144 ShortCommunications [Auk, Vol. 106

biology, vol. 5 (D. S. Farner and J. R. King, Eds.). 1989. flight performance:a practical New York, Academic Press. calculation manual. Oxford, Oxford Univ. Press. --. 1978. Fifteen testablepredictions about bird In press. flight. Oikos 30: 165-176. --, H. H. OBRECHT,& M. R. FVLI.•R. 1988. Em- 1982a. The ornithodolite: an instrument for pirical estimatesof body drag of largewaterfowl collectinglarge samplesof bird flight speedmea- and raptors.J. Exp. Biol. 135: 253-264. surements.Phil. Trans. Roy. Soc.London B 300: ROTHE,H. J., W. BIESEL,& W. N^CHTIC•LL. 1987. Pi- 61-73. geon flight in a wind tunnel: II. Gas exchange 1982b. The flight of petrels and albatrosses and power requirements.J. Comp. Physiol. 157: (Procellariiformes),observed in South Georgia 99-109. and its vicinity. Phil. Trans. Roy. Soc.London B SCHMIDT-NIELSEN,K. 1972. Locomotion:energy cost 300: 75-106. of flying, swimming and running. Science177: 1987a. Flight of auks (Alcidae) and other 222-228. northern seabirdscompared with southern Pro- 1983. physiology: adaptation and cellariiformes:ornithodolite observations.J. Exp. environment. Cambridge,Cambridge Univ. Press. Biol. 128: 335-347. TVClCER,V.A. 1973. Bird metabolismduring flight: 1987b. Cost of transportand performance evaluation of a theory. J. Exp. Biol. 58: 689-709. number, on Earth and other planets.Pp. 371-386 in Comparativephysiology: life in water and on Received5 April 1988, accepted29 August1988. land (P. Dejours,L. Bolis,C. R. Taylor, and E. R. Weibel, Eds.). Berlin, Springer.

The Taxonomic Status of the Small Ground-, (Aves: Emberizidae) of GenovesaIsland, Gal•pagos, and Its Relevanceto Interspecific

JOSEPHVAGVOLGYI AND MARI• W. VAGVOLGYI BiologyDepartment, College of StatenIsland, City Universityof New York,Staten Island, New York 10301USA

From their study of the feeding habitsof Geospiza ficilis]aggregation, but the differentiatingcharacters difficilisand G. fuliginosaon Genovesa,Pinta, and Mar- are such as to make it seem desirable to treat the form chena islands,Galfipagos Archipelago, Schluter and as specificallydistinct." Lack (1945, 1947, 1969) rec- Grant (1982, 1984) concludedthat these speciesprob- ognized the transfer of G. acutirostristo the G. difficilis ably competedin the past.Crucial to this conclusion group,but arguedthat its measurementsoverlapped is the taxonomic identification of the Genovesapop- widely thoseof G. d. difficilisof Pinta Island (Sharpe ulation, classifiedas G. acutirostris,G. fuliginosa,or G. 1888), and combined G. acutirostriswith the latter. difficilisby variousauthors (see below). We present Paynter and Storer (1970) followed Lack'sarrange- morphologicalevidence to indicatethat, contraryto ment. Harris (1973: 265) made "... no attempt... to its current classificationas G. difficilis,the Genovesa discuss the taxonomic status of species." Schluter population may more justifiedly be placed in G. fu- (1984), Grant et al. (1985) and Grant (1987) studied liginosa,as done by Snodgrassand Heller (1904).We the classification of Darwin's , found it solid, also discussthe evolutionaryand ecologicalconse- and suggestedno modifications.Neither of theseau- quencesof this suggestedtaxonomic rearrangement. thorsexamined specifically the statusof the Genovesa The smallground-finch of the genusGeospiza living population.Bowman (1961, 1983) adopted Lack's clas- on GenovesaIsland was first describedby Ridgway sification,but noted (pers.comm.) that the "Genovesa (1894: 363; see also Ridgway 1897) as Geospizaacuti- Geospizasong is quite different from other difficilis rostris,a form "Similar to G. parvula(Gould) [synon- songsas well asfuliginosa songs. Like fuliginosa it lacks ymized sincewith G. fuliginosa],but bill longer,with the 'specialbasic' song of other difficilispopulations, straighteroutlines, and extremelyacute at tip." Roth- and this I think is very significant,indicating alle- schild and Hartert (1899) concurredwith Ridgway's gianceto fuliginosa.(See Bowman 1983, p. 437,fig. 62 view. Snodgrassand Heller (1904:316) characterized and p. 423, fig. 48.)" It thus appearsthat the early the taxon acutirostrisas "Very similar to G. f. fuliginosa, authorsassigned the Genovesaform to the G. fuligi- but bill more acute, with straighter outlines" and nosagroup, on the basis of overall similarities, pri- ranked it as a subspeciesof G. fuliginosaGould, 1837. marily in beak morphology.Swarth and Lack br.oke Swarth (1931: 178) felt that "The Tower [Genovesa] with these views when they assignedthe Genovesa Island acutirostrisis, to my notion, of the Geospizade- form to the G. difficilisgroup. We believe that they bilirostris[currently considered a subspeciesof G. dif- focusedon a single feature,bill length, giving little January1989] ShortCommunications 145 weight to other features. Modern authors adopted Lack's ,some perhaps with reservations, without reanalyzingthe Genovesaissue or any of the difficult casesmentioned by Lack (1947: 18). We studied 762 specimens,identified by museum labelsas either G.fuliginosa or G. difficilis,in the Acad- emyof Natural Sciences,Philadelphia, Pennsylvania, the AmericanMuseum of Natural History, New York, New York, the California Academyof Sciences,San Francisco, California, the Museum of Comparative Zoology, Cambridge,Massachusetts, San Francisco State University, San Francisco,California, and the U.S. National Museum of Natural History, Washing- ton, D.C. Only adult maleswere considered,in order to obtain more homogeneoussamples. The specimens represented31 populationsfrom 27 islands.Most is- landshad either G.fuliginosa or G. difficilis,but Pinta, Santiago, southern Isabela, and Santa Cruz islands supportedboth. We included 20 populations(with n • 10) in the analysis.Following standardprocedures (Baldwin et al. 1931),we measuredbill depth, bill or culmen length, bill width, wing length, and tarsus length. We alsoutilized data from the literature (Lack 1945; Schluter and Grant 1982, 1984; Grant et al. 1985), and Bowman(pers. comm.). We comparedthe mea- surements of the disputed Genovesa population to 22_ thoseof G. fuliginosaand G. difficilis(Fig. 1A to IF). In five of six characters(namely bill depth, bill 21_ width, wing length, tarsuslength and body mass), the Genovesapopulation fell within the rangeof vari- ation of G. fuliginosa.Its affiliationwas clearestin tar- _ suslength and body mass,because it is in thesefea- turesthat G. fuliginosadiffered most decisively from G. difficilis.The Genovesapopulation has an unusually D. Wo Pt Ma G. $g F, Ni $•i HaePaz B •' C•u S•f FalC: ES long bill, which surpassedall other populationsof G. fuliginosa.On thesegrounds, the Genovesapopulation maybest be describedas one of G.fuliginosa, of small- 28: ishdimensions, that resembles G. difficilis in bill length. Therefore, we suggestthat the Genovesapopulation be transferredfrom G. difficilis,to which it is currently assigned,to G. fuliginosa,as first proposedby Snod- grassand Heller (1904). •24A•,• F

Fig. I. Geographicvariation in Geospizadifficilis and G. fuliginosa.The meansof 20 populationsare shown in: A, bill depth; B, bill length; C, bill width; D, wing length; E, tarsuslength; F, body mass.Symbols: tri- Ci = San Crist6bal; Cu = Santa Cruz; Da = Darwin; angles,G. difficilis;ß our own measurements;A from Es = Espanola;Fe = Fernandina;FI = Floreana(Santa Lack (1945); z• from Grant et al. (1985); circles, G. Maria); Ge = Genovesa; He = Los Hermanos; Ma = fuliginosa;ß our own measurements;O from Lack Marchena; Ni = northern part of Isabela; Pt = Pinta; (1945); O from Grant et al. (1985); note that both Lack Pz = Pinz6n; Sf = SantaF•; Sg = Santiago(San Sal- and Grant et al. placed the Genovesapopulation in vador); Si = southern part of Isabela;Wo = Wolf. G. difficilis;• datasupplied by Bowman(pers. comm.). Other abbreviations (See 1F): "a" means data refer to Oversizedsymbols ß O (•) indicatethe Genovesa the whole of Isabela Island; "b," n = 2; "c," locality population;and n = 11-80 per population, exceptas of Bahia Academia, south shore of Santa Cruz Island; noted below. Islands are arranged from northwest to "d," locality at Bahia Borrero,north shore of Santa southeast. Island names are as follows: Ba = Baltra; Cruz Island "e," n = 4; "f," n = 2. 146 ShortCommunications [Auk,Vol. 106

1.0

& &

0.5.

o

0 o.o o d 0 110 100 Relative abundance Fig. 3. The proportionin the diet of Geospizadif- ficilisand G.fuliginosa of invertebrates(arthropods and gastropods),plotted against their relativeabundance, on Pinta, Marchena, and Genovesa islands. Redrawn from Schluterand Grant (1982:fig. 5), exceptthat the Genovesasymbols have been changed. Schluter and Fig. 2. Geographicvariation in characterindex in Grant placedthe Genovesaspecimens in G. difficilis, Geospizafuliginosa and G. difficilis.Symbols: ß G. fuli- and marked them with empty triangles. Symbols:ß ginosa;[] G. difficilis.Character indices computedas G. difficilis,Pinta; ß G. fuliginosa,Pinta; O G. fuliginosa, explained in text; n = number of specimens;numbers Marchena;O Geospizaof Genovesa.Points in the dia- next to histogramsindicate sample size. The Santa gram representstudy sites;the number of specimens Cruz populationof G. difficilisis "no doubt"extinct involved was not clearly stated. (Bowman pers. comm.). Abbreviations as in Fig. 1. have come in contact" (1947: 26). Abbott et al. (1977) alsoconsidered this distributionalpattern as evidence The Los Hermanospopulation, currently assigned for interspecificcompetition. However, Schluterand to G. fuliginosa,has large beakdimensions, body mass Grant (1982, 1984)reported that the two speciesover- (Fig. 1A, lB, 1C, 1F), and characterindex values (Fig. lapped altitudinally on Pinta, Santiago,and Fernan- 2), as well as great variability. In our opinion, this dina islands without showing interspecificaggres- population is a between G. fuliginosaand G. sion.They ruled out presentcompetition between them. fortis(Vagvolgyi and Vagvolgyi in press). They observedthat, when sympatric,the diet of G. We comparedthe Genovesafinches to G. fuliginosa fuliginosaand G. difficiliswas dissimilar,e.g. on Pinta. and G. difficilisin characterindex (Fig. 2), computed When allopatric,however, the diet of the two species from five characters;body masswas omitted for lack wasquite similar, e.g. the dietof the allegedG. difficilis of sufficientdetails. The following formula was used: of Genovesaresembled that of G. fuliginosaof Mar- sum of scores in 5 characters Character Index = chena (Fig. 3). Schluter and Grant concluded that in 2 the absenceof competitionthe Genovesabirds shifted Each character was scored on a scale from 0 to 20. The toward the diet of the absentcompetitor G. fuliginosa, fuliginosacharacter states (the small dimensions)re- andthey inferredfrom this pattern that the two species ceivedlow scoresand the difficilischaracter states (the had competedin the past.Reviewing the case,Grant large dimensions)received high scores.Character in- (1987:307) assertedthat the fuliginosa-likediet of the dexvalues for specimensof G.fuliginosa ranged from Genovesafinches was attributableto the "... greater 4 to 31 (not counting the values of Los Hermanos profit obtainedfrom the fuliginosafoods, and the ab- specimens);for thoseof G. difficilis,18-44. The char- senceof a populationof G. fuliginosa"there. He also acterindices of the disputedGenovesa specimens var- suggested(1987: 307) that "... metabolicefficiency as ied from 11 to 22 and fell in the rangeof G. fuliginosa. well as perching ability were probably major select- Lack(1947) believed that G.fuliginosa and G. difficilis ing factorsin the of small size" in the Ge- always occurredon separateislands, or in separate novesa population. He did not elaborate on these altitudinal zoneson islandswhere they coexisted.He points. inferred from this pattern that G. difficilis"... hasbeen Central to the competitionhypothesis is the pattern eliminatedby G. fuliginosawherever the two species of similarities and dissimilarities in the diet of the January1989] ShortCommunications 147 four populations.This in turn restsupon their correct I. Bowman, Raymond A. Paynter, Mark Robbins, identification.There has been general agreementon FrancoisVuilleumier, and RichardL. Zusi for making the identification of the Pinta and Marchena popu- the collection under their care available to us. lationsof G. fuliginosa,and the Pinta populationof G. difficilis;the identity of the Genovesapopulation, on LITERATURE CITED the other hand, is contested.The competitionhy- pothesisassumed that the Genovesafinches belonged ABBOTT,I. J., L. K. ABsOTT,& P. R. GRA•T. 1977. Com- to G. difficilis.However, aswe have attemptedto show, parative ecologyof Gal•pagosground finches thesefinches are morphologicallyso similar to G. fu- (GeospizaGould). Ecol. Monogr. 47: 151-184. liginosathat we feel they should be assignedto that BALDWIN, S. P., H. C. OBERHOLSER,& L. G. WORLEY. speciesinstead. Such rearrangement would funda- 1931. Measurements of . Sci. Publ. Vol. II. mentally changethe dietary pattern, becausethe sim- Cleveland Mus. Nat. Hist. ilarity in diet of the allopatric Marchena and Ge- BOWMAN,R. I. 1961. Morphological differentiation novesapopulations, if both belongedto G. fuliginosa, and adaptationin the Galfipagosfinches. Univ. would clearly be attributable to their conspecificity. Calif. Publ. Zool. 58: 1-302. Our conclusion is consistentwith the results pro- 1983. The evolution of song in Darwin's vided by the comparisonof the population of Ge- finches. Pp. 237-537 in Patterns of evolution in novesa Island with those of Darwin and Wolf islands. Galfipagosorganisms (R. I. Bowman, M. Berson, Accordingto the competitionhypothesis, Genovesa and A. E. Leviton, Eds.). San Francisco, Pacific supportsG. difficiliswhich, in the absenceof the com- Div. Am. Assoc. Advmt. Sci. peting speciesG. fuliginosa,underwent profound FORD, H. A., A. W. EWING, & D. T. PARKIN. 1974. changesin its diet and morphology to become de- Blood proteins in Darwin's finches. Comp. Blo- ceptively similar to the absentcompetitor. If so,com- chem.Physiol. 47 B: 369-375. parable evolutionary events should occur on com- GOULD,J. 1837. Descriptionofnewspeciesoffinches parable islands. This was not the case, however. collectedby Darwin in the Galfipagos.Proc. Zool. Although G. difficilisis presenton Darwin and Wolf Soc. London 5: 4-7. islands,and G. fuliginosais absent,G. difficilishas not GP,ANT, P. R. 1987. Ecologyand evolution of Dar- assumedany of the morphological attributes of G. win's finches. Princeton, Princeton Univ. Press. fuliginosaon these islands.In fact, these populations --, I. ABBOTT,D. SCh'ccrrER,R. L. CURRY, & L. K. areclearly dissimilar in morphologyfrom G. fuliginosa A13BOTT.1985. Variation in the size and shape (Figs. 1, 2). Unfortunately, the diet of the Darwin and of Darwin's finches. Biol. J. Linn. Soc. 25: 1-39. Wolf populationsis insufficientlyknown for detailed HARRIS,M.P. 1973. The Galfipagosavifauna. Con- comparisons. dor 75: 265-278. In essence,our hypothesisholds that the fuliginosa- Jo, N. 1983. KaryotypicanalysisofDarwin'sfinches. like diet and morphologyof the Genovesapopulation Pp. 201-217 in Patternsof evolutionin Galfipagos are not attributableto dietaryadvantages and absence organisms(R. I. Bowman, M. Berson,and A. E. of competition, rather to the ancestryof this popu- Leviton, Eds.). San Francisco, Pacific Div. Am. lation from G. fuliginosastock. The competitionhy- Assoc. Advmt. Sci. pothesis,on the other hand, holds that the fuliginosa- LACK,D. 1945. The Galfipagosfinches: a study in like diet and morphologyof the Genovesapopulation variation. Occas.Pap. California Acad. $ci. 21: 1- are attributableto dietary advantagesand releasefrom 159. competition;the Genovesabirds appearlike G. fuli- 1947. Darwin's finches. Cambridge, Cam- ginosa,but evolvedfrom G. difficilis.The first hypoth- bridge Univ. Press. esis is supportedby two independent data sets:(1) 1969. Subspeciesand sympatryin Darwin's morphologicalshifts did not occur in G. difficilisof finches. Evolution 23: 252-263. Darwin and Wolf islands,as expectedfrom the com- PATTON,J. L. 1984. Genetical processesin the Ga- petition hypothesis,and (2) the song of Genovesa lfipagos.Biol. J. Linn. Soc.21: 97-111. birds resemblesG. fuliginosamore than it doesG. dif- PAYNTER,R. A., JR.,& R. W. STORER.1970. Check-list ficilis.As far as we know, the competitionhypothesis of the birds of the world, vol. 13. Cambridge, has no independent data supporting it. Studies on Massachusetts, Harvard Univ. mate preference (Ratcliffe and Grant 1983) were in• POI,AlqS,N.O. 1983. Enzymepolymorphism in Ga- conclusive. Cytogenetic (Jo 1983) and biochemical lfipagosfinches. Pp. 219-236 in Patternsof evo- studies(Ford et al. 1974,Yang and Patton 1981, Polans lution in Galfipagosorganisms (R. I. Bowman,M. 1983,Patton 1984)yielded no informationthat could Berson, and A. E. Leviton, Eds.). San Francisco, be used for or against either of the alternative hy- Pacific Div. Am. Assoc. Advmt. Sci. potheses. RATCLIFFE,L. M., & P. R. GRANT. 1983. Speciesrec- We sincerelythank Walter J. Bock,Robert I. Bow- ognition in Darwin's finches (Geospiza,Gould). man, Roger T. MacFadden, and David W. Winkler for II. Geographicvariation in matepreference. Anim. readingthe manuscript,and StephenF. Bailey,Robert Behav. 31: 1154-1165. 148 ShortCommunications [Auk,Vol. 106

RIDGWAY,R. 1894. Descriptionsof twenty-two new or perching birds, in the collectionof the British speciesof birdsfrom the Gal&pagosIslands. Proc. Museum. Fringilliformes: Part III. Cat. Birds Brit- U.S. Natl. Mus. 17: 357-370. ish Mus. XII. 1897. Birdsof the Gal•pagosArchipelago. SNODGRASS,R. E., & E. HELLER.1904. Papers from Proc. U.S. Natl. Mus. 19: 459-670. the Hopkins-StanfordGalfipagos Expedition, ROTHSCHILD, W., & E. HARTERT. 1899. A review of 1898-1899. XVI. Birds. Proc. Washington Acad. the ornithologyof the GalfipagosIslands. Novit. Sci. 5: 231-372. Zool. 6: 85-205. SWARTH,H.S. 1931. The avifaunaof the Galfipagos SCHLUTER,D. 1984. Morphologicaland phylogenetic Islands. Occas.Pap. California Acad. Sci. 18: 1- relationsamong the Darwin's finches.Evolution 299. 38: 921-930. VAGVOLGYI,J., & M. W. VAGVOLGYI.In press. Hy- ß & P. R. GRANT. 1982. The distribution of bridization and evolution in Darwin's finches of Geospizadifficilis in relation to G. fuliginosain the the Gal•pagosIslands. Accad. Naz. Lincei. Atti GalfipagosIslands: tests of three hypotheses.Evo- dei Convegni Lincei. lution 36: 1213-1226. YANG,S. Y., & J. L. PATTON.1981. Genic variability ß& . 1984. Ecologicalcorrelates of mor- and differentiationin Gal•pagosfinches. Auk 98: phologicalevolution in a Darwin'sfinchß Geospiza 230-242. difficilis.Evolution 38: 856-869. SHARPE,g.B. lSSS. Catalogueof the Passeriformes, Received15 January1988, accepted 30 August1988.

Homing Experiment with Leach'sStorm-Petrels

ELIZABETH GARY PIERSON,• CHARLES E. HUNTINGTON, 2 AND NATHANIEL T. WHEELWRIGHT Departmentof Biology,Bowdoin College, Brunswick, Maine 04011 USA

Leach's Storm-Petrels (Oceanodromaleucorhoa) show Kent Island were selected on the basis of nest acces- high rates of return to their nest sites after being sibility and previous breeding experience.The birds experimentallyreleased at variousdistances from the includedmales and femalesthat averaged10.9 yr old breedingcolony (Griffin 1940,Billings 1968).Homing (SD 5.2 yr) and ranged in age from a minimum of 4 at speedsof up to 350 kin/day and navigating dis- yr to at least 22 yr. We estimatedage by adding the tancesof up to 4,800 km acrossunfamiliar territory, number of years since the birds were first banded as manybirds actually gain massalong the way (Billings breedersto the 4 yr needed to achieve reproductive 1968). In somecases storm-petrels, which are almost maturity (Huntington and Burtt 1970). Males and fe- never sightedover land, apparentlycross land to avoid malesdid not differ in age(n = 20 and 12,respectively; muchlonger all-water routes (Billings 1968). Inspired Kruskal-Wallis Test: P = 0.56). None had been used by Griffin'sresearch at the BowdoinScientific Station, in Billings' experiments.For severaldays before the Kent Island, New Brunswick (44ø35'N, 66ø45'W),Bil- experiment,all nestswere checkeddaily to determine lings testedthe hypothesisof overlandnavigation by when each bird had arrived to begin its incubation transporting15 storm-petrelsto the coastaltown of shift. Storm-petrelshave incubationshifts that last up Stephenville, Newfoundland (48ø33'N, 58ø36'W), to 5 days (Gross1935, C. Huntington unpubl. data) which is separatedfrom Kent Island by two major and attend their nests erratically (Boerstoa and land barriers, Prince Edward Island and Nova Scotia. Wheelwright 1979), so it was difficult to find large Billings reported return speedsnearly double those numbersof birds at identical stagesin their incuba- of Griffin in several different homing experiments. tion shifts.Consequently, we usedbirds that had spent Given the discrepancybetween their resultsand the varying periodsof time on the nestat the startof the fact that experimentsin ecologyare too rarely re- experiment. peatedand independentlycorroborated by different The experimental procedure was similar to that of investigators,we report the resultsof a replicationof Billings (1968). Birds were removed from their nest Billings' Stephenvillehoming experiment. burrowsbeginning at 2330,3 July 1974.Each bird was Thirty-two incubatingLeach's Storm-Petrels from weighed with Pesolaspring scales,placed in a cloth bag, and put into a cardboardbox. At 0500 the fol- lowing day, the birds were transportedby boat to 1Present address: RR 2, Box 248, South Harpswell, Grand Manan Island, a distance of 9 kin. Two hours Maine 04079 USA. later the birds were flown to Stephenville, with a brief 2Author to whom reprint requestsshould be ad- stop in St. John, New Brunswick,to changeplanes. dressed. The birds were not fed in captivity. On the afternoon