via
convection limits hovering performance in the cold.
Material and Methods
Anna’s hummingbirds were captured in the wild in Berkeley,
California, and were acclimated to laboratory conditions over
This content downloaded from 152.2.15.253 on Wed, 11 Dec 2013 11:00:44 AM
All use subject to JSTOR Terms and Conditions

Hummingbird Hovering Energetics 407
Table 1: Morphological data for three male Anna’s
hummingbirds (Calypte anna)
Individual Mass (g) Wing Length (m) Aspect Ratio
Adult male 4.6 Ⳳ.3 .050 8.5
Adult male 5.0 Ⳳ.0 .050 9.2
Subadult male 4.7 Ⳳ.2 .059 8.3
Note. Average body mass was 4.8 g. Wing length and aspect ratio were
determined from photographs of the spread wings against a grid. Values rep-
resent mean Ⳳ1 SD.
Figure 1. Representative thermal images showing behavioral response to increased air temperatures. At low to midrange air temperatures (A),
heat loss is primarily from areas around the head, eyes, and pectoral muscles. At high air temperatures (B), heat loss is evident from the entire
body, including the wings. In addition, the feet are extended.
several days. Birds were maintained in individual mesh cages
(90 cm #90 cm #90 cm) with ad lib. access to a commercial
solution designed for nectar-feeding birds (Nektar-plus, Pforz-
heim, Germany). Morphometric data for the three study in-
dividuals (two adult males and one subadult male) are provided
in Table 1. At the completion of the experiments, all birds were
released into the wild, at the point of capture. Measurements
with individual birds were taken in several 4-h periods that
were spread over two and 10 d. Data were considered to be
from separate flight trials if measurements were separated by
at least 15 min of intermittent flight and perching.
Birds were trained to hover at a feeder suspended within a 90
cm #90 cm #90 cm nonhermetically sealed acrylic chamber.
Flight experiments were conducted at each of five nominal air
temperatures (mean ⳲSD: , , ,18⬚Ⳳ2⬚24.0⬚Ⳳ0.9⬚26⬚Ⳳ2⬚
, and C), with the order of presentation30.2⬚Ⳳ0.2⬚33.5⬚Ⳳ0.5⬚
chosen randomly. To obtain air temperatures above 24⬚C, a
small convection heater was used to regulate chamber tem-
perature. Ice baths on the chamber floor were used to cool the
chamber below 24⬚C. Chamber air was mixed regularly via the
bird’s periodic flight bouts; air temperature was measured at
the height of the feeder mask, where metabolic and surface
temperature measurements were also obtained. Through the
use of moving visual cues, birds were regularly stimulated to
fly in order to obtain longer hover-feeding bouts and to reduce
transient changes in body temperature at the start of hovering.
Relative humidity during our experiments was 56% Ⳳ9%
(mean ⳲSD); air temperature effects on humidity were non-
significant ( ). Air temperature was monitored duringPp0.17
experiments and varied no faster than 1⬚Cin10min.
Rates of oxygen consumption during hover-feeding were ob-
tained with an open-respirometry system (see Bartholomew
and Lighton 1986; Chai and Dudley 1995). Expired air was
pulled from the nares at a rate of 0.8 L min
⫺1
through a mod-
ified syringe (attached to the feeder) that functioned as a res-
pirometry mask. The air was then drawn through a column of
desiccant (Drierite, Xenia, OH) to remove water vapor. Oxygen
concentration of the airstream was recorded with a portable
oxygen analyzer (Foxbox; Sable Systems International, Las Ve-
gas, NV). Oxygen depletion was estimated as the difference
between baseline and minimum equilibrium values of oxygen
partial pressure, incorporating the rate of airflow and the effect
of ambient humidity. The calculated volume of oxygen con-
sumed was divided by the duration of the feeding bout to obtain
metabolic rate. A minimum of five hover-feeding bouts per
individual bird were obtained at each experimental air tem-
perature; feeding bouts of less than 2 s in duration were dis-
carded. A standard conversion factor of 20.1 J mL O
2
⫺1
was
assumed.
To determine the rate of heat loss from hovering birds, in-
frared (IR) thermal images of the hummingbirds were obtained
using a thermal imaging camera (Fluke, Everett, WA) operated
through a window cut in one wall of the flight chamber. Hov-
ering birds were filmed from a lateral perspective while at the
feeder (see Fig. 1); orthogonal perspectives were then obtained
by rotating the feeder about vertical through 90⬚and repeating
the procedure. An emissivity eof 0.95 for feathers was assumed
(Cossins and Bowler 1987). From the thermal images, the re-
gion of interest containing the hummingbird was first identified
manually. The image was then cropped and an optimal thresh-
old was chosen to segment the background from theforeground
(see Otsu 1979). Mean surface temperature was computed as
the area-weighted average of temperature pixels in the cropped
This content downloaded from 152.2.15.253 on Wed, 11 Dec 2013 11:00:44 AM
All use subject to JSTOR Terms and Conditions

408 D. Evangelista, M. J. Ferna´ndez, M. S. Berns, A. Hoover, and R. Dudley
Figure 2. A, Life-sized (4-cm-long) physical model was used to obtain heat transfer coefficients at typical induced velocities and surface
temperatures. Average surface temperatures were determined using the same thermal imaging camera and algorithm as for the live birds. B,
Representative surface temperature profile using infrared thermal imaging (red corresponds to 37⬚C, blue corresponds to 20⬚C). Variation in
the surface temperatures is due primarily to model grazing angle effects, which are identical to those in the live bird because they are in the
same position.
Figure 3. Metabolic rate measurements (in W) for all birds at
˙
Q
metabolic
all temperatures. for all birds is roughly constant at
˙
Q1.1 Ⳳ
metabolic
W, or about mL O
2
g
⫺1
h
⫺1
(mixed-mode regression,0.1 42 Ⳳ5
; individual effects not significant, ). For all re-Pp0.17 Pp0.10
maining results, power is expressed in Watts to allow side-by-side
comparison of metabolic rates, rates of heat loss, and mechanical
power.
image. The wings are ignored in the heat-transfer calculations,
as the thermal imaging camera could not reliably resolve the
fast-moving wings. In “Discussion,” we bound the potential
magnitude of heat loss from the wings on the basis of
convection.
The net rate of heat loss to ambient, (in W), was de-
˙
Q
loss
termined from both convective and radiative terms by the fol-
lowing equation (see Incropera and Dewitt 1996; Ward et al.
1999):
44
¯
˙
QphA(T⫺T)⫹jeA[(T⫹273) ⫺(T⫹273) ], (1)
loss s a s a
where is the overall average convective heat transfer coeffi-
¯
h
cient, W m
⫺2
K
⫺4
is the Stefan-Boltzmann
⫺8
jp5.67 #10
constant, and T
s
is the surface temperature and T
a
is the ambient
air temperature (both temperatures in ⬚C). This formulation
ignores evaporative heat loss and assumes that the surrounding
environmental temperature for radiative heat transfer is the air
temperature. Thermal images suggest that surrounding envi-
ronmental temperatures are close to ambient air temperatures.
In any case, radiative heat transfer makes up only about 5%
of the heat loss calculated with equation (1). To estimate the
heat transfer coefficient , a life-sized physical model of an
¯
h
Anna’s hummingbird (see Fig. 2) was constructed of a solid
piece of polymer clay (Sculpey; Polyform Products, Elk Grove
Village, IL) wrapped with 30-gauge nichrome wire and paper
tape (Johnson and Johnson, Langhorne, PA) characterized by
an emissivity eof 0.95 (Incropera and Dewitt 1996). The model
was heated by connecting a power supply to the nichrome wire,
yielding an estimated surface heat flux that is based on the
model’s surface area and the applied voltage and current within
the wire. Convective flow was imposed on the model using a
large fan operated at air speeds comparable to estimated values
of the induced velocity for hummingbirds (4.4–4.6 m s
⫺1
, cal-
culated according to Ellington 1984b; values from 3 to 6 m s
⫺1
were tested). Flow was measured with a hot-wire anemometer
(Kurz Instruments, Monterey, CA), sampling at 25 Hz, located
four body lengths from the model and parallel within the work-
ing section. Flow direction was effectively downward relative
to the model oriented in an appropriate feeding position (see
Fig. 2B). The aforementioned thermal imaging camera wasthen
used to obtain surface temperatures, with the model placed at
the feeder in the same orientation as a hovering bird. Measured
heat transfer coefficients were within a factor of 2 of that for
a sphere, indicating reasonable estimates for the nonstandard
hummingbird geometry (Incropera and Dewitt 1996; Ward et
al. 1999). The model test results were used to estimate con-
vective heat loss for each hovering trial, using the model’s mean
surface temperature and an induced airflow velocity based on
This content downloaded from 152.2.15.253 on Wed, 11 Dec 2013 11:00:44 AM
All use subject to JSTOR Terms and Conditions

Hummingbird Hovering Energetics 409
Figure 4. A, Average surface temperature, T
s
, determined from thermal
images. B, Rate of heat loss, , for all birds and all temperatures.
˙
Q
loss
T
s
increases significantly with ambient air temperature T
a
(Tp
s
, ; individual effects not significant, ).0.78T⫹9.32 P≤0.0001 Pp0.17
a
declines significantly with ambient air temperature as the driving
˙
Q
loss
temperature difference, is reduced (mixed-mode regression,T⫺T
sa
, ; individual effects not significant,
˙
Qp⫺0.06T⫹2.05 P!0.0001
loss a
). Blue arrow (at !20⬚C; a color version of this figure isPp0.34
available in the online edition of Physiological and Biochemical Zoology)
indicates onset of feather fluffing while individual is perched, as ob-
served in bird 2. Red arrow (at 130⬚C) indicates onset of wing spread-
ing, mouth gaping, and foot extension, as observed in all birds.Symbols
are as in Figure 3.
Figure 5. Overall average heat transfer coefficient, , for the physical
¯
h
model of Calypte anna is shown as a function of surface temperature,
T
s
, and airflow velocity ( , ). The
¯
hp137 ⫺3T⫹13.5uPp0.018
sind
star indicates the average of all values used in calculations forhovering
hummingbirds (see text). A color version of this figure is available in
the online edition of Physiological and Biochemical Zoology.
aerodynamic calculations. For all trials, induced velocity based
on kinematics was m s
⫺1
.4.5 Ⳳ0.2
To obtain kinematic data for estimates of mechanical power
output, hover-feeding hummingbirds were filmed ventrally
with a high-speed digital video camera (AOS Technologies, Ba-
den Daettwil, Switzerland) operated at 500 frames s
⫺1
. Video
sequences were analyzed frame by frame to obtain values of
wingbeat frequency and stroke amplitude (see Chai and Dudley
1995, 1996). Air density was determined from measurements
of barometric pressure. Body mass of individual birds was mea-
sured before and after each experimental series; the mean value
was used in aerodynamic calculations. Outstretched wings of
birds were photographed against graph paper and then digitized
using ImageJ (National Institutes of Health, Bethesda, MD) to
obtain morphological parameters relating to wing planform
(see Ellington 1984a).
Kinematic and morphological data were used to calculate the
mechanical power output using a standard model of animal
hovering (Ellington 1984b) modified to incorporate unsteady
drag coefficients as measured on a hummingbird wing in con-
tinuous rotation (Altshuler et al. 2004). Stroke plane angle was
assumed to equal 0, and simple harmonic motion was assumed
for wing movements within the stroke plane (see Chai and
Dudley 1995; Altshuler and Dudley 2003).
The mechanical power output required to hover is the sum
of the power required to overcome profile drag andthe induced
power required for weight support. Net inertial power to ac-
celerate the wings is taken to be 0 on the basis of the assumption
that the hummingbird flight apparatus exhibits full elastic stor-
age of wing inertial energy (Ellington 1984c). Drag on the body
during hovering flight was similarly assumed to be small com-
pared with profile power lost to drag on the fast-moving wings
(Ellington 1984c). Induced power, the power required to sup-
port body weight, follows from momentum balance and was
calculated following Ellington (1984c), as was profile power
using a wing drag coefficient (see Altshuler et
—
Cp0.139
D, pro
al. 2004).
Heat balance for the hovering hummingbird at thermal equi-
librium is given by
˙˙
˙
QpQ⫹W,(2)
metabolic loss mechanical
where positive is the rate of metabolic heat production
˙
Q
metab olic
(as measured via respirometry), positive is the rate of heat
˙
Q
loss
This content downloaded from 152.2.15.253 on Wed, 11 Dec 2013 11:00:44 AM
All use subject to JSTOR Terms and Conditions

410 D. Evangelista, M. J. Ferna´ndez, M. S. Berns, A. Hoover, and R. Dudley
Figure 6. A, Wingbeat frequency decreases with ambient air temper-
ature T
a
(individual effects significant, ). B, Stroke amplitudeP!0.001
increases with ambient air temperature (individual effects significant,
). A color version of this figure is available in the onlineP!0.001
edition of Physiological and Biochemical Zoology.
Figure 7. Mechanical power output required to hover, , is
˙
W
mechanical
roughly constant at W (mixed-mode regression,0.11 Ⳳ0.01 Pp
; individual effects not significant, ). Mass-specific me-0.2611 Pp0.17
chanical power is approximately 4.1 mL O
2
h
⫺1
g
⫺1
. Symbols are as in
Figure 3.
loss to the surrounding air, and positive is the me-
˙
W
mechan ical
chanical work output as estimated aerodynamically. We rep-
resent the overall efficiency h(Josephson et al. 2001) as
˙
W
mechan ical
hp,(3)
1
˙
Q
metab olic
where h
1
is efficiency based on mechanical power estimates and
respirometric measurements of metabolic power. Alternatively,
we compute h
2
as follows using equation (2) for metabolic
power estimates:
˙
W
mechan ical
hp.(4)
2
˙
˙
W⫹Q
mechan ical loss
Purely on the basis of thermodynamics, h
2
can be viewed as
an “efficiency,” but measurements of may miss some heat
˙
Q
loss
loss terms. Discrepancies between these two estimates indicate
supplemental avenues of heat dissipation (such as evaporative
loss) not represented in equation (1).
The effects of variable air temperature on rate of heat loss,
metabolic rate, wingbeat kinematics, mechanical power, heat
balance, and muscle efficiency were evaluated using mixed-
model regression in JMP (SAS Institute, Cary, NC) unless oth-
erwise noted. In all analyses, individual effects and interactions
(e.g., metabolic rate as a function of bird, bird #temperature,
and temperature) were checked by ANOVA. In all cases except
wingbeat frequency and stroke amplitude, which are discussed
below, individual effects were found to be nonsignificant and
so data were pooled in subsequent analyses.
Results
Metabolic rates of hover-feeding hummingbirds averaged
W (mean ⳲSD) and were independent of ambient1.1 Ⳳ0.1
air temperature ( ; see Fig. 3; individual effects notPp0.17
significant, ). The measured metabolic rates corre-Pp0.10
spond to a mass-specific metabolic rate of mL O
2
g
⫺1
42 Ⳳ5
h
⫺1
. By contrast, surface temperature (T
s
) of hovering Calypte
anna increased significantly with ambient air temperature T
a
(Fig. 4A; , ; individual effects notTp0.78T⫹9.32 P≤0.0001
sa
significant, ). Correspondingly, the estimated rate ofPp0.17
heat loss to the environment by convective and radiative heat
transfer declined significantly with air temperature (Fig. 4B;
, ; individual effects not sig-
˙
Qp⫺0.06T⫹2.05 P!0.0001
loss a
nificant, ).Pp0.34
The overall average heat transfer coefficient, , as measured
¯
h
on the physical model of a hummingbird, varied with both
surface temperature and airflow velocity (Fig. 5; ¯
hp137 ⫺
, , ). The model test results
2
3.0T⫹13.5urp0.63 Pp0.018
sind
were used to estimate convective heat loss for each hovering
trial on the basis of mean surface temperature of the bird and
an induced airflow velocity on the basis of aerodynamic cal-
culations. To determine whether the small changes in un-
¯
h
derlay observed trends, we also calculated convective heat loss
with a constant ; estimates differed slightly but do not alter
¯
h
This content downloaded from 152.2.15.253 on Wed, 11 Dec 2013 11:00:44 AM
All use subject to JSTOR Terms and Conditions

Hummingbird Hovering Energetics 411
Table 2: Heat balance terms as a function of air
temperature
T
a
(⬚C) ˙
Q
metabolic
˙
W
mechanical
˙
Q
loss
18 Ⳳ2 1.0 Ⳳ.1 .11 Ⳳ.01 .9 Ⳳ.3
24.0 Ⳳ.9 1.0 Ⳳ.1 .12 Ⳳ.01 .5 Ⳳ.2
26 Ⳳ2 1.1 Ⳳ.1 .11 Ⳳ.01 .6 Ⳳ.2
30.2 Ⳳ.2 1.1 Ⳳ.1 .11 Ⳳ.01 .5 Ⳳ.1
33.5 Ⳳ.5 1.0 Ⳳ.1 .111 Ⳳ.003 .4 Ⳳ.2
Note. All values are in Watts unless otherwise noted, and are
means ⳲSD. In mass-specific form, metabolic rates are approxi-
mately mL O
2
h
⫺1
g
⫺1
and mechanical power estimates are42 Ⳳ5
about mL O
2
h
⫺1
g
⫺1
.4.1 Ⳳ0.5
Table 3: Efficiency estimates as a function of
air temperature
T
a
(⬚C) h
1
h
2
h⫺h
12
18 Ⳳ2 .11 Ⳳ.02 .109 .001
24.0 Ⳳ.9 .12 Ⳳ.02 .194 ⫺.074
26 Ⳳ2 .10 Ⳳ.02 .155 ⫺.055
30.2 Ⳳ.2 .10 Ⳳ.02 .180 ⫺.080
33.5 Ⳳ.5 .10 Ⳳ.01 .217 ⫺.117
Note. h
1
efficiencies are computed using hp
1
.h
2
estimates are computed using
˙
˙
W/Qhp
mechanical metabolic 2
.
˙
˙˙
W/(W⫹Q)
mechanical mechanical loss
our conclusions. For all trials, the average heat transfer coef-
ficient was 107 W m
⫺2
K
⫺1
on the basis of average surface
temperatures of C and average induced airflow veloc-
29⬚Ⳳ4⬚
ities of m s
⫺1
.
4.5 Ⳳ0.2
Wingbeat frequencies decreased with ambient air tempera-
ture by about 10% over the experimental range (Fig. 6A;sig-
nificant individual effects, ). By contrast, stroke am-
P!0.001
plitudes increased with ambient air temperature by about 10%
over the same range (Fig. 6B; significant individual effects,
). Mechanical power output was, however, indepen-
P!0.001
dent of air temperature (Fig. 7; , individual effects not
Pp0.26
significant, ), and it averaged W (or, in
Pp0.17 0.11 Ⳳ0.01
mass-specific form, mL O
2
g
⫺1
h
⫺1
).
4.1 Ⳳ0.5
At low air temperatures, metabolic heat generation was ap-
proximately equal to the summed rate of heat loss via radiation
and convection and the estimated mechanical power output
(Table 2). At midrange and high temperatures, however, the
terms did not balance, demonstrating additional modes of heat
loss, such as evaporative cooling, enthalpy rise of excreted water,
and avenues of convective heat loss not included in equation
(1). Overall efficiencies calculated for hover-feeding on the basis
of either aerodynamic power estimates and respirometry (h
1
)
or aerodynamic power and radiative and convective heat loss
(h
2
) at low temperatures were about 10% (Table 3). The h
1
efficiencies of 10% are comparable to results from other studies
of hovering hummingbirds (9%–11%; see Wells 1993; Chai et
al. 1998). At high air temperatures, h
2
estimates (15%–20%)
are somewhat higher than efficiencies based on aerodynamic
power (h
1
), indicating unaccounted modes of heat dissipation
in the overall energy balance (eq. [1]).
We observed several behaviors corresponding to the loss of
the driving temperature differential at higher air temperatures.
At air temperatures of 18⬚–24⬚C, the main areas of heat loss
were found around the head, eye, and pectoral muscles (Fig.
1A). Above 30⬚C, these regions expanded to include the entire
body, as seen in thermal images (see Fig. 1B), potentially cor-
responding to an increase in blood perfusion over the entire
body. Above 30⬚C, all birds exhibited bill gaping while perched
and one bird exhibited wing spreading while perched. Birds
were also observed to extend their feet while hovering at high
air temperatures (see Fig. 1B).
Discussion
At low air temperatures, application of thermal imaging, aero-
dynamic estimates, and respirometric measurements yielded
consistent results for the energetics of hovering flight. By con-
trast, use of infrared thermography at higher ambient air tem-
peratures underestimated heat lost to the environment because
of unaccounted modes of heat transfer. At higher air temper-
atures, the primary mechanisms available for dissipating heat
are evaporative cooling and enthalpy rise in either exhaled air
or excreted water (Lasiewski 1964; Powers 1992; Lotz et al.
2003). In addition, heat loss from the wings, which have rel-
atively large surface areas and higher convective heat transfer
coefficients due to high flapping velocities, may become rela-
tively more important at higher temperatures.
As ambient air temperatures approach body temperature, the
driving thermal gradient ( ) for convection and radiativeT⫺T
sa
loss is progressively diminished. At high ambient air temper-
atures, flight performance may then become thermally limited,
as is suggested by estimated rates of convective and radiative
heat loss that extrapolate to zero heat loss at approximately
40⬚C (Fig. 4). This diminished shedding of heat is further com-
pounded by a slight decline in the heat transfer coefficient with
increased air temperature (Fig. 5). At temperatures higher than
its thermal neutral zone, Calypte anna may become slightly
hyperthermic, which could augment heat transfer by increasing
the gradient (Powers 1992).T⫺T
sa
Whereas our calculations do not include such heat transfer
modes as evaporative cooling, enthalpy rise, and convection
from the wings, the measurements taken here provide an in-
direct way to estimate these terms. At low temperatures, these
additional heat loss terms, taken as ˙˙
Q⫺W⫺
metabolic mechanical
, are negligible ( W). At the highest air temper-
˙
Q0.0 Ⳳ0.3
loss
ature, these additional terms would have to be W,0.5 Ⳳ0.2
almost half of the metabolic rate. It is interesting to consider
how this could be split among evaporative water loss, enthalpy
rise of ingested liquid, and wing convection. Powers (1992) and
Lasiewski (1964) give evaporative water loss rates for hum-
mingbirds of around 20 mg g
⫺1
h
⫺1
. Assuming that 2,257 kJ is
needed to vaporize 1 kg of water, the associated evaporative
heat loss would be about 0.06 W, corresponding to 6% of the
metabolic rate measured here. The energy needed to warm
ingested nectar could also be significant (Lotz et al. 2003).
This content downloaded from 152.2.15.253 on Wed, 11 Dec 2013 11:00:44 AM
All use subject to JSTOR Terms and Conditions

412 D. Evangelista, M. J. Ferna´ndez, M. S. Berns, A. Hoover, and R. Dudley
Ferna´ndez et al. (2002) give fluid ingestion rates of up to 0.03
g min
⫺1
for a 6-g green-backed firecrown hummingbird (Se-
phanoides sephanoides). The energy required to raise the tem-
perature of such nectar volumes from 20⬚C to typical body
temperatures would be about 0.04 W, or 4% of the measured
metabolic rate for Anna’s hummingbirds. To estimate wing
convection, we assume a convective heat transfer coefficient
for the wings of 200 W m
⫺2
K
⫺1
, on the basis of forced con-
vection in air on a flat plate at a local velocity of 15 m s
⫺1
(Incropera and Dewitt 1996). The patagia are the main vas-
cularized areas of the wings, and the heat transfer surface areas
for the dorsal and ventral surfaces of both patagia are approx-
imately m
2
. For wing temperatures equal to the ob-
⫺4
2.6 #10
served body surface temperatures, convective heat loss from
the wings would be about 0.4 W. As with the body, heat loss
from the wings will become restricted as ambient air temper-
ature increases.
Limited heat dissipation in high air temperatures may impose
constraints on hovering performance. Hummingbirds in hot,
humid conditions in the tropics may raise their body temper-
atures in order to increase convective heat transfer when evap-
orative cooling is limited (see Powers 1992), although further
ecological and behavioral data are needed. Field observations
of the giant Andean hummingbird (Patagona gigas) indicate
decreased activity on hot days (M.J. Ferna´ndez, personal ob-
servation). Mechanisms of heat dissipation will be further lim-
ited under humid conditions, as enthalpy rise associated with
exhalation declines with increasing water content of inhaled
air. For hummingbirds with relatively high ventilatory fre-
quencies, the magnitude of respiratory cooling is potentially
substantial. The fraction of metabolic heat dissipated by evap-
oration in Anna’s hummingbirds is lower than that of other
birds in dry air, but it exceeds that of other birds at high
humidities when air temperature is less than 33⬚C (Powers
1992). Experimental manipulation of both air temperature and
relative humidity for hovering hummingbirds would defini-
tively test the role of thermal constraints on flight performance.
Convection is expected to be the dominant mode of heat
loss at low and midrange air temperatures. However, variation
in wingbeat frequency and stroke amplitude at low ambient air
temperature (see Fig. 6) does not significantly alter either in-
duced velocity or wing relative velocity, which otherwise might
change the convective heat transfer coefficient. Variation in
estimated rates of heat loss with air temperature (Fig. 4B) sug-
gests that metabolic rates must increase to compensate for in-
creased heat loss below 15⬚C, as is observed for ruby-throated
hummingbirds hovering at 5⬚–15⬚C (Chai et al. 1998). At even
lower air temperatures, below the thermal neutral zone, heat
loss becomes large compared with rates of metabolic heat pro-
duction. For example, we observed substantial feather fluffing
by perched birds in very cold air. Studies of otherhummingbird
species have documented increases in foraging and food intake
to maintain energy balance at low air temperatures (Gass et al.
1999). At high elevations in the Andes, hummingbirds often
fly and forage at near-freezing air temperatures; associated
physiological responses may have been an important factor
influencing montane colonization by this lineage.
In spite of significant changes in wingbeat frequency and
stroke amplitude over the experimental range of air tempera-
tures (Fig. 6), we observed no significant changes in hovering
metabolic rates or mechanical power expenditure at the lowest
temperatures used in our experiments (Fig. 3; Table 2). Altered
kinematics could potentially generate more heat at the expense
of mechanical efficiency (Ivanov 1989; Full et al. 1998; Jo-
sephson et al. 2001; Bicudo et al. 2002). Some indication of
this effect was evident at the lowest tested air temperatures, for
which heat loss approached metabolic rate (see Figs. 3, 4B; see
also Chai et al. 1998). Flight efficiency, however, remained re-
markably constant at about 10% (see Table 3). These efficiency
estimates are comparable to those derived in other studies of
hovering hummingbirds (9%–11%; see Wells 1993; Chai et al.
1998). Kinematic variation without reduction in efficiency of
hovering may also be of benefit when hummingbirds visit dif-
ferent floral morphologies (Wells 1993).
The thermal limits on flight performance suggested here
likely influence hummingbird evolution across altitudinal gra-
dients (McGuire et al. 2007), and they may also impinge on
size-based trade-offs in flight performance (see Chai and Dud-
ley 1999; Altshuler and Dudley 2002). The most thermally chal-
lenging conditions for trochilids are likely to involve hovering
in hot and humid air (resulting in insufficient heatdissipation),
as well as forward flight in cold, dry air (which may result in
excess heat loss). Fieldwork assessing flight behaviors in relation
to microclimatic data would enable assessment of the specific
role of thermal limits in influencing ecological distributions of
hummingbirds. Similarly, comparative work on thermal re-
sponses by differently sized species would assess the allometry
of heat production and loss during hovering flight.
Acknowledgments
We thank C. Clark for capturing birds and providing a skin
specimen for the construction of the thermal model. We also
thank R. Full, M. Koehl, Y. Munk, S. Sponberg, and T. Libby
for assistance, equipment, and comments as a part of Berkeley’s
Mechanics of Organisms Lab class (IB 135L) within the Center
for Integrative Biomechanics Education and Research (CIBER).
Several anonymous reviewers provided comments that were
helpful. D.E. was supported by a National Science Foundation
(NSF) graduate fellowship. M.J.F. was supported by a Fulbright
fellowship. Y. Munk provided useful comments on the man-
uscript. Laboratory support was provided by NSF grant DEB-
0543556.
Literature Cited
Altshuler D.L. and R.A. Dudley. 2002. The ecological and evo-
lutionary interface of hummingbird flight physiology. J Exp
Biol 205:2325–2337.
———. 2003. Kinematics of hovering hummingbird flight
This content downloaded from 152.2.15.253 on Wed, 11 Dec 2013 11:00:44 AM
All use subject to JSTOR Terms and Conditions

Hummingbird Hovering Energetics 413
along simulated and natural elevational gradients. J Exp Biol
206:3139–3147.
Altshuler D.L., R. Dudley, and J.A. McGuire. 2004. Resolution
of a paradox: hummingbird flight at high elevation does not
come without a cost. Proc Natl Acad Sci USA 101:17731–
17736.
Bartholomew G.A. and J.R.B. Lighton. 1986. Oxygen con-
sumption during hover feeding in free-ranging Anna’s hum-
mingbirds. J Exp Biol 123:191–199.
Berger M. and J.S. Hart. 1972. Die Atmung beim Kolibri Ama-
zilia fimbriata wa¨hrend des Schwirrfluges bei verschiedenen
Umgebungstemperaturen. J Comp Physiol A 81:363–380.
Bicudo J.E.P.W., A.C. Bianco, and C.R. Vianna. 2002. Adaptive
thermogenesis in hummingbirds. J Exp Biol 205:2267–2273.
Chai P., A. Chang, and R.A. Dudley. 1998. Flight thermogenesis
and energy conservation in hovering hummingbirds. J Exp
Biol 201:963–968.
Chai P. and R. Dudley. 1995. Limits to vertebrate locomotor
energetics suggested by hummingbirds hovering in heliox.
Nature 377:722–725.
———. 1996. Limits to flight energetics of hummingbirds hov-
ering in hypodense and hypoxic gas mixtures. J Exp Biol
199:2285–2295.
———. 1999. Maximum flight performance of hummingbirds:
capacities, constraints and trade-offs. Am Nat 153:398–411.
Cossins A.R. and K. Bowler. 1987. Temperature Biology of An-
imals. Chapman & Hall, London.
Ellington C.P. 1984a. The aerodynamics of hovering insect
flight. II. Morphological parameters. Philos Trans R Soc B
305:17–40.
———. 1984b. The aerodynamics of hovering insect flight. V.
A vortex theory. Philos Trans R Soc B 305:115–144.
———. 1984c. The aerodynamics of hovering insect flight. VI.
Lift and power requirements. Philos Trans R Soc B 305:145–
181.
Epting R.J. 1980. Functional dependence of the power for hov-
ering on wing disc loading in hummingbirds. Physiol Zool
53:347–357.
Ferna´ndez M.J., M.V. Lo´pez-Calleja, and F. Bozinovic. 2002.
Interplay between the energetics of foraging and thermoreg-
ulatory costs in the green-backed firecrown hummingbird
Sephanoides sephanoides. J Zool 258:319–326.
Full R.J., D. Stokes, A. Ahn, and R. Josephson. 1998. Energy
absorption during running by leg muscles in a cockroach. J
Exp Biol 201:997–1012.
Gass C.L. and J.S.E. Garrison. 1999. Energy regulation by trap-
lining hummingbirds. Funct Ecol 13:483–492.
Gass C.L., M.T. Romich, and R.K. Suarez. 1999. Energetics of
hummingbird foraging at low ambient temperature. Can J
Zool 77:314–320.
Hixon M.A. and F.L. Carpenter. 1988. Distinguishing energy
maximizers from time minimizers: a comparative study of
two hummingbird species. Integr Comp Biol 28:913–925.
Incropera F.P. and D.P. Dewitt. 1996. Fundamentals of Heat
and Mass Transfer. Wiley, New York.
Ivanov K.P. 1989. Thermoregulatory chemical metabolism and
muscle work efficiency. J Therm Biol 14:1–18.
Josephson R.K., J.G. Malamud, and D.R. Stokes. 2001. The
efficiency of an asynchronous flight muscle from a beetle. J
Exp Biol 204:4125–4139.
Lasiewski R.C. 1963. Oxygen consumption of torpid, resting,
active and flying hummingbirds. Physiol Zool 36:122–140.
———. 1964. Body temperatures, heart and breathing rate,
and evaporative water loss in hummingbirds. Physiol Zool
37:212–223.
Lotz C.N., C. Martı´nez del Rio, and S.W. Nicholson. 2003.
Hummingbirds pay a high cost for a warm drink. J Comp
Physiol B 173:455–461.
McGuire J.A., C.C. Witt, D.L. Altshuler, and J.V. Remsen. 2007.
Phylogenetic systematics and biogeography of humming-
birds: Bayesian and maximum likelihood analyses of parti-
tioned data and selection of an appropriate partitioning strat-
egy. Syst Biol 56:837–856.
Miller P.J., ed. 1996. Miniature Vertebrates: The Implications
of Small Body Size. Oxford University Press, Oxford.
Otsu N. 1979. A threshold selection method from gray level
histograms. IEEE Trans Syst Man Cybern 9:62–66.
Powers D.R. 1992. Effect of temperature and humidity on evap-
orative water loss in Anna’s hummingbird (Calypte anna).
J Comp Physiol B 162:74–84.
Suarez R.K. 1992. Hummingbird flight: sustaining the highest
mass-specific metabolic rates among vertebrates. Experientia
48:565–569.
Suarez R.K., J.R.B. Lighton, C.D. Moyes, G.S. Brown, C.L.Gass,
and P.W. Hockachka. 1990. Fuel selection in rufous hum-
mingbirds: ecological implications of metabolic biochemis-
try. Proc Natl Acad Sci USA 87:9207–9210.
Ward S., J.M.V. Rayner, U. Mo¨ller, D.M. Jackson, W. Nachtigall,
and J.R. Speakman. 1999. Heat transfer from starlings Stur-
nus vulgaris during flight. J Exp Biol 202:1589–1602.
Weis-Fogh T. 1972. Energetics of hovering flight in humming-
birds and in Drosophila. J Exp Biol 56:79–104.
Welch K.C. and R.K. Suarez. 2008. Altitude and temperature
effects on the energetic cost of hover-feeding in migratory
rufous hummingbirds, Selasphorus rufus. Can J Zool 86:161–
169.
Wells D. 1993. Muscle performance in hovering hummingbirds.
J Exp Biol 178:39–57.
Wolf L.L. and F.R. Hainsworth. 1971. Time and energy budget
of territorial hummingbirds. Ecology 52:980–988.
Zerba E. and G.E. Walsberg. 1992. Exercise-generated heatcon-
tributes to thermoregulation by Gambel’s quail in the cold.
J Exp Biol 171:409–422.
This content downloaded from 152.2.15.253 on Wed, 11 Dec 2013 11:00:44 AM
All use subject to JSTOR Terms and Conditions