FLIGHT SPEEDS

FRANK B. GILL TheAcademy of NaturalSciences, Philadelphia, Pennsylvania 19103 USA

ABSTR^CT.--Long-tailedHermits (Phaethornissuperciliosus) normally fly fasterthan the ve- locity predictedto minimize their costof transport.The averagespeed we measuredfor individualsflying a known40-m course was 11.5 m/s. Rapidflower visitation yields rewards thatcould compensate for the extracosts of fastflight. Received 2 March 1984, accepted 30 July 1984.

How fast should a fly? Intermediate relate to differencesin flight ecology (Fein- flight speeds that minimize instantaneous singerand Chaplin 1975,Feinsinger et al. 1979). power costs (Vmp),or more realistically that Theoretically,forward flight speedsalso should minimize the costof flying a certain distance influencea hummingbird'sflight costsand op- (Vm,),can be predicted from aerodynamicthe- timal wing lengths (Pennycuick1969, Greene- ory that is devoid of ecological context. The wait 1975);however, these speeds have not been potential for minimization of power costsat determined in an ecological context. intermediatespeeds emerges from the U-shaped Here I considerthe flight speedsof a 6-g her- form of the relationshipbetween power costs mit hummingbird, the Long-tailed and flight speed(Fig. 1; Pennycuick1969, 1975; (Phaethornissuperciliosus). Hermit humming- Greenewalt 1975). relate directly to the questionsposed When can birds justify the extra energetic above becauseforward flight is the dominant investment required for flight at speedsslower componentof their routine flight ecology.They or faster than Vm,?The advantagesof power fly considerabledistances between dispersed dives, top-speed chases,and escape flights flowers (Stiles 1979). Some speciesof hermit clearlyoutweigh the energeticsacrifices of brief ,including the speciesfeatured high-speed flights. Stationary flight, such as here,also commute frequently from lek display hoveringat flowers,also may yield net rewards grounds to distant flowers. The estimates of (Pyke1981). However, the naturalflight speeds flight speedsI presentare the first for a hermit of birds in routine contextsrarely are com- hummingbirdand amongthe few naturalflight pared to aerodynamic models. In one excep- speedsrecorded for any hummingbird. tional study Schnell and Hellack (1979) found that gulls and terns (Laridae) cruise near their METHODS coloniesat velocitiesbetween Vmpand V•,, i.e. at intermediate speeds at the bottom of the This study was carried out in the lowland rain for- est on the Osa Peninsula of Costa Rica. In June 1982 power curve. The need for more such information is a single observerwith a stopwatchtimed flights of Long-tailedHermits as they flew 40 m directly from heightenedby the recognitionthat time gained one artificial flower to another along a broad trail. by flying fast may be used profitablyin ways Thesehummingbirds visited the flowersregularly as that compensatefor the extraenergy expended a part of normal, daily trapline foraging. The observ- (Norberg 1981). Hummingbirds should be of er sat near one flower, started the stopwatch when particular interest in this context becauseof the hermit left it, and stoppedthe stopwatchas the their sensitivityto the energeticconsequences bird pulled up to the distant flower. Arrivals were of alternative behaviors (DeBenedictis et al. studiedat the distantflower through 10 x binoculars. 1978). Some slight error in judgment, reaction time, and Previousanalyses of the ecologicalcorrelates parallaxwas inherent in thesemeasurements. Minor acceleration and deceleration of unknown duration of hummingbirdwing lengthshave stressedthe near the beginning and end of a flight make these costsof hovering flight, a hummingbird spe- measurementsslight underestimatesof true flight cialty. The costof hovering is a direct function speed.We excludedtimes of flights during which the of body weight and wing span (Greenewalt hummingbirdveered off the trail and out of sight. 1960a, 1975; Pennycuick 1969, 1975). Slight The wing lengths and massesof Long-tailed Her- variationsin wing lengthsrelative to body mass mits netted and marked on the Osa Peninsula in 1979- 97 The Auk 102:97-101. January1985 98 FRANKB. GILL [Auk, VoL 102

0.200.24

0.16

o.12

0.08

Pmr Pmp 0.04

0.00 0 2 5VmpVmr 8 11Vest 14 Velocity (m/s) Fig. 1. Theoreticalflight power curve for the Long-tailedHermit calculatedfrom Eq. 44 in Greenewalt (1975:39). Minimum power velocity(Vm•) and maximumrange velocity (Vm,) and associatedpower require- ments for this speciesare indicated by dashedlines. H is the cost of hovering projectedfrom Eq. 38 in Greenewalt(1975: 38). The positionof the averageflight speed(Ve,,) presented in this paper for the Long- tailed Hermit is indicated by a star.

1982 were 61.8 + 0.2 mm and 6.3 + 0.4 g (n = 271). The flight times we recordedaveraged 3.47 s These values fall close to the fitted regression for (n = 10, SD = 0.216, range 3.2-3.8 s), which cal- hummingbirdsas a group (Fig. 2). culates to an estimated average velocity of 11.57 + 0.71 m/s (41.8 + 2.56 kph). The Long- RESULTS tailed Hermits flew 40 m in 5.3 s. These hermit hummingbirdsflew not at Vm,or V•,,,but fast- V•, is about5.5 m/s (20 kph) for all hum- er-at speedsof 2.1 x V•, and 1.6 x V•,,, cost- mingbirds;the range of hummingbirdbody ing 72% of the energeticrequirements of hov- massesand wing spansis not sufficientto affect ering at flowers (see Fig. 1). Our slowest thisgreatly (Greenewalt 1975: 39). V•, is higher recordedflight speed(3.8 s = 10.5m/s) was 3.2 [7.5 m/s (27 kph)] for hummingbirds. Using m/s faster than V•, for this . Greenewalt'sEq. 45 (1975: 39) for the Long- Long-tailedHermits also fly fast when they tailed Hermits studied,we projectedVm, to be commutethrough relatively clear midstrataof 5.4 m/s and V•,, to be 7.3 m/s. The power re- the rain forest between lek territories and feed- quirementsof this speciesflying at 1.4m/s and ing areasup to 500 m away. The two flight 12.5 m/s should be about 3 times those at speedsof commuting hummingbirdsthat we and only slightlyless than the costof hovering were able to measure were 10 and 11 m/s. (Fig. 1). Field measurementsof flight speedswith a January1985] HummingbirdFlight Speeds 99

lO

8

7

4

3 2 3 4 5 6 7 9 10 11 12

Weight, grams

Fig. 2. Dimensionsof hummingbirdsfor which samplesizes of more than 15 specimenswere available (from Greenewaltunpubl.). Brokenlines indicate! SD of the leastsquares regression, the equationfor which is log L = 0.5973log M + 3.315,and the standarddeviation of the intercept= 0.0264.The averagedimensions of the Long-tailedHermit are indicatedby a black star,while thoseof Eutoxeresare indicatedby a white star.

stopwatch are estimates subject to several tomobile-pacedRuby-throated Hummingbirds sources of error. Individual errors of reaction (Archilochuscolubris) are not consistent with time or parallax,however, were at the level of subsequentobservations. Ruby-throated Hum- 0.1-0.2 s or less,an order of magnitudeless than mingbirds, for example, could not progress the difference between estimated flight speed against a 12 m/s (45 kph) headwind in wind and V•r. More detailed studies and larger sam- tunnel experiments,suggesting that their max- ple sizes are needed to establish the flight imum air speed was in this range (Greenewalt speeds of Long-tailed Hermits precisely, but 1960b). these data leave little doubt that they fly faster Wolf et al. (1976) reported that Green Violet- than Vmr. ears (Colibri thalassinus)flew 0.2-1.2 m/s (0.7- 4.3 kph) between flowers less than 1 m apart. DISCUSSION The linear relation of flight speed to distance in thesedata suggeststhat accelerationand de- It remains to be establishedthat any hum- celeration constrained flight speed over such mingbird routinely flies at speedsthat mini- short distances.Montgomerie (1979) recorded mize its power costs.The few published rec- speedsof 0.3-1.2 m/s by CinnamonHumming- ordsof hummingbird flight speedsvary greatly birds (Amaziliarutila) flying short distances(less but do not suggestusual flight speedsat Vm,or than 1 m) between flowers. Thus, slow flight Vmr, speedsmay characterizehummingbirds flying Pearson (1961) released Allen's Humming- between adjacentflowers on a bush or hedge- birds (Selasphorussasin) inside a highway drain- row. agepipe and timed their escapespeeds at 8 m/s Norberg's (I 981) ecologicalmodel of optimal (30 kph) and 11 m/s (40 kph). He concluded flight speeds suggeststhat birds should in- that normalcruising speeds of smallhumming- creasetheir speedabove V•, when foraging, if birdswere approximately11 m/s (40 kph). Ear- the travel time saved compensatesfor the in- ly reports of 21 m/s (75 kph) and 25-28 m/s creasedtravel costs.The foraging ecology of (90-100 kph) (Hayes 1929, Allard 1934) by au- Long-tailedHermits providesthis opportunity. 100 FRANK]3. GILL [Auk,Vol. 102

The 1.9 s saved by flying 40 m at 11.6 m/s hummingbirdspecies spanning the family'ssize instead of 7.5 m/s is closeto the time a Long- range (Greenewalt MS) is tailed Hermit takes to extract nectar from one flower (Gill pers. obs.).The extra costof flying log L = 0.5186log M + 0.3891. (I) 40 m between flowers at 11.6 m/s is roughly 0.144 J, whereas I •tl of nectar in flowers the The relativelylonger wings of largehumming- birds compensatefor their weight in such a hermits typically visit (e.g. Costusand Helico- way that their specificper-gram hovering costs nia)contains roughly 5 J. Extendedto hundreds are the sameas thoseof small hummingbirds of flowers visited daily, the accumulatedgain could be substantial. (Hainsworth and Wolf 1972, Greenewalt 1975). There are two other features of the natural Hermit hummingbirdsdo not have atypical wing lengths,despite their variousbody sizes. history of Long-tailed Hermits that potentially The regressionfor hermits of the genera Phae- may be more important than simple energetic thornis, , Threnetes, and Eutoxeresis compensation.First, nectar rewards in the un- defended, dispersed flowers these birds visit log L = 0.4724log M + 0.3966. (2) depend on which bird gets to a flower first. Individuals rarely have exclusiveuse of a par- The lower slopeof this equationdoes not differ ticular flower (Gill pers. obs.). Even slight re- significantly(d = 1.6110,P > 0.10) from that of visitation delay can result in loss of nectar to a the equation for all hummingbirdsexcluding competingindividual. Second,male Long-tailed these hermit genera, namely, Hermits potentially sacrificereproductive per- formance when they leave their lek territory log L = 0.5647log M + 0.3735. (3) (Stiles and Wolf 1979). Minimization of forag- Hermits of the genus Eutoxeresare aberrant, ing time by meansof fastflight and rapid flow- er visits should be advantageous if the time big (9-12 g) hummingbirdswith short wings saved is used instead for lek activities. that correspond in length to those of hum- mingbirds about 7 g in mass.Instead of hov- Our evidence of fast flight in hermit hum- ering, they usually cling to Heliconiabracts mingbirds also bears on the issue of adaptive while feeding. Their wing lengthscertainly do wing-length variations in hummingbirds. not relate to efficient hovering. The regression Traplining hummingbirdsthat visit dispersed of wing length vs. massfor hermits excluding or low-yield flowers have longer wings and Eutoxeresis even more similar to Eq. 3: lower hovering coststhan do territorial hum- mingbirds (Feinsinger and Chaplin 1975, log L = 0.5092 log M + 0.3760. (4) Feinsinger et al. 1979). Paradoxically, some specialized trapliners, such as the hermit The high averagewing discloading of hermit hummingbirds (Phaethorninae), do not have hummingbirds mentioned by Feinsinger et al. long wings and may even have shorter-than- (1979) apparently was the result of including average wings (Feinsinger et al. 1979). Eutoxeresin their sampleof "hermits"; the av- There is no aerodynamic reason to expect eragevalue of their data set excludingEutoxeres high-speedtrapliners, such as the Long-tailed (by my calculations)was not significantly dif- Hermit, to evolve longer-than-averagewings. ferent from other groupsof hummingbirds. Selection should favor longer-than-average Variations in wing disc loading and hover- wings in hummingbirds that fly slowly or that ing power requirementsdiscussed by Feinsing- hover at low-yield flowers for much of their er and Chaplin (1975), Feinsinger et al. (1979), foraging time (Feinsinger et al. 1979). How- and others are departures from the average ever, selection could favor shorter-than-aver- trend, the variance about the regressionrather age wings in hummingbirds that fly fast be- than the slope of the regression itself. How- cause shorter wings reduce power costs of ever, this variance(Eq. I) is not great;the stan- profile drag, which increasewith the cube of dard deviation of the intercept is 0.0393. Con- velocity (Pennycuick 1969, Rayner 1979). ceivably,some traplining hermit hummingbirds Hermit hummingbirdsas a group do not have have higher wing disc loadings than other shorter-than-averagewing lengths. The rela- specieswith which they coexist,a result per- tion between wing length (L, mm) and body haps of local sorting of specieswith different mass(M, g) in a large sample (n = 244) of 130 foraging strategies and corresponding wing January1985] HummingbirdFlight Speeds lengths (Feinsinger et al. 1979). If this proves ergetics,and foraging ecology of tropical hum- to be the case, details on flight speeds would mingbirds. Amer. Natur. 113: 481-497. be needed to determine whether flight power GREENEWALTßC. H. 1960a. Hummingbirds. Garden considerationswere involved in any way. The City, New YorkßDoubleday & Co. short wings of territorial hummingbirds (see 1960b. The hummingbirds.National Geo- graphic118: 658-679. Feinsinger and Chaplin 1975), for example, 1975. The flight of birds. Trans.Amer. Phil. probably relate to the ability to accelerateand Soc. (New Series) 65(4): 1-67. maneuver in chasesand displaysrather than to HAINSWORTH, F. R., & L. L. WOLF. 1972. Power for energetic considerations. hovering flight in relation to body size in hum- mingbirds. Amer. Natur. 106: 589-596. HAYES,S. P., JR. 1929. Speed of flying humming- ACKNOWLEDGMENTS birds. Auk 46: 116. MONTGOMERIE,R. 1979. The energeticsof foraging I am especiallygrateful to C. H. Greenewalt, who and competition in some Mexican humming- proddedmy thinking away from energeticsand who birds. Unpublished Ph.D. dissertation,Montreal, made available unpublished analysesof humming- Quebecß McGill Univ. bird dimensions.R. Montgomerie, F. G. Stilesßand NORBERG,R. A. 1981. Optimal flight speedin birds C. L. Gass generously provided accessto unpub- when feedingyoung. J. Anim. Ecol.50: 473-477. lished manuscripts.P. Feinsinger, C. L. Gass,F. R. PEARSON,O. P. 1961. Flight speedsof some small Hainsworth, J.P. Myers, and L. L. Wolf criticized birds. Condor 63: 506-507. early drafts of this paper. R. Ray, A. MackßD. Wechs- PENNYCUICK,C.J. 1969. The mechanics of bird mi- ler, F. Joyce,P. Marcotullio, and D. Emery assistedin gration. Ibis 111: 525-556. the field. I am indebted also to the staff of the Ser- ß 1975. Mechanicsof flight. Pp. 1-75 in Avian vicio de Parques Nacionales of Costa Rica for their biology, vol. 5 (D. S. Farner, J. R. King, and K. many courtesiesand their assistancein Parque Cor- C. Parkes, Eds.). New York, Academic Press. covado.I extend specialthanks to A. Ugalde, F. Cor- PYKE,G. H. 1981. Why hummingbirdshover and tez, J. Gamboa, H. Delgadoßand M. Fells. honeyeatersperch. Anim. Behav. 29: 861-867. RAYNER,J. M.V. 1979. A new approach to LITERATURE CITED flight mechanics.J. Exp. Biol. 80: 17-54. SCHNELL,G. D., •: J. J. HELLACK.1979. Bird flight ALLARD,H.A. 1934. Speed of the Ruby-throated speedsin nature: optimized or a compromise? Hummingbird'sflight. Auk 51: 84. Amer. Natur. 113: 53-66. DE]•ENEDICTIS,P. A., F. B. GILL, F. R. HAINSWORTH,G. STILES,F. G. 1979. Notes on the natural history of (Musaceae) in Costa Rica. Brenesia H. PYKE,& L. L. WOLF. 1978. Optimal meal size 15(suppl.): 151-180. in hummingbirds. Amer. Natur. 112: 301-316. ß& L. L. WOLF. 1979. Ecologyand evolution FEINSINGERßP., •: S. M. CHAPLINß 1975. On the re- of lek mating behavior in the Long-tailedHer- lationship between wing disc loading and for- mit Hummingbird. Ornithol. Monogr. 27: 1-78. aging strategyin hummingbirds.Amer. Natur. WOLFßL. L., F. G. STILES,& F. R. HAINSWORTH. 1976ß 109: 217-224. Ecologicalorganization of a tropicalßhighland ., R. K. COLWELL,J. TERBORGH,& S. M. CHAPLINß hummingbirdcommunity. J. Anita. Ecol.45: 349- 1979. Elevation and the morphology, flight en- 379.

Nominations for Elective Members and Fellows of the A.O.U. may be made up until 5 monthsprior to the next statedmeeting. Nominations of Fellowsmust be signedby three Fellows.Nominations of Elective Membersmust be signedby three Fellowsor ElectiveMembers. The number of Fellows under 60 yearsof age is restrictedto 75 by the bylaws.Howeverß the numberof new ElectiveMembers each year is in large part determinedby the numbernominated--we can elect up to 75%of the nominees.Prescribed forms may be obtainedfrom the Secretary(see inside front coverof TheAuk for address).Completed forms must be returnedto the Secretaryby 7 May 1985in order to be eligible.

Nominationsfor officersmay be madein writing to the Secretaryat any time prior to the next meeting of Fellowsand ElectiveMembers, which is 7 October1985. The following positionswill be vacantin 1985: Vice President and three Elective Councilors.