Energetic cost of hovering flight in nectar-feeding bats (Phyllostomidae: Glossophaginae) and its scaling in moths, birds and bats

ORIGINAL PAPER
C. C. Voig áY. Win e
Ene ge ic cos o ho e ing ¯igh in nec a - eeding ba s (Phyllos omidae:
Glossophaginae) and i s scaling in mo hs, bi ds and ba s
Accep ed: 10 No embe 1998
Abs ac Th ee g oups o specialis nec a - eede s co -
e ing a con inuous size ange om insec s, bi ds and
ba s ha e e ol ed he abili y o ho e ing ¯igh . Among
bi ds and ba s hese g oups gene ally comp ise small
species, sugges ing a ela ionship be ween ho e ing
abili y and size. In his s udy we es ablished he scaling
ela ionship o ho e ing powe wi h body mass o
nec a - eeding glossophagine ba s (Phyllos omidae).
Employing bo h s anda d and as - esponse espi ome-
y, we de e mined a es o gas exchange in Hylonyc e is
unde woodi (7 g) and Choe onyc e is mexicana (13±18 g)
du ing ho e - eeding ¯igh s a an a i®cial ¯owe ha
se ed as a espi ome ic mask o es ima e me abolic
powe inpu . The O
2
up ake a e _
Vo2in ml g
)1
h
)1
(and
de i ed powe inpu ) was 27.3 (1.12 W o 160 W kg
)1
)
in 7-g Hylonyc e is and 27.3 (2.63 W o 160 W kg
)1
)in
16.5-g Choe onyc e is and hus consis en wi h mea-
su emen s in 11.9-g Glossophaga so icina (158 W kg
)1
,
Win e 1998). _
Vo2a he onse o ho e ing was also used
o es ima e powe du ing o wa d ¯igh , because a e a
ansi ion om le el o wa d o ho e ing ¯igh gas ex-
change a es ini ially s ill e¯ec o wa d ¯igh a es. _
Vo2
du ing sho ho e ing e en s (<1.5 s) was
19.0 ml g
)1
h
)1
(1.8 W) in 16-g Choe onyc e is, which
was no signi®can ly die en om a p e ious, indi ec
es ima e o he cos o le el o wa d ¯igh (2.1 W,
Win e and on Hel e sen 1998). Ou es ima es sugges
ha powe inpu du ing ho e ing ¯igh P
h
(W) inc eased
wi h body mass M(kg) wi hin 13±18-g Choe onyc e is
(n4) as P
h
3544 (2057 SE) M
1.76 (0.21 SE)
and
be ween die en glossophagine ba species (n3) as
P
h
128 (2.4 SE) M
0.95 (0.034 SE)
. The slopes o
h ee scaling unc ions o ¯igh powe (ho e ing, le el
o wa d ¯igh a in e media e speed and submaximal
¯igh powe ) indica e ha : 1. The ela ionship be ween
¯igh powe o ¯igh speed may change wi h body mass
in he 6±30-g ba s om a J- owa ds a U-shaped cu e.
2. A me abolic cons ain (ho e ing ¯igh powe equal
maximal ¯igh powe ) may in¯uence he uppe size limi
o 30±35 g o his g oup o ¯owe specialis s.
Mass-speci®c powe inpu (W kg
)1
) du ing ho e ing
¯igh appea ed cons an wi h ega d o body size ( o he
mass anges conside ed), bu die ed signi®can ly
(P< 0.001) be ween g oups. G oup means we e
393 W kg
)1
(sphingid mo hs), 261 W kg
)1
(humming-
bi ds) and 159 W kg
)1
(glossophagine ba s). Thus,
glossophagine ba s expend he leas me abolic powe pe
uni o body mass suppo ed du ing ho e ing ¯igh . A a
me abolic powe inpu o 1.1 W a glossophagine ba can
gene a e he li o ces necessa y o balancing 7 g
agains g a i a ion, whe eas a hummingbi d can suppo
4 g and a sphingid mo h only 3 g o body mass wi h he
same amoun o me abolic ene gy. These die ences in
powe inpu we e no ully explained by die ences in
induced powe ou pu es ima ed om Rankine-F oude
momen um-je heo y.
Key wo ds Ho e ing ¯igh áGlossophagine
ba s áHummingbi ds áSphingid mo hs áAllome y
Abb e ia ions Mbody mass áP
ho izon al o wa d
¯igh powe áP
min
minimum ¯igh powe áP
h
ho e ing
¯igh powe áP
indRF
minimum induced powe s áP
submax
submaximal ae obic ¯igh powe áRE espi a o y
exchange a io áRQ espi a o y quo ien á_
Vo2oxygen
up ake a e á_
Vco2ca bon dioxide elease a e
J Comp Physiol B (1999) 169: 38±48 ÓSp inge -Ve lag 1999
C.C. Voig (&)
1
áY. Win e
Ins i u u
È Zoologie II, Uni e si a
È E langen,
S aud s asse 5, D-91058 E langen, Ge many
e-mail: [email p o ec ed],
[email p o ec ed]
P esen add ess:
1
Bos on Uni e si y, Depa men o Biology,
5 Cumming on S ee ,
Bos on, MA 02215, USA
e-mail: [email p o ec ed]
In oduc ion
Ho e ing ¯igh is gene ally assumed o equi e mo e
powe han ho izon al o wa d ¯igh because he ae o-
dynamic induced powe componen , necessa y o p o-
ducing he li o ce o esis ing g a i y, comes en i ely
om wing oscilla ions (c .No be g 1990). Fo h ee
g oups o nec a - eede s (sphingid mo hs, humming-
bi ds and glossophagine ba s) which e ol ed hei
dis inguished ho e ing abili y as pa o hei ¯owe -
eeding habi s, he ene ge ic cos o ho e ing has been
de e mined om espi ome y (see e e ences gi en in
Fig. 5). Wha makes a compa ison especially in e es ing
is ha while indi idual powe equi emen s o ho e ing
o e lap be ween he h ee g oups, hey all die in
ho e ing s yles ( on Hel e sen and on Hel e sen 1975;
on Hel e sen 1986; c . No be g 1990; No be g e al.
1993; Willmo and Elling on 1997; Liu e al. 1998).
Wi hin he wo e eb a e classes, bo h hummingbi ds
(2±20 g) and glossophagine nec a - eeding ba s (6±30 g)
comp ise species o ela i ely small body mass. This end
ollows he scale eec ha makes i easie o smalle
species o mobilize powe ese es o ¯igh . This is be-
cause powe equi emen s o ¯igh inc ease wi h mass a
a s eepe slope han maximal me abolic a es (Pennycuick
1968, 1986; No be g 1990; see also Elling on 1991). De-
spi e his gene al scaling eec , de ails o he powe e-
qui emen s o ho e ing ¯igh , ela i e o o wa d ¯igh
and hei ela ion o me abolic and ae odynamic powe
capaci ies, we e p e iously unknown.
Ho e ing ¯igh in glossophagine ba s die s om
ha o hummingbi ds in a numbe o ways.
1. Hummingbi ds oscilla e hei ully ex ended wings in
a ®gu e-eigh mo ion wi h symme ical hal -s okes
(No be g 1990). The ho e ing kinema ics o
Glossophaga gene ally esemble slow ¯apping ¯igh
wi h wings ¯exed du ing he ups oke. Li is s ill
gene a ed du ing ups oke by supina ion o he
handwing ( he `dis al wing iangle'), an adap a ion
ha is possibly unique among ba s ( on Hel e sen
and on Hel e sen 1975; on Hel e sen 1986).
2. Wing loading in hummingbi ds is abou wice as high
as in Glossophaga (G eenewal 1962; Win e 1998), as
hummingbi d wings a e sho e and na owe .
3. The mass-speci®c powe inpu du ing ho e ing in
hummingbi ds is 60% highe han in Glossophaga
so icina (45 ml O
2
g
)1
h
)1
s 27 ml O
2
g
)1
h
)1
o
261 W kg
)1
s 158 W kg
)1
, Win e 1998)
4. While (a leas he smalle ) hummingbi ds a e able o
sus ain ho e ing ¯igh o up o se e al minu es
(Pea son 1950), G. so icina does no sus ain ho e ing
¯igh o ex ended pe iods. In he ®eld, du a ion o
ho e ing o Glossophaga when eeding on nec a a
¯owe s is almos always less han 1 s (C. Voig ,
Y. Win e pe sonal obse a ion) and he longes
ho e ing e en s obse ed in he labo a o y a au o-
ma ic eede s which delayed he nec a - ewa d
(Win e 1998) we e well below 10 s o his species
(only he p esen s udy p o ides new e idence ha ,
unde special ci cums ances, a glossophagine ba may
also sus ain ho e ing ¯igh o longe pe iods). This
sugges s a limi ing ac o on ho e ing abili y e¯ec -
ing ei he ae odynamic o me abolic cons ain s.
The e a e also simila i ies be ween he wo g oups.
Measu ed a es o oxygen up ake _
Vo2by wo small
hummingbi d species du ing o wa d ¯igh in a wind
unnel (Be ge 1985) indica e ha powe inpu du ing
ho e ing in hummingbi ds may be equal o only sligh ly
highe han o o wa d ¯igh a in e media e speeds
(7ms
)1
). Likewise, me abolic da a o G. so icina also
indica e a ac o o a ound only 1.1 be ween powe in-
pu s du ing ho e ing and le el o wa d ¯igh a in e -
media e speed (Win e 1998; Win e and on Hel e sen
1998). In con as , a p e ious quasi-s eady ae odynamic
analysis based on wing kinema ic da a and wake eloci y
measu emen s sugges s ha ae odynamic powe ou pu
o ho e ing ¯igh in Glossophaga exceeds he co e-
sponding alue o ho izon al o wa d ¯igh by a ac o
o 1.7 (pe ec elas ic s o age) o 2.6 (ze o s o age)
(No be g e al. 1993).
In iew o he modes 10% inc ease in me abolic
powe om ho izon al o wa d o ho e ing ¯igh , i is
no ob ious ha me abolic powe is limi ing du ing
ho e ing. In ac , he ex apola ion o submaximal ae -
obic me abolic capaci ies (P
submax
) o medium o la ge-
sized ba s (95±800 g) du ing ¯igh s in wind unnels
(Thomas 1987) o small body masses sugges s ha small
ba s may s ill ha e an 80% powe ma gin abo e he
powe equi emen s o ho izon al o wa d ¯igh a in-
e media e speeds (Win e and on Hel e sen 1998). Fo
an 11.9-g Glossophaga, p edic ed P
submax
equals 3.0 W,
compa ed o he powe inpu o 1.88 W du ing ho e ing
¯igh and 1.65 W o ho izon al o wa d ¯igh a in-
e media e speed. While he accu acy o his p edic ion
o P
submax
is unce ain, i s ill indica es ha me abolic
equi emen s du ing ho e ing ¯igh may no cons i u e
he limi ing ac o o a 12-g glossophagine ba . Simila
o he p edic ion o Glossophaga, small hummingbi ds
ha e conside able me abolic (and ae odynamic) powe
ese es a hei command. O
2
up ake a maximum ¯igh
speeds in he wind unnel has been measu ed a 1.6 imes
he a e du ing ho e ing (Be ge 1985). Simila ly, du ing
expe imen al manipula ions o me abolic and ae ody-
namic powe equi emen s o ho e ing, hummingbi ds
ha e been p o en o possess conside able powe ese es
(Wells 1993; Chai and Dudley 1995, 1996; Chai e al.
1996; Chai and Milla d 1997). Da a o he la ges
hummingbi ds (20-g Pa agona gigas, G eenewal 1960),
howe e , a e no a ailable.
The p esen s udy was conduc ed o: (1) in es iga e
he in a- and in e -speci®c scaling o ho e ing powe
wi h body mass in glossophagine ba s, and (2) compa e
he ela ionships be ween powe inpu du ing ho e ing
¯igh and body mass be ween he glossophagine nec a -
eeding ba s, hummingbi ds and sphingid mo hs. Fo
39
his, ho e ing powe inpu was es ima ed om a es o
gas exchange du ing ho e - eeding a an a i®cial ¯owe
ha se ed as a espi ome y mask o wo species o
glossophagine ba (7 g and 17 g), and ( oge he wi h a
p e ious alue o 12-g Glossophaga) hese esul s we e
compa ed wi h da a a ailable in he li e a u e o
sphingid mo hs and hummingbi ds.
Ma e ial and me hods
Animals and ¯igh cage
Measu emen s o his s udy we e pe o med wi h he wo ba
species Hylonyc e is unde woodi Thomas 1903 and Choe onyc e is
mexicana Tschudi 1844 (Phyllos omidae: Glossophaginae). P io o
he expe imen s, animals we e kep in g eenhouses on an ad libi um
ood supply. A week be o e measu emen s we e aken, we ans-
e ed he animals o an expe imen al oom wi h a ligh : da k cycle
o 12 h:12 h L:D, an a e age empe a u e o 24 °C, and a ela i e
humidi y o 75%, o acclima ize o he condi ions. Fo a de ailed
desc ip ion o he expe imen al p o ocol, see Win e e al. (1998).
Du ing each expe imen , we in oduced one indi idual in o a plas ic
¯igh unnel (leng h 7 m, wid h 1 m and heigh 2 m). Wi hin his
¯igh unnel he only place he ba could es was on a piece o co k
ha was suspended om an elec onic balance (Me le PM 100,
accu acy 1 mg). Body mass and imes o a i al a and ake-o om
he oos we e eco ded wi h a compu e connec ed o he elec onic
balance. The animal was ed wi h a 17% (w /w ) suga solu ion
(mix u e o glucose, uc ose and suc ose) p o ided om a com-
pu e ized nec a - eede . Du ing ho e - eeding ¯igh , he ba in e -
up ed an in a ed ligh beam a he on edge o he eede mask.
This was eco ded by a compu e ha bo h ac i a ed he nec a
pump and eco ded he du a ion o he ho e ing e en (Win e and
on Hel e sen 1998). The eede opening was cons uc ed as a ¯ow-
h ough espi ome ic mask (see d awing in Win e e al. 1998). The
inne diame e o he mask was 2.3 cm du ing he measu emen s
wi h H. unde woodi and 3 cm wi h C. mexicana. The eede opening
was o ien ed wi h an inclina ion o 30° o he g ound and a a heigh
o abou 80 cm abo e he g ound.
Na u al ho e ing du a ions o glossophagine ba s a e usually
less han 1 s. In o de o inc ease he du a ion o ho e ing ¯igh s
we delayed he injec ion o nec a in o he eede opening when a
ba a emp ed o eed and p o ided he nec a ewa d in he o m
o successi e small d ople s (app oxima ely 5 ll each) gi en e e y
0.8±1 s du ing a single ho e ing e en . Typical nec a gains du ing
a single eeding isi a a ¯owe in he wild is 10±50 ll (Howell
1979; on Hel e sen and Reye 1984; Tschapka and on Hel e sen,
pe sonal obse a ion). We ca ied ou wo ypes o espi ome ic
measu emen s which ocused on wo die en aspec s o ho e ing
gas exchange.
Cumula i e measu emen s
We de e mined cumula i e O
2
and CO
2
olumes o single ho e ing
e en s. This allowed us o es o die ences in gas exchange a es
be ween ho e ing e en s ha die ed in du a ion and o de e mine
he espi a o y exchange a io (RE). These measu emen s we e
pe o med wi h wo H. unde woodi ( emale 1: 7.1 g; emale 2: 6.9 g)
and ou C. mexicana (see Table 1). We could no moni o changes
in ins an aneous a es o O
2
consump ion and CO
2
elease du ing a
ho e ing e en , due o he wash-ou cha ac e is ics o he gas an-
alysing sys em.
Gas analysing sys em
Room ai was d awn h ough he espi ome ic mask wi h a
memb ane pump a a a e o 1000 ml min
)1
. As shown p e iously
o Glossophaga (Win e e al. 1998), his ¯ow a e made i possible
o sample all espi a o y gases while he ba was a he eede mask.
Downs eam om he pump he ai was d ied using molecula sie e
(3 A
Ê) and he ¯ow egula ed wi h a mass ¯ow con olle (F 201 C-
FB, B onkho s , Ruu lo, Ne he lands). Nex , he ai s eam was
spli and led o bo h he O
2
analyse (S-3 A/II wi h a dual N-37 M
senso , Ame ek now AEI, Pi sbu gh, USA) and he CO
2
analyse
(URAS 10 E, Ha mann & B aun, F ank u , Ge many). Due o
he high sensi i i y o he CO
2
analyse (0±500 ppm), he sample ai
s eam was dilu ed wi h d y, CO
2
- ee ai (NaOH as desiccan and
CO
2
abso ben ) by a ac o o app oxima ely en. The dilu ion
ac o was moni o ed wi h a mass ¯ow con olle (B onkho s F
201 C-FB) and a mass ¯ow me e (B onkho s F 111 C-HB) and
held cons an du ing he measu emen s. The analog signals o he
gas analyse s we e eco ded wi h a 386-MS-DOS compu e a a
a e o 3 Hz, using Sable Sys ems (Hende son, USA) ha dwa e and
so wa e (16-bi A/D). The analog ou pu signals o he h ee ¯ow
me e s we e checked be o e and a e an expe imen wi h a digi al
ol me e .
Calib a ion and alida ion
The CO
2
analyse was calib a ed by injec ing known concen a-
ions o CO
2
wi h a gas mixing pump (Wo
Ès ho, SA27/2). Fo he
O
2
analyse , no ze o calib a ion was necessa y because o he
echnical cha ac e is ics o i s zi conium-oxide senso . We alida ed
sys em calib a ion egula ly by simula ing he O
2
consump ion and
CO
2
p oduc ion o a ho e ing ba . We injec ed small olumes (100,
150 and 200 ll) o CO
2
(99.998%) in o he espi ome ic mask and
compa ed he measu ed wi h he injec ed CO
2
olumes. Fo he
alida ion o he O
2
analyse we used he N
2
-dilu ion echnique
(Fedak e al. 1981). The a e age de ia ion be ween he measu ed
and expec ed gas olumes was always less han 3.5%.
Da a analysis
Due o he delay in wash-ou o he gas-analysing sys em we we e
con on ed wi h he p oblem o o e lapping signals when he ani-
mal isi ed he eede mask a sho in e als. This p oblem was
especially se e e o he CO
2
signal, because o he la ge olume o
he analyse 's measu ing chambe . Du ing ou ini ial measu e-
men s wi h Choe onyc e is ( emale 1) and Hylonyc e is ( emale 1),
o e lapping signals we e sepa a ed ollowing he p ocedu e de-
sc ibed in Win e e al. (1998). Fo all la e measu emen s we ap-
plied a Z- ans o ma ion o sepa a e o e lapping signals
(Ba holomew e al. 1981) using he Da acan p og am (Sable
Sys ems). The Z- ans o ma ion is used o e-calcula e he o iginal
pa e n o O
2
deple ion and CO
2
inc ease. Because o wash-ou
eec s a s epwise change in O
2
o CO
2
concen a ion is usually
measu ed as an exponen ial change o gas concen a ion. By ap-
plying Eq. 3 om Ba holomew e al. (1981) o he da a, such an
exponen ial cu e can be ans o med o i s o iginal, ins an aneous
shape:
Meq M ÿMD =ZD MD
whe e M
eq
Z- ans o med signal M, M

signal M a ime ,
M
D 
signal M a e D has passed, and Z
D
 ac o Z, desc ibing
he wash-ou cha ac e is ic o he sys em.
On he basis o he Z- ans o med da a, we calcula ed he in-
eg als o he O
2
and CO
2
signals o each ho e ing e en and
mul iplied hem wi h sampling a e and ¯ow a e a he eede
mask. Fo he calcula ion o he CO
2
olume he dilu ion ac o
was aken in o accoun . O
2
olumes we e co ec ed wi h Eq. 3a in
Wi he s (1977) as we used H
2
O abso ban , bu no CO
2
abso ban
o he sample gases. Gas exchange a es (ml s
)1
) we e calcula ed
by di iding he CO
2
and O
2
olumes by he co esponding ho e ing
du a ions o he animal. The mass-speci®c gas exchange a es (ml
g
)1
h
)1
) we e calcula ed by mul iplying he gas exchange a e by
3600 and di iding i by body mass, ob ained sho ly be o e he
ho e ing e en . A co ec ion o STP was no necessa y as mass
¯ow me e s gi e a es in STP.
40
Fas - esponse measu emen s
Animals and expe imen al design
We s udied wo indi iduals o C. mexicana (see Table 1). Ai hu-
midi y in he ¯igh cage was inc eased o mo e han 95% du ing he
measu emen s o he eason desc ibed below.
Gas analysing sys em and alida ion
Fo moni o ing nea -ins an aneous a es o O
2
up ake du ing sin-
gle ho e ing e en s, we modi®ed he analysis se up. To ob ain a as
esponse om he O
2
analyse we minimized wash-ou delay by
ins alling he O
2
senso wi hou desiccan di ec ly behind he eede
mask (Win e 1998). Wi hou a desiccan , he e apo a i e wa e
expi ed by a ba du ing espi a ion could deple e he O
2
concen-
a ion in he sampled ai , which is no accoun ed o in he co -
ec ion o mula ha we used (Eq. 3a in Wi he s 1977). This
p oblem was minimized, howe e , by he high ambien ela i e
humidi y o 95% du ing measu emen s, which g ea ly educed a
ba 's po en ial o u he inc easing he wa e con en o he ai .
Downs eam om he senso he ai was d ied wi h a mo-
lecula sie e (3 A
Ê) and he ¯ow a e was egula ed wi h a mass
¯ow con olle (B onkho s , F 201 C-FB) a 1000 ml min
)1
.We
alida ed he sys em by injec ing die en N
2
olumes in o he
eede mask (N
2
-dilu ion echnique; Fedak e al. 1981). The de-
ia ion be ween measu ed and expec ed O
2
olumes a e aged less
han 5%.
Da a analysis
Fo calcula ing a mean cu e o O
2
deple ion, deple ion cu es o
each ba we e supe imposed om hei o igin up o he poin whe e
he ba had le he eede (i.e. wi hou declining slopes). Mean
alues we e hen calcula ed o each sampling in e al o e all da a
poin s esul ing in one mean O
2
deple ion cu e o each ba
(Fig. 1). These mean O
2
deple ion cu es showed ha a es o O
2
up ake we e cons an a e 5 s o ho e ing (Fig. 1). A single mean
alue o he a e o O
2
up ake was he e o e de e mined o an
indi idual ba by a e aging all alues om he 5 h s o ho e ing o
he end. As be o e, his O
2
a e was co ec ed acco ding o Wi he s
(1977) using a espi a o y quo ien , RQ (RE) o 0.65. This alue o
RE was ob ained in he ollowing way. CO
2
was no measu ed
du ing he as - esponse measu emen s (due o he slow esponse
ime o he CO
2
analyse ) bu was only measu ed du ing he cu-
mula i e measu emen s. Cumula i e (mean) RE was 0.76 o
ho e ing e en s o 5±10 s du a ion (Fig. 2C). As he RE was
a ound 1 a he beginning o a ho e ing ¯igh (Fig. 2C) i mus
Fig. 1 Mass-speci®c a es o O
2
up ake du ing ho e ing ¯igh o
C. mexicana (indi iduals
2
,n38, m
1
n44) measu ed wi h a
as - esponse O
2
analysis sys em. The da a show he change o a es
du ing a ho e ing e en . This is die en o he da a in Fig. 2 which
a e mean alues calcula ed om o al consump ion du ing ho e ing
e en s o die ing du a ions. Da a a e mean alues 1 SD. See
Table 1 o o e all mean
Fig. 2A±C Mean a es o mass-speci®c gas exchange du ing ho e -
eeding e en s o die en du a ions in ou Choe onyc e is mexicana
(means 1 SD). The da a a e AO
2
up ake, BCO
2
elease and C
espi a o y exchange a io (RE). No e ha he highe alues o
indi idual
1
in Aa e associa ed wi h a highe body mass (see also
Table 1). The ough in RE du ing ho e ing e en s o medium
du a ion mos likely e¯ec s he ime lag in he inc ease o CO
2
ou pu a e a s epwise inc ease in me abolic ac i i y du ing he
change om le el o wa d o ho e ing ¯igh . All ho e ing e en s
included in his analysis we e p eceded by a ¯igh in e al o a leas
4 s p io o ho e ing
41
ha e d opped o a alue o 0.65 wi hin 5±10 s o ho e ing o gi e
an o e all mean RE o he comple e ho e ing e en o 0.75. Fo
con e sion o mass-speci®c a es o O
2
up ake (ml O
2
g
)1
h
)1
) o
me abolic powe inpu (W), da a we e di ided by 3600/
21.1 170.64, assuming ca bohyd a e ca abolism and a me abolic
RQ o 1.
Resul s
Choe onyc e is mexicana
All ba s we e ac i e h oughou he whole nigh . Ac i i y
bou s no mally las ed up o 15 min and we e in e up ed
by es ing pe iods o app oxima ely equal du a ion.
To al ¯igh ac i i y du ing a nigh a e aged 3.0 h (2.7
SD, minimum 0.58, maximum 9.3) and o al ho e ing
¯igh ime a e aged 0.38 h (0.13 SD, minimum 0.10,
maximum 0.58). On a e age a ba consumed 17.5 ml
(11.8 SD, minimum 1.5, maximum 36.8) o suga
wa e pe day. Ba s p olonged hei ho e ing du a ions
a he eede o a ying deg ees, due o he delayed ood
deli e y (Table 1). One single indi idual wi h a e y low
body mass (m
1
) ho e ed o up o 17 s.
A e he ansi ion om es ing o ¯ying, a ba needs
se e al seconds o each i s new espi a o y s eady-s a e
(Win e 1998). In his con ex we ound ha _
Vo2in
Choe onyc e is was signi®can ly dep essed when he
du a ion o he ¯igh bou p eceding a ho e ing e en
was less han 4 s. Fo his eason da a analysis was e-
s ic ed o ho e ing e en s p eceded by a ¯igh ac i i y
o a leas 4 s du a ion.
All Choe onyc e is inc eased hei _
Vo2du ing he ® s
5 s o a ho e ing e en . This is illus a ed bo h by he
esul s o he as - esponse measu emen s o _
Vo2(Fig. 1)
and by he cumula i e measu emen s made du ing
ho e ing e en s o die ing du a ions (Fig. 2A). O he
ou indi iduals, only wo ba s ho e ed o long enough
o each a pla eau in _
Vo2(Fig. 2A). This inc ease in _
Vo2
du ing a ho e ing ¯igh was also shown in he mean
a es o _
Vo2du ing bo h sho (<1.5 s) and long ho -
e ing ¯igh s (>5 s) (Table 1). In con as o _
Vo2, he
ca bon dioxide elease a e _
Vco2was almos cons an
du ing he ® s 8 s o ho e ing ¯igh (Fig. 2B). Only in
m
1
did _
Vco2inc ease du ing ho e ing ¯igh s o longe
han 9 s. As a consequence o he une en change in _
Vo2
and _
Vco2du ing a ho e ing e en , he RE showed a
ma ked de ia ion om 1 du ing ho e ing e en s o 4±
10 s du a ion (Fig. 2C).
Fo calcula ing a species mean o me abolic ho e -
ing powe inpu in Choe onyc e is we only included
da a om longe ho e ing ¯igh s (>5 s, Table 1).
Mean O
2
up ake o all h ee indi iduals was 25.2 ml
O
2
g
)1
h
)1
a a mean body mass o 15.4 g, which co -
esponds o a powe inpu o 2.25 W o 146 W kg
)1
.I
he alues o m
1
a e excluded because o i s a ypically
low body mass (see Discussion) hen mean O
2
up ake
was 27.3 ml O
2
g
)1
h
)1
a a mean body mass o 16.5 g,
which con e s o a powe inpu o 2.63 W o
160 W kg
)1
.
Hylonyc e is unde woodi
The wo emale Hylonyc e is unde woodi showed a mean
o al ¯igh ac i i y o 1.26 h pe nigh (0.17 SD,
minimum 0.22, maximum 1.55) and a mean o al ho -
e ing ¯igh ime o 0.22 h pe nigh (0.05 SD, mini-
mum 0.07, maximum 0.87). We measu ed gas exchange
a es du ing 206 ho e ing e en s o emale 1
(2.11  1.02 s, minimum 0.15, maximum 8.77;
mean  SD) and 38 ho e ing e en s o emale 2
(3.08  1.31 s, minimum 0.88, maximum 4.97;
mean  SD). A K uskal-Wallis es e ealed no signi-
®can die ence in mass-speci®c gas exchange a es be-
ween he wo indi iduals (P> 0.05), and we he e o e
pooled he da a. The mass speci®c a es o O
2
up ake a e
shown in Fig. 3. In e es ingly, in he small Hylonyc e is
we ound no change in _
Vo2 o ho e ing ¯igh s o di -
e ing du a ions (slope o linea eg ession did no die
om ze o,
2
0.008, )1.365, P0.17,
n242). The mean O
2
up ake o emale 1
(7.1  0.1 g; mean  SD) was 28.8 ml g
)1
h
)1
(7.1
SD) and o emale 2 (6.9  0.08 g; mean  SD) i was
25.7 (9.6 SD). Con e ed o powe his gi es a species
mean o 1.12 W o a 7.0-g ba o 160 W kg
)1
.
Discussion
Powe inpu du ing ho e ing
The powe inpu du ing ho e ing ¯igh (P
h
), as es i-
ma ed in his s udy om oxygen consump ion, was
1.12 W o 160 W kg
)1
in he small H. unde woodi (7.0 g)
and be ween 1.69 W and 2.84 W (o 128±162 W kg
)1
)in
he la ge C. mexicana (13.2±17.5 g). These alues a e in
line wi h ou ea lie ®nding o Glossophaga so icina
Fig. 3 Mean a es o mass-speci®c gas exchange du ing ho e ing
e en s o die ing du a ions o wo Hylonyc e is unde woodi ba s.
The da a combine measu emen s (n242) om wo indi iduals
(6.9 g and 7.1 g) ha did no die signi®can ly om each o he
(K uskal-Wallis es , P> 0.05). Ra es o O
2
and CO
2
exchange
die ed signi®can ly o ho e ing e en s las ing less o longe han
2s ( - es , P< 0.05), bu he wo o e all means did no die
signi®can ly om each o he ( - es , P> 0.7). O
2
®lled symbols,
CO
2
open symbols
42

(1.88 W a 11.9 g o 158 W kg
)1
, Win e 1998). This
s eng hens he e idence o he conclusion ha me a-
bolic powe inpu inc eases only mode a ely du ing
ho e ing ¯igh as compa ed o ho izon al o wa d ¯igh
in glossophagine ba s. Acco ding o Win e and on
Hel e sen (1998), me abolic ¯igh powe du ing le el
o wa d ¯igh (P
) a in e media e speed in small ba s
ollows he ela ion P
(W) 50.2 M(kg)
0.771
whe e M
is body mass which yields 1.09 W and 2.22 W o a 7.0 g
and 17.5 g ba , espec i ely. Wi hin he ange o body
masses in es iga ed he e, powe inpu du ing ho e ing
¯igh amoun ed o only 1.03±1.28 imes he cos o
ho izon al o wa d ¯igh a in e media e speeds. Con-
side ing ha his eg ession equa ion may o e es ima e
minimum ¯igh powe (P
min
) by 10% (Win e and on
Hel e sen 1998), hen ho e ing ¯igh powe inpu would
amoun o 1.14±1.42 imes P
min
o he 7.0 g and 17.5 g
ba s. This die ence in powe inpu be ween ho e ing
and ho izon al o wa d ¯igh is well below he p edic ed
ae odynamic powe ou pu based on an ae odynamic
and kinema ic analysis in Glossophaga so icina, assum-
ing quasi-s eady condi ions (No be g e al. 1993). No -
be g and co-wo ke s p edic ed ha ae odynamic powe
ou pu du ing ho e ing ¯igh would exceed he co e-
sponding alue o ho izon al o wa d ¯igh by a ac o
o 1.7 (pe ec elas ic s o age) o 2.6 (ze o elas ic s o -
age). This is ma kedly highe han he ange o 1.1±1.4
ob ained in he p esen s udy and he ac o o 1.2 ob-
ained p e iously o Glossophaga so icina (Win e
1998). No be g e al. (1993) used a quasi-s eady ap-
p oach based on wing kinema ic da a and wake eloci y
measu emen s o Glossophaga so icina o de i e ae o-
dynamic pa ame e s. Howe e , i uns eady eec s a e
in ol ed, a quasi-s eady app oach may unde es ima e
he induced powe , while he o he powe componen s
emain unaec ed (No be g e al. 1993). Since acco ding
o heo e ical calcula ions he induced powe componen
in ho e ing ¯igh is h ee imes highe han in ho izon al
o wa d ¯igh (No be g e al. 1993), he conside a ion o
possible uns eady eec s in he ae odynamic models
migh lead o an e en la ge die ence in he heo e i-
cally expec ed ae odynamic powe ou pu be ween hese
wo modes o ¯igh . I seems unlikely ha die ences in
muscula eciencies be ween ho e ing and ho izon al
o wa d ¯igh can accoun o his de ia ion as his
would mean ha muscle eciency a he egula o ag-
ing ¯igh speed which accoun s o mo e han 90% o
daily ¯igh ene gy expendi u es (Win e and on He-
l e sen 1998) would be especially low. Thus a dis inc
and p esen ly un esol ed die ence be ween he esul s
o me abolic measu emen s and hose o he quasi-
s eady ae odynamic analysis exis s (c .Elling on 1991).
Powe inpu du ing o wa d ¯igh
The cos o ho izon al o wa d ¯igh in small ba s
(<35 g) has hus a only been de e mined by me abolic
ene gy balance me hods. Using his p o ocol, he o al
ene gy u no e o an animal was asce ained o e a
leng h o ime consis ing o pe iods o bo h ¯igh and
es . Fligh cos was de e mined by sub ac ing he es-
ima ed cos en ailed du ing he non-¯igh pe iod om
he o al ene gy u no e (Speakman and Racey 1991;
Win e e al. 1993; Win e and on Hel e sen 1998). The
esul s o he measu emen s ca ied ou o da e a e ca.
20±25% lowe han p edic ed by he majo i y o he
allome ic equa ions o bi d ¯igh (Masman and
Klaassen 1987; Rayne 1990), which indica es a me a-
bolically less cos ly o wa d ¯igh by small ba s a hei
slowe speeds.
App oxima ions o o wa d ¯igh cos can also be
de i ed om measu emen s conduc ed in he p esen
s udy, allowing es ima es o ¯igh cos ob ained o da e
o be compa ed wi h esul s o espi ome y. Immedi-
a ely ollowing he change in pe o mance le el om
o wa d ¯igh o ho e ing, a es o gas exchange mea-
su ed a ® s co espond o hose o o wa d ¯igh be-
cause o he espi a o y delay. Such an eec was
obse ed in Choe onyc e is (Fig. 2, Table 1). The a e o
oxygen up ake du ing sho ho e ing bou s (<1.5 s)
die ed clea ly om he alue ollowing longe bou s o
ho e ing. This alue, a 19.0 ml O
2
g
)1
h
)1
o 1.8 W
(a e age mass o 16.0 g), was lowe han, bu no sig-
ni®can ly die en om he expec ed alue o 2.1 W o
he cos o o wa d ¯igh . The espi ome ic da a a e
hus consis en wi h he o wa d ¯igh cos s ob ained
using he ene gy balance me hod, as al eady shown in
p e ious s udies o Glossophaga so icina (Win e e al.
1998; Win e 1998). Mo eo e , hese esul s suppo he
®ndings ha o wa d ¯igh cos s o small ba s a e ela-
i ely low. Only in Hylonyc e is did we ®nd no die ence
in he _
Vo2be ween sho and longe ho e ing ¯igh
(Fig. 3). Howe e , his is no expec ed since in his
species (a 7 g, he smalles o hose es ed) he die ence
in ¯igh powe be ween ho e ing and o wa d ¯igh was
ela i ely small (see Fig. 4B). The du a ion o he e-
spi a o y ansi ion pe iod is assumed o be pa icula ly
sho , so ha adjus men o he ho e ing me abolic le el
occu s e y quickly.
Scaling o ho e ing powe wi h body mass
In a-speci®c scaling
Ho e ing powe inc eases wi h body mass. He e, ho -
e ing powe inpu in Choe onyc e is inc eased om
1.69 W a 13.2 g o 2.84 W a 17.5 g (Table 1), p o-
po ional o M
1.76 (0.21 SE)
(n4, Fig. 4A). The he-
o e ically expec ed change o ¯igh powe wi h body
mass depends on he associa ed change o body geom-
e y (i.e. wing a ea and body shape). When body mass
inc eases bu wing a ea emains cons an (wi hin a single
indi idual) hen ¯igh powe inc eases mo e han when
bo h mass and wing a ea inc ease oge he (as be ween
43
die en ly sized indi iduals) (No be g and Rayne
1987). In he p esen case, he h ee Choe onyc e is in-
cluded in his analysis (ho e ing >5 s, Table 1) we e o
nea ly iden ical linea body dimensions despi e he 40%
a ia ion in mass du ing he measu emen s. Bo h o e-
a m leng h (a sui able in a-speci®c index o body size in
ba s) and wing a ea a ied only by 1% be ween he h ee
indi iduals measu ed (Table 1). Powe inpu o ho e -
ing in his case should scale wi h mass acco ding o he
expec a ions o in a-indi idual body mass changes. Fo
his eason, we used he wo measu emen s o m
1
a
die en body masses as wo da a poin s.
A heo e ical expec a ion o he inc ease in ae ody-
namic powe ou pu is dicul o de i e wi hou
knowing how kinema ic and ae odynamic pa ame e s
changed. While he induced powe componen o ho -
e ing ¯igh should inc ease wi h an exponen o 1.5 o
body mass, he change in ine ial powe depends (besides
he un esol ed ques ion o elas ic s o age) on wing ip
eloci y and wing bea ampli ude (No be g e al.1993)
which we e no measu ed he e. Beside hese wo powe
componen s, p o®le powe is o mino impo ance du -
ing s a iona y ho e ing (No be g e al. 1993).
A lowe mass exponen han measu ed in he p esen
s udy was ob ained om an analysis o ae odynamic
powe ou pu o a i®cially loaded hummingbi ds du ing
ho e ing ¯igh (Chai and Milla d 1997). When he
masses o ou species o 3±7-g hummingbi ds we e in-
c eased o 2±3 imes body mass, es ima ed ae odynamic
powe ou pu changed wi h an exponen o mass o
a ound 1.4.
To ob ain an es ima e o ypical body masses o ou
expe imen al animals, we ecap u ed and weighed hem
6 mon hs a e he measu emen s. All ou
Choe onyc e is had inc eased hei body masses and
weighed be ween 16.8 g and 18.3 g; m
1
inc eased in mass
by 30%. These alues co esponded o alues eco ded
in he animal's na u al habi a . Six Choe onyc e is wi h a
mean o ea m leng h o 45.7 mm had a median mass o
17.0 g, wi h ® e indi iduals be ween 16.8 g and 17.5 g
and one indi idual a 14.8 g (Mexico, Tapan i, O. on
Hel e sen pe sonal obse a ion). Fligh mo phological
and physiological pa ame e s in Choe onyc e is may
ha e become adap ed du ing e olu ion o suppo adul
body masses be ween 16 g o 18 g. Only he wo indi-
iduals,
1
and
2
, o which da a we e ob ained om
ho e ing ¯igh s o longe han 5 s ell in he mass ange
ha is ypical o his species. The e o e, he species
mean alue o Choe onyc e is used o he in e speci®c
compa ison below was based on hese wo indi iduals
only o ensu e ha i mo e likely e¯ec ed he alue
expec ed unde na u al condi ions.
In e -speci®c scaling
Da a a ailable o glossophagine ba s p o ide he op-
po uni y o de e mine a ® s p o isional es ima e o he
in e -speci®c scaling o ho e ing ¯igh cos , and possible
ene ge ic limi s o ho e ing in ba s. Ho e ing powe o
h ee species wi h mean body masses anging om 7 g o
16.5 g inc eased in p opo ion o M
0.95
(Fig. 4B). Based
on n3 species, his mass exponen , howe e , is nei-
he signi®can ly die en om he p e iously de i ed
slope o he cos o ho izon al o wa d ¯igh (0.771,
Win e and on Hel e sen 1998) no om he alue o 1,
hus any in e ences based on his die ence mus be
conside ed p o isional.
An in e es ing compa ison can be made wi h he e-
la ionship es ablished by Thomas (1987) be ween body
size and submaximal ae obic pe o mance o ba s (95±
800 g). He used maximum me abolic a es measu ed
du ing wind unnel ¯igh s o de i e he equa ion ( e-
calcula ed o SI-uni s) P
submax
(W) 65.7 M(kg)
0.70
(Fig. 4B). Fo an 11.9-g Glossophaga p edic ed P
submax
equals 3.0 W, compa ed o 1.88 W ho e ing powe in-
pu . While i mus be kep in mind ha he accu acy o
Table 1 Oxygen up ake du ing ho e - eeding ¯igh s in ou in-
di iduals o Choe onyc e is mexicana ba s de e mined by wo di -
e en me hods (cumula i e and as - esponse, see Ma e ials and
me hods). Values a e means  1 SD. The species mean me abolic
ho e ing powe inpu o C. mexicana was de e mined as 27.3 ml
O
2
g
)1
h
)1
(2.63 W o 160 W kg
)1
) a 16.5 g. This is he mean o
he alues o
1
and
2
du ing ho e ing ¯igh s o longe han 5 s;
m
1
was no included as i s mass du ing he measu emen s was
un ypically low o his species (see Discussion). In bo h
1
and m
1
,
O
2
up ake a e was signi®can ly die en be ween sho (<1.5 s)
and long (>5 s) ho e ing ¯igh s (Mann-Whi ney-Rank-Sum-Tes ,
P< 0.05). Wing a eas (A) we e de e mined acco ding o No be g
and Rayne (1987) and include body and ail memb ane a eas
which make up abou 7% o he o al alue. Ra es o O
2
up ake
(ml O
2
g
)1
h
)1
) we e con e ed o powe (W) by mul iplying by
mass and di iding by 3600/21.1 = 170.64, assuming ca bohyd a e
ca abolism and he espi a o y quo ien = 1. All ho e ing e en s
included in his analysis we e p eceded by a ¯igh in e al o a leas
4 s p io o ho e ing. (bwing span, FA o ea m leng h, Ind. in-
di idual ba , _
Vo
2
oxygen up ake a e)
Ind. Body
mass
FA
mm
b
m
A
m
2
Ho e
e en s
Du a ion _
Vo
2
< 1.5 s _
Vo
2
> 5 s Powe > 5 s
W
Type o
measu emen
gnmean max. ml g
)1
h
)1
nml g
)1
h
)1
n
ss
1
17.50.3 45.4 0.304 0.0145 252 2.61.2 8.8 18.16.8 40 27.75.1 23 2.84 0.52 Cumula i e
2
16.31.3 46.0 0.312 0.0147 292 2.90.8 4.9 18.17.5 71 Cumula i e
2
15.40.1 46.0 0.312 0.0147 38 3.62.3 8.6 26.82.3 38 2.420.21 Fas - esponse
m
1
14.40.6 45.6 0.315 0.0145 414 6.74.6 16.7 21.75.9 71 24.32.7 236 2.05 0.23 Cumula i e
m
1
13.20.2 45.6 0.315 0.0145 44 6.41.7 10.1 21.82.1 44 1.69 0.16 Fas - esponse
m
2
15.70.5 45.0 0.314 0.0153 224 2.11.2 5.3 18.28.5 6 Cumula i e
44
his ex apola ion o P
submax
o he small body sizes o
glossophagine ba s is unce ain, he wide powe ma gin
p edic ed may indica e ha ho e ing ¯igh capaci y o
he majo i y o small o medium sized glossophagine
species is no limi ed by me abolic equi emen s. In his
con ex i is in e es ing o no e ha e en p egnan e-
males o Glossophaga a 130±140% adul body mass eed
om ¯owe s while ho e ing (Voig 1998). The die en
mass exponen s be ween P
submax
(0.70) and P
h
(0.95)
lead o he expec ed esul ha la ge animals mus in-
es p opo ionally mo e in ho e ing ¯igh han small
animals. Howe e , as can be seen om Fig. 4B, all
ho e ing glossophagine ba s should be able o sus ain
ae obic ho e ing ¯igh up o hei maximum size o 30±
35 g. E en ually, me abolic cons ain s may limi he
body size o a ho e ing glossophagine ba and, inci-
den ally, he la ges glossophagine species, Lep onyc e is
ni alis, has a body mass o up o 30±35 g.
Sus ained ho e ing
A e we obse ed ha glossophagine ba s mos ly ho -
e ed o e y sho pe iods, o o a ew s a mos , we
assumed ha hey we e undamen ally unable o main-
ain ho e ing ¯igh o longe pe iods. Howe e , du ing
he p esen in es iga ion we obse ed o he ® s ime
ha a glossophagine could ho e o a ime in e all o
17 s (Table 1); in addi ion, 7.5-g Saccop e yx bilinea a
males (Chi op e a, Emballonu idae) we e obse ed o
ho e o 14.4 s du ing display ¯igh s in on o emales
(C. C Voig and O on Hel e sen pe sonal obse a ion).
This demons a es ha he basic chi op e an s uc u e
pe mi s sus ained ho e ing. The sus ained ho e ing o a
Choe onyc e is was possible unde he ollowing condi-
ions: while Choe onyc e is no mally weighs ca 17 g,
indi idual m
1
had a empo a ily educed body mass o
13.2 g o 78% o his alue. Wing loading was he e o e
educed by he same amoun , om 11.5 N m
)2
o
8.9 N m
)2
( hese alues inc ease by 7% when body a ea
and ail memb ane a e excluded om he `wing a ea'; see
Table 1). Simila ly, wing disc (pb
2
/4) loading dec eased
om 2.15 N m
)2
o 1.67 N m
)2
. Ho e ing powe de-
c eased om 2.7 W o 1.7 W, a alue well below he cos
o o wa d ¯igh a in e media e speed (2.2 W), expec ed
o a ypical 17-g Choe onyc e is. Thus, ene ge ic and
ae odynamically ele an pa ame e s changed in con-
co dance and i p esen ly emains un esol ed which was
mos c i ical o sus ained ho e ing. Tha sus ained
ho e ing is no obse ed in ba s mo e o en is p obably
explained by he dominance o o he selec ion p essu es
on ¯igh abili ies.
Bi ds, ba s and mo hs
Th ee g oups o nec a - eede s wi h nea ly o e lapping
size anges wi hin he Insec a, A es and Chi op e a ha e
e ol ed specialized adap a ions o ho e ing ¯igh :
glossophagine ba s, hummingbi ds and hawkmo hs.
Figu e 5B compa es he mass-speci®c me abolic ene gy
expendi u e equi ed o ho e ing in he h ee g oups.
Al hough all eg ession slopes appea ed o be sligh ly
nega i e, in no case did he mass exponen die om
ze o o a s a is ically signi®can deg ee (P> 0.05).
Thus, o he ange o body masses conside ed he e, he
da a did no indica e any changes in he mass-speci®c
cos s o ho e ing wi hin g oups. Mean mass-speci®c
Fig. 4A, B Scaling o me abolic powe inpu o ho e ing wi h
body mass in AC. mexicana and B h ee species o glossophagine
ba s. Da a in Aa e om Table 1 (ho e ing du a ions >5 s). Da a
in B,solid line wi h ®lled symbols, combine he species means om
he p esen s udy o Hylonyc e is (n2) and Choe onyc e is
(n2, only
1
and
2
, because o low body mass o m
1
, see
Discussion) and a published alue o Glossophaga so icina (11.9 g,
1.88 W, Win e 1998). Addi ional unc ions in Bwe e included o
compa ison wi h ho e ing ¯igh cos s. Dashed line ene gy cos o
le el o wa d ¯igh a in e media e speed (P
) (W) 50.2 body
mass (M) (kg)
0.771
(Win e and on Hel e sen 1998); dashed-do ed
line es ima ed minimum powe ¯igh cos ( 0.9 ´dashed line);
do ed line: ex apola ion o submaximal ae obic ¯igh capaci y
(P
submax
) in 95±800-g ba s du ing wind unnel ¯igh s, P
submax
(W) 65.7 M(kg)
0.70
(Thomas 1987, ecalcula ed o SI-uni s).
Open symbols a e, o compa ison, he indi idual da a poin s o
Choe onyc e is om A. The eg essions gi en in he ®gu es we e
based on mass in kg; s anda d e o s o eg ession coecien s we e
A3544  2057 and 1.76  0.21, n4, B128  12.4 and
0.950  0.037, n3
45
ho e ing ¯igh cos s we e 393 (98.8 SD, n13 spe-
cies) W kg
)1
in sphingids, 261 (23.6 SD, n10
species) W kg
)1
in hummingbi ds, and 159 (0.76 SD,
n3 species) W kg
)1
in glossophagine ba s. These
alues we e signi®can ly die en om each o he
(ANOVA on anks, P< 0.001). Thus, glossophagine
ba s expend only 40% o he ene gy o a hawkmo h and
61% o a hummingbi d o main ain ho e ing ¯igh . A a
me abolic powe inpu o 1.1 W a glossophagine ba can
gene a e he li o ces necessa y o balancing 7 g
agains g a i a ion, whe eas a hummingbi d can suppo
4 g and a sphingid mo h only 3 g o body mass
(Fig 5 A). Wi h he same expendi u e o me abolic en-
e gy, a glossophagine ba can he e o e gene a e a con-
side ably g ea e li o ce han membe s o he o he
wo g oups.
Minimum induced powe
The essence o ho e ing ¯igh is he p oduc ion o a
e ical o ce o balance he body mass o an animal
agains g a i a ion. The main powe d ain o ho e ing
is he induced powe , and he equi emen s o he in-
duced powe can be educed by inc easing he wing
span. This is a classical solu ion o bo h animals and
helicop e s, and glossophagine ba s wi h hei long
wings make use o i . Fo a ela i e compa ison be ween
he h ee g oups, minimum induced powe s (P
indRF
)
we e calcula ed om Rankine-F oude momen um-je
heo y (No be g 1990). We de e mined he alues o
hose ep esen a i es o he h ee g oups wi h an iden-
ical me abolic powe inpu o ho e ing o 1.1 W
(Fig. 5A, Table 2). This ® s app oxima ion o a com-
pa ison be ween powe inpu and ou pu du ing ho e -
ing indica es ha die ences in induced powe can pa ly
explain he die ence in me abolic powe equi emen s
be ween ba s and hummingbi ds bu none be ween ba s
and mo hs. Wi h he same ene gy inpu , ba s appea ed
o be able o p oduce an ae odynamic powe ha is 30%
highe han in hummingbi ds and 130% highe han in
sphingid mo hs. To explain hese die ences, an ae o-
dynamic model ha is mo e de ailed han he classical
quasi-s eady ae odynamic app oach o ho e ing ¯igh
may be equi ed.
Manoeu abili y
One consequence o he die ences in ¯igh kinema ics
be ween he g oups is an appa en ly educed ¯igh ac-
cele a ion om ho e ing in glossophagine ba s as com-
pa ed o hummingbi ds o sphingid mo hs (al hough
his has no been quan i®ed ye ). Selec ion o high
manoeu abili y in hese wo g oups has been ela ed o
i s impo ance o in e e ence in compe i ion and e -
i o iali y in hummingbi ds (Feinsinge and Chaplin
1975; Feinsinge e al. 1979) and o p eda ion p essu e
du ing nec a - eeding in sphingid mo hs (Wasse hal
1993, 1997, 1998). Te i o iali y o he ype obse ed in
hummingbi ds is unknown and unlikely in glossopha-
gine ba s. In addi ion, hei main an i-p eda o s a egy
du ing nec a - eeding appea s o lie in he b e i y o
hei ¯owe isi s ( ypically less han 0.75 s), as hei
abili y o eac o an ambush p eda o is se e ely limi ed
when eyes (and ea s) a e inse ed in he ¯owe co olla
du ing eeding (see pho og aphs in Tu le 1991 and
Win e 1998).
Fig. 5A, B Scaling o me abolic powe inpu o ho e ing wi h
body mass in sphingid mo hs, hummingbi ds (T ochilidae) and
nec a - eeding ba s (Phyllos omidae: Glossophaginae). Da a in A
a e indi idual and da a in Bmass-speci®c alues. The dashed line in
Aconnec s he body masses wi h equal me abolic powe inpu o
ho e ing o 1.1 W. Da a o sphingid mo hs om Ba holomew
and Casey (1978); da a o hummingbi ds om Be ge and Ha
(1972), Be ge (1985), Ep ing (1980), Ba holomew and Ligh on
(1986), Sua ez e al. (1990), Chai and Dudley (1996); da a o
glossophagines as in Fig. 4. When se e al alues we e epo ed o
a species, da a we e combined in o a single mean alue o each
species. Reg ession equa ions in Awe e: P(W) aM(kg)
b
Sphingidae: a 185 (17.7 SE), b 0.882 (0.104 SE)
(n13 species), T ochilidae: a 186 (17.8 SE), b 0.933
(0.049 SE) (n10 species), Glossophaginae: a 128 (12.4
SE), b 0.950 (0.037 SE) (n3 species), all exponen s
signi®can ly die en om ze o (P< 0.001) bu no signi®can ly
die en om 1 (P> 0.05). In Bmass exponen s b did no die
signi®can ly om ze o (P> 0.05) o he mass anges conside ed
he e
46