242 GP•ANTET AL. [Auk, Vol. 99

--, A. OLSZOWKA, & A. AR. 1974. The avian WHITTOW.1976. Regulationof incubation water egg: surfaceareaß volume and density. Condor lossin sevenspecies of terns. Physiol.Zool. 49: 76: 319-325. 245-259. PETTIT, T. N., G. S. GRANT, G. C. WHITTOW, H. RICKr•FS, R. A., & H. RAHN. 1979. The incubation RAHN,• C. V. PAGANELLI.In press.Respiratory temperature of Leach's Storm-Petrel. Auk 96: gas exchangeand growth of Bonin Petrel em- 625-627. bryos. Physiol.Zool. ROB•I•TS,B. 1940. The life cycle of Wilson's Petrel RAHN,H., • A. AR. 1974. The avian egg:incubation Oceanitesoceanicus (Kuhl). Scient. Rep. Br. Gra- time and water loss. Condor 76: 147-152. ham Ld Exped. 1934-37, 1: 141-194. , & --. 1980. Gas exchangeof the avian ROUDYBUSH,T., L. HOFFMAN, & H. RAHN. 1980. egg: time, structure, and function. Amer. Zool. Conductance,pore geometry and water loss of 20: 477-484. eggsof Cassin'sAuklet. Condor82' 105-106. , & W. R. DAWSON. 1979. Incubation water TUrL•TT, S. G., & R. G. BoAar>. 1977. Determinants loss in eggsof Heermann'sand Western Gulls. of avian eggshellporosity. J. Zool., London 183: Physiol. Zool. 52: 451-460. 203-211. ß C. V. PAGANErrS, & A. AR. 1975. Relation VL•CK, C. M., & G. J. KœNAG¾.1980. Embryonic of avian egg weight to body weight. Auk 92: metabolism of the Fork-tailed Storm-Petrel: 750-765. physiologicalpatterns during prolongedand in- , R. A. ACKERMAN,& C. V. PAGANELLI. 1977. terruptedincubation. Physiol. Zool. 53: 32-42. Humidity in the avian nest and egg water loss WHITTOW,G. C. 1980. Physiologicaland ecological during incubation. Physiol. Zool. 50: 269-283. correlatesof prolongedincubation in seabirds. , C. V. PAGANELLI, I. C. T. NISBET, & G. C. Amer. Zool. 20: 427-436.

The sixthannual meeting of the ColonialWaterbird Group will be held 4-7 November1982 in Washington, D.C. A symposiumon the feedingbiology of waterbirdsis planned.Papers given at the meetingare eligible, after refereeing,for publicationin ColonialWaterbirds. Anyone wishing to contributeto either the sym- posium (deadline 1 September)or general session(deadline 15 September)should contactDr. Michael Erwin, U.S. Fish and Wildlife Service,Patuxent Wildlife ResearchCenter, Laurel, Maryland 20708.In- formationconcerning registration can alsobe obtained from Dr. Erwin. THE ROLE OF MIGRATION AND WINTER MORTALITY IN THE LIFE HISTORY OF A TEMPERATE-ZONE MIGRANT, THE DARK-EYED JUNCO, AS DETERMINED FROM DEMOGRAPHIC ANALYSES OF WINTER POPULATIONS

Errv.• D. KETTERSONAND VAL NOLA• JR. Departmentof Biology,Indiana University, Bloomington, Indiana 47405 USA

ABSTR•CT.--Thisstudy of migratoryDark-eyed Juncos (Junco h. hyemalis)investigates dif- ferencesin winter populationstructure and dynamicsassociated with north-southvariation in wintering site and considersthe relationshipsbetween these winter differencesand population dynamics at other seasons.Banded juncos were captured and recapturedin several winters at four latitudes (42ø-33.5ø),and results were integrated with long-term United StatesFish and Wildlife Service recordsof recoveriesthroughout the winter range. The data suggest: (1) Populationswintering in the north experience higher winter mortality than populationswintering in the south. (2) Sex and age do not affectthe prob- ability of overwintersurvival; therefore, dominance may not affectthat probability.(3) Some individuals do and others do not return to the samewintering site year after year. Individ- uals that do not return tend to be those that have spent the first winter of life in the north, and theseshift southward.(4) Annual mortality of northernand southernwinter populations probablyis the same. This conclusionis based in part on a comparisonof north-south annual return rates and on logic arising out of the geographicaldistribution of the sexes during winter. In juncos, becausemales tend to winter north of females,greater annual mortalityof northernwinter populationswould be expectedto skewthe primary and sec- ondarysex ratiosin favor of males.No evidenceof suchskews has been detected.(5) If annual mortality is the same,it is likely that southernpopulations suffer heavier mortality duringtheir longermigrations and that thesemigration losses offset the advantagedescribed in (1), above. In north temperate-zonebreeding species that migratesouthward into an extensivewinter range, distancetraveled probably is correlatedclosely and positively with both mortality rateduring migrationand survivalrate duringwinter. If theserates offset each other, annual survivorshipamong winter migratory populations(and among sedentaryand migratory populationsof partially migrant species)can be equal. Therefore, it is unnecessaryto pos- tulate for different winter populationseither unequal reproductivesuccess (von Haartman's hypothesis)or unequalyear-to-year survival rates that balanceout overthe longterm (Lack's hypothesis). Migration can causebreeding populations to be redistributedin variousways in winter; e.g. they may mix randomly, or northern breedersmay leapfrogsouthern breeders. Each variationin migrationpattern would producewinter populationswith predictablecharac- teristics.We considerthe extent to which winter junco populationshave these character- istics.Received 26 January1981, accepted30 September1981.

THE Dark-eyed Junco(Junco h. hyemalis),an ers, by migrating farther, move into progres- abundant ground-feedingemberizid, migrates sively more moderateand even mild climates. between a breeding range lying largely in the Male juncostend to winter north of females boreal forests of Canada and, in eastern North (Ketterson and Nolan 1976, 1979). Assuming America, a winter range reachingfrom south- an equal sex ratio on the breedinggrounds-- ern Ontario and the northern boundary of the and both field study and theoreticalconsider- United States almost to the Gulf of Mexico ations of sex ratio and mating systemssupport (Bent 1968). Becausethe winter range is so ex- this assumption(see below)--the winter sexual tensive, northern-winteringjuncos spend that separationnecessarily implies that members of critical season(Lack 1954, Fretwell 1972) in cold southern-winteringpopulations have migrated and often snow-coveredregions, whereas oth- a greateraverage distance than membersof

243 The Auk 99: 243-259. April 1982 244 KETTERSOtaAtaD NOLAta [Auk, Vol. 99 northem populations. Further, in both sexes In this paper we present data on minimum individuals produced in the preceding breed- overwinter survivorshipand minimum annual ing seasonand adults exhibit differences,on survivorship in northern- and southem-win- the average, in winter distribution. The pat- tering populations that we sampled by cap- tern of age distribution is stable from year to ture-recapturetechniques at the beginning and year and is complex (Kettersonand Nolan in end of single winters and also in successive prep.), but its most interestingfeature is that early winters. Interpretation of such data re- the young of eachsex tend to concentratenorth quires that death and dispersalbe separated of their adult counterparts.As with the sexual out as factorsaffecting the recapturedata. We separation, this pattern implies differential thereforeanalyze United StatesFish and Wild- migratory behavior associatedwith age, with life Service(USFWS) Bird BandingLaboratory adults traveling farther. When age classesof recordsbearing on the questionof north-south bird speciesseparate in the winter range, it is variation in the winter dispersaltendency of usual for adults to make the shortermigration juncos in the eastern United States. We also (Gauthreaux1978), although the pattern we examinethese USFWS records for time elapsed find in the junco has been observed in the betweenbanding and recovery,which we take American Goldfinch (Carduelistristis) (A. Mid- as an estimateof survivorship,and compare dleton pers. comm.) and the White-crowned them with a long-term estimateof survival of Sparrow(Zonotrichia leucophrys gambelii) (King an Indiana sample. et al. 1965). Our results, like the USFWS data that we Predicting that juncos that winter in the consider, are limited to easternwinter popu- north would experience greater overwinter lations. The breeding sitesof thesebirds were mortality than those that migrate into the unknown, and all references to northern and south, we also expectedthat annual survivor- southernjuncos are to winter populations,ex- ship of southernwinter populationswould be cept where it is explicitly statedotherwise. higher. A corollaryof this latter prediction is that members of such southern populations METHODS should have lower reproductive success,as- Capture-recapturedata.--Juncos end their autum- suming that fitnessof juncosdoes not vary ac- nal migration about 1 Decemberand begin spring cordingto the latitude of the wintering site. As migration about 1 March (evidence for this is re- we here report, the expectationabout unequal viewed in Ketterson and Nolan 1976), with indica- overwinter mortality appearsto be correct,but tions that few move very far in the interim (see be- the evidence does not support the prediction low). Therefore, individuals found from December that wintering in the southconfers greater an- through February we consider to be winter resi- nual survivorship.This implies that southern dents.We capturedjuncos at a seriesof sitesin early winterers have higher mortality than northern winter (December, except that an occasionaleffort winterers at some other stage of the annual extended into the first few days of January), then cycle, and the most probableoccasion for such made recaptureefforts in late winter (February,ex- cept that one effort extendedinto early March) and/ compensatorymortality is during their longer or in the subsequentDecember; all unmarked jun- migrations. cos, whenever caught, were given USFWS bands. Recent evolutionary speculations have Early-winter samplingwas at Kalamazoo,Michigan stressedhow little is known of the impact of (42øN); Bloomington, Indiana (39øN); Clemson, migration on survival and reproduction and South Carolina (34.5øN);and Birmingham,Alabama have urged that more attention be devoted to (33.5øN) in the winters of 1976-1977, 1977-1978, acquiringdata that will further understanding 1978-1979,and, exceptfor Alabama,1979-1980 (Fig. of thesesubjects (Baker 1978, Gauthreaux 1979, 1). Late-winter sampling was restrictedto the In- Krebs 1979, Dingle 1980). For such inquiries diana and South Carolina sites in 1977-1978 and 1978--1979. the Dark-eyedJunco provides an exceptionally good system.Its extensivewinter distribution Long-term mean temperaturesfor the period De- cember-February and annual snowfall at the four and sex-agedifferences in distancemigrated capture-recapturesites follow: Michigan, -4.0øC, permit comparisonof seasonaland annual sur- 139.9 cm; Indiana, -0.6øC, 59.9 cm; South Carolina, vivorship amongconspecific populations that 6.5øC, 8.4 cm; Alabama, 7.8øC,4.6 cm (U.S. Weather differ greatlyin the way they balancethe risks Bureau 1932). of migration againstthe rigors of winter. Each capture effort was limited to a 3- to 6-day April 1982] Migrationand Winter Mortality in Juncos 245

consideredin calculatingrecapture rates. A few were marked in one effort, not caught in the next, then recapturedin still the next. Thesewe treated as hav- ing been presentduring the interveningeffort, even thoughwe had not caughtthem, but, having count- ed them as presenton that one occasion,we elimi- nated them from subsequentrecapture calculations. Samples collectedunder the time constraintsde- scribedare smallerthan ideal for statisticalpurposes, and for some comparisonswe have been compelled .IN to pool acrossyears. We have alsocombined the data from Michigan and Indiana, which we regard as northern wintering locations, and from South Car- olina and Alabama, which we regard as southern. .SC We tested for statisticalhomogeneity before such .AL combining;despite nonhomogeneity in onecase, we combined, becauseto do so produced a conservative result that did not affect the argument. Other Indiana data.--We draw also on the results of two independent studiescarried out near Bloom- ington at locationsother than thosereferred to above and involving other juncos. (1) In December1978 and 1979, as the first part of an investigation(in prep.) of dispersalwithin singlewinters, we caught,band- Fig. 1. Sitesat which juncoswere captured. ed, and releasedjuncos at eight stations,each 1.5-2 km apart, along a transect. In February 1979 and 1980, we again trapped and netted at these stations. period of intensivefield work that had been preced- Sexingand aging were as describedabove. We pres- ed by severaldays of baiting at the spots(hereafter ent these data here in order to show the sex-age called subsites)at which nets and traps were set up. structure of the Bloomington population at the be- One to four subsiteswere sampled at each site. At ginning and end of two winters and to draw conclu- the end of all captureefforts, baiting stopped,except sions about the relative overwinter mortality of the in Michigan, at one subsitein South Carolina, and different sex-ageclasses there. It is important to note at one subsitein Alabama. We made every attempt that the structureat all eight stationswas about the to standardize capture methods from site to site, same;we detectedno sexor agesegregation in early with the exceptionthat weather differencesaffected or late winter. Recapture rates are not shown, al- the ease of capture and therefore causedthe length though they were very much like the Indiana rates of captureefforts to vary. For example,snow usually presentedin Table 1, becauseour methodswere dif- coveredthe ground in Michigan and did so in late ferent from those described above. (2) For a more winter in Indiana and once even in South Carolina; accuratemeasure of the annual survivorshipof jun- this led juncosto concentrateat our bait and let us cos that winter in Indiana, we selected 128 site-faith- abbreviateour operations.At all siteson the last day ful individuals caughtbetween 1961and 1977at two of an effort, the proportionof recapturedindividuals sitesat which trapping and netting were carried out that had been caughtand banded during that effort nearlydaily from Octoberinto May (theapproximate was approximatelythe same. limits of the junco'soccurrence in Indiana). Male and We sexedjuncos by examinationof plumagetraits femalenumbers in this samplewere representative and wing length (Kettersonand Nolan 1976), per- of the proportionsin the Bloomingtonwinter pop- forming laparotomieswhen in doubt. The degreeof ulation; the sexesdid not differ significantlyin re- skull ossificationin early winter revealedwhich in- tums and accordinglywere pooled. All 128 were dividuals had hatched in the preceding breeding present in at least two winters and three were pres- season (herein called young) or in some earlier ent in six; not every bird was caughtin an uninter- breedingseason (herein adult). In latewinter, juncos ruptedseries of winters.Ignoring calendar years, we caughtfor the first time were aged on the basis of formed a composite in which each individual's plumage, wing length, and eye color (Ketterson bandingyear was the baseline and subsequentyears 1979b). This method has proved 93% accuratein were +1, +2, etc. As is typical in avian survivorship casesin which we tested late-winter age determi- studies (Nolan 1978:466 and citations therein), the nationsof previouslybanded individuals against our proportionof any year's birds that were presentthe original age determinationsbased on skull exami- followingyear did not changeas the birds aged.We nation. Only birds releasedin goodcondition were applied Caughley's(1977: 151) method to estimate 246 KœxxœgsotqAND NOLAN [Auk,Vol. 99

TABLœ1. Rateof recapturein latewinter of juncosmarked in earlywinter accordingto sex,age, year, and location. a,b,e

Male Female Classes Adult Young Adult Young combined Indiana 1977-1978 6/27 2/36 0/4 0/10 8/80 1978-1979 4/41 6/42 1/9 3/19 16/118 Years combined 15% 10% 8% 10% 12% (68) (78) (13) (29) (198) South Carolina 1977-1978 8/31 6/15 11/22 5/13 30/81 1978-1979 7/15 3/7 12/21 8/12 30/55 Years combined 33% 41% 53% 52% 44% (46) (22) (43) (25) (136) a samplesizes are in parenthesesand represent the number of juncosmarked in earlywinter and released in goodphysical condition at a locationthat was sampled in latewinter. The size of thecombined sample is sometimesgreater than the sum of samples ofthe respective sex- ageclasses, because a few individualsmarked were of unknownsex or age. bLate-winter recapture rates of thesex-age classes at a locationwere compared after combining data for the two different years. In Indiana, thesex-age classes were equally likely to be recaptured IX2 = 0.94,df = 2, n.s.(female age classes combined because samples small)]. The same was true of South Carolina 0(2 = 4.72, df = 3, n.s.). c Late-winterrecapture rates of a sex-ageclass at thedifferent locations were compared after combining data for the2 yr. Eachsex-age class wassignificantly more likely to berecaptured in SouthCarolina than in Indiana(adult males: adj. X • = 4.15,df = 1, P < 0.05iyoung males: adj. X2= 9.36,df = 1, P < 0.01;adult females: adj. X2 = 6.77,df = 1, P < 0.01;young females: adj. X 2 = 9.26,df = 1, P < 0.01).

the annual mortality rate of the group. Use of site- cations were the same, we noted the interval in years faithful birds was necessary,of course;we recognize between the two events. the possibilitythat nonfaithfulindividuals may have Becausejuncos cannot be sexedreliably without a different mortality rate. considerableexperience and the methodof agingby USFWSdata.--The Bird BandingLaboratory stores skull ossificationis both relativelyrecent and not information on the date and location of initial band- used by all banders,we could not analyzeUSFWS ing and recoveryas well as on the methodof recov- data for possible variation associatedwith sex and ery, but it recordslocations only accordingto their age. We recognizethat the USFWS data set is small, 10-min blocks of latitude and longitude (roughly with few cases from the south. Further, the areas speaking100 km• in the easternUnited States).Re- encompassedby 10-min blocks of latitude and lon- coveriesmade by bird bandersare identified as such. gitude are much largerthan the areasof our capture The fact that bandersare concentratedin more pop- sites. This presentsa difficulty, describedin detail ulous areas, especiallyin the northeasternUnited below, in combining USFWS recoverydata with our States, would bias the estimates made herein if re- recapture data. Of course, the size of the 10-rain coveries by banders were included in our analysis, blockshas no effecton the validity of comparisons and we therefore consideredonly juncos recovered of northern and southern juncos when these com- dead by the generalpublic (i.e. we omitted cases parisonsare restrictedto the USFWS data. with the Laboratory'srecovery code numbers 10, 16, 29, 51-53, 87, 89, 99). From the resulting records for RESULTS the years1921 through August 1976,we selectedthe 368 cases of individuals that were both banded and Recapturein winter of banding.--Minimum recovered in winter (December-February) east of overwinter survivorship, as estimated by the 106øWlongitude. Each case we categorizedaccording percentageof juncos that were recaptured at to (1) latitude of initial capture location, (2) whether the end of winter, differed geographically.A recoverywas during the winter of bandingor during muchlarger proportion of SouthCarolina birds a subsequentwinter, and (3) whether recoverywas than of Indiana birds (44% vs. 12%) was re- at or away from the initial capturelocation. We then examined the results for latitudinal trends in ten- captured in late winter (Table 1). A February snow storm, an unusual event in South Caro- dency to dispersefrom the banding location, either in the winter of initial capture or a later winter. lina, may have been partly responsiblefor the When banding and recoverywere in different win- high recapturerate there. Even in the winter ters and at different locations, we noted the direc- in which Februaryweather in South Carolina tional relationship between the locations. If the lo- was normal (1977-1978), however, the recap- April1982] Migrationand Winter Mortality inJuncos 247

TAB• 2. Sex-agea structure (percentages of individ- TABL•3. Percentagesof juncosrecorded by United uals)of early- and late-wintersamples of juncosin StatesFish and Wildlife Service (1921-1975)as re- Indiana and South Carolina. coveredin winter at the place of original capture, accordingto latitude of place of capture:an esti- Male Female mate of the relationshipbetween dispersal and lat- itude. a,b n AdultYoungAdultYoung Recovered in Indiana a.• subsequent Recoveredin December 1978 395 28 42 10 21 winter same winter February 1979 400 '24 45 9 22 December 1979 346 17 52 10 21 Latitude Percent- Percent- February 1980 320 22 44 13 22 of place age at age at of Total place of Total place of South Carolina a'c capture cases capture cases capture December 1977 87 37 17 28 18 February 1978 170 23 15 34 28 45-44 2 100 3 100 43-42 25 52 73 95 December 1978 132 26 17 33 24 41-40 45 53 75 91 February 1979 104 34 10 41 15 39-38 42 74 60 90 a Indiana sampleswere caughtin a studyof dispersalat eight sites; 37-36 6 83 6 83 individuals in them differ from those in all other tables. South Carolina 35-34 8 88 9 77 juncoswere caughtin the overwintersurvival study. See text. b Indiana seasonalcomparisons of sex-ageratios, Decembervs. 33-32 3 100 11 100 February 1978-1979:X •= 1.90, df= 3, n.s.; December-February 131 237 1979-1980: X2 = 5.10, df = 3, n.s. e SouthCarolina comparisons of sex-ageratios, December vs. Feb- a Data were takenfrom fliesof the BandingLaboratory of the United ruary1977-1978: )? = 7.96,df = 3, P • 0.05;December vs. February StatesFish and Wildlife Serviceand restrictedto juncosbanded in the 1978-1979,)? = 5.95, df = 3, n.s. monthsDecember-February and recoveredin those months in the sameor in any laterwinter. Banding and recoverysites were located betweenlongitude 70øW and 105øW. The Laboratory records locations only as within 10-minblocks of latitudeand longitude;thus, recovery at theplace of capturemeans within the same block. To avoidpossible bias associatedwith the northward concentrationof bird banding sta- ture rate was significantlyhigher than the In- tionsin theUnited States, the sampleis confinedto juncosrecovered diana rate, and all Indiana late-winter recap- other than by a bird bander.Note that mostinitial captureswere in ture efforts were conducted after heavier snow the northeasternUnited States(Fig. 2). b In correlatingthe percentageof recoveryat the placeof original storms than the one in South Carolina. capture(i.e. fidelityto a 10-minblock as definedin footnotea) and latitudeof placeof originalcapture, intervals of latituderepresented Comparingsex-age classes within eachsite, by fewer than five recoveriesare ignored. For recoveriesin a winter the late-winter recapturerates in Indiana were subsequentto thatof originalcapture, r, (Spearman'srank) between latitude and fidelity = 1.0, n = 5, P < 0.05. For recoveriesin same statisticallyindistinguishable for all sex-age winter as that of original capture,r, = 0.14, n = 6, n.s. classes,and the same was true in South Car- olina (Table 1). In accordancewith this evi- dence that at neither latitude did age or sex affect the probability of overwinter survival, USFWSdata: recovery in winterof banding.- data from our studyof winter dispersalin In- When analysisis limited to USFWS recoveries diana (Table 2) reveal no statisticalchange in during the winter of initial banding, no geo- sex-ageratios between early and late winter graphicvariation in overwinterdispersal is evi- and certainlysuggest no disadvantageassoci- dent. Only 18 juncosamong 237 that were thus atedwith beingimmature or female.When the recovered had moved to a different 10-min block, structureof December and February samples and suchdispersal was as frequent among south- caughtin SouthCarolina is compared(Table ernas among northern birds (Table 3). We inter- 2), we see that in one winter it remained the pret this belowas supportfor theview that the sameand in one it changed.In the latter,adult difference in overwinter recapture rates in In- maleswere lessfrequent in late February-early diana and South Carolina reflects, at least in March, and all femaleswere more frequent. If part, differential winter mortality. However, if this wasnot samplingerror, it mightmean that juncos that dispersefrom northern banding old maleshad begunto migrate(in which case sitestend to movevery shortdistances whereas our already high overwinter recapture rate those from southernsites move long distances, might not be high enough);but it is mostun- the USFWS method of recording only 10- likely that it meansold malessurvive winter min blocks would underestimate the relative in South Carolina less well than do females. numbers of northern dispersersand make the 248 KETTERSONAND NOLAN [Auk, Vol. 99

TABLE4. Rate of return from one Decemberto the next accordingto sex, age, and location.a'b,½

Males Females Classes Adult Young Adult Young pooled

North Michigan 19/141 4/165 1/43 2/64 26/413 Indiana 6/79 5/95 0/16 1/36 12/226 Combined 11% 3% 2% 3% 6% (220) (260) (59) (100) (639) South South Caro- lina 5/62 7/40 4/78 5/44 21/224 Alabama 1/9 1/6 4/22 3/14 9/51 Combined 8% 17% 8% 14% 11% (71) (46) (100) (58) (275) a The numbersof birdsrecaptured from among those released in goodcondition at the capturesite in the precedingwinter arepresented for each location.A few juncoswere recapturedonly after an interveningwinter, and thesewere consideredto have been presentin the first postbandingwinter. Juncoswere markedin four winters, and recaptureefforts were made in threeat Michigan, Indiana, and SouthCarolina. Only two recaptureefforts were made at Alabama.The percentageswere computed on the combinednorthern and combinedsouthern locations. b Comparisonsof return of eachsex-age class to northern(Michigan, Indiana) and southern(South Carolina, Alabama) locations follow: adult males,north vs. south,adj. X2 = 0.23,df = 1, n.s.;young males, north vs. south,adj. X2 = 11.28,df = 1, P • 0.001;young females, north vs. south,adj. X2 = 4.85, df = 1, P < 0.05. For all classesp•oled, rate of returnto the southwas significantlyhigher (adj. X2 = 7.07, df = 1, P < 0.01). c Frequencyof return of the sex-ageclasses differed at northern sites (X2 = 17.87, df = 3, P • 0.001) but not at southernsites (X2 = 3.80, dœ= 3, n.s.).

data uselessfor our purposes.At the present statistical comparison of northern female age time, we can say only that we have no reason classes(2% vs. 3%), because so few adult fe- to expect such a geographicaldifference. males winter in the north. Similarly, adult fe- Recapturesubsequent to winter of banding.- males cannot be compared statisticallywith The percentageof birds recapturedin a winter adult males, although the percentageof adult subsequentto that of banding, i.e. the return- male returnswas higher (11% vs. 2%). recapture or minimum annual survival rate, In contrast to the variation between northern showed both geographicalvariation and vari- age classesand also between northern sex ation accordingto sex-age class. When data classes,all sex-ageclasses in the south were pooledfrom Michigan and Indiana without re- recapturedwith statisticallyequal frequency. spect to sex or age and combined acrossall Turning to sex-age comparisons between years are comparedto similarly pooled and north and south, young northern juncos of combined data from South Carolina and Ala- each sex returned and were recaptured at the bama, return-recaptureat the northern sites banding site less often than young southern (6%) was significantlylower than at the south- juncos of the same sex (males: 3% vs. 17%, ern sites (11%) (Table 4). females: 3% vs. 14%). Adult male returns were When the data are broken down by sex and the same, north and south; again the small age, the frequencyof returnsat the combined northern sample of adult femalesprevents sta- northern sites differed for the four classes tistical comparison (males: 11% vs. 8%, fe- (P • 0.001), but this was attributable to vari- males: 2% vs. 8%). ation between the two most numerous classes: Because we believe the low return rate of adult males returned and were caught more northern adult females to be largely a product frequently than young males (11% vs. 3%, of samplingerror--recall that the Indiana sam- P < 0.001) (Table 4). This difference, in con- ple of site-faithful individuals showedno sex- junction with the preponderanceof males in ual bias--we summarize and interpret Table 4 the north, also causes the northern adult rate as follows. The difference in the pooled return- to exceedthe rate for northern young (9% vs. recapturerates between northern and southern 3%) significantly.Sample sizes do not permit locationsis apparentlylargely or entirely due April1982] Migrationand Winter Mortality in Juncos 249

Fig. 2. Initial capturelocation and recoverylocation of juncos shown by USFWSrecords to have been capturedand recoveredin different winters and at different places.For clarity, movementsof lessthan 30 rain are not plotted. See text for other details. Each line representsan individual junco. The dot is at the originalcapture location and the arrowpoint is at the recoverylocation. either to lower survivorship or lesser site fi- cases), 38-39 ø (31 returns in 42 cases), and delity among young northern juncosof both 32-37 ø (15 returns in 17 cases):X 2 = 9.16, df = sexes. These alternative possibilities are dis- 3, P •< 0.05. Both approachesindicate that site cussed below. fidelity was greater to more southerly loca- USFWSdata: recovery subsequent to winter of tions. In addition, a directional component is banding.--We considerthe extent to which the apparent in cases in which individuals chose juncos in the USFWS sample tended to select a new sites. Most tended to move southward, new location in the later winter. Of 131 individ- sometimesgreat distances,in later years (Fig. uals banded in one winter and recovered in 2); among28 casesof latitudinal shiftsequal to another, 46 had moved away from the 10-min or greaterthan 30 min, i.e. shortermovements block in which they had first been caught(Ta- and east-west movements excluded, 24 were ble 3). This considerableshift was largely con- southward. fined to birds first capturedat higher latitudes This interestingresult caused us to reanalyze (Table 3). The concentration of data from more the initial capture(banding) dates of the juncos northerly locations makes statistical compari- that shifted to new sites. Our question was son somewhat arbitrary. Table 3 employs a whether these individuals really had been Spearman'srank coefficient,but here we com- winter residentswhere first caught:if the cap- pare the subsets from latitudes 42-45ø (15 re- ture dates had fallen in early December, this tums in 27 cases), 40-41 ø (24 returns in 45 might supportthe view that these birds were 250 KœTTœRSO•A•D NO•,A• [Auk, Vol. 99 still migrating southward when caught, i.e. T•s•.œ5. Years elapsedbetween banding and recov- had not necessarilyresided in different places ery of juncosrecovered in somesubsequent winter at the banding location, accordingto latitude: an in the two winters. No such support was estimateof the relationshipbetween wintering lat- found, however.Only two banding dateswere itude and life expectancy(USFWS data). prior to 18 December (on 11 December of dif- ferentyears), and sevenand nine, respectively, Recovered more were in Januaryand February. than 1 yr USFWSdata: life expectancy.--Weexamined (winter) after the number of whole years elapsed between Total bandingb banding and recoveryof the 85 USFWSjuncos Latitudea casesa n Percentage that were taken in two different winters in the 450-42 ø 15 8 53 same 10-min block and that had therefore been 41ø-39 ø 44 26 59 exposedto the same climates,if not weather, 380-36 ø 16 7 44 350-32 ø 10 5 50 in both thosewinters (and probablyin any in- tervening winters as well). Treating years 85 46 elapsedas an estimateof life expectancyat the a Sampleconsists of juncosrecovered in same10-min block at least 1 yr (one winter) after winter of banding. Extremenorthern and ex- time of banding, we looked for variation ac- beemesouthern bands of latitude are 4ø wide; middle intervals are 3ø cordingto latitude. No variation is detectable wide in order to make samplesizes comparable. b Comparinglatitudes for number of juncosrecovered more than (Table 5). Approximatelyhalf the juncosthat 1 yr (winter)after banding and numberrecovered only 1 yr afterband- survived to be recovered in a winter subse- ing, X2 = 1.20,df = 3, P > 0.5; time elapsed(survived) between band- ing and recoveryis independentof latitude. quent to that of banding had lived at least2 yr after first capture,regardless of the latitude of the banding site. Translatingthe datain Table5 into a survival These results have interesting implications for rate and making the usual assumptionof con- theories about the importance of dominance, stantannual mortality with advancingage, the which until quite recently (Ketterson1979b, survival rate is 54%. Baker et al. 1981, Rohwer and Ewald 1981) Survival of site-faithful Indiana juncos.-- has been generally acceptedas affecting sur- Caughley'smethod estimatesthe annual sur- vival. vival rate of the juncos site-faithful to Bloo- The sex-ageclasses of juncosdiffer in dom- mington to be 53%, as follows: inance rank when winter flocks assemble at food (Balph 1977;Ketterson 1979a, b), and in c•=1- ml-•- m2 -•- ms -•- m4 + m5 rno + rn• + rn2 + rna + rn4 juncosdominance rank hasbeen related to sur- vival in nature (Fretwell 1969) and in the lab- 99 oratory(Baker and Fox1978). These facts would q = I 187 lead to the prediction that overwinter survi- vorship variesaccording to the rank associated q = 0.47 with each sex-ageclass, that is, in descending where• = finiterate of annualmortality and order from adult males to young males, adult rn. = number of marked juncos caught at females,and young females.Survivorship dif- time n. ferencesof this kind might havebeen expected to be especiallyconspicuous in the severewin- DISCUSSION ters of Indiana. Our observations to the con-

OVERWINTER SURVIVORSraP AND trary suggesteither (a) that dominanceplayed no role in the evolutionof geographicvariation DOMINANCE STATUS in winter sex-ageratios (seeMyers 1981), be- Indiana samples,which were large as well cause it does not influence overwinter survivor- as diverse in method and location of collection ship, or (b) that dominancedid play a roleand (TablesI and 2), were alike in indicating that hasproduced a patternof varyingratios in which sexand age did not influencethe probability subordinate classessuffer differentially only that an individual would survive the winter. when populationstructures or densitiesshift As discussedabove, we interpret the less nu- from thosethat obtainedduring this study. Avi- merousdata from South Carolinaas indicating ary experimentsin which sex-ageratios are var- that the same was true there (Tables 1 and 2). ied, food restricted,and behavior and survivor- April 1982] Migrationand Winter Mortality in Juncos 251 ship noted are underway in an effort to pursue survivorshiptoward equality, or (2) greaterre- this problem. productive success,or (3) a combination of these. Investigation of the nature of the putative DIFFERENTIAL OVERWINTER MORTALITY IN advantage accruing to northern winterers en- NORTH AND SOUTH counters, at the threshold, the fact that data on The lower overwinter recapture rate in In- comparative reproductive success are non- diana as compared to South Carolina, when existent and probably unobtainable. Getting considered with the USFWS data suggesting them would require that members of winter no geographicdifference in overwinter dis- populations be traced to their summer loca- persal, points to greater winter mortality tions in a vast breeding range. Good estimates among northern juncos. That juncos do die of long-term annual survival rates of northern during severeweather is a fact. Their emaciat- and southern winter populations can be de- ed carcassesare found after periods of pro- veloped, however, and comparison of these longed snow coverand low temperature(Rose- will give an idea whether, and to what extent, berry 1962, Johnston1962, Graber and Graber differential reproductivesuccess is operating 1979, pers. ohs.), which are frequent in the to equalize fitness. If long-term annual survi- north and rare or absent in the south. Further- vorship is found not to vary with the latitude more, our 2-yr study of dispersalin Indiana of the wintering site, the assumptionof sta- and over 20 yr of intensive winter banding bility in the existingwinter distributionwould there at several sites about 0.1-0.3 km apart exclude the possibility of unequal long-term have revealed little dispersal and none on a reproductivesuccess. There would remainonly scale that could produce the substantial dif- the alternativethat seasonalsurvivorship is un- ference in Indiana and South Carolina over- equal, not simply in winter but during other winter recapture rates. segmentsof the year as well. We conclude that very probably northern The data to be discussedbelow suggestto juncos of all sex-ageclasses die more frequent- us that the average annual survival rates of ly in the averagewinter than do southernjun- northern and southern winter populations are cos. Becauseour data were obtained during the same. We argue that southern winterers only two winters, we do not proposethat they probably suffer heaviermortality during, and are necessarilyrepresentative of the long-term becauseof, their longer migrations and that difference between Indiana and South Caroli- this equalizestheir survivorshipwith that of northern winterers. Evidencefor equalannual mortality.--(1) The FACTORS OFFSETTING DIFFERENTIAL life expectancyof juncos banded in one winter and recovered in the same 10-rain block in a OVERWINTER MORTALITY IN NORTH AND SOUTH later winter, as recorded by the USFWS, was invariant with latitude. The sample is not Theoretical considerations.--If the conclusion large, but it gains strengthfrom having been just reached is correct and if, as we take for compiledover many years. Probablythe result granted, choice of winter site is governed by is not much distortedby random annual vari- natural selection,the mortality difference as- ation. sociatedwith wintering in the north must be (2) Our estimated annual survival rate of offset by advantagesthat tend to equalize the 53% amongsite-faithful Indiana juncos agrees fitness of members of northern and southern very closely with the independent estimate, winter populations. (We assume that fitness 54%, from the USFWS data. It is importantto does not vary accordingto locationof winter- note that 53% is well within the rangeof an- ing site, i.e. that directional selectionis not in nual survivalrates reported for temperate-zone the processof shifting the junco'swinter dis- passerines(Farner 1955, Greenberg 1980, Sear- tribution either southward or northward.) The cy and Yasukawa1981). If our estimateis rep- compensatoryadvantage(s) of wintering in the resentative of juncos that winter in Indiana, north must take the form of (1) higher survi- there is no a priori reasonto expectSouth Car- vorship than that of southern juncos during olina juncosto have higher survivorship,nor some other season(s),thus bringing annual is it necessaryto supposethat Indiana popu- 252 KETTERSO•qANDNOLAN [Auk,Vol. 99 lations require some unusual reproductivead- latitudes revealed in the USFWS data. We em- vantage to maintain equal fitness with south- phasize that the junco's age distribution in ern juncos. winter would have led us to predict a lower (3) The return-recapture rates in Table 4 return-recapturerate of northern young even show equal minimum annual survival of adults before that lower rate was found. That is, a in northern and southernwinter populations. distribution in which young tend to winter The 53% Indiana survival rate obtained by north of adultscould be maintainedyear after continuousefforts throughoutthe junco's res- year only by lower survivorshipof youngthan idence there shows that in Indiana sampling of adultsbetween December sampling opera- lasting only a few days fails to recapturemany tions (a possibility just consideredand dis- banded individuals that are in fact present. missed) or by choice of a different and more That this is also true at our other sites is in- southerly secondwinter site by some individ- dicatedby the factthat at all of them we some- uals that have spent their first winter farther times caughtbanded juncos that had escaped north. Notably, Spaans (1977) has reported capturein the precedingeffort. We assumethat similar between-winter shifts toward milder the probability of failing to recapturean indi- climatein the Starling(Sturnus vulgaris). vidual even though it was presentwas equal In sum, the return-recapture rates, the at all sites. If this conservativeassumption is USFWS records,and our winter demographic correct,annual survivorshipof adults in those data converge from independent directions samplesdid not vary with latitude. and point to equal annual mortality of north- In contrast to the adult rates, the return-re- em and southernwintering populations. capture rate of northern young fell below the Consequencesof unequalannual mortality.- rate for northern adults, for southernyoung, Without regardto the affirmativeevidence for and for southern adults, all of which were sta- equal annual mortality, unequal mortality of tisticallyindistinguishable. Despite this differ- winter populationswould have demographic ence, for two reasons we believe that northern consequencesthat can be predicted from, or young survived the year between sampling tested against, theory. All the expectedcon- and resamplingas well as did the other classes. sequences would be observable in the field, (1) The survival of northern adults from De- and none has been observed. cember to Decemberwould not be expectedto Any hypothesisthat northernwinter popu- be higher than that of northernyoung if it were lationshave lower annual survivorshipmust no higher in the first 3 monthsof that 12-month also take account of, and be reconciled with, period. Any survival disadvantageassociated five points that we regard as factsor fair as- with inferior age and experienceshould show sumptions about the junco. (1) More males up early, particularlybecause the early part of winter in the north, more females in the south. the year used in our calculationswas winter. (2) Overwinter survival of the sexes in the But Tables I and 2 reveal no disadvantage; north is equal. (3) Southernjuncos are more northern young were as likely as adults to be likely to survive winter than are northern; recaptured in late winter, and in each of two therefore,because of points 1 and 2, fewer fe- winters rather large Indiana samplesrevealed malesthan malesdie duringwinter. (4)Mortal- no changein ageratios between December and ity of adults,i.e., experiencedmigrants, is prob- February.Evidently, young juncos have passed ably no higher duringautumn migration than beyond the "initial juvenile period of higher during spring migration, because weather is mortality" (Farner 1955: 403)by December. In lesscertain at the timeof springmigration, and this they are like many other north temperate- the precedingyear's seed crop is depletedafter zone passerines,for which, as a rule of thumb, having supportedconsumers throughout the I Januaryis often taken as the date at which winter. (5) The hazardsof migrationper unit life expectancystabilizes at the adult rate (see of distancetraveled are probablythe samefor discussion and citations in Farner 1955). (2) femalesas for males;if the two sexesdepart The low return-recapture rate of northern from the same spot for the same destination, young can be explained plausibly by suppos- femalesshould be at leastas likely as malesto ing that somesurvivors of thatage class choose survive and arrive. This assumption seems new sites in their second winter and thus are reasonablein view of the absenceof any sex responsiblefor the lower fidelity to northern role associated with the behavior involved in April 1982] Migrationand Winter Mortality in Juncos 253

Greater winter mortality in North lygynously(Fig. 3). The first of these possibil- ities resultsin greater reproductivesuccess for males,the minority sex,than for females(sensu Fisher 1930). Such a situation, in which a float- 1 ing population of females exists, will not be More (2 (2 than • at stable: instead, the sex ratio at conception start of breeding (hereafter,primary sexratio) would be expect- ed to evolve toward a skew in favor of males and this skew would persistin the sex ratio at the end of parental care (hereafter,secondary Floating(2 (2 Polygyny sex ratio). The second alternative, polygyny, supposesthat this mating systemwould evolve as a responseto an unbalancedtertiary (breed- Lower • reproductive ing) sex ratio. Such an evolutionarysequence successof (2 (2 was at one time accepted as a satisfactory than • • mechanismto accountfor avian polygyny,but it is now generally rejected, becauseit prob-

Evolution of skew ably confusescause and effect (Selander1965, in primary and Orians 1969, Emlen and Oring 1977; but see secondary sex ratios Wiley 1974). favoring• • Field data bearing on the alternativepossi- FzG.3. Possibleevolutionary consequences of a bilities just presented are as follows. Mating situation in which winter mortality, and as a result system:Juncos are reportedto be, like the great annual mortality, of males exceedsthat of females, majority of birds (Lack 1968a,Verner and Will- producing an excessof females among potential son 1969), monogamous. During 8 weeks breeders. See text. (spreadover 4 yr) of intensivestudy of juncos in July and early August, when most females have dependentyoung, we have found 30-35 migrating.If a bias doesexist, it is likely to be broods,all of them closelyattended by an adult against males, becausemales migrate earlier male and an adult female. Sexualdimorphism (spring only, pers. obs.), when weather con- of juncos is slight, which supportsthe view ditions are both less predictableand less mod- that monogamyis their normalmating system erate. (Emlen and Oring 1977). In addition to the five points just enumer- Secondarysex ratio: At Wawa, Ontario, we ated we make another assumption, which, netted randomly for 3 days in early September though it is conservative,must be justified. In 1979and for 10 daysat the end of July1980 and the next sectionwe presentour reasonsfor as- caught43 newly independentyoung that were suming(6) that survivorshipof the sexesis ap- still in completejuvenal plumage. Of these, 22 proximatelyequal during the breedingseason, were male and 21 female. (The birds were held but for the presentwe simplyproceed on the in captivity and sex determined in the usual assumption. way after postjuvenalmolt was completed.) Now supposea breedingpopulation, its sex Breedingsex ratio: We have seenno evidence ratio at unity and its membersabout to migrate of a floating nonbreeding female population. varyingdistances into the winter range.Under Instead, all females that we have netted ran- the hypothesisthat annualmortality rates of domly, even early in the breeding season winter populations are unequal because of (June), have had incubation patches.Twice we higher survivorship of southern winterers, have observed males, known to be unmated, more femalesthan males from this breeding advertise for at least 2 weeks in June. population will be alive at the start of and dur- In sum, the hypothesisthat southern-win- ing the next breedingseason. The imaginable tering juncoshave greaterannual survivorship consequencesof a breeding-seasonsurplus of implies a sexualdifference in mortalitybecause females are either (1) that some females will not of the clinal variation of the winter sex ratio. get matesor (2) that somefemales will mate po- None of the consequencesof a sexual differ- 254 KETTtRSON^•D NOL^• [Auk, Vol. 99 ence in annual mortality has been observed, equalized during their migrations,in an av- and what field evidencethere is tendsto falsify erage year approximately the same number of the hypothesis. males and females will arrive on and leave the Mortality at other seasons.--If further evi- breeding grounds.Males will predominatein dence should establishthat annual mortality the winter rangeat the beginningof winter, be is invariant for northern and southern popu- in the minority just before spring migration lations sampled in December, during what begins, and be in parity again after migration season(s)is it likely that southernwinterers mortality has reducedsouthern winterers, with compensateby experiencinghigher mortality their disproportionatefemale numbers. than northern? It may be objectedthat a substantialincrease We turn first to the breedingseason. It is in mortalityduring migration into the southis most unlikely that higher mortality during unlikely when the differencein distancetrav- breeding would be causedby wintering in the eledto southernsites is probablyat mostabout south, but there are two possiblereasons for 1,000 km, approximatelythe distanceseparat- a correlation.(1) If breeding location deter- ing our Michigan and Alabama sites. It is true mined winter location in the sense that mem- that some juncos have been found (Johnston bers of breeding populationswintered togeth- 1962, Helms et al. 1967) to carry migratory fat er, e.g. if northern breeders were northern reserves sufficient to permit them to cover winterers, latitudinal variation in some factor 1,000 km in only two flights, with a stopover causingsummer mortality might offsetlatitu- to fatten up, althoughwe seefew fat juncosin dinal variation in winter mortality. Although spring and autumn in Indiana. Theoretical ar- we can conceive of clinally varying summer gumentsand limited evidence(reviewed by mortality (e.g. as the result of distribution of Baker 1978: 679) suggest,however, that over- predators),we reject this first alternative, be- land migrantsdo not make long flights, what- causebreeding populationsdo not remain in- ever their fat reserves.Instead, they are be- tact in winter: the partial sexual separation lieved to migrate in numerousshort moves and, probably, the partial age separation at separatedby stopovers,a conclusionthat John- that season eliminate this possibility. (2) ston (1962)drew for juncosafter field and ex- Higher survivorship by southern winterers perimentalphysiological study of North Car- would tend to be offset by higher summer olina populations. At each stop the migrant mortality if breeding females had a greater may confront dangersarising out of unfamil- mortality rate than breeding males. We know iarity with locationsof food, shelter, and ref- of no evidence bearing on a sexual difference uge from predators.We imaginesuch dangers in breeding mortality, exceptthat we have not to be considerableand to increasein stepwise seen dependentyoung being tended by only fashion with the number of stops.So far as we one parent (seeabove). If, however, excessive know, the extent to which hyperphagialeads male mortality in winter were balancedby ex- birds to run greaterrisks of predationhas not cessive female mortality in summer, a non- been investigated, but the increase may be breedingsurplus of femaleswould arrive--and important. Certainly, hungry, free-living for a time survive•on the breedingrange, un- juncos will tolerate the closeapproach of hu- less the primary and secondarysex ratios shift- mans before interruptingtheir foraging(e.g. ed to adjust for the greater reproductive suc- when snow coversthe ground), and we have cess of males or unless some mating bonds seen many apparent migrants with such low were polygynous.These points have been ex- energy reservesthat they can be assumedto amined above, where it was concluded that have been hungry. For example,juncos caught existing evidence does not point to a surplus in midafternoonin Indiana during the migra- of females, or a skewed secondaryratio, or a tion seasonssometimes starve before morning polygynousmating system.We thereforepro- if held at room temperature without food. In posethat southernwinter populationscompen- contrast, Indiana juncos caught in winter nor- sate for their greater overwinter survivorship mally can fast for 40-45 h at 4øC and still re- by greatermortality during, and becauseof the cover (Ketterson and Nolan 1976, 1978; Stuebe increasedrisks involved in, their longermigra- and Ketterson 1982). tions. The magnitudeof the disparityin migration If annual mortality of southernwinterers is mortality that would be necessaryto equalize April 1982] Migrationand Winter Mortality in Juncos 255

A B C migration mortality is positively correlated Leapfrog Latitudinal Winter-Mix Diaplacement with distancemigrated, and we now deal with that aspect of the subject. The question is whether the model for winter populations-- equal annual mortality rates and therefore no need to supposeunequal reproduction--suf- ficesfor breedingpopulations or must be mod- ified to permit variation in reproduction among these. The answer, we think, depends on what migration pattern (definedbelow) has evolvedin the junco. Figure 4 depicts three basic ways (here- after patterns)in which breedingpopulations can arrange themselves on the wintering grounds. (Intermediate patterns can be imag- Fig. 4. Three possiblemigration patterns in the ined, but consideration of them would add Dark-eyedJunco: leapfrog, latitudinal displacement, nothing to this analysis.)In the leapfrogpat- and winter-mix. Two breeding populations,each tern (Greenberg1980) (A), the southernmost consistingof males and femalesmaking long- and breeders make the shortest migration and short-distancemigrations to their wintering sites, therefore winter farthest north, etc., and are shown for each pattern. Hatching on lefthand populations from different latitudes do not in- vertical axis indicatesarbitrary units of distance.The termingle in the winter range. In the latitudinal predictedconsequences of the patternsare summa- rized in Table 6. displacementpattern (B), all breeding popula- tions travel the same distance; the northern- most breeders are the northernmost winterers, and here again breeding populationsremain in- annual survival is easy to overestimateif one tact in the winter range. In the winter-mix pat- thinks only in terms of the overall annual rate. tern (C), no relationship exists between lati- As an example,consider two adult populations tudes of breeding and of wintering sites, and of equal size leavingthe samebreeding ground, breeding populations mingle randomly in one heading for a northern winter site and the winter. other for a southern. In order to reduce each Beforepredicting the different consequences by 50% at the start of the next fall migration these patterns would have for the dynamics and to impose on the northern population a and other characteristicsof junco populations, winter mortality rate twice as great as that of we statethose respectsin which we assumeall the southern population, seasonal mortality patternswill be alike: (1) membersof breeding rates during the year could be as follows: and winter populationshave equal averagefit- northern population--fall migration, 13%; ness; (2) females from each breeding popula- winter, 30%; spring migration, 13%; breeding tion tend to winter south of males from the season,5%; southern population--fall migra- same population; (3) becausemigration mor- tion, 21%; winter, 15%; spring migration, tality and winter survivorshipcorrelate posi- 21%; breeding season,5%. tively with distance migrated and can offset Consequencesof patterns of migration.--To each other, southern winterers from each this point, consideration has been virtually breeding population are equal in fitness with confined to winter junco populations and to northern winterers from that population; (4) why and how we believe theseprobably arrive breeding-seasonmortality does not vary cli- at equal survivorship, on average, from De- nally with latitude. (This last assumptioncan cember to December. Breeding populations be discarded with predictable consequences, have been referred to only when necessaryto but we see no reasonto complicatematters by project the effects that various mortality pat- doing so here.) terns might have on sex ratio or the mating Table 6 presents consequenceswe expect system. These expectedeffects would be in- each pattern to produce. In A, the southern dependent of the latitude of the breeding site. breedingpopulation migrates the shortestdis- But breedinglatitude will have important con- tance and trades its higher migration survi- sequencesif we are correct in proposing that vorship for higher winter mortality; equal fit- 256 KETTERSONANY NOLAN [Auk, Vol. 99

TABLE6. Predictedconsequences of possiblemigration patterns in juncos, based on the conclusionthat migration mortality increaseswith distancetraveled, while winter mortality decreases.

Migratory pattern of breeding populations

B A Latitudinal C Leapfrog displacement Winter-mix Northernvs. southernbreeding populations a Annual mortality = %, N > S %, N > S Migration mortality •, N • S = •, N • S Winter mortality %, S • N %, N • S = Reproductive success = %, N > S %, N • S Northern vs. southernwinter populationsa Annual mortality = %, N • S = Migration mortality %, S • N = %, S • N Winter mortality %, N > S %, N • S %, N • S Reproductivesuccess = %, N • S = Variance in genetically based characters greaterin winter than in breeding populations? No No Yes Preservesduring winter any clinal variation occurringon breeding grounds? Yes Yes No (but cline (and cline reversed) in same direction)

northern; S = southern.

ness across populations (both winter and be attainedbecause each population would con- summer) can be achieved without inequality sist of two elements: (1) southern breeders with of reproduction. In B, migration mortality is lower migration mortality and lower reproduc- equal for all populations, but winter mortality tion, and (2) northern breeders with higher mi- of northern breedersis higher; thesemust have gration mortality and higher reproduction.As higher reproduction if they are to be as fit as for winter sexratios, in our view only a migra- southern breeders. Because C mixes northern tion that incorporatedsome degree of pattern C and southern breeders on the wintering would produce clinal variation. Finally, to the ground, winter mortality is invariant with extentthat breedingpopulations vary morpho- breeding latitude, but northern breedersmake logically(Miller 1941:317 describessome geo- longer migrations,die at a higher rate for that graphicalvariation in Juncoh. hyemalis),A and reason,and must have greaterproductivity if B would preserve such variation in the winter their fitnessis to equal that of southernbreed- range, whereas C would tend to abolish it. ers. Variance in relevant morphologicalcharacters Each pattern predicts its own set of charac- of winter populations(or in other genetically teristics of winter populations, and these can basedtraits) would in that casebe greaterthan be comparedwith what is known to date about that of breeding populationsunder C but not winterinõ iuncos. Annual survival rates mea- under the other two patterns. sured from December to December (i.e. at start It will be apparentthat if the junco satisfies of winter) couldbe equal under A but not under the assumptionsset forth above,its partial sex- B. They could alsobe equal under C, but the ual segregationin winter suggeststhat its mi- situation would be somewhat complex:equal- grationpattern approachesC moreclosely than ity acrosswinter populationswould exist de- A or B. The evidence favoring equal annual spite inequality acrossbreeding populations. mortality from December to December is con- Restatedin terms of fitness, the equal average sistentwith C but also with A. Our body-size fitnessof membersof winter populationswould data for winter and summer populations are April 1982] Migrationand Winter Mortality in Juncos 257 still being analyzed,but we can statenow that mortality associatedwith migration, assumes if there are latitudinal body-size clinesin sum- that the benefitsof migrating are self-evident, mer they are not preserved in winter. and considersthe theoretical problem to be In conclusion, it is obvious that adequate accounting for the continued existence of the breeding studies in northern and southern nonmigrants. The other model is that of Lack Canada, although they would be difficult to (1954, 1968b), who startsfrom the proposition carry out, would reveal whether reproductive that migration is costly. In some years migra- success varies with latitude and would thus tion mortality is higher than the winter mor- contribute to an understandingof the junco's tality of sedentaryindividuals, in other years migration pattern. lower. Over time, the annualrates for migrants and nonmigrants balance out, and both ele- THE ROLE OF MIGRATION IN LIFE HISTORY ments continue to exist. Productivity appar- ently is taken as unvaryingwith migratorysta- HOW do our evidence and interpretations re- tus. late to current theory concerningthe signifi- Lack's view, because it weighs migration canceof migration for individual fitness?Most losses against winter losses, resembles the considerationof variation in distancemigrated model proposedherein for winter junco pop- by conspecificshas approached the problem ulations. It differs, however, in that its concern from the point of view of breeding popula- is with unequal year-to-year mortality rates tions. In particular, the question has been the that fluctuatearound equal long-termmeans. identification of the advantagesand disadvan- We do not disagreewith Lack, in that we do tages of remaining on the breeding ground not supposethat the annual mortality rates of throughoutthe year among so-calledpartially northern- and southern-winteringjuncos are migratory species.The logic of theories about identicalfrom year to year, althoughin theory systemsin which fewer than all individuals are annual rates could be invariant despite large migratory, however, appearsapplicable alsoto and important differences in their seasonal fully migratory speciesin which elementsof the components.Nevertheless, it seemsclearly de- population travel varying distancesand there- sirable to analyze population dynamics over fore settle in different winter sites. the shortestpossible, biologically meaningful Baker (1978: 635) summarizes the literature time intervals, even though other temporal on partial migrantsand derivestwo modelsto perspectivesmay be equally important. With account for the fact that selection has not elim- specialreference to the subjectof this paper, inated either the migrant or the nonmigrant study of geographicdifferences in winter mor- element. The models are alike in assuming (1) tality of conspecificpopulations may permit a stable system(i.e. equal fitnessfor migrants estimatesof the scale of mortality during mi- and nonmigrants),in which (2) the nonmigrant gration if annual and breeding-seasonsur- element in some or all years suffers higher vivorshipsof the populationsare known. winter mortality than the migrant (3) as the result of failure to leave the breeding range. One model proposesthat a higher reproductive ACKNOWLEDGMENTS rate compensatesfor the greater winter mor- We thank the many people who have helped make tality of nonmigrantsand attributesthis repro- our capture efforts possible, particularly the follow- ductive gain to priority of accessto limiting ing: Ray Adams, Pat Adams, Sidney Gauthreaux,Jr., breeding resources(von Haartman 1968) or to a Paul Hamel, Carl Helms, JamesV. Peavy, Jr., Anna longer period in which to reproduce.[Compare Ross, Ruth Schatz, and Paul Schatz. Members of our Nolan (1978) on the advantages in terms of own group who accompaniedus and whose field work was indispensable were Sue Braatz, L. Jane breedingtime gained by male Prairie Warblers Clay, Mary Latham-Weeks, Dorothy Mammon, that return early to the breeding ground, and CatherineMeyer, J.P. Myers, Cindy B. Patterson,Toy seeGreenberg (1980), who arguesthat temper- S. Poole, Richard Rowlands, Michael Shelton, Robert ate-zone breeding species wintering in the Steel, and Ken Yasukawa. Sidney Gauthreaux, Jr., tropics pay for enjoying a benign winter cli- John M. Emlen, and JamesBlank made valuable sug- mate by losing time for reproduction.] This gestionsabout an earlier version of this paper. Bowl- view, as we read it, does not concern itself with ing Green State University, Indiana University, 258 KETTERSO•AND NOLAN [Auk, Vol. 99

Clemson University, the KalamazooNature Center, GREENBURG,R. 1980. Demographicaspects of long- and the Banding Laboratoryof the U.S. Fish and distance migrations. Pp. 493-504 in Migrant Wildlife Service supported us in important ways. birds in the neotropics:ecology, behavior, dis- This researchwas funded in part by NSF DEB-78- tribution, and conservation (A. Keast and E. S. 11982 and DEB-81-10457. Morton, Eds.). Washington,D.C., Smithsonian Inst. Press. HAARTMANN, L. VON. 1968. The evolution of resi-

LITERATURE CITED dent versusmigratory habit in birds. Somecon- siderations. Ornis Fennica 45: 1-7. BAKER, M. C., & $. F. Fox. 1978. Dominance, sur- HELMS, C. W., W. H. AUSSIKER,E. B. BOWER, & $. vival, and enzymepolymorphism in Dark-eyed D. FRETWELL.1967. A biometricstudy of major Juncos,Junco hyernalis. Evolution 32: 697-711. body componentsof the Slate-coloredJunco, --, C. $. BELCHER,L. C. DEUTSCH,G. L. $HER- Juncohyernalis. Condor 69: 560-578. MAt4,& D. B. THOM?SOt4.1981. Foraging success JOHNSTON,D. W. 1962. Lipid deposition and gonad- in juncoflocks and theeffects of socialhierarchy. al recrudescencein responseto photoperiodic Anim. Behav. 29: 137-142. manipulationsin the Slate-coloredJunco. Auk BAKER,R. R. 1978. The evolutionaryecology of an- 79: 387-398. imal migration. New York, Holmes and Meier, KETTERSON,E. D. 1979a. Aggressivebehavior in Inc. wintering Dark-eyed Juncos: determinants of BALPH, M. H. 1977. Winter social behaviour of Dark- dominanceand their possiblerelation to geo- eyed Juncos:communication, social organiza- graphic variation in sex ratio. Wilson Bull. 91: tion, and ecologicalimplications. Anim. Behav. 371-383. 25: 8594384. --. 1979b. Status signaling in juncos. Auk 96: BE•t, A. C. 1968. Life histories of North American 94-99. cardinals,grosbeaks, buntings, towhees,finch- --, & V. NOLAN,JR. 1976. Geographicvariation es, sparrows and allies; order Passeriformes: and its climatic correlates in the sex ratio of east- family Fringillidae (O. L. Austin, Jr., Ed.). U.S. ern-winteringDark-eyed Juncos (Junco hyernalis Natl. Mus. Bull. 237. hyernalis).Ecology 57: 6794593. CAt•GHLEY,G. 1977. Analysis of vertebrate popu- --, & --. 1978. Overnight weight loss in lations. London,John Wiley & Sons. Dark-eyed Juncos (Juncohyemalis). Auk 95: DINGLE,H. 1980. Ecologyand evolution of migra- 755-758. tion. Pp. 1-103 in Animal migration, orientation --, & --. 1979. Seasonal,annual, and geo- and navigation(S. A. Gauthreaux,Jr., Ed.). New graphic variation in sex ratio of wintering pop- York, Academic Press. ulations of Dark-eyed Juncos(Junco hyemalis). EmLEN,S. T., & L. W. ORIt4C. 1977. Ecology,sexual Auk 96: 532-536. selection,and the evolution of mating systems. KING, J. R., D. 5. FARNER,& L. R. MEWALDT. 1965. Science 197: 215-223. Seasonalsex and age ratios in populations of FARNER,D. $. 1955. Birdbanding in the study of White-crowned Sparrows of the race garnbelii. population dynamics. Pp. 397449 in Recent Condor 67: 489-504. studiesin avian biology (A. Wolfson, Ed.). Ur- KR•BS,J. R. 1979. Survival value of migration (book bana, Illinois, Univ. Illinois Press. review). Nature 278: 671-672. FISHER,R. A. 1930. The geneticaltheory of natural LACK, D. 1954. The natural regulation of animal selection. London, Oxford Univ. Press. numbers. London, Oxford Univ. Press. FREtWELL, S. D. 1969. Dominance behavior and --. 1968a. Ecologicaladaptations for breeding winter habitat distributionin juncos(Junco hye- in birds. London, Methuen & Co. Ltd. rnalis).Bird-Banding 40.' 1-25. --. 1968b. Bird migration and natural selection. 1972. Populations in a seasonalenviron- Oikos 19: 1-9. ment. Princeton, New Jersey,Princeton Univ. MILLER,A. H. 1941. Speciationin the avian genus Press. Junco. Univ. California Publ. Zool. 44: 173434. GAUTHREAUX,$. A., JR. 1978. The ecologicalsignif- MYERS,J. P. 1981. A test of three hypothesesfor icance of behavioral dominance. Pp. 17-54 in latitudinal segregationof the sexesin wintering Perspectivesin ethology,vol. 3. (P. P. G. Bate- birds. Can. J. Zool. 59: 1527-1534. son and P. H. Klopfer, Eds.). New York, Plenum NOLA•, V., JR. 1978. Ecologyand behaviorof the Publ. Corp. Prairie Warbler Dendroica discolor. Ornithol. --. 1979. Priorities in bird migration studies. Monogr. 26: 1-595. Auk 96: 813-815. ORIANS,G. H. 1969. On the evolutionof mating sys- GRABER,J. W., & R. R. GRABER.1979. Severe winter tems in birds and mammals. Amer. Natur. 103: weather and bird populationsin southern Illi- 5894503. nois. Wilson Bull. 91: 88-103. ROHWER,$., & P. W. EWALD. 1981. The cost of April 1982] Migrationand Winter Mortality in Juncos 259

dominance and the advantageof subordinance STUEBE,M. M., & E. D. KETTERSON.1982. A study in a badge signalingsystem. Evolution 35: 441- of fasting in Tree Sparrows(Spizella arborea) and 454. Dark-eyedJuncos (Junco h. hyemalis):ecological ROSEBERRY,J. L. 1962. Avian mortality in southern implications. Auk 99: 229-308. Illinois resulting from severe weather condi- U.S. WEATHER BUREAU BULLETIN W. 1932. Clima- tions. Ecology43: 739-740. tography of the United States.Sections 1-106. SEARCY, W. A., & K. YASUKAWA. 1981. Sexual size VERNER,J., & M. F. WILLSON.1969. Mating systems, dimorphism and survival of male and female sexual dimorphism, and the role of male North blackbirds (Icteridae). Auk 98: 457-465. American passerinebirds in the nesting cycle. SELANDER,R. K. 1965. On mating systemsand sex- Ornithol. Monogr. 9: 1-76. ual selection. Amer. Natur. 99: 129-141. WILEY, R. H. 1974. Evolution of socialorganization SPAANS,A. L. 1977. Are Starlings faithful to their and life-history patterns among grouse. Quart. winter quarters?Ardea 64: 83--87. Rev. Biol. 49: 201-227.

The AOU Student Awards Committee issues the following reminders to all students interested in com- peting for the AOU Student Paper Award and the two Nellie JohnsonBaroody Student Paper Awards at the 1982 AOU meetings:(1) To qualify for the competition,each student must submit an expandedab- stract of the paper directly to the chairpersonof the AOU Student Awards Committee (Douglas Mock, Departmentof Zoology, University of Oklahoma, Norman, Oklahoma 73069)by the "Call for Papers" deadline in the meeting announcement.(2) Appropriate applicationswith expanded abstractsmust be submitted separatelyfor the student paper awards and for the Marcia Brady Tucker travel awards. (3) A studentis ineligiblefor studentpaper awardsif she/heshares authorship of a paper, i.e. the studentmust be the exclusive author.

The AmericanOrnithologists' Union will hold its annualmeeting in Chicago,Illinois from 10-14 October 1982. A formal call for papers is enclosedwith this issue of The Auk, with a deadline for submission of abstractsfor papers and poster sessionsof 20 May 1982. For further information on the scientific program, contactJames R. Karr, Program Chairman, Department of Ecology, Ethology, and Evolution, Universityof Illinois, 606 E. Healey, Champaign,Illinois 61820USA. COMMUNITY ORGANIZATION OF RIPARIAN BREEDING BIRDS: RESPONSE TO AN ANNUAL RESOURCE PEAK

KENNETH V. ROSENBERG,ROBERT D. OHMART, AND BERTIN W. ANDERSON Departmentof Zoologyand The Centerfor EnvironmentalStudies, Arizona State University, Tempe,Arizona 85287 USA

ABSTRACT.--Spatialdistribution, foraging behavior, and diets of 12-13 insectivorousbird specieswere measured during two breeding seasonsin a cottonwood- and willow-domi- nated riparian forest in Arizona. Spatially, no negative associationsexisted between any speciesin either year. Althoughforaging behavior varied considerably,diets overlapped broadly, even betweenmembers of different foragingguilds. Eight speciespreyed heavily on cicadas(38-80% of diet). Sampling of emergingcicadas in cottonwood-willowhabitat indicatedthat numbersexceeded the metabolicneeds of the bird communityby as much as tenfold. Peak cicada numbers coincidedwith fiedging young in all eight species.We proposethat this seasonallysuperabundant and predictableresource permits coexistence without niche segregationand may, in part, explain the high density and diversity of eco- logicallysimilar breeding birds in desertriparian communities.It appearsthat niche vari- ables such as bill shapeand foraging behavior may not be true predictorsof actualresource use within a community. Assumptionsof resource-basedinterspecific competition among breeding birds are not supported.Received 4 May 1981, accepted1 October1981.

AMONG the forces that operate to shape and berry 1979, Rotenberry 1980a). Furthermore, maintain patterns of avian community orga- resource-availabilitypatterns are poorly known nization, competition for limited resources in even the most well-studied communities. (e.g. food) has been frequently invoked as a In the deserts of western North America, ri- primary selectiveforce (MacArthur 1972, Cody parian forests (the alluvial floodplain and its 1974, Cody and Diamond 1975). Such compe- attendant vegetation)have been shown to be tition has rarely been unequivocallydemon- of great ecologicalimportance to bird popula- strated for birds. An essentialassumption be- tions (Carothers et al. 1974, Anderson and hind competitiontheory is that resourcesare Ohmart 1977, Stamp 1978). These ecosystems in short supply. Although White (1978)argued typically supportextremely high densitiesand that all organismsare limited by an inadequate diversities of breeding bird speciesand pro- supplyof nitrogenousfoods, this has not been vide vegetationally and geographicallywell- properly demonstrated for many bird com- defined systemsfor study. munities. Recently,strong criticism of the ap- Riparian systems,like other temperate for- plication of competitiontheory in community ests, undergomarked seasonalchanges in pri- ecologyhas been expressed(Wiens 1977, Con- mary and secondaryproductivity. It might be ner and Simberloff 1979), and older arguments predicted that the avian community present against competition are not resolved(e.g. An- during the period of peak productivity would drewartha and Birch 1954). be structuredin part by aspectsof that resource Measurementsof avian diets are lacking in peak. most community studies involving resource We studied an avian community on a large use. Instead, diets are most often inferred from riparian forestplot over two completebreeding morphological (Schoener 1974, Hespenheide seasons.The generalapproach of the studywas 1975) or behavioral measures(MacArthur 1958, to examine densities, microhabitat prefer- Cody 1974, Rabenold 1978). The few studies ences,foraging behaviors, and dietsof all diur- that have examined actual resource use em- nal insectivorousspecies on the plot. This re- phasized the inadequaciesof such inferences port documents a response within this bird and have not supported the premises of the community to a summer resourcepeak and competition argument (e.g. Wiens and Roten- discussesthe evolutionary consequencesof

26O The Auk 99: 260-274. April 1982 April 1982] ResourceUse by RiparianBirds 261

this responsein light of existing community aging measures were computed by subprogram theory. CROSSTABSin SPSS (Nie et al. 1975). We present data for the two seasons combined. The degreeof overlapbetween species pairs (i, j) METHODS for each foraging measurewas calculatedas'. Studyarea.--A 20-ha plot was establishedin a con- 5' P•aP•a tinuous forest stand along the Bill Williams River Oa -- near its delta at Lake Havasu in Yuma County, Ar- (• PS,)(ZPSi) ' izona, elevation 100 m. The dominant tree species where P•aand p•, are the proportionaluse of resource are cottonwood(Populus frernontii) and willow (Salix state "a" by species i and j, respectively (?ianka gooddingii),with a patchy understory of exotic salt 1974, May 1975). cedar (Tamarix chinensis),cattails (Typha latifolia), To assessthe combined effectsof overlap in several and bulrushes(Scirpus spp.). Daily temperaturevar- foraging measures, two additional matrices were ied from -6øC to 33øCin early spring, and from 8øC computed. The first representsthe total overlap in to 43øC in summer. Other details are described in foragingbehavior and is the averageof overlapsfor Rosenberg(1980). eachspecies pair in foragingmethod, substrate, and Densityand breedingphenology.--We gridded the branch diameter, Because these measures were rea- 20-ha plot at 20-m intervals with surveyor's tape and soned to be correlated, the summation-alpha ap- marked each grid point with a coordinate number. proach (Cody 1974) appearedsatisfactory. Similarly, Detections of all territorial breeding specieswere re- the secondmatrix representsthe total overlap in for- cordedon daily field mapsfrom Januarythrough July aging spaceand is the averageof overlapsin use of 1977 and February through August 1978. We gave tree speciesand tree portion multiplied by the over- special attention to locating nests, following known lap in foraging height, The use of product-alpha pairs, and determining territorial boundaries. All (Cody 1974) seemed appropriate in the case of for- parts of the grid receivedapproximately equal cov- aging height, becausethis measureintroduced con- erage. We determined the number of pairs and the ditions that were independent of other space mea- relative territory size from the composite of field sures, i.e. two species that did not overlap in maps for each season. foragingheight couldnot overlapin overallforaging In addition, we assigned each bird detection its space.The hazards of this approachwere discussed closestgrid coordinatenumber from the daily field by May (1975), but these matrices seemedto repre- maps. Totals of 1,482 bird detections in 1977 and sent a realisticapproximation of the patternsof sim- 1,800 detections in 1978 of 12 specieswere used in ilarity within this bird community. We performed this analysis. Associations between pairs of bird a group averageclustering on these matricesto con- specieswere testedseparately for each seasonbased struct dendrograms for visual display (Cody 1974). on the frequencyof co-occurrenceat each of the 281 To test for actual differences between species for points. The Cramer's V statisticwas calculatedas: each foraging measure, we compared the distribu-

ad - bc tions of the state frequencies of each measure for V- (mnrs)•' goodness-of-fitusing GH (Sokaland Rohlf 1969:575). The lack of a significant difference between two where a, b, c, and d are the observed cells of a 2 x 2 speciesis assumedto imply a biologicallyimportant contingency table, and m, n, r, and s are the row and overlap.When the statisticalsignificances associated column totals, respectively(Pielou 1977: 201). This with different degreesof overlapbetween speciesare is essentiallya correlationcoefficient and its signif- compareddirectly, the biologicalrelevance of each icance was tested by simultaneously computing approach can be assessed. "X2," which approximatesa Chi-squaredistribution Diet/rnorphology.--DuringJuly and August 1978, with one degreeof freedom. 106 birds of 13 specieswere collectedfrom riparian Foragingbehavior.--We observed foraging by birds forest stands similar to the study plot, and their on the study plot and in similar foreststands along esophagiand stomachswere immediatelypreserved the Bill Williams River. A foraging observationwas in a formalinsolution. We recordedage, sex, weight, defined as an actualattempt to procureprey, and we bill size (exposedculmen, depth, and width), wing recorded the following data for each observation: chord, fat condition, and condition of molt for each bird species,foraging method, height of bird in tree, specimen.A referenceseries of skins was prepared. tree species, tree height, branch diameter, portion Esophagealand stomachcontents were dried and of tree, substratefrom which prey was captured, and weighed, and their volumes determined. Length, type of prey (if observed).All statesof foraging mea- frequency, and percent-volume of each individual suresare listed and defined in the Appendix. A total prey item were recorded. All arthropod prey were of 2,122 observations of 12 specieswas included in identified to order and many were identified to fam- this analysis. ily. Identifiable insect parts contributed to the per- Frequenciesand proportions of all statesof all for- cent-volumeof that categorybut not to its frequency. 262 ROSENBERG,OHMART, AND ANDERSON [Auk, Vol. 99

TABLE1. Densities and breeding characteristicsof 13 speciesof birds on a riparian forest plot. Speciesnames are followed by abbreviationsused in subsequenttables and figures. Density is based on estimatesfrom two breedingseasons. Solid line indicatesduration of stay;F indicatestiming of fledgingof broods.

Breeding chronology Density Jan- Feb- Au- Species Code (n/40ha) uary ruary March April May June July gust Yellow-billed Cuckoo (Coccyzusamericanus) YC 28-30 F Common Flicker (Colaptesauratus) CF 2-4 F F Gila Woodpecker (Melanerpesuropygialis) GW 36-40 F F Ladder-backedWoodpecker (Picoidesscalaris) LW 28-36 F Wied's Crested Flycatcher (Myiarchustyrannulus) WF 20-24 F Ash-throated Flycatcher (Myiarchus cinerascens) AF 4--6 F F Vetdin (Auriparusfiaviceps) VN 20-24 F F Common Yellowthroat (Geothlypistrichas) YT 80 F F Northern Oriole (Icterusgalbula) NO 64--80 F F F Summer Tanager (Piranga rubra) ST 24 F F Blue Grosbeak (Guiraca caerulea) BG 4 F Abert's Towhee (Pipilo aberti) AT 50-56 F F F F Song Sparrow (Melospizarnelodia) SS 80-100 F F

Unidentified parts were not included in volumetric prey-size and prey-taxafrequency distributions of analysis. eachspecies pair, as was done for foragingmeasures. The importanceof selectingrelevant categorical Percent-volumedata were not appropriate for such divisions in the computation of diet diversity and comparisons. overlaps has been emphasized by Hespenheide From the morphologicalmeasurements, the ratio (1975)and others. Becausea majority of insectorders of the values of each measurefor any pair of species shows consistent features of form, habit, and catch- was used as an index of similarity. Overall morpho- ability, this level was used in most casesto group logicalsimilarity for each speciespair was the av- prey items in this study. A few families(e.g. Cica- erageof the similarityvalues for eachmeasure. We didae), which were frequently identified and were performedgroup average clustering on the matrix of consideredsufficiently different from othersin their similarity values, as for foraging overlaps. order, compriseddistinct categories.We considered soft-bodiedlarvae of severalorders as one group and RESULTS all spiders as another. Overlap matrices were pre- pared for all speciespairs, as described.We also Density amd breedingphenology.--Table 1 groupedprey itemsinto 10-mmsize classes for com- lists the breeding densities (n/40 ha) and sum- putationof prey size overlaps,as above.The GH sta- marizesthe breedingbiology for the 13 species tistic was used to test for differences in the mean in this study. Densities were nearly identical April 1982] ResourceUse by Riparian Birds 263

TABLE2. Summary of spatial associationsamong 12 riparian bird speciesin 1977 and 1978. + = significant positive association(Cramer's V; P • 0.05); ¸ = no association;n = number of grid points where a speciesoccurred each season.Species codes from Table 1.

1977 n 108 152 99 109 22 46 49 136 143 24 25 72

n Species YC GW LW WF AF VN YT NO ST BG AT SS

72 YC O O O O O O + O O O O 144 GW O + + + O O O O + O O 133 LW O + O O O O O O O O O 114 WF + + o o + + o o o + o 38 AF O O O O + O O O O O O 32 VN O O O O + O O O O O O 93 YT + + + O O O O + O O O 148 NO o + + + o o + o o o o 95 ST + o o o o o + o o o o 11 BG o o o o o o o o o o o 84 AT o o o o o o o + o o o 136 SS O O O + O O O + O O O

in both seasonsand are presented as a range in Fig. 1. Qualitatively,three major functional of values, representingbreeding adults. Five groupingsof specieswere recognized.These speciesWere permanent residents on the plot. groups may be termed foraging guilds (Root The Verdin (Auriparusfiaviceps) is a resident 1967). in the region but was only a summervisitor to The two speciesof woodpeckersforaged pri- the studyarea. All speciesexcept the Common marily on bark, on larger branches (e.g. Flicker (Colaptesauratus) nested on the plot. trunks), and in the inner portion of trees. Flickersused saguarocacti (Carnegieagigantea) Abert's Towhees(Pipilo aberti) and SongSpar- in the adjacentdesert for nestingbut regularly rows (Melospiza melodia) foraged on the foraged in the riparian forest. This species, ground. The Common Flicker foragedprimar- however, was not detectedfrequently enough ily like other woodpeckers,but it also fed on on the plot to be includedin the spatialanal- the ground. ysis. The remaining sevenspecies can be gener- If bird specieswere dividing spaceso that ally groupedinto a canopy-feedingguild. All they were avoiding or in someway excluding tended to use leaves as a substrate and to feed one another, species pairs should have oc- on the outer portions of trees. This group can curredtogether less frequently than by chance be divided further into specialist gleaners in a given season.No significantnegative as- [Verdin, CommonYellowthroat (Geothlypis tri- sociationswere found for any speciespair in chas), and Northern Oriole (Icterusgalbula)] either year (Table2). Nine significantpositive and those that tend to hover or hawk for prey associations were found in 1977, however, and [Yellow-billed Cuckoo (Coccyzusamericanus), 16 in 1978. With 66 comparisons, as many as two flycatchers,and SummerTanager (Piranga four positive and four negative associations rubra)]. Sufficientforaging data for the Blue would be expectedif the specieswere distrib- Grosbeak (Guiracacaerulea) were not obtained. uted at random with respect to one another Anecdotalobservations, however, suggestthat (Kirk 1968: 197). It can be concluded,then, that this speciesalso hovered and gleanedfrom fo- there was a tendencyfor somespecies such as liage. cavity nestersto be clumped together on the Average niche overlapsand their associated plot, althoughthe pattern of positive associa- levels of statistical significancebetween all tion was not consistent. speciespairs for eachmeasure were presented Foragingbehavior.--Species-specific use pat- by Rosenberg(1980). A summaryof thesere- terns for each foraging measureare illustrated lationships is presentedin Table 3. In general, 264 ROSENBERG,OHMART, AND ANDERSON [Auk, Vol. 99

A METHOD

N 48 116 408 464 82 84 109 32 231 109 160 189 B GLEAN B HOVER ] HAWK • PECK []PROBE

Sp. YC CF GW LW WF AF VN YT NO ST AT SS

B FORAGING HEIGHT

B GROUND I•1•0-6.2 m 1-•i6.2-12.3 m B 12.3-21m

Sp YC CF GW LW WF AF VN YT NO ST AT SS

C TREE SPECIES

B COTTONWOOD B WILLOW B DEAD r• SALTCEDAR I OTHER

Sp. YC CF GW LW WF AF VN YT NO ST AT SS April 1982] ResourceUse by RiparianBirds 265

D BRANCH DIAMETER

I :1 crn D 1-2.5cm D 2.5-5cm B 7.5-10 cm D 12.5-20cm B >20cm Sp. YC CF GW LW WF AF VN YT NO ST AT SS

E TREE PORTION

I INNER D OUTER

Sp. YC CF GW LW WF AF VN YT NO ST AT SS

F SUBSTRATE

• BARKLEAF

• AIRROUND D FLOWER/ FRUIT [•] WATER Sp. YC CF GW LW WF AF VN YT NO ST AT SS Fig. 1. Foragingbehavior of 12 riparian birds for six measures(A-F). Barsindicate percentage of obser- vations.Sample size (N) appliesto all measures.Species codes from Table 1. 266 ROSENSERO,OHMART, AND ANDERSON [Auk, Vol. 99

TAULE3. A demonstrationof the lack of significant differencesamong 12 riparian birds on three measures of foragingbehavior (abovediagonal) and three measuresof foragingspace (below diagonal).M = meth- od, B = branch size, S = substrate,H = height, T = tree species,P = tree portion. Speciescodes from Table 1.

Species YC CF GW LW WF AF VN YT NO ST AT SS YC ------BS- BS- S-- BS- BS- B CF -- MB ...... GW -- P-- S LW HT- P-- HTP WF HTP -- H-- H-- MBS B-- B-- B AF P .... HP- B-- B-- BS .... VN HTP ------HP- HP- MB- MBS -- M-- M-- YT P ..... HT- P-- MBS -- M-- M-- NO HTP -- H-- H-- HTP HP- HP- P-- -- M-- M-- ST HTP -- -- H-- HTP HTP HP- P• HTP -- -- AT -- HP- P--- P .... T------M- SS P------P--- P--- P-- P-- P--- P-- H--

Song Sparrowsrarely used branches;their use of this measurewas not tested.

pairs with the highest calculatedoverlaps in ing methodsbut useddifferent substrates; the each matrix (usually>0.900) did not differ sig- SongSparrow often gleanedinsects from the nificantly. All differencesstated below were surfaceof shallowstanding water, whereasthe significant(P < 0.05). Abert's Towhee gleanedfrom dry ground and As described, overlap matrices for the six litter. foraging measureswere combined into two With respectto foraging space, again the sets of overlap patterns: one for foraging be- greatestoverlap was within the canopy-feed- havior and one for foragingspace (Fig. 2). In ing guild. The Northern Oriole, Wied's Crest- general,the clusteringof speciesparalleled the ed Flycatcher(Myiarchus tyrannulus), and Yel- previous separationinto three guilds, although low-billed Cuckoodid not differ in any spatial membership by several specieswas now less measure.The Verdin and SummerTanager dif- clear. fered only in regard to tree speciesuse. The Behaviorally, the most similar specieswere two flycatchersoverlapped less spatially than within the canopy-feedingguild. The two con- behaviorally,with the Ash-throatedFlycatcher generic flycatcherswere statisticallyinsepara- (Myiarchuscinerascens) foraging more in salt ble in all three behavioral measures. Both fed cedar. Overall, there were fewer differences in primarily by snatching insects from foliage spatialthan in behavioralmeasures within this during short flights through the canopy. The guild (43 versus29), and there was a nonsig- Northern Oriole, Verdin, and Common Yel- nificant tendencyfor these to be complemen- lowthroat were specialistleafgleaners and did tary (r = -0.323, P • 0.05). The CommonYel- not differ in eight of nine possible compari- lowthroat was widely separatedspatially from sons. The Yellow-billed Cuckoo was interme- the other foliage feeders and from all other diate in behavior between these two groups, speciesas well. whereas the Summer Tanager differed from all The two woodpeckerswere inseparablein all species except the cuckoo on all measures. spatial measures.The Common Flicker'sfor- The Gila (Melanerpesuropygialis) and Lad- aging spacewas most similar to the Abert's der-backed woodpeckers (Picoidesscalaris) Towhee's, with no differences in foraging were similar only in their use of bark as a sub- heightor use of the inner portionof trees.The strate; they differed with respectto foraging SongSparrow and Abert'sTowhee did not dif- method and branch size. The Common Flicker fer in foraging-height distributionsand were differed from the Ladder-backed Woodpecker widely separatedfrom most other species. in all measures but differed from the Gila Morphology.--Table 4 lists the means and Woodpecker only with respect to substrate. standarddeviations of all morphologicalmea- The two groundforagers used identicalforag- suresfor the 13 speciesused in this study. The April 1982] ResourceUse by RiparianBirds 267

NO VN YT YC WF AF ST AT SS LW GW CF AF VN ST NO WF YC GW LW YT CF AT S$

809

L .245

137

a BEHAVIOR 1 b. SPACE Fig. 2. Foragingsimilarity among12 riparian birds. Behavior(a) = averageof overlapsin method,branch size, and substrate.Space (b) = averageof overlapsin tree speciesand tree portionmultiplied by overlap in height (see text). Speciescodes from Table 1.

smallestspecies was the Verdin (6.5 g, 8.4-ram Overall morphologicalsimilarity based on bill length), and the largestwas the Common ratios of the measuresis illustratedin Fig. 3. Flicker (110 g, 32.7-mm bill). Seven species In general,species were arrayedby body size were between 27 and 46 g, with bill lengths and bill length, forming three major size varying from 15.7 mm to 23.8 min. groups. The seven medium-sized speciesdif-

TABLE4. Morphologicalmeans of 13 speciesof riparianbirds. Standard deviations are in parentheses.Bill lengthis exposedculmen; width and depth measuredat nares.Species codes from Table 1.

Bill Weight Wing Length Width Depth Species n (g) (mm) (mm) (mm) (mm) YC 6 66 (4.8) 149 (4.0) 29.3 (0.9) 7.8 (0.3) 8.9 (0.8) CF 2 110 (2.6) 145 (3.1) 32.7 (2.7) 9.2 (0.1) 8.0 (0.3) GW 11 62 (6.9) 128 (1.8) 27.4 (2.8) 7.2 (0.6) 7.3 (0.5) LW 2 33 (3.0) 106 (8.5) 20.4 (0.6) 6.5 (0.2) 5.9 (0.2) WE 11 38 (11.6) 108 (5.8) 23.8 (2.5) 9.7 (0.6) 8.1 (0.6) AF 13 27 (1.8) 93 (3.4) 19.0 (0.7) 7.5 (0.4) 6.3 (0.3) VN 12 6.5 (0.4) 49 (3.1) 8.4 (0.4) 3.7 (0.4) 3.9 (0.4) YT 10 10 (0.7) 54 (1.3) 11.9 (0.7) 3.6 (0.3) 3.6 (0.4) NO 10 31 (2.7) 95 (3.0) 18.1 (0.5) 5.7 (0.5) 6.9 (0.4) ST 7 33 (3.3) 100 (3.7) 20.3 (0.8) 9.4 (0.4) 9.7 (0.4) BG 9 27 (1.9) 85 (3.6) 16.3 (0.9) 8.5 (0.8) 11.3 (0.8) AT 10 46 (3.8) 89 (2.8) 15.7 (0.4) 6.7 (0.5) 8.8 (0.4) SS 10 19 (2.4) 64 (2.1) 11.9 (0.5) 5.4 (0.4) 6.2 (0.3) 268 ROSENBERG,OHMART, AND ANDERSON [Auk, Vol. 99

CF YC GW LW AF NO ST WF AT BG YT VN SS low-billed Cuckoo, ate very small insects as well. There was much overlap among both the larger and smallerbird specieswith respectto prey size (Fig. 5a). Within eachgroup, the can- opy-foragingsummer visitors were nearly in- separable statistically (Table 5). Among per- .911 manent residentsand the other two foraging •5 guilds, the only nonsignificantdifference was between the Gila Woodpeckerand the Abert's Towhee. Distribution of prey types in the diets of the .508 13 speciesis shown in Fig. 6. Eight species preyed heavily on cicadas.Among thesebirds, grasshopperswere an important food for the Yellow-billed Cuckoo, Blue Grosbeak, Abert's Towhee, and possibly the Northern Oriole. Summer Tanagerswere the only birds to feed on bees and wasps, and Gila Woodpeckersate many ants. Abert's Towhees fed heavily on beetles, and the two flycatcherstook a wide variety of food items in addition to cicadas. The smaller bird speciesfed on a variety of soft-bodiedprey, primarily spidersand larvae of several insect orders. The small samplesof Common Flicker and Ladder-backed Wood- Fig. 3. Morphologicalsimilarity among 13 ripar- ian birds. Representsaverage of ratiosof five mea- pecker stomachssuggest that their diets may sures for each speciespair (Table 4). Speciescodes have been very different from most other from Table 1. species. Of the three species sometimes re- ferred to as granivores,the SongSparrow and Abert's Towhee ate very few seeds, and the nine stomachs of Blue Grosbeaks contained fered primarily with respectto bill width and none. depth. The Ladder-backed Woodpecker, Overlaps in diet are illustrated in Fig. 5b for Northern Oriole, and Ash-throated Flycatcher percent-frequencydistributions. Among the had relatively narrow or flattened bills; the eight speciesthat preyed upon cicadas, 23 of Wied's Crested Flycatcher and Summer Tana- 28 pairs did not differ significantlyin diet (Ta- ger had the widest bills, and the Abert's Tow- ble 5). Among the remaining five species,only hee and Blue Grosbeak had very deep as well two of 10 speciespairs were different: the Ver- as wide bills. din with both the Common Yellowthroat and Diet.--Contents of 106 stomachs,represent- Song Sparrow. ing midsummerdiets of 13 species,were used for this study. The small sample size for the DISCUSSION Common Flicker and Ladder-backed Wood- peckerprevents conclusive analysis; they were Speciesusage patterns indicated a high de- included for completeness,however. gree of overlap with respectto both spaceand Distribution of size classes in the diets is food. Although diet sampleswere small, these shown in Fig. 4. The largest and smallest are believed to be adequate for most species, species (Common Flicker and Verdin) were based on the small variation in individual diets specialistson the smallestsize class,illustrat- and their correspondencewith field observa- ing the dangerof inferring prey size indirectly tions. Wiens and Rotenberry (1979) similarly from bill measurements.The most frequently justified small samplesas being reasonablein- eaten size classamong the larger specieswas ventories of avian diets. 21-30 mm. All these species, except the Yel- The failure of avian species to avoid or ex- April1982] ResourceUse by Riparian Birds 269

31/6 225/2 76/13 75/2 26/9 26/13 22• 3•8 21/9 22/7 19/7 31/10 79/11 lO0

75

ß 50 o

25

Sp. YG OF GVV LW WF AF VN YT NO ST BG AT Fig.4. Preysize distributions of 13riparian birds. N representsnumber of fooditems/number of stom- achs. Speciescodes from Table 1. clude one another spatiallysuggests that, for partitioning was unnecessary.An assumption each species, microhabitat selection on the of interspecificcompetition is not necessaryto study plot was independent of the other explainthe observedspatial distribution. speciespresent. The lack of distinct microhab- In nearlyevery study investigating actual re- itat preferencesamong these species(Rosen- source use among a community of breeding berg et al. MS) further suggeststhat spatial birds, a high level of dietary overlaphas been

YC NO AF WF AT BG ST GW LW SS YT VN,CF WF AF NO GW ST CK AT BG CF LW YT VN 9994

929

300

2

a PREY SIZE b PREY TYPE Fig. 5. Similarityin preysize (a) and diet (b) among13 riparianbirds. Overlaps based on percent- frequencydistributions. Species codes from Table 1. 270 ROSltI•BEI•G,OIlMARt, AI•D AI•l•ERSOI• [Auk, Vol. 99

TAnLE5. Nonsignificantdifferences in prey size (S) and prey type (T) among 13 riparian birds. Species codes from Table 1.

Species YC CF GW LW WF AF VN YT NO ST BG AT SS

YC ------S ...... CF ..... S ...... GW .... S -- -- S -- S S -- LW ..... S ...... WF T -- T -- S -- -- S S S S -- AF -- -- T -- T -- -- S S S S -- VN ...... S .... S

NO T -- T -- T T -- -- S S S -- ST T ------T T -- -- T S S -- BG T ------T T -- -- T T S -- AT T ------T T -- -- T T T -- SS ...... T .....

discovered.This is true for suchdiverse groups (1981). These consisted of successiveweekly as shorebirds (Holmes and Pitelka 1968), black- counts of cicada exuvia in cottonwood-willow- birds (Orians and Horn 1969), finches(Pulliam salt cedarhabitat along the San Pedro River in and Enders 1971), owls (Herrera and Hiraldo southeastern Arizona in summer 1978. The 1976), grasslandinsectivores (Wiens and Ro- speciesof cicada(Diceroprocta apache) was the tenberry 1979), and riparian insectivores(this same as on our study plot, and qualitative study). Thesefindings are in contrastto theo- comparisonsof habitat and cicadapopulations retical predictions of resourceallocation and (noise levels)suggest that thesedata are rele- limiting similarity and would most likely be vant to the present study. consideredexceptions to the "generalrule" by Cicada numbers peaked at nearly 700,000per these models. As pointed out by Wiens (1977), 40 ha on 21 July and declined steadily until late however, there is no justificationin assuming August (Fig. 7). The averagenumber of adult a priori that resource-basedcompetition exists cicadasemerging through the samplingperiod or even that a community is at equilibrium was 250,000per 40 ha per week. We feel that with respectto resourceuse. these numbers, superimposedon the sweep- It is often suggestedthat high dietaryover- sample data, indicate the availability of large lap, especially among breeding birds, most insects to birds in midsummer. often resultsfrom temporarysuperabundances Daily expenditures of birds were estimated of food. The magnitude and duration of these using the equation DDE = 11.87 (massø'6ø8) superabundancesare rarely measured, how- (Walsberg 1980) for all species that preyed ever. heavily on cicadas(Table 6). The energy need- Arthropod sweep samples from the study ed to maintain this segment of the bird com- area (Anderson and Ohmart unpubl. data) in- munity was estimatedas 200,019kJ per 40 ha dicate a peak in biomassand numbers in May. per week. Assumingthe metabolizableenergy This peak is made up of mostly tiny insects from insectsto be 4.94 kJ/g(Ricklefs 1974) and (e.g. Cicadellidae),however, that are appar- using a weight of 1.5 g for a live cicadaand a ently unrepresented in the avian diets. Al- 0.92 efficiencyfor birds (Thompsonand Grant though a July-August peak in larger insects 1968), the energy gained per cicada was cal- such as Orthopteranscan be shown, abundant culatedas 6.82 kJ. If all thesespecies depended prey such as Cicadidae, ants, and termites entirely on cicadasfor food (a conservativeas- were generallyabsent from sweepsamples. sumption), then approximately 29,328 cicadas Becausecicadas were so important to a large were needed each week. This is far fewer than segmentof the bird community,including sev- were presentin any week during which stom- eral similarly sized and potentially competing ach samples were taken. Becausecicadas are species, the question of food abundance was conspicuousand easy to catchby a variety of addressedusing data from Glinski and Ohmart methods, it is reasonable to assume that most April 1982] ResourceUse by RiparianBirds 271

100

75

'• 5o

25

SPECIES YC CF GW LW WF AF VN YT NO ST BG AT SS (N) 6 2 13 2 9 13 4 10 9 7 9 11 11

ß Cicada [] Ant and Termite [] Odonata I• Hymenoptera [] Grasshopper [] Hemlptera [• Lepldoptera [] Dermaptera [] Mantls [] Coleoptera [] Spider [] Seed I•] Larvae [] Diptera • Homoptera Fig. 6. Dietsof 13riparian birds in summer.Percent-volume on left;percent-frequency on right. Species codes from Table 1. flying adult cicadaswere readily available to aging, respectively.These two speciescould birds. have "avoided competition" with the similar- Although the above calculationis crude, and sized coexistingspecies but instead shared in the additional energy demandsof growing the exploitationof the most abundant prey. A young need to be considered,the magnitude similar pattern has been reported for breeding of the cicadasurplus and the additional abun- shorebirdsdiffering greatly in body size and dance of grasshoppersand other prey justify bill morphology(Baker 1977). the conclusionof a nonlimiting food supply. Thus, measured niche variables such as bill This is consistentwith Rotenberry's(1980b) conclusionfrom a morecomplete bioenergetics study of shrubsteppebirds. Furthermore, ci- cada irruptions are of annual occurrencein this region, suggestingthat this surplusis predict- 600- able. All speciesexploiting cicadas fledged either their first or secondbroods in Julywhen food was potentially most abundant. This is indirect evidence of a responseto a resourcepeak. In addition, environmentalconditions were pos- sibly most stressfulat that time, and most other speciesof the region breed much earlier. The consequenceof this unlimited resource wasthe convergencein diet of speciesdiffering greatly in behavior and morphology. Most strikingwas the predationon cicadasby Gila Fig. 7. Density of emerging cicadas in cotton- Woodpeckers and Abert's Towhees, two wood-salt cedar habitat during summer 1978. Data speciesspecialized for bark and ground for- from Glinski and Ohmart (1981). 272 ROSENSEaO,OHMART, AND ANVERSON [Auk, Vol. 99

TABLE6. Energeticsof eight medium-sized riparian The remaining smaller species are rather insectivores. DEE = 11.87 massø.6øs (Walsberg widely separatedecologically, and an apparent 1980). Population energy expenditure (PEE)= DEE-density. Speciescodes, densities from Table peak in small arthropod biomassseems unex- 1; body mass from Table 4. ploited. In this region, somesmall insectivores have experienced local extinctions [Yellow DEE PEE Warbler (Dendroicapetechia), Bell's Vireo (Vir- Species (kJ/day) (kJ/week/40ha) eo bellii)] or are near the periphery of their YC 152 3O,775 breeding range [Bewick's Wren (Thryornanes GW 146 38,836 bewickii)].Thus, this segmentof the commu- WF 108 16,694 nity is depauperatefor reasonsprobably un- AF 88 3,080 NO 96 46,924 related to food supply, and discussionof these ST 100 16,712 species can neither refute nor support the BG 88 1,848 above conclusions. AT 122 45,151 Finally, it should be noted that this discus- Community 200,019 sion has been based only on a period of re- source superabundance. Indeed, theory pre- dicts the consequencesof such competitive release situations (MacArthur 1972). What is of shape and foraging behavior may not be true greaterinterest is that this may not be a unique or even a rare situation. The evolution of di- predictorsof actual resourceuse within a com- munity, and all community studies that have verse breeding bird communities in response made such inferences may have drawn spu- to seasonallysuperabundant and predictable rious conclusions.For example,there is no jus- resourcesmay be quite common,especially in tification for assumingthat coexistencewithin temperateregions. In fact, many communities a communityis due to any of the observeddif- studied in light of competition models (e.g., ferencesamong the speciespresent. If each in- MacArthur 1958, Cody 1974)may be examples of such evolution (Rabenold 1978). dividual bird is opportunisticallyexploiting its Given the abundanceand variety of theoret- environment as efficiently as is necessaryto survive and reproduce, irrespective of the ical considerationsof community organization, it is clear that empirical evidence lags far be- presenceof other species,then differencesmay exist within a community, and local interspe- hind. Specifically,the exact dietary relation- cific competition need not be invoked as a ships of speciesassemblages and the relative seasonalresource productivity of various hab- cause. Indeed suchopportunism was implicat- itats need to be measured and not inferred. We ed by Rotenberry (1980a) as determining rela- maintain that such a sound empirical founda- tionships among a simple community of arid shrubsteppebirds. tion is essential if speculationabout commu- This is not to say that resourcelimitation and nity patterns is to be more meaningful. competitioncould not have been the selection pressuresthat affectedthe species-specificdif- ACKNOWLEDGMENTS ferences observed in this and other commu- We are grateful to Tim Brush and Alton Higgins nities. Selection pressures from competition for helping significantlywith field data collections. may only be intermittently important to pres- The expertiseof JeanneWielgus, Bill Warner, and Ed ent-day populations(Wiens 1977).Such selec- Minch were invaluable to the analysisof insect and tion, however, may still be reflectedin current bird stomach samples. Richard Glinski provided distribution and behavioral repertoiresof bird data on cicadas. Kurt Webb was our needed liaison species. The important distinction to be made with the computer,Cindy D. Zisner typed the manu- is between those factors governing the pres- script, and Kay Cunningham, Duna Strachan, and Elaine Hassinger helped with the figures. Critical ence of species over large geographic areas advice on analysesand early draftsof the paper was through evolutionary time and those that given by Jim Collins, SusanM. Cook, JaneR. Dur- structure short-term, sympatric species inter- ham, Andrew Laurenzi, Julie K. Meents, Jake Rice, actions. and ScottTerrill. This researchwas largely supported Until now, we have concentratedheavily on by the U.S. Bureau of Reclamation Contract No. the cicada-eating subset of this community. 7-07-30-V0009. April 1982] ResourceUse by RiparianBirds 273

LITERATURE CITED ORIANS, G. H., & H. S. HORN. 1969. Overlap in foods of four speciesof blackbirds in the pot- ANDERSON,B. W., & R. D. OHMART. 1977ß Wildlife holes of central Washingtonß Ecology 50: use and densities report of birds and mammals 930-938ß in the lower ColoradoRiver valleyßU.S. Bureau PIANKA,E. R. 1974. Niche overlap and diffuse com- of Reclamation Contract No. 14-06-300-2531 petitionß Proc. Natl. Acad. Sci. USA 71: Rept. 2141-2145ß ANDREWARTHA,H. G., & L. C. BIRCHß 1954. The PtELOU, E. C. 1977. Mathematical ecologyß New distribution and abundance of animalsß Chica- York, Wiley-Interscience,Inc. go, Univ. ChicagoPressß PUtLIAU, H. R., & P. ENDERS.1971. The feeding BAKER,M. C. 1977. Shorebird food habits in the east ecologyof five sympatricfinch speciesßEcology Canadian arcticß Condor 79: 56•52. 52: 577-586ß CAROTHERS,S. W., R. R. JOHNSON,& S. W. AITCHI- RABENOtD,K. N. 1978. Foraging strategies,diver- SON. 1974. Population structure and social or- sity, and seasonalityin Appalachian spruce-fir ganization of Southwestern riparian birdsß bird communitiesßEcol. Monogr. 48: 397-424. Amer. Zool. 14: 97-108. RICKLEFS,R. 1974. EcologyßNewton, Massachu- COD¾,M. L. 1974. Competition and the structure of setts, Chiron Pressß bird communitiesß Princeton, New Jersey, Root, R. B. 1967. The niche exploitation patterns of Princeton Univ. Pressß the Blue-gray GnatcatcherßEcol. Monogr. 39: , & J. M. DIAMOND(EDs.). 1975. Ecologyand 317-350. evolution of communitiesßCambridge, Massa- ROSENBERG,K. V. 1980. Breeding bird community chusetts, Harvard Univ. Pressß organizationin a desertriparian forestßUnpub- CONNER,E. F., & D. SIMBERLOFF.1979. The assem- lished M.S. thesis, Tempe, Arizona, Arizona bly of speciescommunities: chance or compe- State Univ. tition? Ecology 60: 1132-1140. ROTENBERlly,J. T. 1980a. Dietary relationships GLINKSI, g. L., •r g. D. OHMART. 1981. Breeding among shrubsteppe passerine birds: competi- ecologyof the Mississippi Kite (Ictinia mississip- tion or opportunism in a variable environment? piensis)in Arizona. Report submitted to U.S. Ecol. Monogr. 50: 93-110ß Bureau of Reclamation in fulfillment of Contract ß 1980b. Bioenergeticsand diet in a simple No. 14-06-300-2674.Boulder City, Nevadaß communityof shrubsteppebirdsß Oecologia 46: HERRERA, C. M., & F. HIKALDO. 1976ß Food-niche 7-12. and trophic relationships among European SCHOEN•R,T. W. 1974. Resourcepartitioning in eco- Owls. Ornis Scandinavica 7: 29-41. logical communitiesßScience 185: 27-39ß HESPENHEIDE,H. A. 1975. Prey characteristicsand SOKAL,R. R., & F. J. ROHL•. 1969. Biometry. San predator niche-widthß Pp. 158-180 in Ecology Francisco, California, W. H. Freeman and Co. and evolution of communities (M. L. Cody and STAMP,N. E. 1978. Breedingbirds of riparian wood- J. M. Diamond, Eds.). Cambridge, Massachu- land in south-central Arizonaß Condor 80: 64-71ß setts, Harvard Univ. Press. THOMPSON,R. D., & C. V. GRANT. 1968. Nutritive HOLMES,R. T., & F. A. PITELKA.1968. Food overlap value of two laboratory diets for StarlingsßLab. among coexistingsandpipers on northern Alas- Anim. Care 18: 75-79ß kan tundra. Syst. Zool. 17: 305-318. WALSBE•G,G. E. 1980. Energy expenditurein free KIRK, R. E. 1968. Experimental design; procedures living birds, patterns and diversityßActa XVII for the behavioral sciencesßBelmont, California, Cong. Intern. Ornithol., Berlin: 300-305ß Brooks/Cole Publ. Co. WHITE, T. C. R. 1978. The importanceof a relative MACARTHUR,g. H. 1958. Population ecology of shortageof food in animal ecology.Oecologia some warblers of northeastern coniferous for- 33: 71-86. ests. Ecology39: 599-619ß WIENS,J. A. 1977. On competition and variable en- --. 1972. Geographicalecology. New York, Har- vironments. Amer. Sci. 65: 590-597. per and Row. , & J. T. ROTENBERRY.1979. Diet niche rela- MAY, R. g. 1975. Some notes on estimating the tionshipsamong North Americangrassland and competition matrix. Ecology56: 737-741ß shrubsteppebirds. Oecologia42: 253-292ß NIP, N.H., C. H. Hurt, J. G. JENKINS,K. STEIN- BRENNER,& D. H. BENT. 1975. Statistical pack- age for the socialsciencesß New York, McGraw- Hill, Inc. 274 ROSENBERG,OI-IMART, AND ANDERSON [Auk, Vol. 99

APPENDIX.Definitions of foraging measuresand states.

Measure State Definition

Method 1. Glean Bird perched;prey capturedfrom surface of substrate 2. Hover Bird 'in flight; prey capturedfrom surface of substrate 3. Hawk Both bird and prey in flight 4. Peck Bill struck against substrateto exposeprey below surface 5. Probe Bill enterssubstrate to captureprey below surface Foragingheight 1. Ground Ground and litter 2. 0.0--6.2m Understory 3. 6.2-12.3 m Mid-canopy 4. 12.3-21.0 m Upper-canopy Tree species 1. Cottonwood Populusfrernontii 2. Willow Salixgooddingii 3. Dead Dead cottonwood or willow 4. Salt cedar Tamarix chinensis 5. Other Honey mesquite(Prosopis glandulosa), cat- tails (Typhalatifolia), etc. Branch diameter 1. 41 cm 2. 1.0-2.5 cm 3. 2.5--5.0 cm 4. 7.5--10.0 cm 5. 12.5--20.0 cm 6. •20.0 cm Tree portion 1. Inner Larger branchesclose to and including trunk 2. Outer Smallerbranches and twigs, containing most foliage Substrate 1. Leaf 2. Bark 3. Ground 4. Air 5. Flower or fruit 6. Surface of water NESTING ECOLOGY OF ROSEATE SPOONBILLS AT NUECES BAY, TEXAS

DONALD H. WHITE, 1 CHRISTINE A. MITCHELL, 1 AND EUGENE CROMARTIE 2 •U.S. Fishand WildlifeService, Patuxent Wildlife Research Center, Gulf CoastField Station, P.O. Box 2506, Victoria, Texas 77902 USA, and zU.S. Fishand WildlifeService, Patuxent Wildlife Research Center, Laurel, Margland20708 LISA

ABSTWACT.--Weconducted a study in 1978-1980of the nestingecology of RoseateSpoon- bills (Ajaiaajaja) in a relativelypolluted environment at NuecesBay, Texas.For 154marked nests,the averageclutch size was 3.0 eggs;73% of the eggshatched, and 87% of the nests were successful(hatched at leastI young). The averagenest success rate (totalfledglings:total eggs)was 50% or 1.5 fledglingsper total nests.Incubation began the day after the first egg was laid, and the incubationperiod for each egg was 22 days. Eggshatched in the order that they were laid; the first and secondeggs hatched on consecutivedays, and the second, third, and fourth eggshatched every other day. Nest compositionand size were highly consistent,but nest placementvaried considerablyand was dependenton the vegetative configurationof the dredge-materialislands. Growth rate of nestlingsconformed to a "stan- dard" growth curve, where body weight of nestlingsat fledging equaled that of adults. There was no differencein weight gain among siblings based on actual age. Nestlings fledgedat about6 weeksof age,when featherdevelopment was complete.At fledging,the bills of juvenileshad almostreached adult width, but bill lengthwas only 67% that of adults. In general,environmental pollutants were low (<2 ppm) in spoonbilleggs, though a few eggscontained elevated concentrations of DDE (up to 15 ppm, wet weight). Someeggshells were 5% thinner than thosein museumcollections, but the degreeof thinning was not within the range known to causepopulation declines. We concludethat organochlorine pollutantsare not adverselyaffecting spoonbill reproduction at NuecesBay, Texas,though the areais surroundedby industriesand agriculturallands. Apparently, spoonbills are less sensitiveto thesecontaminants than are other aquaticspecies. Received 8 April 1981,accepted 2 September1981.

ALTHOUGHRoseate Spoonbills (Ajaia ajaja) Other than the early work of Allen (1942), probably never were abundantin Texas,they very little is known of the reproductivebiology were virtually extirpated there between 1850 of RoseateSpoonbills on the TexasGulf Coast. and 1919 becauseof encroachingcivilization A preliminary study in 1977 indicated that and exploitationby the millinery trade (Allen DDE (the major metabolite of technicalDDT) 1942). With the advent of laws to protect co- in a few spoonbilleggs from NuecesBay was lonial nestingbirds, however, RoseateSpoon- within the range known to causereproductive bill numbershave graduallyincreased in Texas problemsin other avian species(Stickel 1973, to about 2,500 breeding pairs (Blacklocket al. 1975).Therefore, we began a 3-yr study in 1978 1979).About 150-200pairs of spoonbillshave to learn more of the nestingecology of Roseate nested in Nueces Bay at Corpus Christi, Texas Spoonbillsand to evaluatethe effectsof or- in recent years. In addition, 17 other species ganochlorinepollutants on reproduction. of marine and estuarine birds regularly nest there. NuecesBay is bounded on the southby METHODS an industrial complexincluding oil refineries, chemical plants, and a zinc-smelting plant; Our study area in NuecesBay (27ø52'N,97ø30'W) (Fig. 1) consistedof threeunnamed dredge-material other industries and agriculturallands border islands from I to 2 ha in size, composed mainly much of the northern shore. Thus, the poten- of oystershell and sand. Low vegetation(<0.5 m), tial for pollution problemsis great in Nueces dominatedby coastbecopa (Bacopa monnieri), glass- Bay becauseof industrialeffluents and pesti- wort (Salicornia),and sea oxeye (Borrichiafrutescens), cide runoff from surrounding crop lands. covered most of each island, and some small (1-3 m)

275 TheAuk 99:275-284. April 1982 276 WHITE, MITCHELL,AND CROMARTIE [Auk, Vol. 99

-.-':.'...... Nueces Bayß Pehc a•n Is

Corpus Christi Bay

Kilometers

Fiõ. 1. RoseateSpoonbill study area, NuecesBay, Texas. trees and shrubs [retama (Parkinsonia aculeata), fore fledging),we countedall the young that were paloverde (Cercidiurnmacrum), saltbush (Baccharis presentand the eggsin neststhat failed to hatch. angustifolia)]were present on two of them. Associ- Thus, we comparednest successon a relatively un- ated nesting birds were Great Egrets (Casrnerodius disturbed island with that on the three islands that albus), Snowy Egrets (Egrettathula), Cattle Egrets were visited weekly. (Bubulcusibis), Great Blue Herons (Ardea herodias), We collectedthe intact eggsthat failed to hatch in Louisiana Herons (Hydranassa tricolor), Black- our marked nestsand one egg eachfrom a seriesof crowned Night Herons (Nycticoraxnycticorax), and clutcheson islandsadjacent to our study sites for Laughing Gulls (Larusatricilla ). organochlorineand lipid analyses.Organochlorine In the breedingseasons of 1978-1980we marked and lipid analyseswere conductedat the Patuxent spoonbillnests with numberedstakes at the onset Wildlife Research Center, Laurel, Maryland of egglaying;eggs within nestswere markedwith (1977-1979 samples)and at Raltech Scientific Ser- the assigned nest number and the egg sequence vices, Madison, Wisconsin (1980 samples). A de- number if known. Nests were visited weekly there- tailed descriptionof Patuxent'sanalytical methods after, and observations and measurements relating is included in Cromartie et al. (1975) and Kaiser et to nestingecology were recorded.During the week al. (1980).Similar analytical procedures were usedat of peak hatchingin 1980,we visited nestsevery other Raltech. Residues in 5% of the sampleswere con- day, exceptonce on 2 consecutivedays, to mark firmed by massspectrometry. Quantification limits hatchlingsso that body measurementsof known-age were 0.1 ppm for organochlorinepesticides and 0.5 nestlingscould be recordedthroughout the early de- ppm for polychlorinatedbiphenyls (PCBs) on a wet velopmentalstage. Body weights were taken with a weight basis.We comparedour eggshellthicknesses dietetic scaleaccurate to +2 g, and feather develop- with those of prepesticide-eraeggs from the Texas ment was documentedby taking colorslides of the coast housed at the Welder Wildlife Foundation to various age groups.Our data for feather develop- determine whether or not shell thinning had oc- ment from hatchingto fledgingof known-agespoon- curred. Eggshellswe collectedwere dried at room bill chicksfollow the terminologyof Welty (1962). temperaturefor at least 30 days, then weighed and In 1980 we estimated productivity or nest success measured with the shell membranes intact. Thick- (totalfledglings: total eggs) on an islandadjacent to nessmeasurements of all eggswere determined with our study sites in NuecesBay as a measureof in- a modified Starrett Model 1010 M micrometer accu- vestigatorbias. This island was visited only twice. rate to 0.01 mm. Externalegg dimensionswere mea- On the firstvisit (earlyincubation stage), we counted suredwith calipers,and volumeswere estimatedfol- all nestsand the eggs.On the secondvisit (just be- lowing Hoyt (1979). April 1982] SpoonbillNesting Ecology 277

TABLE 1. Roseate Spoonbill nest measurements, ter months each year, but by mid-October Nueces Bay, Texas. most of the birds had dispersed, presumably southward into Mexico (Allen 1942). Parameter n œ + SE Range Nest compositionand placement.--Spoonbills Nest width (cm) 16 55.6 + 1.8 43-71 constructedtheir nestsatop low vegetationand Nest depth (cm) 10 10.7 + 0.3 10-13 in trees and shrubs with plant material that Nest height (crn) (atop low vegetation) 28 23.9 + 2.4 10-51 was abundant on the nesting islandsor nearby Nest height (cm) shoreline. The base of each nest consisted of (in trees/shrubs) 12 70.6 + 13.3 33-168 large dead twigs of saltbush, paloverde, and retama loosely fitted together to form a crude platform about 6 cm deep. Upon this platform, live and dead stemsof seaoxeye, saltbush, and RESULTS AND DISCUSSION grasses (Paspalum) were intertwined into a Nestingchronology.--Adult spoonbillsin full coarselining about 5 cm deep. The activities breeding plumage began congregatingon the of the birds on the nest eventually formed a Nueces Bay dredge islands around mid-March wide, shallow depressionin which the eggs each year; by late March, 100 or more adults were laid. The external dimensions of nests were present. Birds were not paired at this (width and depth) were consistent,as seen by time, and they loafed in large groups on the the small standard errors in Table 1. A brief shores of the islands. By mid-April, several description of spoonbill nests is given in Bent hundred adults were present in the area, and (1963) and Oberholser (1974), but no mention some pairing had occurred. We never saw is made of specific vegetative components or breeding spoonbills in other than full adult nest measurements. Nest placement on the plumage, though Dunstan (1976)reported sub- breeding islands was determined by the veg- adult-plumaged birds nesting in Florida. Ru- etative profile; spoonbills nested where low dimentary nestswere found on 5 May in 1978, vegetation was most prevalent or in and and egglaying began around 20 May; in 1979 around trees and shrubs. The height of nests and 1980 the averagenest building and egg- atop low vegetation was much less variable laying dateswere 15 April and 25 April. than that of nests in trees and shrubs (Table 1). Allen (1942)estimated the incubationperiod This was due to the uniform configurationof to be 23-24 days for Texas spoonbills, but he low vegetation, i.e. density and height, in con- was unsure of the deposition rate and when trast to the highly variable arrangement of incubation began. Our data from marked eggs limbs in trees and shrubs. in 29 nests indicate that laying occurredat the Nest success.--Clutchsize in 1978 was sig- rate of one egg every other day, with no vari- nificantly (P • 0.001, analysis of variance) ation from this pattern. The time in the nest lower than in 1979 and 1980 because of fewer until hatching was 23 days for the first egg and 4-egg clutchesand more 2-egg clutchesin the 22 days for the second,third, and fourth eggs. 1978 sample(Table 2). Six nests(4%) contained Becauseincubation began the day after the first 1 egg,20 (13%) contained2 eggs,92 (60%) con- egg was laid, the incubation period for each tained 3 eggs,34 (22%) contained4 eggs, and eggwas 22 days.Eggs hatched in the order that 2 (1%) contained5 eggs.Allen (1942)recorded they were laid; the first and second eggs a lower average clutch size of 2.6 for a Texas hatched on consecutivedays, and the hatching spoonbill colony in 1940, but the proportion of interval between the second, third, and fourth 3-egg clutchesin his 1-yr study was identical eggswas every other day. to ours. By early July, young spoonbillswere able to Hatching success varied significantly fly severalhundred meters to adjacentislands; (P < 0.02, Chi-square test) among years, rang- thus, we estimated that fledging occurred at ing from58% to 88% (Table2). The 58% hatch- about 6 weeks of age. By early August, adults ing successrecorded in 1980 was low when and young began to dispersefrom the islands comparedto the successrates of the other 2 yr, and by September few birds of any age re- but therewas no differencein hatchingsuccess mained there. We observed small mixed-age between 1979 and 1980. We were able to ac- flocks of spoonbillsin Nueces Bay and other count for 92% of the eggs that were laid in areasof the south Texas coastduring the win- 1980. Lowered hatching in 1980was due main- 278 WHITE, MITCHELL,AND CROMARTIE [Auk, Vol. 99

TABLE2. Summaryof reproductivesuccess of RoseateSpoonbills at NuecesBay, Texas, 1978-1980.

1978 1979 1980 Total Parameter (n = 72) (n = 35) (n = 47) (n = 154)

Clutch size Mean 2.8 3.3 3.3 3.0 Mode 3 3 3 3 Range 1-4 1-5 2-5 1-5 Survival rates Percentagesuccessful nests a 88 97 79 87 Percentageeggs hatched 79 88 58 73 Eggs hatched/nest 2.2 2.9 1.9 2.2 Young to 1 week/nest 2.1 2.7 1.4 2.0 Young to 2 weeks/nest 1.9 2.3 1.3 1.8 Nest success Rate (%)b 54 52 36 50 Fledglings/totalnests 1.5 1.7 1.2 1.5 Percentagenests that hatched at least one young. Total fledglings/totaleggs. ly to death of embryos during incubation 1979, and 1980, respectively(• = 32%; n = (31%). In 1978this categoryaccounted for 7% 41). Mortality of nestlingseach year was great- of the egg lossand 1% in 1979. Of the embry- est during the first 2 weeksafter hatching; 14% onic deaths,78% occurredin the late stagesof of the nestlingswere lost during this period in development. There was no evidence that the 1978, 21% in 1979, and 32% in 1980 (• = 18%, last egg in the clutch contributed more to the n = 23). Half of the birds that died within the overall mortality than earlier eggs, because first 2 weeks during the study period were mortality was evenly divided among eggs of a found dead in or near their nests; the other half clutch. The combined proportion of eggs that disappearedbetween our visits to the colony, disappeared and rolled out of nests each year and their fate is unknown. Only once did we was similar, averaging 13%. Most of the see evidence of predation; two nests in 1980 spoonbill pairs were able to hatch at least one contained the partial remains of severalhatch- young, as evidencedby the high percentageof lings that appearedto have been removedfrom successfulnests for each year (Table 2). There the eggs as they were pipping. were fewer (P < 0.005, Chi-square test) suc- Spoonbills fledged fewer (P < 0.01, Chi- cessfulnests in 1980 than in the other 2 yr be- squaretest) young in 1980than in 1978 or 1979 cause of lowered hatchability in 1980; no dif- becauseof lowered hatchability in 1980 (Table ference (P > 0.05) existed between 1978 and 2); there was no difference (P > 0.05) in pro- 1979. Clutch size had no significant effect ductivity between 1978 and 1979. Though (P > 0.05, Chi-square test) on the ability of hatchability was lower in 1980, the adults lost spoonbillsto hatch at least one young: 100% only 0.7 young per nest that did hatch; this of the pairs with 1-egg clutches,80% with 2- was identical to the average loss per nest in egg clutches,87% with 3-egg clutches,88% 1978 but much lower than the 1.2 young per with 4-egg clutches, and 100% with 5-egg nest that failed to fledge in 1979(Table 2). clutcheshatched at least one young. In addi- Nest success in 1980 on the island we visited tion, spoonbillshaving various clutchsizes did only twice (see Methods) was estimated at 1.3 not differ significantly (P > 0.05) in their abil- fledglings per nest (n = 18 nests), similar to ity to raise at least one young to 2 weeks of the 1.2 fledglingsper nest producedin 1980on age: 100% of those with 1 egg, 89% with 2 our more frequently visited islands. Because eggs, 93% with 3 eggs, 87% with 4 eggs, and there was no difference (P > 0.05) in nest suc- 100% with 5 eggshad at leastone live nestling cessrates between the two areas, our frequent 2 weeks after hatching. visits to a portion of the breeding colony in Total mortality of nestlingsfrom hatching to 1980 did not appear to impair reproduction. fledging was 32%, 41%, and 37% for 1978, Growth and developmentof young.--The Ro- April1982] SpoonbillNesting Ecology 279

¸ ¸ ¸ 0 0 0 0 0 280 WHITE,MITCHELL, A•V CROMARTIE [Auk,Vol. 99

TABLE4. Bill measurementsof known-ageRoseate Spoonbill chicks and adultmuseum specimens.

Cumulative Cumulative Age (days) n Length (mm) increase(%) Breadth(mm) a increase(%) I 2 20.0 + 0.0 b 12 8.8 + 0.0 15 9 4 38.5 + 1.9 23 11.8 + 0.8 22 12 3 48.3 + 1.7 28 14.3 + 0.3 27 16 6 61.6 + 1.2 36 19.5 + 0.5 37 17 3 66.0 + 2.7 39 22.3 + 0.9 42 19 2 67.0 + 3.0 39 22.0 + 1.0 42 39 3 114.3 + 1.3 67 44.7 + 0.9 85 56c 1 124 73 50 95 Adult '• 8 170.0 + 2.4 100 52.7 + 1.2 100

Breadthmeasurements were takenat widestpart of bill near tip. b• -+SE. Measurementswere taken from an individualapproximately 8 weeksold founddead on studyisland in 1980. Museumspecimens at theNational Museum of NaturalHistory, Washington, D.C. seate Spoonbill is a semi-altricial species.The caudaltracts are light pink; the spinal, fernoral, skin, mandibles, and legs of newly hatched ventral, and capital tracts are white. The tips chicks are a uniform, deep salmon pink; the of the primaryfeathers are blackish at fledging, feather tracts are covered with thick white first appearingabout 28 days after hatching. down that becomesdry and fluffy about 12 h RoseateSpoonbills do not usually attain full after hatching, giving the appearancethat the adult plumage until they are 3 yr old, but entire body is covered with down. About 2 plumagedevelopment may varywidely among days after hatching, the legs and feet begin to individuals (Allen 1942). darken, so that by fledging, they generallyare The bills of newly hatchedspoonbill chicks a dark gray-brown.The bill remainsbasically are not spatulate as in the adults; instead, the flesh-coloreduntil fledging, though there is a chickshave short stubby bills that are soft and darkeningfrom the mid-bill to the basebegin- leathery.At 9 daysof age, the bill near the tip ning about 5 days after hatching. The eyes beginsto broadenand flatten;at 16 daysit be- open when the chicksare about 2 days old; at gins to take on the characteristicform of the this time the iris is entirely black and remains adult. The cumulative increase (% of adult so until fledging. Our accountof eye and bill size) in length and breadth is almostidentical color in nestlingsdiffers somewhatfrom that up to about19 daysof age(Table 4). At 39 days, of Allen (1942), who stated that the bills of late however, when the birds are near fledging (6 nestlingsare light yellow and the irisesare yel- weeks), there is a significant(P < 0.01, t-test) low. bill growth differential; the breadth is 85% of Allen (1942) and Bent (1963) discussed at the adults' but the length is only 67%. This length plumagesand molts in RoseateSpoon- growth differentialapparently continues after bills afterfledging, but little attentionwas giv- the chicksfledge; the bill of one freshly dead en to feather development in nestlings. specimenabout 8 weeksold had grownto 95% Sheathsof the flight feathersand their coverts of the adults' in breadth, but to only 73% in are the first to emerge, beginning around day length (Table 4). 5 (Table3). The rest of the body remainsmostly At hatching, chicksweighed an averageof down-covereduntil about day 21. At this time, 50 g and gained an averageof only 9 g for the the sheathsof the alar tract begin to rupture, next 3 days (Fig. 2); on day 4, however, weight and feather sheaths of the other major tracts beganto increaserapidly. By day 16 the chicks begin to emergeand rupture. Feathergrowth had reachedabout half the adult weight (Ob- continues steadily for the next 3 weeks, and erholser1974, J. Remsenpers. comm.). Because development is complete about 42 days after nestling weights were not obtained beyond hatching;birds then are capableof sustained day 19, we estimatedthe overall growth curve flight and are regardedas fledglings.At fledg- for Roseate Spoonbills using the Gompertz ing, feathersof the alar, humeral, crural, and growth equation(Ricklefs 1967). The estimated April1982] SpoonbillNesting Ecology 281

1800 ...... lOOO 1600 Asymptøte/ Nest 118 750

1400 50o _ 1200 25o .? :• 1000 lOOO • 800 Nest 244

•o 600 ..c 750 .? 4OO • 500 2OO • 250 o 0 4 8 1216 20 24 28 32 36 40 I 44 Hatching Fledging 1000 Nest237 •' Days 750

Fig. 2. Growth curve of RoseateSpoonbill nest- 5OO lings. Some data points representa single chick, but others may be an averageof up to 8. 250

0 2 4 6 8 10 12 14 16 18 20 asymptote,or fledging weight, closelyresem- Days After Hotching bled known adult weight; therefore, the Fig. 3. Weight gain of RoseateSpoonbill siblings growthcurve may be referredto as "standard" from three nests. Dashed, solid, and dotted lines (i.e. body weight of nestlingsequaled that of representfirst, second,and third chicks,respective- adults at fledging: Ricklefs 1968). Initially, there was a weight differenceamong siblings, as hatching occurredasynchronously; in most instances,however, siblings were visibly in- ment, although most of the eggs were either distinguishablein size after about21/2 weeks. fresh and unincubated (n = 33) or incubated Data from known-age siblings in three nests about 5 days or less (n = 17). Although there are illustrated in Fig. 3. During develoment, was considerableinterclutch variability among the first chick in nest 118 weighed slightly some egg parameters,most of the measure- more than the secondchick basedon days after ments were clustered closely around the hatching, but the reversewas true in nest 244; means, as evidenced by the extremely small in nest 237 the third chick weighed slightly standard errors (Table 5). Variability in whole more than the others up to about 14 days. Be- egg weightswas due mainly to water lossdur- causeour growth data on siblingswere sparse, ing incubation(Ar and Rahn 1980).Fresh eggs we did not attempt a statisticalanalysis; our averaged5.5 g heavierthan advancedeggs (in- limited resultssuggest, however, that weight cubated about 16 days or more), a significant gain differenceswere negligible on an actual difference (P < 0.05, analysis of variance). age basis. Thus, siblings probably had an There was no difference (P > 0.05), however, equal chance for survival, because fledging in dry shell weights of fresh and advanced weights were similar. eggs. Percentagelipid showed the greatest Egg measurements.--gveragemeasurements variation of the egg parametersmeasured (Ta- of Roseate Spoonbill eggs from Nueces Bay, ble 5), but there was no difference (P > 0.05) Texas are given in Table 5. Lengths and in lipid levels of fresh and advanced eggs. breadthshave been published previously (Bent There also were no differences (P > 0.05) in 1963, Oberholser 1974) and are almost identical egg measurementsor lipid contentamong the to our data. We found no published record, first, second, third, or fourth eggs (n = 22 however, of whole weights, volumes, shell nests). In all instancesthere was less variation weights, or lipid contentsof spoonbill eggs. in egg measurementswithin clutches than The averagesin Table 5 represent measure- amongclutches, indicating that individual fe- ments of eggs in various stagesof develop- maleswere more inclined to lay eggsof similar 282 WHITE, MITCHELL, AND CROMARTIE [Auk, Vol. 99

TABLE5. Measurementsof RoseateSpoonbill eggs from NuecesBay, Texas, 1977-1980.

Interclutch vari- Intraclutch vari- Parameter • + SEa Range ability (%)a ability (%)b Length (mm) 64.4 + 0.3 56.9-71.2 20 13 Breadth (mm) 43.7 + 0.1 41.347.9 14 8 Whole weight (g) 62.1 + 0.5 52.4-77.0 32 24 Volume (ml) 61.8 + 0.5 52.0-75.4 31 18 Shell weight (g) 5.7 + 0.1 3.8-7.0 46 37 Percentagelipid c 5.5 + 0.1 2.9-7.5 61 38 one egg each from 70 separateclutches. 10 completeclutches, 33 eggs. Lipid reportedonly for 1977, 1978,and 1979 •amples;n = 45 eggs. size and composition than were females as a rine pesticidesdetected in spoonbilleggs were whole. dieldrin and chlordane isomers, but these Organochlorineresidues and shellthickness.- compounds occurred in concentrations less Chemical analysis of egg contents indicated than 0.5 ppm in all instances. that DDE and PCBs were the predominant or- Therewas no significantdifference (P) 0.05, ganochlorinepollutants occurring in spoonbill analysisof variance) in eggshellthickness of eggs;DDE was presentin all of the samples our collectionsand that of the 1920's (Table 7). and PCBswere found in 90% of the eggs(Table Eggshellsin 1977-1979were 5% thinner than 6). In general, organochlorine concentrations thosefrom the pre-DDT era, but the percentage were low (•2 ppm), though a few of the eggs differencewas within the range of intraclutch contained fairly high levels of DDE (up to 15 variation (11%) and is not biologicallysignif- ppm). Becausethe residue data were highly icant. Ohlendorfet al. (1979)reported no sig- skewed, medians rather than means are pre- nificant eggshell thinning in spoonbill eggs sented in Table 6 as a better indicator of pol- collected from Louisiana and Texas between lutant levels in eggs. There was no difference 1947and 1973.Embryonic development during (P • 0.05, t-test) in DDE concentrations be- incubation is believed to decrease shell thick- tween addled and fresh eggs that would im- ness in some species (Kreitzer 1972), because plicate DDE in egg failure. Other organochlo~ the developingchick derivesthe major portion of its calciumfrom the shell (Simkiss1967). We found no difference (P)0.05), however, in TABLE 6. Organochlorine residues (ppm, wet eggshellthickness of spoonbilleggs in various weight) in Roseate Spoonbill eggs from Nueces stagesof development (unincubated, 1/4, 2/4, Bay, Texas. 3/4, and 4/4). Lindvail and Low (1980) found that incubationstage had no measurableeffect Year n a DDE PCBs on eggshell thickness in Western Grebes 1977 5 2.8 b 0.85 (Aechmophorusoccidentalis). (5)c (4) 1.1-15 d NEF-1.7

1978 24 2.5 0.83 TABLE 7. Eggshell thicknesses (mm) of Roseate (24) (18) Spoonbilleggs from the Texascoast. 0.7-7.5 ND-5.3

1979 16 1.4 0.92 Per- (16) (16) centage 0.4-3.8 0.4-3.5 Year na • + SE change 1980 25 1.3 0.47 1923-1924 • 11 0.44 + 0.01 -- (25) (25) 1977 10 0.42 + 0.01 -5 0.3-5.8 0.2-3.1 1978 24 0.42 + 0.01 Number of clutchesfrom which one egg eachwas analyzed. 1979 16 0.42 + 0.01 Median. 1980 25 0.44 + 0.01 0 Number of eggscontaining detectable residues. Range. • Number of clutches. ND = not detected. • Pre-DDT era; measurementsfrom both years were combined. April1982] SpoonbillNesting Ecology 283

CONCLUSIONS ulation is in no immediate danger of decline. The populationstatus of this relativelyuncom- Under normal circumstances,the proportion mon speciesshould be monitored closely, of eggsthat producesflying young (nest suc- however,especially because industrial and ag- cess)varies widely among arian speciesand riculturaldevelopment is increasingalong the is dependent upon many factors (Ricklefs Texas Gulf Coast. 1973).Nest successis known to vary from year Environmental pollutants (including DDE to year within somespecies (Maxwell and Kale and other organochlorinecompounds) proba- 1977, Schreiber 1979), and we found this to be bly had little effecton spoonbillreproduction, true of Roseate Spoonbills. Allen (1942) re- thoughsome eggs had fairly high concentra- ported only a 2% nesting successfor a Texas tions of DDE. Apparently, spoonbillsare less spoonbillcolony in 1940because of heavytick sensitiveto DDE than other species,such as infestation, but in 1941 he estimated that re- the Brown Pelican. Blus et al. (1974)presented production was excellent in this colony, convincingevidence that Brown Pelicans suffer thoughno figureson nest successwere given. nestfailure if eggscontain more than about2.5 Nest successin our study colonyranged from ppm (wet weight)DDE; 34% of our spoonbill 36% to 54%, averaging50%, or 1.5 fledglings eggs had DDE levels of this magnitudeor per nestfor the 3-yr period.Failure of the eggs higher.Some eggshell thinning was detected, to hatch and lossof young during the first 2 but it was not to the extent (15-20%) known weeksafter hatchingappeared to be the great- to causepopulation declines in other avian est limiting factorsin spoonbillreproduction. species(Anderson and Hickey 1972). Levels of Rodgers(1980) reported that mortality among other pollutants,such as PCBs,dieldrin, and heron nestlingsalso was greatestwhen the chlordane,were low in spoonbilleggs and are chickswere 1-2 weeksof age.In 1980,the year not suspectedof causingreproductive prob- we made frequentvisits during the early part lems. of the nesting season,productivity was lowest becauseof poor hatching success.Human dis- ACKNOWLEDGMENTS turbancecan adverselyaffect reproduction, es- pecially in the pre-egg and incubation stages We thank the Welder Wildlife Foundation, Sinton, (Schreiber 1979). This was not the case in our Texasfor allowingus to measureeggshells in their study, however; there was no difference in collection. James Remsen, Louisiana State Univer- productivitybetween a portion of the colony sity, kindlyprovided information on adultspoonbill visited frequently and one visited only twice. weights, as did M. Ralph Browning, USFWS, on adult bill measurements. Cathy Lefever, Sharna Our visits were most frequentafter the chicks King, and Tammy Preissprovided technical assis- had hatched, and this probably lessenedour tance, and ClementineGlenn typed the manuscript. impact, as Schreiber(1979) showed for Brown EugeneDustman, Harry Ohlendorf,James Rodgers, Pelicans (Pelecanusoccidentalis ). Jr., and William Robertsonprovided many helpful Reproduction of Roseate Spoonbills in commentsthat greatlyimproved this paper. NuecesBay appearedto be good, based on an overallnesting success of 50%. It is difficult to LITERATURE CITED assesswhether or not this figure is adequate ALLEN, R. P. 1942. The Roseate Spoonbill. New to maintain a stablepopulation, not only from York, Dover Publ., Inc. a logisticsstandpoint, as discussedby Schrei- ANDERSON,D. W., & J. J. HICKEY. 1972. Eggshell ber (1979), but also becauseof a scarcityof changesin certainNorth Americanbirds. Proc. comparablereproductive data on this species. 15th Intern. Ornithol. Congr.: 514-S40. Spoonbillsare absentfrom many of the areas AR, A., & H. RAHN. 1980. Water in the avian egg: in Texaswhere they formerlybred (Allen 1942), overallbudget of incubation.Amer. Zool. 20: 373-384. and population numbers probably are far be- BENT,A. C. 1963. Life historiesof North American low what they once were. Nevertheless,the marsh birds. New York, Dover Publ., Inc. present population appears to be holding BLACKLOCK,G. W., R. D. SLACK,D. R. BLANKINSHIP, steadyat about2,500 breeding pairs, according A. H. CHANEY, K. A. KING, J. C. SMITH, & L. to the Texas Colonial Waterbird Census (Black- MULLINS. 1979. Texas Colonial Waterbird Cen- lock et al. 1979). For this reason, we believe sus. Proc. 1st Welder Wildl. Found. Symp.: that nestsuccess is adequateand that the pop- 252-259. 284 WroTE, MITCHELf, AND CROtaARTIE [Auk, Vol. 99

BLUS,L. J., B. S. NEELY, JR., A. A. BELISLE,& R. M. OHLENDORF, H. M., E. E. Kr•s, & T. E. KAISER. PROUTY. 1974. Organochlorine residues in 1979ß Environmental pollutants and eggshell Brown Pelican eggs: relation to reproductive thickness: Anhingas and wading birds in the successß Environß Pollution 7: 81-91. easternUnited States. Washington, D.C., Spec. CROMARTIE,E., W. L. REICHEL, L. N. LOCK•, A. A. Sci. Rep., Wildl. No. 216. BELISLE, T. E. KAISER, T. G. LAMONT, B. M. RICKrEFS,R. E. 1967. A graphical method of fitting MUrHERN, & D. M. SWINEFORD.1975ß Residues equationsto growth curves.Ecology 48: 978-983. of organochlorine pesticides and poly- ß 1968. Patternsof growth in birds. Ibis 110: chlorinatedbiphenyls and autopsydata for Bald 419-451. Eagles,1971-72. PesticideMonit. J. 9: 11-14ß 1973. Fecundity, mortality, and avian de- DUNSTAN,F. M. 1976ßRoseate Spoonbill nesting in mography. Pp. 366-447 in Breedingbiology of Tampa Bay, Florida. Florida Field Natur. 4: birds (D. S. Farner, Ed.). Washington, D.C., 25-27ß Nat. Acad. Sci. HOYT, D. F. 1979ßPractical methods of estimating RODGERS,J. A., JR. 1980. Reproductive successof volume and fresh weight of bird eggs.Auk 96: three heron specieson the west coastof Florida. 73-77ß Florida Field Natur. 8: 37-40. KAISER,T. E., W. L. REICHEL,L. N. LOCKE,E. CRO- SCHREIEER,R. W. 1979. Reproductiveperformance MARTIE, A. J. KRYNITSKY,T. G. LAMONT, B. M. of the eastern Brown Pelican, Pelecanusocciden- MULHERN, R. M. PROUTY, C. J. STAFFORD,& D. talis. Contrib. Sci. Nat. Hist. Mus. Los Angeles M. SWINEFORD.1980ß Organochlorine pesti- County 317: 1-43. cide, PCB, and PBBresidues and necropsydata SIMKISS,K. 1967. Calcium in reproductivephysiol- for Bald Eagles from 29 states•1975-77. Pesti- ogy. New York, Reinhold. cide Monit. J. 13: 145-149. STICKEr, L. F. 1973. Pesticide residues in birds and KREITZ•R,J. F. 1972. The effect of embryonic de- mammals. Pp. 254--312in Environmental pollu- velopment on the thicknessof the shells of Co- tion by pesticides (C. A. Edwards, Ed.). New turnix Quailß Poultry Sci. 5: 1764-1765ß York, Plenum Press. LINDVALL, M. L., & J. g. Low. 1980. Effects of DDE, STICKEr,W. H. 1975. Some effectsof pollutants in TDE, and PCBs on shell thickness of Western terrestrial ecosystems.Pp. 25-74 in Ecological Grebe eggs, Bear River Migratory Bird Refuge, toxicologyresearch (A.D. Mcintyre and C. F. Utah--1973-74. Pesticide Monit. J. 14: 108-111. Mills, Eds.). New York, Plenum Press. MAXWEft, G. R., & H. W. KATE. 1977. Breeding WErTY, J. C. 1962. The life of birds. Philadelphia, biology of five speciesof heronsin coastalFlor- W. B. Saunders Co. ida. Auk 94: 689-700. OEERHOrSER,H. C. 1974. The bird life of Texas, vol. 1. Austin, Texas, Univ. Texas Press. HYBRIDIZATION AND BREEDING SUCCESS OF COLLARED AND PIED FLYCATCHERS ON THE ISLAND OF GOTLAND

RAUNO V. ALATALO, LARS GUSTAFSSON,AND ARNE LUNDBERG Departmentof Zoology,Uppsala University, Box 561, S-75122 Uppsala,Sweden

ABSTRACT.--Twoclosely related Old World flycatchers,the Pied Flycatcher(Ficedula hy- poleuca)and the Collared Flycatcher(F. albicollis),are allopatricon most of the European mainland but have overlappingranges in centraland easternEurope and on the island of Gotland in the Baltic. On Gotland, the Collared Flycatcheris approximately 10 times as abundantas the Pied Flycatcher.The two specieshybridize (4% of all matings)at frequencies lessthan those predicted for random mating (13%). Mixed pairs produceas many offspring as pure Pied Flycatcherpairs and more offspring than Collared Flycatchers.The competence at courtshipand/or viability of hybrid offspring, however, is probably reduced, because fewer hybrids breed than would be expectedfrom the proportion of hybrid fledglings.

SPECIES are defined as intrafertile but non- onthe Baltic islands of Gotland and Oland (Fig. interhybridizing populations,and interspecif- 1). Sympatric populations hybridize to some ic hybrids are rarely encounteredin most sam- extent (L6hrl 1955, Alerstam et al. 1978). ples of animals under natural conditions. Collared and Pied flycatchersare closely re- Among birds, interspecifichybrids occur at a lated. Both are highly sex-dimorphicin plum- frequencyof one in about 50,000 (Mayr 1970). age. Malesof the two speciesare easilydistin- We assume that numerous isolating mecha- guishedfrom one another, the male Collared nisms owe their existenceto selectionagainst Flycatcherdiffering from the Pied Flycatcherin interspecific hybridization, which wastes having a white collar and rump. On the other genes.Yet, somebird speciespairs remain dis- hand, femalesare almost indistinguishablein tinct over much of their commonrange despite the field by plumage characters.The songsand hybridization in certain regions, perhaps es- alarm calls, however, are highly species-spe- pecially those regions in which habitats have cific. recentlybecome greatly modified and/orwhere On the island of Gotland, where this study sympatry is recent (e.g. West 1962, Short 1969, was performed, the total populationsof Col- Mayr 1970, Gill 1980). Similarly, largely allo- lared and Pied flycatchershave recently been patric speciesmay hybridize in narrow zones estimated at 4,000 and 500 pairs, respectively of overlap, in which case it is sometimes a (Gustafssonand H6gstr6m in press).The near- matter of taste whether the two are best re- est CollaredFlycatcher population is found ap- garded as speciesor subspecies.While situa- proximately 600 km southeast of Gotland, tions of this kind are commonly described in while the Pied Flycatcherbreeds abundantly the literature, quantitative information about on the mainland of northern, central, and east- the rate of hybridization and the reproductive ern Europe (see Fig. 1). Collared Flycatchers output of mixed, as opposed to pure, pairs is thus predominate in this peripheral, isolated almost always wanting. Here we report such area within the range of the more northerly details for a case of two European allospecies Pied Flycatcher. interbreedingin a zone of sympatry. In the present paper, we attempt to deter- The speciesinvolved are the Pied Flycatcher mine whether or not (1) birds mate assorta- (Ficedulahypoleuca) and the Collared Flycatch- er (F. albicollis).These Old World flycatchers tively and (2) breeding successof interspecific occuras allospecies(semispecies), with ranges pairs is reduced,as would be expectedfor true overlappingin centraland easternEurope and species.

285 The Auk 99: 285-291.April 1982 286 ALATALO,GUSTAFSSON, AND LUNDnEa6 [Auk, Vol. 99

• Pied flycatcher :.•!• Collaredflycatcher

0 500 1000 k m

Fig. 1. Breedingranges of Collaredand Pied Flycatchers(based on Alerstam et al. 1978, but modified from Creutz 1970 and Glowacinski 1974). The island of Gotland is encircled.

STUDY AREA AND METHODS Male hybrids can easily be identified from their vague whitish collars. In the hand, hybrid females Collared and Pied flycatchersare especially suit- can be distinguishedby the amount of white on their able for study becausethey use tree holes as nest sites. With nest boxes, one is able to control the neck feathers(see Svensson 1975). Collared Flycatch- er males were classifiedas old or yearlings on the whole population,because the birds seldomnest in basis of the color of all primary coverts (Svensson natural cavities when nest boxes are present. We 1975), while Pied Flycatchermales were aged from erecteda total of 435 nest boxesin two major forest the innermost primary covert (Alatalo et al. MS). habitats in southern Gotland (57ø10'N, 18ø20'E). In deciduouswoodlands the nest box density was 6/ha and in coniferous forest, 1.5 nest boxes/ha. The de- RESULTS ciduous woodlands are dominated by oak (Quercus robur)and ash (Fraxinusexcelsior), with a denselower Hybridizationfrequency.--The occupancy of layer of hazel (Corylusaveliana) and hawthorn(Cra- nest boxes and the breeding densities of dif- taegusspp.). The coniferousforest is dominatedby ferent pair combinations in the two habitats pine (Pinussylvestris), but birch(Betula pubescens) is are given in Table 1. The breeding density of also common. flycatcherswas 20 times higher in deciduous The arrival of the flycatcher males and their oc- than in coniferousforest, and many more nest cupationof territorieswere monitoredby censuses boxes remained unoccupiedin the latter hab- everysecond day. For each nest we recordedthe on- itat. The Collared Flycatcher was almost 30 set of laying, clutch size, hatching success,and times as abundant as the Pied Flycatcher in fledglingnumbers. We capturedand bandedall fe- deciduous forest, while in the coniferous area males and almost all males during the nestling pe- riod. In some caseswe did not succeedin capturing we found only three times as many Collared the male,because polygamous males seldom feed the Flycatchers.The proportionof mixed pairs was youngof their secondnests. The speciesidentity of much higher in coniferous forest than in de- these males was clear, however, from our census rec- ciduousforest (17% and 2.5%, respectively). ords. In addition, we observed four nests in which April1982] Hybridizationin Flycatchers 287

TABLE71. Breeding densities of Pied (PF)and Collared(CF) flycatchersand of mixed pairs in differenthabitats on the island of Gotland.

Percentage Size of occupancyby Density of breeding study Number Density of pairs/ha areas of nest Fly- All flycatcher Habitat (ha) boxes catchers species pairs/ha CF x CF CF x PF PF x PF Deciduous woodland 37 243 48.6 75.7 3.19 3.00 0.08 0.11 Coniferous woodland 116 192 9.9 24.0 0.16 0.10 0.03 0.03

one parentbird apparentlywas a hybrid; these habitat, so comparisonsmust take the time fac- pairs are not included in Table 1. tor into account. Our data are extensive for the If the individuals of the two speciesmated Collared Flycatcher, and we therefore use randomly, one would expect 13% of the pairs regressionequations for clutch size and fledg- to be mixed (Table2). The observedproportion ling numbers versus the laying date of this of mixed pairs was 4.4%, however, which de- species as a reference line. The regression viates significantly(X 2 = 7.66, P < 0.01) from equation for clutch size is: y = -0.078x + 6.77 random. Thus, there is an overrepresentation (SD = 0.66, n = 123), where x is laying date of intraspecificmatings, interbreedingoccur- (day 1 = 20 May). Pied Flycatchershad, on av- ring only in 36% of the number of casesex- erage,0.71 (SD = 0.84, n = 7) more eggsthan pectedif randommating is assumed.We may simultaneously laying Collared Flycatchers expect underrepresentationof mixed pairs, (Fig. 2). This differenceis significant(t = 2.70, simply becausethe frequenciesof occurrence P < 0.01). The clutch size of mixed pairs was of the two speciesare not identicalin the two intermediatebetween that of pure pairs of the habitats. In each habitat, however, the ob- two species(0.46 higher than for the Collared served number of mixed pairs was 29% (de- Flycatcher, SD = 0.65, n = 7, t = 1.81, ciduous: X2 = 5.11, P < 0.05) and 44% (conif- P < 0.10). erous:X 2 = 2.51,P > 0.10), lower than expected Hatching successof the flycatcherswas high, if random mating is assumed. In 5 of the 7 and no difference emerged among the pair mixed pairs (including 1 pair outside the main combinations(Table 3). The regressionline for study areas),the male was a CollaredFlycatch- fledgling numbers of the Collared Flycatcher er and the female a Pied Flycatcher.Moreover, decreasedmore rapidly with time than the line of all Pied Flycatchers,30% were involved in for clutch size (Fig. 3; regressionequation for interspecificmatings, whereas the correspond- number of fledglings: y = -0.128x + 6.43, ing figure for the CollaredFlycatcher was only SD = 1.71, n = 123). Early breeders presum- 2%. Of malesinvolved in interspecificmating, ably gained from an outbreakof caterpillars, 6 out of 7 (86%) were yearlings. For all intra- which levelled off in the second half of June. specificpairs this proportionwas lower (55%), but the differenceis not significant(Fisher ex- act, P = 0.12). Breedingsuccess.--Average clutch size and TABLE2. Observed and expected mating combina- tions of flycatchers. fledgling numbers of intra- and interspecific pairs are given in Table 3. We have combined Pure Pure the data from the two habitats, because we Collared Pied found no differencesin breeding successbe- Fly- Fly- tween habitats. Lundberg et al. (1981) found catcher catcher Mixed higher breedingsuccess for the Pied Flycatcher pairs pairs pairs in deciduous than in coniferous habitat, but Number found 123 7 6 the differencewas small. Laying date affects Percentagefound 90.4 5.1 4.4 breeding successmuch more profoundly than Percentageexpected 86.3 1.0 12.7 288 AL^•r^LO,GUS•r^FSSON, ^tad LUNDBERO [Auk, Vol. 99

•S4 i *%,0>%

ß M,xed pa,r

o Pied flycatcher

ß Mrxed parr

LAYING DATE

24 29 3 8 13 18 MAY JUNE

LAYING DATE Fig. 3. Number of fledglingsof Pied Flycatchers and of mixed pairs in relation to the regressionline for fledgling numbers versus laying date in the Col- Fig. 2. Clutch size of Pied Flycatchersand of lared Flycatcher. mixed pairs in relation to the regression line for clutch size versus laying date in the Collared Fly- catcher. to be significantusing the Mann-Whitney rank test (z = 2.17, P < 0.05). Because there is no differencein the mean laying date of the two Pied Flycatchersproduced, on average, 1.32 groups (Table 3), one could comparefledgling (SD = 1.32, n = 7) more fledglings than si- numbersdiscounting the laying date, in which multaneously breeding Collared Flycatchers case the difference is still significant (Mann- (Fig. 3). Unlike clutch-sizedistributions, the Whitney; z = 2.10, P < 0.05). Interspecific brood-size distributions clearly deviated from normal because of the total failure of some pairswere alsofo_und to producesignificantly more fledglings (y = 1.51, SD = 0.60, n = 7, broods.Therefore, t-tests of significancecould not be used, and, instead, we calculated for z = 2.84, P < 0.01) than simultaneously breeding Collared Flycatchers.No significant each Collared and Pied flycatchernest the de- difference was found between pure Pied Fly- viation in fledgling number from that to be catcher pairs and mixed pairs. expectedfrom the regressionline in Fig. 3. In this case,the regressionequation is used only DISCUSSION to standardizethe generaldecrease in breeding successwith season. Between-speciesdiffer- In the two flycatcherspecies, though mixed ences in terms of such deviations were found pairs occur, there is a clear preference for con-

TABLB3. Breeding data of Pied and Collared flycatchersand of mixed pairs.

Mean laying Clutch Hatched Species n date size SD eggs(%) Fledglings SD Collared Flycatcher 123 2 June 5.76 0.81 94.1 4.64 1.90 Pied Flycatcher 7 1 June 6.57 0.79 95.7 6.14 0.69 Mixed pairs 7 3 June 6.14 0.69 100 6.00 0.58 April1982] Hybridizationin Flycatchers 289 specific mates (Table 2). In two other cases of parent introgressionbetween severalallospe- allospecieshybridization, pairs alsomated as- cific pairs (for examples,see Anderson1977) sortatively [Gill and Murray 1972 (Vermivora implies that hybrids in many cases are not pinus/V. chrysoptera),Hoffman et al. 1978 (La- completelysterile. Brewer (1963;see also Ris- rus glaucescens/L.occidentalis)]. In the latter ing 1969)suggested that, in natural popula- case,quantitative data were provided. Because tions, hybrids of Parusatricapillus and P. car- only 25% of the Laruspairs were pure, the in- olinensisproduce less viable offspring. He cidenceof assortativemating was clearlylower found a very low fledglingsuccess in six nests than in this study. In contrast, Ingolfsson in an areaof sympatry.Examples of allospecific (1970) found that Larus hyperboreusand L. ar- pairsthat produce only F• hybridsare given by gentatusin Iceland mated randomly. Anderson (1977). In this study, 2 out of 4 pairs One would expect marked positive assorta- in which hybridswere involvedfailed to hatch rive mating if reproductivesuccess of interspe- their eggs,while 2 producedyoung (4 and 6, cific pairs or of hybrid offspringwere low and respectively).In one studyplot on the Swedish randommating if hybridizationdid not reduce mainland, (59ø50'N, 17ø40'E)where the Col- reproductivesuccess (Short 1972). Mixed pairs lared Flycatcheris absent,we found one hy- of Collared and Pied flycatchersproduced vi- brid male, which was paired with a Pied Fly- able offspring. In fact, mixed pairs produced catcherfemale. This pair raised four fledglings significantlymore offspringthan CollaredFly- from six eggslaid. L6hrl (1955)and Veps•l•i- catchers(Fig. 2), presumablybecause the Pied nen and J•rvinen (1977) each found a fertile, Flycatcherhas a largerclutch size than doesthe breeding, hybrid flycatchermale, but in the Collared Flycatcher(Fig. 2), and most of the latter casethe authorsmentioned the possibil- mixed pairs involved Pied Flycatcherfemales. ity that the broodcould have been adopted by What happens to the offspring of mixed fly- the hybrid male.Thus, in summary,flycatcher catcherpairs? Hybrids comprised6.4% of all hybridsare not fully sterilebut apparentlydo flycatcher fledglings but only 1.4% of the suffer from reduced fertility. Our data are too breedingadults (see also Alerstam et al. 1978). few to estimatethe viability and fertility of This implies reduced survival or incompetence hybrids accurately, but, because the two at courtship. It is not likely that hybrids are specieshave retained their specificcharacters less efficient foragers than pure individuals on Gotland so well, introgressioncan be as- (see Mayr 1970), because the two species are sumed to be slight. very similar in terms of their foraging habits Alerstamet al. (1978)suggested that hybrid- (Alerstam et al. 1978). On the other hand, the ization actsas an "agent of competition"be- two flycatcher species breeding on Gotland tween the two flycatcherspecies on Gotland. have largely different wintering areas(Moreau Becausethe Pied Flycatcheris less abundant, 1972), which may causeproblems for the hy- a greaterproportion of its populationwill suf- brids during migration. Hybrid males may be fer the risk of hybridization (see also Veps•- relativelyunsuccessful at courtship(Ficken and l•inen et al. 1975, Anderson 1977, Veps•l•inen Ficken 1968, Mayr 1970), becausetheir vocal- and J•rvinen 1977). The considerablerate of izations and courtship behavior may differ hybridization should prevent the Pied Fly- from the parental species.In fact, the songsof catcherfrom building up its population.The hybrid flycatchermales were clearly different hypothesisis based on the assumptionthat from the songsof the pure speciesand often hybrids are practically sterile. Mixed pairs includedspecies-specific elements of both the have high reproductivesuccess, but their hy- parental species.Also, hybrid femalevocaliza- brid offspringseem to be at a reproductivedis- tions differed from those of the parental advantage.Our datathus support the hypoth- species. esis of Alerstam et al. (1978) in that Offspring from interspecificmatings may be hybridizationmay help preventthe Pied Fly- fully fertile, may have reducedfertility, or may catcherfrom becoming established on Gotland. be sterile (Mayr 1970). For example, captive Conversely,hybridization may limit the ability Anas platyrhynchosand A. acuta procreated of the CollaredFlycatcher to colonizethe north fully fertile hybrids. Hoffman et al. (1978) and eastEuropean mainland, becausethe like- foundhigh breedingsuccess for hybridsof two lihood of hybridization will be very high for Larus species under natural conditions. Ap- a few immigrants(Veps•l•inen et al. 1975).An 290 ALATALO,GUSTAFSSON, AND LUNDBERG [Auk,Vol. 99 isolated island like Gotland would be easier to ACKNOWLEDGMENTS colonize than the mainland, if the Pied Fly- We are gratefulto T. Alerstam,F. B. Gill, O. Jar- catcher were established in Scandinavia before vinen, S. Ulfstrand, and K. Vepsalainen for valuable the CollaredFlycatcher. criticismof the manuscript.The work was supported Alerstam et al. (1978)proposed that the small by grants(to S. Ulfstrand)from the SwedishNatural Pied Flycatcherpopulation on Gotland proba- Science Research Council. bly is maintained by immigration from the surroundingmainland. We have examined this LITERATURE CITED possibility with the following simple model. ALERSTAM, T., B. EBENMAN, M. SYLV/•N, S. TAMM, We assumed a constant population level of & S. ULFSTRAND.1978. Hybridization as an 4,000 Collared Flycatcherpairs and 500 Pied agent of competition between two bird allospe- Flycatcherpairs (Gustafssonand H6gstr6m in cies: Ficedulaalbicollis and F. hypoleucaon the press),a hybridization frequencyequivalent to island of Gotland in the Baltic. Oikos 31: that found in the presentstudy, a 50% adult 326-331. survival rate to the next year (unpubl. data; see ANDœRSON,R. F. V. 1977. Ethologicalisolation and also Anderson 1977), and the same number of competitionof allospeciesin secondarycontact. offspring per pair of each speciessurviving to Amer. Natur. 111: 939-949. the next breeding season.Then, each year, 73 BREWER,R. 1963. Ecologicaland reproductive rela- tionships of Black-cappedand Carolina chicka- pairs (=15% of the total population) of Pied dees. Auk 80: 9-47. Flycatchershave to immigrate to Gotland to CREUTZ,G. 1970. Dringt der Halsbandschn•pper compensatefor the hybridization losses(if hy- nach Norden vor? Falke 10: 334-339. bridization leads to total reproductivefailure). FICKEN,M. S., & R. W. FICKEN.1968. Courtship of On the other hand, Pied Flycatchershad 32% Blue-winged Warblers and their hybrids. Wil- more fledglingsthan did the CollaredFlycatch- son Bull. 80: 161-172. ers (Table 3). If this difference were maintained GILL, F. B. 1980. Historical aspectsof hybridization until the next breeding, an immigrationrate of betweenBlue-winged and Golden-wingedwar- only 16 Pied Flycatcherpairs per year would blers. Auk 97: 1-18. be sufficient to retain a stable situation. Be- --, & B. G. MUm•A¾, JR. 1972. Discrimination cause huge numbers of Pied Flycatchersare behaviorand hybridizationof the Blue-winged and Golden-winged warblers. Evolution 26: captured and banded on Gotland during 282-293. spring migration (Rosvall 1979), an immigra- GLOWACINSKI,Z. 1974. Ekspansjamucholowki bi- tion of this modest magnitude may easily be aloszyjej,Ficedula albicollis (Temm.) w Europie achieved. õrodkowej.Przeglad Zool. 18: 471-484. The situation is complicated, however, by GUSTAFSSON,L., & S. H/JGSTR/JM.In press. Hur the fact that the Collared Flycatcheris socially mfinga ffiglar h•ckar pfi Gotland?Bl•cku. dominantover the Pied Flycatcher(L6hrl 1955). HOFFMAN, W., J. A. WIENSt • J. M. SCOTT. 1978. Studies on the Swedish mainland showed that Hybridizationbetween gulls (Larusglaucescens the Pied Flycatcherclearly prefers deciduous and L. occidentalis)in the PacificNorthwest. Auk 95: 441-458. forest (Lundberg et al. 1981), but the Collared INGOLFSSON,A. 1970. Hybridization of Glaucous Flycatcherseems to be able to monopolizethe Gulls Larus hyperboreusand Herring Gulls L. favorable deciduous habitats on Gotland and argentatusin Iceland. Ibis 112: 340-362. to relegate the Pied Flycatcherto suboptimal L6HRL, H. 1955. Beziehungenzwischen Halsband- habitats (Table 1). Even if there were no hy- und Trauerfliegenschn•pper(Muscicapa albicol- bridization, we believe that the Collared Fly- lis und M. hypoleuca)in demselbenBrutgebiet. catcherwould survive in its isolatednortherly Acta 11th Congr. Intern. Ornithol. 1954 (Basel): refuge. On the other hand, hybridization may 333-336. be the prime factor preventing the Collared LUNDBERG,A., g. V. ALATALO, A. CARLSON, & S. Flycatcherfrom establishinga population on ULFSTRAND.1981. Biometry, habitat distribu- the surroundingmainland. One can only spec- tion and breedingsuccess in the Pied Flycatcher Ficedula hypoleuca. Ornis Scandinavica 12: ulate about the origin of the presentstable dis- 68-79. tribution of the two species in Scandinavia, MAYR, E. 1970. Animal speciesand evolution. Cam- but the isolated position of Gotland and the bridge, Massachusetts,Harvard Univ. Press. occurrenceof hybridization may well be of im- MOREAU, g. E. 1972. The Palaearctic-African bird portance. migration systems.London, AcademicPress. April1982] Hybridizationin Flycatchers 291

R•s•tac, J. D. 1969. A multivariate assessmentof in- SVEtaSSON,L. 1975. Identification guide to European terbreedingbetween the chickadees,Parus atri- passerines. Stockholm, Naturhistoriska Riks- capillus and P. carolinensis. Syst. Zool. 17: museet. 160-169. VEPS•L•IIqEN, K., • O. J•RVINEN. 1977. Ein Bastard- ROSVALL,S. 1979. HalsbandsflugsnapparensFice- M•nnchen Halsband-Trauerschn•pper(Ficedula dula albicollisankomst till Gotland och (•land albicollisx F. hypoleuca)briitet in Finnland. J. v•ren 1979. Bl•cku 5: 73-78. Ornithol. 118: 436-437. SHORT,L. L. 1969. Taxonomic aspectsof avian hy- , K. KEYIq•S, & P. SALMINEIq. 1975. bridization. Auk 86:84-105. S;pelsiepotFicedula albicollis Hankoniemell• --. 1972. Hybridization, taxonomy and avian 1971-74. Lintumies (Helsinki) 10: 118-122. evolution. Ann. Missouri Bot. Garden 59: WEst, D. A. 1962. Hybridization in grosbeaks 447-453. (Pheucticus)of the Great Plains. Auk 79: 339-424.

REVIEWERS FOR THE AUK, 1980-1981 TheAuk is fortunateto benefitfrom the servicesof many individuals who act as reviewersof manuscripts that are submitted. Their efforts are considerable,and their care and constructivenesscontinue to impress me as an Editor and as a sometimeauthor. The individuals listed below helped make publication of a journal with high scientificand scholarlystandards possible; my thanks to all of them. Individuals who have contributed reviews of two or more manuscriptsare indicated by an asterisk. Ian Abbott, David G. Ainley, JohnAlcock*, Thom- Fitch, JohnW. Fitzpatrick*, Brian K. Follett, Eric D. as Alerstam, Dean Amadon, Daniel W. Anderson, Forsman*,W. J. Francis,Leigh Fredrickson,Herbert Ted R. Anderson, C. Davison Ankney, Einar Arna- Friedmann, Carl Gans, C. Lee Gass*, A. J. Gaston, son, John C. Avise, Robert O. Bailey, Allan Baker, Sidney A. Gauthreaux, W. George, Frank B. Gill*, Myron C. Baker, Robert J. Baker, George S. Bakken, Robert E. Gill, P. D. Gingerich, BradleyM. Gottfried, Russell P. Balda, Thomas G. Balgooyen*, David F. Gilbert Gottlieb, Walter D. Graul, RussellGreenberg, Balph, Martha Balph*, Richard C. Banks*, Luis F. Paul J. Greenwood, T. C. Grubb, Jr., JosephA. Grzy- Baptista*,George F. Barrowdough*,George A. Bar- bowski, Svein Haftorn, Caldwell D. Hahn, F. Reed tholomew, Jr., ThomasS. Baskett,Range Bayer*, Jeff Hainsworth, William J. Hamilton, III, M.P. Harris, Baylis, Michael Beecher*, Colin Beer*, Marc Bekoff, Richard J. Harris, JeremyJ. Hatch, Helen Hays, Rob- Frank C. Bellrose, Jr.*, Louis B. Best*, T. R. Birkhead, ert E. Hegner, Dennis Heinemann*, Carl W. Helms*, CharlesR. Blem, Peter Boag,Walter Bock*, William Charles J. Henny*, Wayne Hoffman*, A. J. Hogan- R. P. Bourne,John H. Brackenbury,Richard A. Brad- Warburg, Richard T. Holmes, Dominique G. Hom- ley*, Eliot Brenowitz, I. Lehr Brisbin*, Pierce Brod- berger,John L. Hoogland*,Henry F. Howe*, Donald korb*, Richard G. B. Brown*, Charles J. Burke, M. F. Hoyt*, JohnP. Hubbard, George L. Hunt, Jr.*, R. Byrd, William A. Calder, Jr.*, Thomas Caraco, Cyn- Hutton, Jerome A. Jackson*, Frances C. James*, Dan- thia Carey*, Lynn F. Carpenter, Susan M. Case, iel H. Janzen*, Olli Jarvinen*, JosephR. Jehl, Jr.*, Graeme Caughley, Mary H. Clench, Charles T. Col- Paul A. Johnsgard*,Douglas H. Johnson*,David W. lins*, Michael Collopy, Edward F. Connor, Peter G. Johnston, Richard F. Johnston*, James R. Karr*, Janet Connors, Michael R. Conover*, Fred Cooke, John Kear, J. Allen Keast, Bretton W. Kent, Paul Kerlinger, Cooper, Kendall W. Corbin, Richard Cowie, Joel Lloyd F. Kiff, Lawrence Kilham, James R. King*, Cracraft*, Robert B. Craig, Richard D. Crawford, Warren B. King, Erwin E. Klaas*, Richard W. Knap- John Croxall, E. Curio, Thomas W. Custer, John Da- ton, Carl E. Korschgen, Gary Krapu*, John R. vis, David DeSante, Diane De Steven, Robert W. Krebs,DonaldE. Kroodsma,James A. Kushlan, Mary Dickerman, James J. Dinsmore, David S. Dobkin, K. LeCroy, Louis Lefebvre,Ross Lein*, Robert E. Richard A. Dolbeer, Hilary Dow, R. D. Drobney, Lemon*, K. Lessells,J. David Ligon*, C. D. Little- William H. Drury, Alfred M. Dufty, Jr., Arthur E. field*, Clare S. Lloyd, William J. Mader, Chris Marsh, Dunham, Erica H. Dunn*, Thomas J. Dwyer*, Eu- Carl D. Marti*, Dennis J. Martin, Elden W. Martin, gene Eisenmann,Phillip F. Elliott, John T. Emlen, L. D. Martin*, Steven Martindale, Barbara W. Mas- Robert Epting, Anthony J. Erskine, R. Michael Er- sey, Harold F. Mayfield*, Donald A. McCrimmon, win*, P. G. H. Evans,Roger M. Evans*,David Ewert, Jr., ChristopherMcGowan, Judith W. Mcintyre, Ann C. J. Feare*, J. Alan Feduccia, Peter Feinsinger, Mil- McNabb, A. H. Meier, Charles Meslow, L. Richard licent Ficken*, JamesS. Findley, Hans Fisher, Walter Mewaldt, David Miller*, Don E. Miller*, DouglasW.

(continuedon p. 315) NESTING SUCCESS OF DICKCISSELS (SPIZA AMERICANA) IN PREFERRED AND LESS PREFERRED HABITATS

JOHN L. ZIMMERMAN Divisionof Biology,Kansas State University,Manhattan, Kansas66506 USA

ABsTRaCT.•Habitatselection in the Dickcissel(Spiza americana) was evaluatedby a com- parisonof over 500 nest historiesfrom the preferredoldfield and the lesspreferred prairie communitiesin Kansas.Females are equally successfulin either habitat, and their produc- tivity is not adverselyaffected by their being matedto polygynousmales. Thus, thesetwo habitats are similarly suitable for females, and females are distributed simply accordingto the availabilityof adequatenesting environments and not in any density-dependentmanner or in responseto densitycues. No differencesexist in survivalrates of nestsassociated with individual males or in the males' individual productivities between prairies and oldfields when monogamousand bigamous males are compared.The oldfieldhabitat is moresuitable for males, however, becausethe potential for sequesteringmore nesting sitesis greaterin this more heterogeneoushabitat. This permits higher levels of polygyny, which increases the productivityof individualmales, even thoughdensities of birds in oldfieldsare greater. Received3 February1981, accepted 23 October1981,

GRASSLANDSand variouslyaged oldfields are also has demonstratedthe importanceof nest- the two major habitatsselected by Dickcissels site microclimatefor the Lark Bunting (Cala- (Spizaamericana) for establishmentof territo- mospiza melanocorys),the short-grass plains ries by malesand nestsby females.The density counterpart of the Dickcissel. of males is higher in oldfieldsthan in prairies, Using the predictions based on the models and the frequencyof polygynyand the average developedby Fretwell and Lucas(1969), which number of females per male is also greater in differentiatebetween the threenotions relating oldfields (Zimmerman 1971). The greater suit- territoriality to population regulation (i.e. ability of preferredoldfields over other habitats Huxley 1934, Kluyver and Tinbergen 1953, appears to be a function of the vegetation, Lack 1964),one may infer that oldfieldhabitats which is more heterogeneousand taller in old- are more suitable, even though densities are fields,with greatercoverage by forbs.The den- higher there, becausefemales prefer oldfield sity of malesin a particularhabitat and the males to prairie males. Male Dickcisselsthus number of femalesper male in this polygynous illustrate the "ideal dominance" distribution speciesare significantlyand positively related (Fretwell and Calver 1969). K. L. Petersen (MS) to the amount of vegetation present there has shown that the temporal sequenceof uti- (Zimmerman 1971). 1 do not know which at- lization of different habitats by male Dickcis- tribute of the vegetation structureparticularly selsfollows the pattern expectedfrom the ideal enhances the attractiveness of oldfield habitats. dominance model. Early in the season,when It could be food. Even though female Dickcis- male densities in oldfields are low, no birds selsregularly forage beyond the boundariesof settle in grasslands,but, as densitiesincrease the territory of their mate (Zimmerman 1966), in the more preferred oldfields, Dickcisselsin- Wittenberger(1980) has suggestedseveral rea- vade upland prairies, peaking in density a sonswhy the proximity of high-quality food week or so after the oldfield populations have resourcesto the nest still may be crucial. The reached maximum numbers but never attain- vegetationprobably does not protectthe nest ing the high densities characteristicof old- from lossto predators(Wittenberger 1976), but fields. the importanceof vegetationfor nest sitesmay The ultimate measureof habitat suitability, be through the improvementof the microcli- of course,is the productivity of birds. As Fret- mate surrounding the nest and the critical im- well and Lucas write (1969: 33), "If a species pact of this mitigation on the growth of the has an ideal territorial (dominance) distribu- young (Blankespoor1970). Pleszczynska (1978) tion, then the success rate in habitats with

292 The Auk 99: 292-298. April 1982 April 1982] HabitatSuitability in Dickcissels 293

higher densities of residents will be higher." ries. Additionally, daily survival rates of nests were Whitham (1980) has pointed out that the Fret- computed on a per-male basis for males with differ- well-Lucas hypotheses have been largely un- ing levels of polygyny by summing all days of nest tested becauseof the difficulty of measuring exposurefor all nests of males with just 1 nest on their territories (including the monogamousmales fitnessin habitats of known quality. The pur- involved with severalsingle nests sequentially dur- pose of this paper is to use the Fretwell-Lucas ing a single season)and of polygynousmales with hypothesesto identify the habitat distribution 2-6 nests active on their territories. This number of systems of male and female Dickcissels nests is not exactly the same as the male's harem through a comparison of nest survival rates size, becausenot all nests for a given male were si- and the production of young from over 500 multaneous. Rather, this value is used as a measure nests in oldfields and prairies, habitats clearly of the suitability of a male's territory, as reflectedby differing in quality. the use of his area for nestingby different females. This procedurecompensates for differencesin repro- ductive effort among males with different levels of METHODS polygyny and isolatesthe effect that differences in Nest data were gathered from 1965 through 1979 the numbersof nestsmight have on the assessment of males' nesting successes.In all analyses,samples from populations on the Ft. Riley Military Reserva- tion and the Konza Prairie Research Natural Area in for males with 5 and 6 nests were pooled to make Riley and Geary counties, Kansas, deep within the the total sample size similar to that of the other nest center of the area with highest breeding densitiesof categories. this species in North America (Robbins and Van While recognizingMayfield's (1961)caveat regard- Velzen 1969). Except for a few populations for which ing the biases inherent in determining the numbers nests were visited every 3 or 4 days, most data were of fledglings produced per nest found, ! have cal- culated this value for nests in both habitats as a mea- obtained by checkingnest contentsonce a week. The fatesof nestsin which youngwere expectedto fledge sure of the averageproductivity per female and have during the intervening period and that were found also converted this value into an average per male empty at the next visit were easily determined by by summing the production for all nestspresent in the behavior of the females.A female with fledged a male's territory. young gives agonisticdisplays and continuesto feed the young in the vicinity of the nest for severaldays RESULTS before moving further away with the brood. A fe- male whose nest has sufferedpredation immediately Nestingphenology.--The week of initiation disappears. In some populations birds were color- of nest construction was known or, in most banded, so that male territories and the densities and cases, estimated for 397 oldfield nests and 130 mating patterns of males and females could be de- prairie nests. Nesting begins earlier in old- termined weekly. In these situations, then, it also fields than in prairies and spansa longer sea- was possiblefor me to relate the successof individ- son (Fig. 1). The peak week of nest starts,how- ual neststo the specificmale presumedto be genet- ever, is the same for both habitats (15-21 June), ically involved,by observingeither copulationor the but 50% of all prairie nest startsoccur in the companionatebehavior of the pair during the early 3-week period from 15 June to 5 July, while days of nesting. Although "stolen" copulations among marked birds have not been observed, they 50% of all nest starts in the oldfield span a may occurfrom time to time. For the purposeof this broader period, 8 June-5 July. The difference analysis, however, I have assumedthat the observed in the varianceof thesetwo populationsis sig- mate is probably the biological parent, so that these nificant (F = 1.65, df• = 396, df2 = 129, nests could then be used to measure the success and P • 0.01). These data are consistent with the productivityof malesas well as of femalesaccording population data gatheredby Petersen(MS) and to the polygynousstatus of the male. parallel the difference in settlement dates of Johnson (1979) has validated the efficacy of the Indigo Buntings (Passerinacyanea), another Mayfield (1961, 1975) nest-exposuremethod of esti- polygynousspecies, in high-density and low- mating nesting successand has effectivelyanswered density habitats (Carey and Nolan 1979). the critics of this procedure. Furthermore, he has Clutch size.--A total of 149 nonparasitized provided a formula by which confidencelimits for survivalrates may be calculated,thus permitting un- nests was known to have complete clutches. biased comparisons between the probabilities of The 15 prairie nestsin this total contained from nest survival in different populations.Accordingly, 3 to 5 eggswith a mean of 4.0 (SE = 0.14), and daily nest survival rates were determined for the sep- the 134oldfield nests contained from 2 to 6 eggs arate stagesof the nesting cyclein oldfieldsand prai- with a mean of 4.0 (SE = 0.07). 294 Joi• L. Zr•a•a•a•aA• [Auk, Vol. 99

Although it is not possibleto determinewith certainty the number of Dickcissel eggs ac- tually laid in nests parasitizedby the Brown- headed Cowbird (Molothrus ater), due to host egg removal by the cowbirds, there is no sig- nificant difference between habitats in the number of Dickcisseleggs present in parasit- 15 ized nestsat the beginning of incubation. For 193 oldfield neststhe mean was 2.4 eggs (SE = 0.09), and for 96 prairie neststhe mean was 2.5 eggs (SE = 0.12). There is, however, a significantdifference in the mean number of cowbird eggs per nest in parasitized oldfield nests (2.3, n = 201, SE = 0.10) compared to those in the prairie, (2.9, n = 96, SE= 0.16) (Student's t= 3.28, df= 295, P < 0.01). The higher intensityof cowbird parasitism in the prairie (Elliott 1978) results in the total clutch of Dickcissel and cowbird O eggs in prairie nests (• = 5.2, SE = 0.16, n = 111) being significantlydifferent from the total clutch in oldfield nests (i = 4.5, SE = 0.08, n = 330) (Student's t = 4.34, df = 439, WEEKS P < 0.01). The larger combined clutch size of Fig. 1. Percentagefrequency distribution of old- prairie nests, however, does not in itself de- field nest starts (open bars) and prairie nest starts crease the chancesfor survival of the young; (shadedbars) with the weeksof the nesting season. within the same habitat there is no significant relationship between clutch size and survival layed initiation of nesting. For the purposeof rate (Zimmerman MS). this analysis,therefore, I have consideredthe Female success.---Female Dickcissels are sin- probability of nest survival and the productiv- gle-brooded. For all the nest histories of ity per nest attempted to approximatethe suc- marked individuals, no two nests that success- cess of the average individual female in each fully fledged young were produced by the habitat. same female in the same year. The length of The daily survival rates computed for 137 nest life (average26 days) and the postfiedging prairie nests and 413 oldfield nests are pre- care of the brood (about 2 weeks), when com- sented for each phase of the nesting cycle (Ta- pared with the period during which reproduc- ble 1). The daily survival rates in each habitat tion is energetically feasible (Zimmerman were not significantlydifferent among phases 1965), precludessecond successful broods. In- of nesting. The overall probability for survival deed, ! have never observed a successful fe- can be calculatedby raising the probability of male initiate a second attempt. each phase to the power equal to the number Females do renest after the failure of an ear- of days in that phase and then multiplying lier attempt. This sometimesoccurs within the these separate probabilities together. Thus, territoryof the samemale (17% of the time) or oldfield nests have a 14.3% chanceof being with anothermale in the samelocal population successful,while prairie nests have a similar (10% of the time), but most frequently the fe- value of 15.2%. Although female densities are male disappearsand is not seen again. These higher in oldfields than in prairies (Zimmer- unsuccessfulfemales may seekout other fields man 1971), the estimated survival rate of nests of the samehabitat type, or perhapsthey select of individual females in each of these two hab- different habitats. Thus, population data from itats is the same. preferredoldfield study plots show a continued Even though the survival rates are similar, increase in female density until mid-July, as during the nest-building and egg-laying pe- females enter the habitat after presumed fail- riods the values for prairie nestsare lower than ures elsewhere or for some reason have de- those for oldfield nests, while during the in- April 1982] Habitat Suitabilityin Dickcissels 295

T^BI•E1. Daily survival rates of nests.

Building Egg-laying Incubation Brooding Oldfield Prairie Oldfield Prairie Old field Prairie Oldfield Prairie

Mean days in phase 2 2 3 3 12 • 12 a 9 b 9b Total days of nest exposure 185 23 405 81 2,117 624 1,056 534 Number of nests lost 13 2 49 15 164 34 53 25 Daily survival rate 0.930 0.913 0.879 0.815 0.922 0.946 0.952 0.955 +95% C.L. 0.04 0.12 0.03 0.09 0.01 0.02 0.01 0.02

Incubationstarts with day the last egg is laid. Includesthe day of hatching.

cubationand broodingphases they are higher. productivity of femalesinvolved, nor is there This is explainedby the greaterimpact of cow- a significantdifference between habitats in the bird parasitism in the prairie, which leads to productivity of femalesmated to males with nest abandonment during construction and equivalentnumbers of mates.It shouldbe not- egg laying, and the higher incidenceof pre- ed that the number of mates used for this com- dation in the oldfield (Zimmerman MS). parison is not strictly equivalent to the harem Productionof Dickcisselsper nest attempted size of the males, but, as in other polygynous is similar in oldfield (• = 0.72, n = 396, SE = species(Holm 1973, Weatherheadand Robert- 0.068) and prairie (• = 0.87, n = 115, SE = son 1977, Carey and Nolan 1979),the success 0.122).There is, however,a significantdiffer- of the femaleis not lessenedby her being mat- ence in the total numbers of Dickcissels and ed to a polygynousmale. cowbirds fledged per nest attemptedbetween Male success.--Somepolygynous males were oldfield (• = 0.94, n = 396, SE = 0.081) and involved with as many as five simultaneous prairie (• = 1.42, n = 125, SE = 0.171) (Stu- mates in oldfields, while two were the maxi- dent's t = 2.774, df = 519, P • 0.01). This re- mum in the prairie (Table 3). The survival rates suit is relatedto the significantlylarger clutch of a male's nests were analyzed according to size in prairie nestsdue to the heavier cowbird the total numbers of females he attracted (Table parasitism. 4). Again, this ranking is not exactlythe same For a subsetof these nests,mating relation- as the harem size. The probability of survival ships were known. It is thereforepossible to of a male's nests, if he has just one or two, is comparethe meanproduction of fledglingsper the same regardlessof habitat. Furthermore, nest in both habitats accordingto different no differenceexists between the daily survival levelsof territorysuitability as indicated by the ratesof nestsaccording to the numberof nests numbersof matesattracted by the males(Table per male in either habitat, nor doesa signifi- 2). No significantassociation occurs between cant association occur between the number of the number of mates a male attracts, that is, nests per male and the daily survival rates for the quality of his territory, and the average the nests of these males.

T^BI•E2. Productionof fledglings/nestattempted ff + SE) accordingto the number of femalesattracted to the territory.

Old field Prairie

Number Number Dickcissel Number Dickcissel of mates of nests Dickcissel + Cowbird of nests Dickcissel + Cowbird

I 83 0.84 + 0.16 1.01 + 0.18 22 1.18 + 0.31 1.73 + 0.45 2 52 0.92 + 0.22 1.31 + 0.26 20 1.10 + 0.38 1.45 + 0.46 3 75 0.88 + 0.16 1.08 + 0.19 4 72 0.50 + 0.14 0.67 + 0.18 5 and 6 46 0.63 + 0.21 0.72 + 0.22 296 JomaL. Z•mm•am• [Auk,Vol. 99

T,•BLE3. Numbers of males accordingto maximum and total young fledged/male in the oldfield number of simultaneousmates/male (percentage within each habitat). habitat (Spearmanr = 0.9, n = 5, P < 0.05).

Number DISCUSSION of females Prairie Oldfield The female Dickcisselmost certainly has a One 23 (74.2) 75 (51.0) set of criteria that she uses to selectan appro- Two 8 (25.8) 43 (29.2) Three 0 24 (16.3) priate nesting location. As they arrive during Four 0 4 (2.7) spring migration and presumablyare attracted Five 0 1 (0.7) by the singingand other displaysof territorial males, females discover suitable nest sites and begin to build. I do not know what involve- ment the male has in nest-site selection, but I The meannumber of Dickcisselyoung/male, do know that the behavioral interaction of the or the total number of Dickcissel and cowbird male with the femaleduring this criticalperiod young/male,is not significantlydifferent when is intense (Schartz and Zimmerman 1971). It is maleswith one or two nestsare comparedbe- also not known whether the female chooses a tween habitats, nor do any significantdiffer- territory on the basis of male phenotype ences exist among prairie males in the num- (Weatherhead and Robertson 1979), and thus bersof Dickcisselsor Dickcisselsplus cowbirds incidentally obtains accessto a suitable nest produced. In the oldfield, on the other hand, site, or selectsthe nesting situation directly there are significant differences between the using criteria of microhabitat quality (Searcy mean numbers of young produced by males 1979). Of course, it could be both, as Weath- with just one nest and males with two nests, erhead and Robertsonsuggest. Yet, if females both for Dickcissels fledged (Student's t = were choosingsolely on the basisof the quality 3.194, df = 94, P < 0.05) and for Dickcissel of the male rather than that of the nesting hab- and cowbird young together (Student's t = itat, somefemales would certainlyend up with 3.194, df = 94, P < 0.01). Males with 3 or more unsuitable nest sites as harem size increased, nests also produce significantly more young assuminga limited number of nest sites, and than males with 1 nest, but the numbers of there would be a decline in nesting success. young producedby maleswith 2, 3, 4, 5, and This does not happen, however (Tables2 and 6 nestsare not significantlydifferent from each 4). other. Furthermore,there is a significantpos- Becausethere is no significantdifference be- itive association between the number of nests tween survival rates of oldfield and prairie of a male and the numbers of both Dickcissels nests (Table 1) nor any difference between

T,•BLE4. Daily nestsurvival rates (+95% C.L.) andproduction of fledglings/male(x + SE)according to the number of femalesattracted to the territory.

Number of different females/male

I 2 3 4 5and6

Oldfield Number of males 70 26 25 18 8 Survival rate 0.930 + 0.02 0.940 + 0.02 0.944 + 0.02 0.913 + 0.02 0.912 + 0.03 Dickcisselyoung 1.00 + 0.19 1.85 + 0.45 1.64 + 0.50 2.00 + 0.48 3.62 + 1.13 Dickcissel and Cowbird young 1.20 + 0.21 2.62 + 0.47 3.24 + 0.57 2.67 + 0.62 4.12 + 1.12 Prairie Number of males 22 9 Survival rate 0.954 + 0.03 0.931 + 0.04 Dickcisselyoung 1.18 + 0.31 2.44 + 0.82 Dickcissel and Cowbird young 1.73 + 0.45 3.22 + 1.23 April1982] HabitatSuitability in Dickcissels 297 these two habitats in the number of Dickcissel bird (Agelaiusphoeniceus) males to obtain nu- young producedby each female, the two hab- merous mates (Searcy 1979, Yasukawa 1979), itats have equal suitability for the female. Any but supporting data are weak or lacking. Sim- individual nestinglocation in the prairie is just ilar factorsmay assistmale Dickcisselsin ob- asgood as a nestingsite in an oldfield.Females taining good territories, and they are being fill up the habitat or, more precisely, occupy tested (Finck MS). adequate nest sites accordingto their avail- Females,on the other hand, simply distrib- ability. The distribution of femalesis not de- ute themselvesas a function of the availability termined by the territory boundaries of the of nesting situations, with nests built in any males nor in any density-dependent manner habitat and in any territory being equally suit- by the numbersof femalesthemselves (for ex- able to the female as long as the site is adequate ample, female-female aggression has never for nesting. She loses nothing by choosinga been observed). nest site within the territory of a male that is An important corollary to this condusion is alreadymated. She only losesif she choosesto that femalesmated to polygynousmales are not mate with a male that is defending an area at a disadvantage(Tables 2 and 4). Such a re- lacking a suitable nesting environment. The lationship is to be expectedfrom the threshold regular occurrenceof bachelormales (Zimmer- model for the evolution of polygyny (Verner man 1966) suggeststhat females do not make 1964, Verner and Willson 1966, Orians 1969). this mistakevery often. If femalesspread themselves across the land- scape simply as a function of the distribution ACKNOWLEDGMENTS of suitablenesting sites, it is importantfor the This work was funded by grant GB-6087from the male to include as many suitable sites as pos- National ScienceFoundation and NSF Undergradu- sible in his territory. Prairie males do just as ate ResearchParticipation programs during several well at this as oldfield males when they are successivesummers. Support was also provided by monogamousor bigamous(Table 4). The en- the ChapmanMemorial Fund of the American Mu- hanced suitability of the oldfield habitat for seumof Natural History. Permissionto conductthis males, however, lies in its providing for higher work on the preferred study sites was generously ordersof polygyny, which augmentmale pro- granted by the Commander, HeadquartersFort Ri- ductivity (Table 4) even though male density ley, and the Director of the Konza Prairie Research is greater. These data supportthe ideal domi- Natural Area. Such an extensivesample of nest rec- ords could never have been accumulated without the nance model for the distribution of male Dick- diligent effortsof Gil Blankespoor,Jeff Fergen, Elmer cissels (Fretwell and Lucas 1969). Finck, Scott Hatch, Harvard Townsend, and Steve Thus, three different patterns interact in the Wiegert. Numerousdiscussions with SteveFretwell distribution of Dickcissels across the available have been essentialin the development of this work habitat space.The heterogeneityof the vege- and reviews of the manuscriptby L. B. Best, J. F. tative substrate results in an uneven distribu- Wittenberger, and P. J. Weatherhead were quite tion of potentially suitable nesting situations helpful. within habitats as well as between habitats (Zimmerman 1971). The males' territories are LITERATURE CITED then superimposed upon the habitats, but BLANKESrOOR,G. W. 1970. The significanceof nest suitable nesting sites are not equitably appor- site and nest site microclimate for the Dickcissel, tioned among the males. Not all the territories Spiza americana.Unpublished Ph.D. disserta- in the same habitat are equal in quality, and tion, Manhattan, Kansas, Kansas State Univ. territories in different habitats also differ in CAREY,M., & V. NOLAN, JR. 1979. Population dy- quality (Zimmerman 1971). A male that con- namics of Indigo Buntings and the evolution of trols a high-quality territory with multiple avian polygyny. Evolution 33: 1180-1192. nesting situations will attract more mates and ELLIOTT,P. 1978. Cowbird parasitismin the Kansas tallgrassprairie. Auk 95: 161-167. gain greater fitness. The male's size, the sig- FRETWELL,S. D., & H. L. LUCAS. 1969. On territorial nalling value of his plumage, the aggressive- behavior and other factors influencing habitat nessof his behavior, his experience,and other distribution in birds. I. Theoretical develop- attributes have been suggestedas possiblefac- ment. Acta Biotheoretica 19: 16-36. tors that might explain observeddifferences in , & J. S. CALVER.1969. On territorial behavior the abilities of polygynousRed-winged Black- and other factorsaffecting habitat distribution 298 JOHNL. ZIMMERMAN [Auk, Vol. 99

in birds. II. Sex ratio variation in the Dickcissel. ing successin Red-winged Blackbirds.Auk 96: Acta Biotheoretica 19: 37-44. 353-363. HOLM, C. H. 1973. Breeding sexratios, territoriality, VERNER,J. 1964. Evolution of polygamyin the Long- and reproductive successin the Red-winged billed Marsh Wren. Evolution 18: 252-261. Blackbird. Ecology54: 356-365. --, & M. F. WILLSON. 1966. The influence of HUXLEY,J. 1934. A natural experiment in territorial habitats on mating systemsof North American instinct. Brit. Birds 27: 270-277. passerinebirds. Ecology47: 143-147. JOHNSON,D. H. 1979. Estimating nest success:the WEATHERHEAD,P. J., & R. J. ROBERTSON.1977. Har- May•ield method and an alternative.Auk 96: em size, territory quality and reproductive suc- 651-661. cess in the Red-winged Blackbird (Agelaius KLUYVER,H. N., & L. TINBERGEN.1953. Territory phoeniceus).Can. J. Zool. 55: 1261-1267. and the regulation of density in titmice. Arch. --, & --. 1979. Offspring quality and the Neerl. Zool. 10: 265-289. polygyny threshold: "the sexy son hypothesis." LACK,D. 1964. A long-term study of the Great Tit Amer. Natur. 113: 201-208. (Parus major). J. Anim. Ecol. 33 (suppl.): WHITHAM, T. G. 1980. The theory of habitat selec- 159-173. tion: examined and extended using Pemphigus MAYI•IELD,H. 1961. Nesting successcalculated from aphids. Amer. Natur. 115: 449-466. exposure. Wilson Bull. 73: 255-261. WITTENBERGER,J. F. 1976. The ecologicalfactors se- 1975. Suggestionsfor calculatingnest suc- lecting for polygyny in altricial birds. Amer. cess. Wilson Bull. 87: 456-466. Natur. 110: 779-799. ORIANS,G. H. 1969. On the evolutionof mating sys- --. 1980. Vegetative structure,food supply and tems in birds and mammals. Amer. Natur. 103: polygyny in Bobolinks(Dolichonyx oryzivorus). 589-603. Ecology61: 140-150. PLESZCZYNSKA,W. K. 1978. Microgeographic pre- YASUKAWA,K. 1979. Territory establishmentin Red- diction of polygynyin the Lark Bunting. Science winged Blackbirds: importance of aggressive 201: 935-937. behavior and experience.Condor 81: 258-264. ROBBINS, C. S., & W. T. VAN VELZ•N. 1969. The ZIMMERMAN,J. L. 1965. Bioenergeticsof the Dick- Breeding Bird Survey 1967 and 1968. Bur. Sport cissel, Spiza americana. Physiol. Zool. 38: Fish. Wildl. Spec.Sci. Rept., Wildl. No. 124. 370-389. SCHARTZ,g. L., & J. L. ZIMMERIViAN.1971. The time --. 1966. Polygyny in the Dickcissel. Auk 83: and energy budget of the male Dickcissel(Spiza 534-546. americana). Condor 73: 65-73. --. 1971. The territory and its density depen- SEARCY,W. A. 1979. Male characteristicsand p•ir- dent effectin Spizaamericana. Auk 88: 591-612. A STUDY OF FASTING IN TREE SPARROWS (SPIZELLA ARBOREA) AND DARK-EYED JUNCOS (JUNCO HYEMALIS): ECOLOGICAL IMPLICATIONS

MIKI M. STUEBE1 AND ELLEN D. KETTER$ON Departmenti•f Biology,Indiana University, Bloomington, Indiana 47405 USA

ABSTm•CT.--Despitesimilarities in winter distribution, habitat selection,and food choice, Dark-eyed Juncos(]unco hyemalis) and Tree Sparrows (Spizellaarborea) differ in the extent to which they storefat duringwinter, with juncosaccumulating greater stores. Anticipating that Tree Sparrowsmight have somemeans of conservingenergy during fastingand thus sufferno disadvantagewhen weatherprevents feeding, we comparedthe speciesfor weight loss, body temperature,and locomotoractivity during fasting and noted relative fasting endurance.Because both speciesexhibit geographicvariation in sexratio during winter, we also made sexualcomparisons, anticipating that maleswould be able to fast longer than females.The speciesresponded similarly to fasting by (a) lowering body temperature,es- peciallyat night, and (b) becominghyperactive, progressively more so as fasting time in- creased.Tree Sparrowsdid not exhibit theseresponses to a greaterdegree (although they becamehyperactive sooner) and were not ableto fastas long as juncos. No sexualdifferences in fastingability were observed.Because the species-specificdifference in tendencytoward fat accumulationcannot be attributedto differencesin energyexpenditure while fasting,at leastin the laboratory,other explanationsare considered.Received 22 June1981, accepted 21 September1981.

TRaE Sparrows and Dark-eyed Juncos that and despitethe fact that populationswintering winter in the northern half of the eastern at the same location encounter identical weath- United States have similar ground-feeding er, the two speciesdiffer in the reservesof en- habits and overlapping diets and sometimes ergy stored as fat. Juncoshave been reported forage together in mixed flocks (West 1967, to store fat in an amount that is 12-17% of their Bent 1968, Willson 1971, Pulliam and Enders mean wet winter weight (Helms et al. 1967); 1971, Coulter pers. comm.). Their ability to by comparison,the figure for Tree Sparrowsis meet energy requirements is surely severely only 5-12% (Helms and Smythe 1969). This tested during winter at times when cold is in- differenceled us to testexperimentally the pre- tense, nights are long, and snow covers food diction that Tree Sparrows either (1) cannot for prolonged periods or stormsprevent feed- endurefasting for as long asjuncos, or (2) com- ing. Like many other speciesthat winter in the pensatefor their reducedenergy stores by con- north-temperate zone, juncos and Tree Spar- serving energy in ways not used by juncos. rows increase deposits of body fat in winter Fastingendurance is determined by the ratio (Helms et al. 1967, Helms and Smythe 1969), of energy stored to the rate of energy utilized, a response that is commonly accepted as an and, becauseper-gram metabolicrate is neg- adaptation to provide energy during forced atively correlatedwith body size, fasting en- fasting (e.g. King 1972, Ketterson and King durance is theoreticallyaffected by body size 1977). In both species,the sexestend to sepa- (Calder 1974). Thus, larger species,or larger rate by latitude in winter, with femalessettling size classeswithin a species,might be expect- in milder climates (Ketterson and Nolan 1976, ed to be able to fast longer than smaller ones 1979, unpubl. data), a behavior that could be if the individuals being compared store fat in related in part to sexualdifferences in the abil- similar proportion to their body weight. Be- ity to withstand severe weather. Yet despite cause juncos are somewhat larger than Tree the similarities of these two small sparrows Sparrows, they might therefore be expectedto show greater fasting endurance, even in the • Present address: Department of Biology, Utah absenceof their disproportionate fat stores al- State University, Logan, Utah 84322USA. ready mentioned. Further, both species are

299 The Auk 99: 299-308. April 1982 300 $TVEBEAND KETTERSON [Auk, Vol. 99 sexuallydimorphic in size, and as no signifi- 1949, Hart 1962, King and Farner 1966, Helms cant sexual differences in fat stores(as a per- and Smythe 1969)suggests that studiesof fast- centageof body weight) have been observed ing in birds newly taken from the wild may (Helms et al. 1967, Helms and Smythe 1969), yield more reliable results. fasting endurancein both speciesmight also We note here (and discuss below) that the be greaterin the larger sex,the male. ability to fast is only one factorthat might af- Decreasesin body temperatureare for birds fect winter survivaland that carryingexcess fat a commonmeans of savingenergy during fast- may have its costs(Helms and Smythe 1969). ing (Baldwin and Kendeigh 1932, Biebach Evidence that winter fat levels may sometimes 1977, deGraw and Snelling pers. comm., Ket- be regulated below maximum capacity is pro- tersonand King 1977). Hyperactivity, presum- vided by reportsthat somebird speciesare fat- ably the resultof hunger, is alsotypical of fast- ter preceding migration than during midwin- ed, cagedbirds and in the field would lead to ter (Linsdale and Sumner 1934, Baumgartner more active searchingfor food (Wagner 1937, 1938, Wolfson 1945, Odum 1949, King et al. Eyster1954, Merkel 1966,deGraw and Snelling 1963, King 1972) and by the observationthat pers. comm., Ketterson and King 1977). The juncos of equal body dimensions are fatter in extent to which hyperactivityincurs an energy winter at northern than at southern latitudes cost is not known, becauseheat generatedby (Ketterson and Nolan unpubl. data). Thus, a exercisemay fully or partially substitute for final objectiveof this paper is to considercir- thermoregulatorycosts, depending on the level cumstancesthat might selectagainst maximal of activity, air temperature,and perhapsnu- fat storage. tritional state (Ketterson and King 1977, Pala- METHODS dino 1979). These considerationssuggested that Tree Sparrowscould employ marked body Ten Tree Sparrowsand 18 juncoswere confinedin temperaturelability to saveenergy during fast- a windless room at 8-10øC on a 10L:14D photo- ing but might or might not conserveenergy by period and deprived of food until death seemedim- diminished locomotor activity. In any case, a minent (as defined below). Body temperature(To) of difference between the two speciesin locomo- somebirds and weights of all were taken every 4 h [or sometimes8 h, becauserecordings (hereafter RT, for responseto fasting would call for further recording time) at 0330 were irregular], while loco- investigation. motoractivity was monitoredcontinuously. Controls In the work reported here, we first deter- were birds treated similarly except that they were mined that juncos and Tree Sparrows at given unlimited accessto a mixture of turkey starter Bloomington,Indiana conformedto the gen- mash, millet, ground eggs, beef, and carrots. All eral patternof greaterfat storageby juncosand birds were provided with water and were cagedin- then investigatedthe following questions: dividually so that none could seeany other bird. Some controls did not eat and lost weight at the 1. Did fasting endurance of the two species same rate as experimentals, a response to captivity differ? that we have occasionallyobserved in various other 2. Regardlessof the answer to 1, did fasting seed-eatingfringillids. To eliminatethe confounding Tree Sparrows develop hyperactivity to a effects of data from such individuals, we have omit- greater or lesserextent than fasting juncos ted their weights, Tb's, and activity from all calcu- or exhibit greater or less lability in body lations. Controls, therefore, included only individ- temperature? uals whose percentageweight loss, if any, was at least 20% less than that of the experimentalbird of 3. Recallingthe sexualdifference in distribu- the same species that exhibited the smallest per- tion in each speciesin winter, did fasting centageweight loss.All controls,so defined,gained affectthe sexesdifferently in either or both weight at least once between successiveRTs, indi- species? cating that they were eating. Seven Tree Sparrows These questionswere investigatedusing new- and 14 juncos satisfiedthe criteria for inclusion as controls,and 3 Tree Sparrowsand 8 juncosdid not. ly caughtbirds (as in Kendeigh1945, Biebach We now think that had the birds not been visually 1977)rather than captive birds (as in Ivacic and isolated, this loss of data could have been avoided. Labisky 1973, deGraw and Snelling pers. Six experimentswere conductedbetween 13 Jan- comm., Ketterson and King 1977). Evidence uary and 15 February1981. On the day eachwas to that captive birds differ from wild birds in fat begin, we capturedas many juncosand Tree Spar- depositionand cold acclimatization(Kendeigh rows as possible (extremes4 and 16) between 1450 April 1982] Fastingin TreeSparrows and Dark-eyed Juncos 301

(141

{0) • (0) (0) Class 0 Class Class 2 Class 3 Class 4 Class 5

Less Fat .• Visible Fat I. More Fat Fig. l. Percentageof juncosand Tree Sparrows according to fat classat capture.Sample sizes as indicated in parenthesesinclude both the experimentalsubjects and thosecontrols that satisifedcriteria for inclusion (seetext). Light bars = Dark-eyedJuncos, dark bars = Tree Sparrows.

and 1830. These we weighed immediately in the as "initial weight." Birds were transferredto the cold field (Pesolaspring balance, 50-g capacity, nearest room between 2045 and 2215. Eachwas placed in a 0.1 g) and transportedto the laboratory,where we small cage(22 cm x 26 cm x 28 cm) equipped with measuredthe flattenedwing and classedfor fat ac- a singleperch attached to a microswitch;each hop cording to a modified version of the scaledescribed on the perch was recordedon an Esterline-Angus by Helms and Drury (1960).Juncos we sexedat the event recorder. In order to minimize the effects of time of captureusing the criteria of Kettersonand disturbance,Tb measurementswere made on only a Nolan (1976), whereasTree Sparrowswere sexedby subsetof subjects(maximum, 10) in each experi- laparotomyafter the completionof all experiments. ment; subjectswere taken from the cold room one Experimentswere conductedtoo late in the year to at a time in a prescribed order and a Schultheis permit us to age subjects.Birds destinedto be fasted quick-respondingthermometer was insertedinto the ot fed were assignedto theseclasses randomly. cloacafor 30 s (nearest0.1øC). In the firstexperiment, Each bird was weighed again 4 h after capture an experimental bird's Tb was measured first, then when the gut may be assumedto have been empty a controlbird's, then an experimental's;in the sec- (Kontongiannis1967), and this weight we designate ond experiment, a control bird was measured first.

TABLE1. Comparisonsof fastedDark-eyed Juncos and TreeSparrows in a 10øCcold room, sexespooled.

Juncos Tree Sparrows Measurement (n = 18)a (n = 10)a t P Wing length (mm)•' 79.7 + 0.72 75.8 + 0.84 3.34 <0.01 Initial weight (g)C 20.78 + 0.39 18.27 + 0.32 4.36 <0.001 Final weight (g)d 15.63 + 0.22 14.74 + 0.25 2.54 <0.05 Weight loss (g)e 5.16 _+0.24 3.44 + 0.19 3.52 <0.01 Percentageweight losst 24.5 + 1.25 17.6 + 0.90 3.18 <0.01 Rate of loss (g/h)• 0.120 + 0.004 0.120 + 0.008 0.07 n.s. Fasting endurance(h) h 43.19 + 2.79 29.51 + 2.02 3.38 <0.01

Mean 4- 1 SE. Flattenedwing. Four h after capture(i.e. gut contentsvoided). At time bird believed to be within 4 h of death. (Initial weight - final weight). (Weightloss/initial weight) x 100. (Weight 1oss/hendured). Numberof hourselapsed between time initial weight taken (i.e. 4 h aftercapture) and termination of experiment(see text). 302 $TVEBEAi•ID KITtERSOi•I [Auk, Vol. 99

80

7C

• 5c

,? 4a

20

15.0 16.0 17.0 18.0 19.0 20.0 21.0 22.0 23.0 24.0 25.0 26.0

Initial Weight Fig. 2. Relationshipbetween initial weight and fastingendurance for Tree Sparrows(*) and juncos (0). In regressionequations, Y standsfor fasting enduranceand IW for initial weight (seetext).

The alternationof experimentalsand controlsand the RTs, specifically,the 1.5-h blockbeginning 2.5 h be- order in which the first subjectwas selectedwas con- fore the next RT. Blocks are assumed to have been tinued for the remainder of the experiments. When representative of the 4-h intervals. Becausecaged T•,measurement was completed,body weights of all birds develop idiosyncratic patterns of movement individuals in the experiment were taken as before that may only occasionallyinclude landing on the and recorded. perch, within each 1.5-h block we countedonly the For experimental birds, the experiment was ter- number of 30-s intervals in which the individual ac- minated at the RT when death appeared imminent tivated its perch. Thus, activity scorescould range using the standardsof Ketterson and King (1977). between 180 and 0. Occasionallya perch would stick For controls, the experiment was terminated when or a pen would fail to write; consequently,the num- the last experimental was removed. Ketterson and ber of individuals comprising a sample sometimes King (1977)report that deathcould be expectedwith- varied among RTs. in 4 h in White-crowned Sparrows (Zonotrichialeu- cophrysgambelii) that erected the feathers of the head, breast,and back, were very lethargic,held the RESULTS eyes partly closed,and sometimesrefused food when Fat class.--Aspredicted from the findingsof presented.Fasted juncos and Tree Sparrowsbehaved Helms and Drury (1960), Helms et al. (1967), similarly and also tucked the head into the scapular feathers when severely stressed. Tree Sparrows and Helms and Smythe (1969), juncos showed showed all signs of stressto a greater degree than more visible fat when capturedthan did Tree juncos. The applicationof the foregoingsubjective Sparrows (Fig. 1). The actual values of the fat standard to determine when death was approaching classesreported here are lower than those of introduces some error into our results, but we be- Helms and Drury (1960), but we believe that lieve that it is similar for the two species.Equal ef- this is due to a difference in application of the forts were made to restore experimental birds at the 0-5 scaleand not to a differencein fat. In any end of experiments(they were warmed and supplied case, the difference between the species is with food, including hand feeding of glucoseand comparable. water), but two juncos (11%) and four (40%) Tree Sparrows died within 24 h after removal from their Fastingendurance and weight loss.--The mean cages.Fasting endurance was calculatedas the num- duration of fastingof the experimentaljuncos ber of hours elapsed between the time the bird's gut was 43 h and the extremes 34 h and 67 h; mean could be assumedempty (4 h after capture) and the weight losswas 25% of initial weight and the time that death was judged imminent. extremes 19% and 36% (Table 1). In contrast, To summarize activity data, we selectedblocks of mean fasting enduranceof the Tree Sparrows time 1.5 h in length from each 4-h interval between was 30 h and the extremes 21 h and 40 h; mean April1982] Fastingin TreeSparrows and Dark-eyed Juncos 303

42.0

41.0

11

40 .C

39.0

38.0

37.0 2330 0330 07•0 1130 1530 1930 2•30 0330 0 •30 11•301530 1930 2330 Time

Fig. 3. Mean body temperatureof fasted(¸) and fed (0) juncosand fasted(/N) and fed (&) Tree Sparrows kept in a 10øCcold room. Vertical lines represent +2 SE. Sample size is indicated below the line. Open horizontal bars indicate daylight hours; closedbars indicate dark hours. weight loss was 18% and the extremes14% When the effectsof fasting first became ap- and 24% (Table 1). parent (after 0730 on the first day of fasting), The mean rate of weight loss per hour was fasted birds of both species tended to drop identical in the two species. In each species, their temperaturesbelow thoseof controls,and initial weight significantly predicted fasting the differenceswere greater at night (Fig. 3, endurance:in the juncos,fasting endurance= Table 2, experimental vs. control body tem- 5.65 (initial weight) - 74.30, r 2 = 0.62; in Tree peraturesdiffered significantlyin both species Sparrows, fasting endurance= 3.38 (initial exceptjuncos, 1530 day 1, 0730 and 1530 day weight) - 41.37, r 2 = 0.39 (Fig. 2). The differ- 2; Tree Sparrows 2330 night 2, one-tailed ence in the slope of the lines suggeststhat in Mann-Whitney U-tests). juncosa given weight changereflects a greater Fastedbirds showed no species-specificdif- increase or decreasein fasting endurance than ference in the extent to which T• dropped: T•'s in Tree Sparrows, but the difference was not differed significantly only at 1530 on day 1 significant(in juncos, sb= 1.11, in Tree Spar- when the mean for fasted Tree Sparrows was rows, sb = 1.73 where s• is the standard error below that of fasted juncos. (For some un- of the regressioncoefficient). known reason,juncos had lower body temper- Body temperature.-•Controljuncos and Tree atures than Tree Sparrowsat 2330 night 1, be- Sparrows had very similar Tds. Differences fore the effectsof prolongedfasting could have were recorded at 2330 on nights 1 and 2 (Fig. set in.) 3, Table 2), but the differences were inconsis- The extreme Tb's recorded in fasted juncos tent. In one case, the mean for Tree Sparrows were 35.2øCand 43.4øC;the comparablefigures was greater;in the other, the meanfor juncos for control juncos were 38.9øC and 43.7øC. In was greater. fasted Tree Sparrows, extreme Tds were 35.6øC 304 STUEBEAND KETTERSON [Auk, Vol. 99

TABLE2. Body temperature comparisonsof fasted (experimental) and fed (control) Dark-eyed Juncosand Tree Sparrows.a'b

Between species Within species Experimentaljuncos Experimental vs. Experimental vs. Control juncos vs. vs. experimental control juncos control Tree Sparrows control Tree Sparrows Tree Sparrows Medians/ne, nc Medians/ne, nc Medians/nj, n.• Medians/nj, ns Time (U,P) (U,P) (U,P) (U,P) 2330 39.4, 39.6/6, 8 41.0, 40.6/6, 4 39.6, 40.6/8, 4 39.4, 41.0/6, 6 (15, n.s.) (9, n.s.) (3, P < 0.014) (0, P < 0.001) 0730 41.4, 41.4/8, 9 42.2, 42.2/7, 4 41.4, 42.2/9, 4 41.1, 42.2/8, 7 (36, n.s.) (13.5, n.s.) (14, n.s.) (18.5, n.s.) 1130 41.1, 42.2/8, 4 41.7, 42.7/7, 4 42.2, 42.7/4, 4 41.1, 41.7/8, 7 (4.5, P < 0.030) (1, P < 0.006) (5, n.s.) (15, n.s.) 1530 42.4, 42.4/7, 3 42.0, 43.0/6, 3 42.4, 43.0/3, 3 42.4, 42.0/7, 6 (10, n.s.) (0.5, P < 0.018) (2, n.s.) (7, P < 0.026) 1930 39.3, 40.8/8, 4 38.5, 41.4/7, 4 40.8, 41.4/4, 4 39.3, 38.5/8, 7 (4.5, P < 0.030) (0, P < 0.003) (6.5, n.s.) (17, n.s.) 2330 38.8, 41.4/11, 5 39.4, 40.2/6, 4 41.4, 40.2/5, 4 38.8, 39.4/11, 6 (4.5, P < 0.010) (6, n.s.) (2, P < 0.032) (29, n.s.) 0330 38.6, 40.5/7, 3 38.6, 40.0/7, 2 (2, P < 0.033) (2, n.s.) 0730 42.0, 42.2/11, 4 42.2, 41.8/4, 4 (16, n.s.) (8, n.s.) 1130 41.0, 42.3/6, 6 42.3, 42.1/6, 3 (5.5, P < 0.026) (8.5, n.s.) 1530 40.9, 42.6/3, 6 42.6, 42.8/6, 2 (4.5, n.s.) (5, n.s.) 1930 -- 42.2, 41.7/4, 2 (4, n.s.)

a n• = samplesize of experimentalbirds of indicatedspecies; n• = sample size of controls;n• = samplesize of juncosof indicated category (i.e. experimentalsor controls);ns = samplesize of Tree Sparrows. b U values are the Mann-Whitney statistic,P values are one-tailedprobabilities from Siegal (1956).

and 42.8øC, and in control Tree Sparrows, full day of fasting. For both species,once the 38.1øC and 43.8øC. difference between experimentalsand controls Activity.--We had expectednocturnal activ- appeared,it remained until the final measure- ity in fasted individuals, because previous ments were made and even tended to increase workers had reported it [Ketterson and King progressively. 1977, C. W. Helms pers. obs. of White-throated Sexualdifferences.--In Table 3 there is a com- Sparrows(Zonotrichia albicollis)], but virtually parison for each speciesof experimentalbirds none was observed.During the day the fasted accordingto sex. Males of both specieshad birds of both species were more active than longer wings than females, male juncos had controls(Fig. 4 and 5, Mann-Whitney U-tests). significantlyheavier initial weights and final When the activity of experimentalsand con- weights, and male and female Tree Sparrows trols is comparedfor each 1.5-h block, fasting did not differ significantly in weight either at Tree Sparrows became significantlymore ac- the outset or the termination of experiments. tive than their controls beginning at 1300 on In neither speciesdid the sexesdiffer signifi- the first full day of fasting. By comparison, cantly in absolute weight loss, percentage fasted juncos first becamesignificantly more weight loss, or rate of weight loss. Further- active than their controls at 1300 on the second more, and contrary to expectation (Ketterson April 1982] Fastingin TreeSparrows and Dark-eyed Juncos 305

150

130

70

8o 7o 6o

5O 09001030 13001430 ,'• 09001030 Day 1 30 Fig. 5. Daytime activity of fasted and fed Dark-

10 eyed Juncos in a 10øC cold room. Dark bars = fasted juncos, light bars = fed juncos. Sample size 0900-1030 'r300-1430 0900- 'r030 is indicated in parentheses.

Fig. 4. Daytime activity of fasted and fed Tree Sparrows in a 10øC cold room. Dark bars = fasted food becameavailable again during or imme- Tree Sparrows,light bars = fed Tree Sparrows.Sam- diately after the second night of fasting, ple size is indicated in parentheses. whereasTree Sparrows,even if they had access to food late in the day of the first day of fasting, would be forced to endure anotherlong night and presumablywould die. and Nolan 1976, 1978; Ketterson and King We found no evidence that Tree Sparrows 1977), no significantsexual difference in fast- reducemetabolic expenditure during fastingto ing endurancewas detected. a greater degree than juncos. Unlike Black- DISCUSSION capped Chickadees(Parus atricapillus), which regulate nocturnal T• about 10-12øC below The finding that the average experimental diurnal T• (Chaplin 1974), our Tree Sparrows junco could fast for about 43 h at an ambient dropped their T• by only about 1-2øC, i.e. no temperaturea few degreesabove freezing and morethan juncos. Neither was therean activity that the averageTree Sparrowcould last about responsethat would seem to result in energy 30 h is consistentwith the determination by saving: Tree Sparrows, like juncos, became Kettersonand King (1977) that the mean fast- hyperactive. Interestingly, the timing of the ing enduranceof another sparrow,the White- hyperactivity--later in juncos--implies that it crowned Sparrow, is about 38.6 h. White- is not triggeredby the time sincefeeding oc- crowned Sparrows typically carry fat reserves curred, but rather at some level of remaining (as a percentageof body weight) larger than energy stores. Although, as indicated above, those of Tree Sparrows but smaller than those the activity response shown by a fasted bird of juncos (King and Farner 1966, Helms et al. does not necessarilyimply a concurrent in- 1967, Helms and Smythe 1969). crease in energy expenditure, the Tree Spar- Becauseour experimentsbegan late in the row's early increaseof activity cannot be con- day when fat stores are near their peak in the struedas an unusualenergy-saving adaptation daily fat cycle,our resultssuggest that free-liv- to permit it to carry lower fat storesthan jun- ing juncoswhose food first becameunavailable COS. in the late afternoonmight be expectedto last The reasons for the difference in fat levels of that night and the following day and night. Tree Sparrowsand juncos remain obscure,but Tree Sparrows under the same conditions four possibilitiesmay deserveattention. The might last that night and the following day. first two relate to the relative need for stored Juncosunder conditions comparableto those energy reservesin the two species. of the experiment would probably survive if First, under natural conditions, the species 306 STVE•EAur• KœTtœRSOU [Auk,Vol. 99 may differ in the actual cost of thermoregula- tion due to differences in microhabitats select- ed for roostingor a tendencytoward nocturnal huddling, i.e. differencesthat would not be apparentwhen housedsingly under laboratory conditions. We find no evidence in the litera- ture that juncos or Tree Sparrows huddle at night, however, as do somespecies (Kendeigh 1945, Frazer and Nolan 1959, King and Farner 1966, King 1972, Haftorn 1972, Chaplin 1974), and both Kendeigh (1945) and Bent (1968)de- scribe roosting behavior that seemsmuch the samefor juncosand Tree Sparrows.Kendeigh +1 +1 +1 +1 +l +1 +1 reports that "the slate-coloredjunco and tree groundsparrowvegetation normally spendor in the thickets."night inWe dense ob~ servedboth juncos and Tree Sparrows inwin- ter roostsin a large outdooraviary; the behav- ior and locationschosen by the two seemed aboutthe same,and no huddlingwas observed in eitherspecies. As a secondconsideration regardingtherelative need for stored energy in thetwo species, the feeding behavior of Tree Sparrowsmay be sufficientlyflexible to permit them to forage and find food in situations in whichjuncos must fast. For example,Tree Sparrowsmay turn more readily than juncos from ground-feedingto foragingfrom plants emerging from snow cover. Reports of the feeding habits of the two species(Knappen 1934,Baumgartner 1937,Bent 1968, Coulter pers.comm.) do not revealwhether Tree Spar- rowshave such an advantage,but the subject warrantsinvestigation. The third and fourth possibilitiesthat may accountfor the difference in stored fat are re- fatedto the idea that fat storagemay impose costsand that the magnitudeof the costmay vary as a functionof the habitatsselected by aspecies oras a function ofits body size. That +1 +1 +l +1 +1 +1 +l is, it maybe disadvantageous to betoo fat (Helms and Smythe1969). Thus, if the foraging locations of Tree Sparrows make them more vulnerable than juncosto avian predatorsand if an increasein fat storeslowers the proba~ bilityof escape, perhaps through lowered agil- ity, TreeSparrows might sacrifice a measure of fasting endurancefor the sake of increased agility. Accordingto Helms and Drury (1960), Tree Sparrowsdo feed in moreopen areasthan juncos,and we too have found that juncos stay nearer to coverthan do Tree Sparrows.Studies on the relative agility of the two speciesat var- ious levels of fat deposition are currently un- April1982] Fastingin TreeSparrows and Dark-eyed Juncos 307 derway. Fourth, the cost of locomotion may ratio must be viewed with increased skepti- increase with level of fat deposition, and the cism (Myers 1981).In any case,it seemsworth extent of the increasecould be greaterto the emphasizing that, becausefat stores for a smaller-bodied Tree Sparrow. Paladino and speciesdiffer from time to time and place to King (1979)have estimatedthe costof transport place, it is not possibleto characterizejust one in terrestriallocomotion (cm 3 O2 (g - km) •) value for a species'fasting endurance. as equal to 3.2 (gramsbody mass)-ø'2•. Thus, Althoughthe reasonsfor the differentlevels if an individual'sbody weight were to increase of winter fat in Tree Sparrows and Dark-eyed from 15 to 20 g, per-gram cost of transport Juncoshave not been clarified, what is clear is would decreasefrom 1.56 to 1.45, but per-bird that the two speciesrespond to fastingin the cost would increase from 23.7 to 29.4 cm 3 O2 same manner and that Tree Sparrows, when km-•, i.e. 19.2%. A comparableincrease in deprived of food, becomestressed sooner than weight from 21 to 26 g would alter the per-bird juncos unlessthey resort to someenergy-sav- cost from 30.5 to 35.7 cm • O2 km -• or 14.6%. ing behaviorthat we havenot yet been ableto Use of a similar equation for the cost of trans- detect. port describedby Fedak and Seeherman(1979) gives comparable results (18.7%, 14.3%). ACrNOWLEr)C•aEN?S Thesedifferences are obviouslysmall and may The senior author would like to thank the mem- have little biologicalsignificance. The relative bers of her undergraduatehonors thesis committee, cost, however, of hopping on a treadmill to P. A. McClure, V. Nolan Jr., and D. R. Whitehead, lean and fat individuals of both speciesis cur- all of whom read a version of this paper and made rently under investigation. useful suggestionsfor revisionßMonetary assistance The absenceof the expectedsexual differ- received from Indiana University Honors Division, ence in fasting endurance remains unex- the Department of Biology, and the College of Arts plained. Two points regarding our methods, and Sciencesis greatlyappreciatedß The work also however, make us uncertainabout the gener- received partial support from the National Science Foundation, grantsDEB-78-11982 and DEB-81-10457. ality of our data on this point. First, samples J. Blank, V. Nolan Jr., and R. Shermanprovided were small, and we note that the largersexual technical advice and/or assistance in the field; V. difference among juncos was in the expected Nolan Jr. also read and improved various drafts of direction.Second, the initial weight of subjects the manuscriptßC. W. Helms and J. R. King served differed between trials, presumably because as reviewersand made severalvery helpful sugges- the weather in the days precedingcapture ef- tionsß forts was, as is usual for Indiana, quite vari- able. When prior weather was cold and snowy (trials 1 and 4), subjectstended to be fatter than BArOWXN,S. D., & S.C. KEtqOE•CH.1932. Physiol- when prior temperatures were mild (trials 2, ogy of the temperature of birdsß Sci. Publ. 3, 5, and 6). This fact did not affectour species Cleveland Mus. Nat. Histß 3: 1-196. comparisons, because equal proportions of BAUMGARTNER,A.M. 1937. Food and feeding habits of the Tree Sparrow. Wilson Bull. 49: 65-80ß juncos and Tree Sparrows began each trial. ß 1938ßSeasonal variations in the Tree Spar- Had morefemales than malesbeen the subjects row. Auk 55: 603-613. of trials 1 and 4, while more males than females BENT, A. C. 1968ß Life histories of North American had been the subjectsof the other trials, the cardinals,grosbeaks, buntings, towhees,finch- failure to observe a sexual difference could es, sparrows, and alliesß U.Sß Nat. Mus. Bullß have been attributed to chance. Within No. 237, part 2. species,however, both sexeswere represented BIES•Ctt, V. H. 1977ßReduktion des Energiestoff- in each trial, so there was no obvious bias. Techselsund der K6rpertemperaturhungernder Still, the effect of the differences in initial Amseln (Turdus merula). J. Ornithol. 118: 294-300ß weight among trials was to increasethe vari- ance in fasting endurance, making statistical C•Lr)ER,W. A. 1974ßConsequences of body size for avian energeticsßPp. 86-151 in Avian energetics significanceless likely. For this reasonwe feel (R. A. Payntner, Ed.). Publ. Nuttall Ornithol. the question of a sexual difference in fasting Club, No. 15. endurance deserves further attention, but note C•AI'LIN, S. B. 1974ßDaily energeticsof the Black- that a causalrelationship between fasting en- cappedChickadee, Parus atricapillus, in winterß duranceand geographicvariation in winter sex J. CompßPhysiol. 89: 321-330ß 308 $TVEBEANr• KETtERSON [Auk, Vol. 99

E¾STER,M. B. 1954. Quantitative measurement of K•N•, J. R. 1972. Adaptive periodic fat storage in the influenceof photoperiod,temperature, and birds. Pp. 200-217 in Proc. 15th Ornithol. Congr. seasonon the activity of captive song birds. (K. H. Voous, Ed.). Leiden, Brill. Ecol. Monogr. 24: 1-28. , • D. S. FARHER.1966. The adaptive role of FEDAK,M. A., & H. J. SEEHERMAN.1979. Reapprais- winter fattening in the White-crowned Sparrow al of energeticsof locomotionshows identical with commentson its regulation. Amer. Natur. costin bipeds and quadrepedsincluding ostrich 100: 403-418. and horse. Nature 282: 713-716. , S. BARKER,& D. S. EARNER. 1963. A com- FRAZER,g., & V. NOLAN, JR. 1959. Communal roost- parisonof energyreserves during autumnaland ing by the Eastern Bluebird in winter. Bird- vernal migratory periods in the White-crowned Banding 30: 219-226. Sparrow,Zonotrichia Ieucophrys gambeIii. Ecolo- HAFTORN,S. 1972. Hypothermiaof tits in arcticwin- gy 44: 513-521. ter. Ornis Scandinavica 3: 153-166. KNAPPEN, P. 1934. Insects in the winter food of the HART, J. S. 1962. Seasonal acclimitization in four Tree Sparrow. Auk 51: 93. speciesof small wild birds. Physiol. Zool. 35: KONtOGIANNIS,J. E. 1967. Day and night changes 224-236. in body weight of the White-throatedSparrow, HELMS, C. W., & W. H. DRVRY, JR. 1960. Winter Zonotrichia aIbicoIIis. Auk 84: 390-395. and migratory weight and fat: field studies in LINSD^LE, J. M., & E. L. SUMNER. 1934. Winter some North American buntings. Bird-Banding weights of Golden-crowned and Fox sparrows. 31: 1-40. Condor 36: 107-112. , & R. B. SMYT•-IE.1969. Variation in major MERI•EL,F. W. 1966. The sequenceof eventsleading body componentsof the Tree Sparrow(SpizeIIa to migratory restlessness.Ostrich (suppl. 6): arborea)sampled within the winter range. Wil- 239-248. son Bull. 81: 281-291. MYERS,J.P. 1981. A test of three hypothesesfor , W. H. AVSSIKER, E. B. BOWER, & S. D. FRET- latitudinal segregationof the sexesin wintering WELL. 1967. A biometric study of major body birds. Can. J. Zool. 59: 1527-1534. componentsof the Slate-coloredJunco, Junco Or•vM, E. P. 1949. Weight variations in wintering hyemalis.Condor 69: 560-578. White-throated Sparrows in relation to temper- IvACIC, D. L., & R. F. LABISKY. 1973. Metabolic re- ature and migration. Wilson Bull. 61: 3-14. sponsesof Mourning Doves to short-termfood PALADINO,F. V. 1979. Energeticsof terrestrial lo- and temperaturestresses in winter. Wilson Bull. comotion in White-crowned Sparrows (Zono- 85: 182-196. trichialeucophorys gambeIii). Unpublished Ph.D. KENDE•CH,S.C. 1945. Resistanceto hungerin birds. dissertation.Pullman, Washington,Washington J. Wildl. Mgmt. 9: 217-226. State Univ. --. 1949. Effect of temperatureand seasonon , & J. R. K•Ng. 1979. Energeticcost of terres- energy resourcesof the English Sparrow. Auk trial locomotion: biped and quadruped runners 66: 113-127. compared.Rev. Can. Biol. 38: 321-323. KETTERSON,E. D., & V. NOLAN,JR. 1976. Geograph- PVLLIAM,H. R., & F. ENDERS.1971. The feeding ic variation and its climatic correlates in the sex ecologyof five sympatricfinch species.Ecology ratio of eastern-wintering Dark-eyed Juncos 52: 557-566. (Juncohyemalis hyemalis). Ecology 57: 679-693. S•EgAL,S. 1956. Nonparametric statisticsfor the , & --. 1978. Overnight weight loss in behavioral sciences. New York, McGraw-Hill. Dark-eyed Juncos(Junco hyemaIis). Auk 95: WAGNER, H. O. 1937. Der Einfluss von Aussenfak- 755-758. toren auf den Tagesrhythmusw•hrend der Zug- , & ---. 1979. Seasonal,annual, and geo- phase. Vogelzug 8: 47-54. graphicvariation in sexratio in wintering popu- W•LLSON, M. F. 1971. Seed selection in some North lationsof Dark-eyedJuncos (junco hyemaIis). Auk American finches. Condor 73: 415-429. 96: 532-536. WEst, G. C. 1967. Nutrition of Tree Sparrowsduring --, & J. R. KINe. 1977. Metabolic and behavioral winter in central Illinois. Ecology48: 58-67. responsesto fastingin the White-crownedSpar- WOLFSON,A. 1945. The role of the pituitary, fat de- row (ZonotrichiaIeucophrys gambeIii). Physiol. position, and body weight in bird migration. Zool. 50: 115-129. Condor 47: 95-147. NOTES ON THE MYOLOGY OF THE PELVIC LIMB IN KIWI (APTERYX) AND IN OTHER BIRDS

JAMESC. VANDEN BERGE IndianaUniversity School of Medicine,Northwest Center for MedicalEducation, 3400 Broadway,Gary, Indiana46408 USA

ABSTRACT.--Anatomicalvariants of the musculatureof the pelvic limb in Apteryx, based on a review of the published literature, are considered from the aspect of (1) errors in descriptivemyology, (2) probable misidentificationor misrepresentationof muscles,and (3) variationsin a given muscleset that have not been thoroughlyexamined in terms of a form-functioncomplex. A more accurateinterpretation of these relationshipsis suggested as a basisfor further studiesof the pelvic limb myologyin birds. Received2 March 1981, accepted30 June1981.

THE first account of the somatic musculature laeognathous" birds (see, e.g. Garrod and of any speciesof kiwi (Apteryx australis,A. Mitchell). The pelvic musculatureof the latter haastii,A. owenil) was apparentlybased on a is described by Gadow (1880) and Pycraft paper read by Richard Owen on 22 February (1900).More recently,Hudson et al. (1972) de- 1842 and published as a "descriptive portion scribe the appendicular musculature of tina- of this communication"in the Proceedingsof mous, a group that some considerto be allied theZoological Society of Londonof the sameyear with ratites. (There is some indication that the (pp. 22-41). The principal muscles described latter investigatorsalso dissectedat least one included the extensive, striated Musculi sub- specimenof Apteryx; see p. 248.) cutanei, the trunk musculature, shoulder mus- Although McGowan completed his recent cles, and the musclesof the pelvic limb. Later study in an attempt to resolve certain "short- some of this descriptive material was incor- comings"in descriptivemyology (what he lat- poratedinto a longer paper with the addition er equateswith "lapsi calami," p. 64), I pro- of plates of illustrations (1849). The latter was pose to review and discussat least three of his essentially reprinted in Owen's "Memoirs" descriptionsof morphologicalvariation in the (1879). pelvic musculatureof A. australisthat seem to In the 35 yr subsequentto these memoirs, differ significantlyfrom anatomicalvariants there were at least five additional papers that describedin other specimensof A. australis,in consideredthe somaticmusculature of Apteryx other speciesof kiwi, and in at least some other based, at least in part, upon independent dis- avian groups.I will alsoshow that certaincom- sections.These include, in chronologicalorder, ponents of the pelvic musculatureare appar- the well-known papersof Garrodon the thigh ently not describedby McGowan, or anyone muscles(1873, see Fig. 6, A. owenil)and on the else, in any specimensof the kiwi. deep plantar tendons in birds [1875, see Fig. One of theseanatomical variants, namely M. 2, A. "mantelli" (=australis)],the generalanat- iliotrochantericus medius, is a constituent of omy and embryologypaper of Parker (1891)as the expandedGarrod muscleformula (symbol well as that of Beddard (1899), selected com- C; see Hudson 1937: 12). I suggestthat this ments by Pycraft (1900), and finally the com- muscleis incompletelydescribed in many avi- prehensivestudy of Mitchell (1913)on the pe- an groups and is, therefore, not sufficiently roneal musculaturein birds ("three examples" well-known in terms of morphologicalrela- of Apteryx, p. 1043).The mostrecent paper on tionships to justify its use in terms of conclu- the subjectis that of McGowan (1979) on the sions regarding phylogeneticaffiliation. I will musculature of the pelvic limb in A. australis show that two other anatomical variants, in the based on a dissectionof two specimens. peronealmusculature and in the deep digital In many of the preceding studies, the so- flexor muscles,are most probably described matic musculature of kiwis is often described both inaccuratelyand incompletelyby Owen with reference to that of the "ratites" or "pa- and McGowan and, in the latter study, are the

309 The Auk 99: 309-315.April 1982 JAMESC. VA•DE• BERGE [Auk, Vol. 99 basis for erroneousstatements regarding pos- basis upon which Hudson (1937: 12) sug- sible phylogenetic affinities among Apteryx, gestedthat the variants of medius be recog- "ratites," and other birds. Finally, I will sug- nized by formula letter "C." Owen initially gest that a restudy of at least one muscle set, stated that "[medius] is peculiar to Apteryx, M. caudo-iliofemoralis,may serve as a test of and the precedingportion [i.e. cranialis]is ab- the evolution of functional modalities in a sent in most birds" (1842: 34), but later stated muscle complex and as a related test of recent that "one of them (i.e. either cranialis or medi- theoriesof paleognathmonophyly (Bock1963, us) is absent in most birds" (1879: 55). Intra- Cracraft 1974). specific variation in relative development of My review of these anatomicalvariants, from certain musclesdoes occur among birds (see, the aspectof descriptivemorphology, is based e.g. Raikow 1975,Raikow et al. 1979)and may on my interpretation of descriptive data and have occurredthen in McGowan's two speci- illustrationsin all previously cited references mens of A. australis. regarding Apteryx, my own previous dissec- I suggestthat the passageof a neurovascular tions of specimens exemplifying numerous bundle (N. coxalis cranialis, Breazile and Ya- avian groups (Hudson et al. 1966, 1969, with suda 1979) between the respective proximal Hudson's unpublished notes on Falconiformes attachments of the cranialis and medius mus- as well as Charadrii; Vanden Berge1970, 1975, cles(see George and Berger1966, Klemm 1969, 1976), and my own participation in the sub- Vanden Berge1970) is a principal morpholog- stantial review of avian musculature incorpo- ical criterion for assessingthe anatomical sta- rated in the first attempt to formulate a stan- tus of both musclesin any avian specimen.In dardized nomenclature (see Myologia in my experience, this criterion seems to be uni- "Nomina Anatomica Avium," Baumel et al. formly applicable among birds and is more 1979). "predictable" than suggested variations in continuity of the tendons of insertion in terms Mm. iliotrochantericus caudalis (="posterior"), of "presence" or "absence" of either muscle. cranialis (="anterior"), medius. Synonymy: Owen (1842, 1849, 1879), glutaeusmedius, minimus (cran- Until such time as this interrelationshipof a ialis and mediusincluded); Beddard (1899), glutaeus neurovascularbundle to proximal attachments secundus,tertius, quartus. of the two musclesis more fully described,the addition of M. iliotrochantericus medius as a M. iliotrochantericuscaudalis is apparently component of the Garrod leg-muscleformula the largest of these three musclesin Apteryx, will be of limited significancein terms of des- as in most other birds (but seeGeorge and Ber- ignating any degree of phylogenetic affinity ger 1966: 390); the insertion on Trochanter amongbirds (seeTable IX in Georgeand Ber- femoris (Baumel 1979a)is subtendedby a Bur- ger 1966). sa synovialis (Vanden Berge 1979) in A. One other componentof the expandedGar- australis(Owen 1842: 34) and most likely in rod leg-muscleformula, namely "V," the Vin- other birds as well in view of the probable culum tendinure flexorum, which binds the functional role of this muscle (Cracraft 1971). tendonof M. flexorperforans et perforatusdig- According to McGowan (1979: 51), M. ili- iti III to that of M. flexorperforatus digiti III as otrochantericus cranialis is absent in one of these two tendons traverse the foot, is of un- two specimensof A. australis;medius is pre- certain status in the kiwi: present, according sumably present in both. This variant seems to Beddard (1899: 398); absent, accordingto to be rather unusualamong birds generally.M. McGowan (1979:62); no relationshipdescribed iliotrochantericusmedius is more frequently in any of Owen's three papers. Furthermore, described as "absent" in birds, if it is not in- M. flexorhallucis brevis and M. adductordigiti dependentlydeveloped with respectto M. ili- II are two musclesthat no one describesin any otrochantericus cranialis, and, instead, is char- specimenof kiwi, at least to my knowledge. acterizedby some continuity of the respective Both muscles are described in some tinamous tendons of insertion if not the actual aponeu- (Hudson et al. 1972); M. adductordigiti II is rotic sheath of the contractile tissue itself. This apparently present in Rhea and Casuariusbut is the basis upon which Beddard (1899: 395) absent in Struthio (Gadow 1880). described the variation of the two muscles in A. australis and A. haastii, and this is also the Mm. fibularis(peroneus) longus and brevis. Syn- April 1982] KiwiMyology 311 onymy: longus is equivalent to "superficialis" and "Nomina Anatomica Avium," Annot. 121; see brevis to "profundus," as described by some in- also Frewein 1967 and Greenlee et al. 1975 for vestigators(see, e.g., Mitchell 1913). a discussionof the digital flexor tendons and their tendon sheaths as described for Gallus). According to McGowan (p. 72), the tendon Confusion in the proper use of the descriptive of insertion of M. peroneuslongus does not termsseems to be most probable.For example, form a conjoinedtendon with that of M. flexor Owen's description of the conjoined tendon perforatusdigiti III in his specimensof A. aus- and his illustrations in the 1849 paper differ tralis, althoughthis statementappears to con- from those in the memoirs of 1879. In all of the tradict his description of both muscles (pp. above papers,however, the tendonof M. pero- 60-61 and pp. 64-65) as well as his illustrations neuslongus does become conjoined with a tendon (Figs. 17, 18, 22, and 23). McGowan also states of a digital flexor muscle,most probably that that this anatomical variant represents a mor- of M. flexor perforatusdigiti III, in sharp con- phologicaldistinction between ratites and car- trast to McGowan's statement as previously inates and that this distinction is supportedby cited. the original descriptionin Gadow (1880). In contrastto the morphologicalvariation in The five principal morphologicalcharacter- M. peroneus longus in Apteryx, M. peroneus isticsof M. peroneuslongus in termsof a gen- brevis is said to be very much reducedin terms eral avian pattern (as exemplified in Chauna of relative development (Mitchell 1913: 1043) chavaria, Anhimidae) include (1) a large, and apparently absent in some specimens fleshy, superficialbelly on the craniolateralas- (Beddard 1899: 397). Owen, in fact, does not pect of the crus; (2) a deep attachmentalong describea "typical" brevis in any of his three the shaft of the fibula; (3) a tendinous connec- papers. tion (aponeurosis)to the Cartilagotibialis (see McGowan describesa short peroneal muscle Baumel 1979a); (4) a tendon that traverses a in his specimens of A. australis. In the next bony sulcuson the lateral aspectof the proxi- section, however, I will attempt to show that mal end of the tarsometatarsus;and (5) the for- "peroneusbrevis" as describedby McGowan mation of a conjoined tendon where the ten- is a synonym for "peroneus medius" in don of insertion unites with that of M. flexor Owen's papers (1849, 1879, but not 1842) and perforatusdigiti III on the plantaraspect of the that, in both cases, the muscle so described has foot. In the kiwi and in ratites, these five char- no morphologicalrelationship to the peroneal acteristicsdiffer from Chauna only in "minor musculature described in other birds. In fact, detail," accordingto Mitchell (1913). Gadow it is probably not a componentof the peroneal (1880: 45) describes the conjoined tendon in musculature at all. Struthio, Rhea, and Casuarius as follows: "Diese Sehne [M. peroneuslongus]--sich mit M. flexor hallucislongus. Synonymy:Owen (1842, der Sehnedes M. flexorperforatus digiti medii 1849, 1879), M. flexor perforans digitorum, tendon (M. flexor perforatusdigiti III) verbindet und 1; McGowan (1979), M. flexor digitorum longus. daher zur Zehenbeugung beitr•gt." On the M. flexor digitorum longus. Synonymy: Owen basis of these two references, the anatomical (1842), M. flexor perforans digitorum; Owen (1849 and 1879), "Peroneus medius Cuvier, Accessorius variant of M. peroneus longus in Apteryx is flexorisdigitorum Vicq d' Azyr"; Beddard(1899), M. similar to that in other ratites and similar to flexor profundus; McGowan (1979), M. peroneus that in at least some other avian groups, de- brevis. spite the statementsto the contraryas given in McGowan. According to McGowan (p. 65), M. flexor Although both Owen (1849:295-296, Plates hallucislongus, as a deep digital flexor of the 31, 32, and 35, tendon 7; 1879: 59-60, Plate II) hallux, is not presentin either specimenof A. and Beddard (1899: 397) describea conjoined australisthat he dissected,although he states tendon, they either describe a most unusual that the muscleis representedby a specificten- intraspecificmorphological variation in Apteryx don of "flexordigitorum longus" to the hallux. australis,or they inadvertently introduce con- The anatomicalrelationships of this latter mus- siderable confusionin use of the term "perfor- cle as described in his specimensof Apteryx atus" and "perforans" in terms of the associ- differ very little from thoseof M. flexorhallucis ated digital flexor muscle (see Myologia, longusdescribed in other specimensof kiwi 312 JAMESC. VANDENBERGE [Auk, Vol. 99

(Beddard 1899). Furthermore, the morpholog- peronealmusculature and deep digital flexor ical relationshipof the long (deep)flexor of the muscles as described in Apteryx by Owen, hallux to the deep flexor of the digits in A. Beddard, Mitchell, and McGowan, and the australisis illustrated as a definitive example same relationships in ratites and other birds, of Type II morphologicalvariant in the original McGowan's statement (p. 72) that "the fusion paper on this subject(Garrod 1875:341, Fig. 2; of the tendons of insertion of the Mm. pero- see also Gadow and Selenka 1891). neus brevis and flexor digitorum longus is not If the muscle described by McGowan as unique to Apteryx but is a condition which is "flexor digitorum longus" is, in fact, M. flexor shared with other ratites," is not substantiat- hallucis longus, then either M. flexor digito- ed. rum longusis itself absentin the samespeci- The preceding review of the status of the mens of kiwi, or the deep digital flexor muscle peroneal musculature, and the deep digital is likewise rnisidentified. The latter is almost flexor muscles,suggests that neither of these certainlythe casehere, based on McGowan'smuscle sets is predictiveof phylogeneticrela- descriptionof the anatomicalrelationships of tionships between Apteryx and ratites, or be- "peroneus brevis" with respectto its proximal tween ratites (collectively) and other birds. attachmentsin the crus, the position of the ten- They do not constitutea "distinction between don as it traversesthe tibial cartilage, the for- ratites and carinates," as suggestedby Mc- mation of a conjoined tendon (with M. flexor Gowan (1979: 71-72). hallucislongus) in the foot, and a trifurcation into a deep flexor tendon to each of the three M. caudo-iliofemoralis: M. caudofemoralis, M. ili- toes. M. peroneus brevis has never been so ofemoralis.Synonymy: Owen (1842, 1849, 1879), M. describedin any other birds (seeMitchell 1913, adductorbrevis femoris, M. adductorlongus; Garrod (1873),Beddard (1899), femoro-caudal and accessory Hudson 1937, and numerous other investi- femoro-caudal (including also "superficial femoro- gators),and the above mentionedmorpholog- caudal"?); McGowan (1979), M. piriformis, pars ical characters are those most often associated caudofemoralis,pars iliofemoralis. with M. flexor digitorum longus. As previously mentioned, Owen does not Both axial and pelvic components of this describeM. peroneusbrevis in any of his three muscle complex are present in Apteryx. The papers, nor any other musclehaving the ex- axial component,M. caudofemoralis,is differ- pected morphologydescribed for other birds. entiated from the pelvic component even Furthermore, the anatomical description of though it is "slender, straplike, of only mod- "flexor perforans digitorum" (1842: 40-41) erate size" (McGowan 1979: 53). This muscle is virtually identical to that of "Peroneus is also described as the "femoro-caudal" and medius Cuvier, Accessoriusflexoris digitorum is illustratedin A. owenii(Garrod 1873, Fig. 6). Vicq d' Azyr" of the later papers(1849, 1879). Beddard (1899), however, indicates that there In all three papers, Owen statesthat the tendon is no clear separationbetween the two muscle of insertion "receivesa strongaccessorial ten- componentsin other specimensof kiwi. don from the musclewhich bendsthe innermost M. caudofemoralisis clearlydifferentiated in toe, and finally divides into three strong per- most tinamous (exceptEudromia, Hudson et al. forating tendons" (emphasis mine). Gadow 1972), but much less so in ratites. The caudo- (1880:51) gives an interestingfootnote to what femoralisis very weakly differentiatedin Stru- he describesas the "outer head of flexor pro- thio (Gadow 1880) and Casuarius(Gadow 1880, fundus." In that footnote he quotes directly Garrod 1873, Pycraft1900) and absentin Rhea from Owen's paper of 1849 and clearlystates and Dromaius(Garrod 1873, Pycraft 1900). Oth- that Owen's use of "Peroneus medius Cuvier, er variants of this muscle in carinate birds are Accessoriusflexoris digitorum Vicq d' Azyr" well known in the literature(George and Ber- as a synonym for M. flexor digitorum longus ger 1966,Table IX, symbol A). in Apteryx was erroneousand that Cuvier, at Of particular interest, however, is the ap- least, was describingM. peroneuslongus! I am parent hypertrophicdevelopment of the pelvic not able to confirm this reference to the work constituent, M. iliofemoralis, in Apteryx (A. of Cuvier, however. owenii, Garrod 1873, Fig. 6; A. australis,Mc- In view of the apparent discrepancies be- Gowan 1979,Figs. 10 and 11) aswell as in some tween the morphologicalrelationships of the ratites (Casuarius,Gadow 1880, Plate III, Fig. April 1982] Kiwi Myology 313

1; Pycraft 1900, Fig. 6), and in tinamous (Hud- or incompleteanatomical descriptions may be son et al. 1972, Fig. 10). In all of these speci- acceptedas authoritative, more so than in hu- mens, as illustrated, the proximal "fixed" at- man or veterinary anatomy, for example.In the tachmentof M. iliofemoralisis "perforated"by previous discussion of a series of anatomical peripheral branches of Plexussacralis and ac- variants of selectedmuscles, as exemplifiedin companying vessels, principally Arteria is- Apteryx, I have attemptedto demonstratesome chiadica (Baumel 1979b; not "femoral vessels" aspects of the "incomplete" status of avian as labelled in McGowan's Figs. 9-11), as this anatomybased on (1) descriptivemyology, (2) neurovascularbundle emergesthrough Fora- probable misidentification or misrepresenta- men ilioischiadicato enter the postacetabular tion of muscles,and (3) morphologicalvaria- region of the thigh. The distal (functional)at- tions in a muscle set that have not been thor- tachmentof this musclein theseavian groups oughly examined in terms of a form-function includes an extensive linear attachment on the complex. shaft of the femur, apparentlyseparate from Studiesof the musculatureof the pelvic limb (but contiguouswith) the femoralaponeurosis in birds should continue to be a source of in- of M. caudofemoralis(see, e.g. McGowan,pp. formation on anatomical variation in somatic 52-53). musclepatterns among vertebrates if the prop- If the M. caudo-iliofemoralis complex in er anatomical questions are first proposed. birds is derived from the coccygeofemoralesThen the relative developmentof individual musclesystem in reptiles,as suggestedby Ro- musclesand/or functional groups of muscles, mer (1927), then M. caudofemoralismay be musclestructure (pennation, fiber types, apo- assumedto be the "primitive" componentand neuroses, innervation, etc.), and musculoskel- to be of common occurrenceamong birds. etal parametersof form and function should all Also, the principal morphologicaland func- contribute to the informational value of ana- tional relationships of this axial component tomical studies and to the interpretation of an- would be correlated with the musculature of atomical variation at several levels of scientific the tail (see Fisher 1957, Owre 1967, Baumel inquiry (seeCracraft 1971, Helmi and Cracraft 1971, Cracraft 1971), rather than the pelvic 1977, Raikow 1978, Zusi and Bentz 1978, for limb. several different applications).Unfortunately, M. iliofemoralis would then representa sec- until very recently the nomenclaturefor avian ondarily derived, appendicular component anatomy, especiallythe musculature,has been (see Romer 1927) whereby the muscle set a source of considerable confusion, but the gained a stationaryattachment to the post- standardization of the anatomical nomencla- acetabularilium (andischium, in somespecies). ture should help to alleviatethis problem. This new morphologicalrelationship may have What is the presentstatus of our knowledge been a primary anatomicaladaptation that per- concerningthe myologyof the pelvic limb in mitted subsequenthypertrophic development Apteryx relative to that in ratites and carinates? of the appendicularcomponent and directly Somehow,it seemsironic that a commentby contributed to the evolution of a new function- T. J. Parker (1891: 103-104) on "deficiencies" al modality for the caudo-iliofemoralismuscle in the original papers of Owen might still be complex. I hope to test this hypothesisfor a applicablenearly 100 yr later: "It is a curious myological correlateof a "form-function com- circumstancethat whenever subsequent ob- plex" (see Bock 1974) in a new study of this servershave had occasionto correctthe origi- musclecomplex in a sampleof specimensfrom nal descriptionof Apteryx, the result has been selectedavian groups. to show the bird to be less aberrant and more typicallyavian than it was consideredto be by the distinguished anatomist [i.e. Owen] to CONCLUSIONS whom we owe our first knowledgeof its struc- As Homberger (1980) mentions in her review ture." of the "Nomina Anatomica Avium," our knowledgeof avian anatomyis still incomplete ACKNOWLEDGMENTS in the sense of not having been checked and I have attemptedto verify the morphologicalre- recheckedby countlessresearchers, and, in- lationshipsof the musclesdescribed by any investi- deed, there is a very real dangerthat inaccurate gator cited, with one or more illustrations of the 314 JAMESC. VANDENBERGE [Auk, Vol. 99

same in that text or some other related publication. GEORGE,J. C. & A. J. BERGERß1966. Avian myology. I wish to thank W. J. Bock, J. Cracraft, S. L. Olson, New York, Academic Pressß R. J. Raikow, and R. L. Zusi for their respectivecri- GREENLEE,T. K., JR., C. BECKHAM,& D. PIKE. 1975. tiques of a muchlonger and more detaileddraft of A fine structuralstudy of the developmentof the this manuscript. chick flexordigital tendon:a model for synovial sheathed tendon healingß Amer. J. Anat. 143: 303-314. LITERATURE CITED HELMI, C. & J. CIdCIVET. 1977. The growth patterns BAUMEL,J. J. 1971.Morphology of the tail apparatus of three hindlimb muscles in the chicken. J. in the pigeon (Columbalivia). 19th World Vet- Anat. 123: 615-635. erinary Congress(Mexico City) 3:849 (Abstract)ß HOMBERGER,D. G. 1980. Review of "Nomina Ana- 1979a. Osteologia. Pp. 53-121 in Nomina tomica Avium." Auk 97: 912-914. anatomica avium (J. J. Baumel, A. S. King, A. HUDSON, G. E. 1937. Studies on the muscles of the M. Lucas, J. E. Breazile, & H. E. Evans, Eds.). pelvic appendage in birdsßAmer. Midi. Natur. London, Academic Pressß 18: 1-108. ß 1979b. Systemacardiovasculare. Pp. 343-407 , R. A. PARKER,J. VANDENBERGE, & P. J. LAN- in Nomina anatomica avium (J. J. Baumel, A. S. ZILLOTTI. 1966. A numerical analysis of the King, A.M. Lucas,J. E. Breazile,& H. E. Evans, modifications of the appendicular muscles in Eds.). London, Academic Press. various genera of gallinaceous birdsß Amer. , A. S. KING, A.M. LUCAS,J. E. BREAZILE,& Midi. Natur. 76: 1-73. H. E. EVANS (Eds.). 1979ß Nomina anatomica , K. M. HOFF, J. VANDEN BERGE,& E. C. TRIV- avium. London, Academic Pressß ETTE.1969. A numericalstudy of the wing and BEDDARD,F. E. 1899. Notes on the anatomyof the leg musclesof Lari and Alcae. Ibis 111:459-524ß genusApteryx. Novit. Zool. 6: 386-402. --, D. O. SCHREIWEIS,S. Y. CHEN WANG, & D. BocK, W. J. 1963. The cranial evidence for ratite af- A. LANCASTER.1972. A numericalstudy of the finities. Proc. 13th Intern. Ornithol. Congr. 1: wing and leg musclesof tinarnous(Tinamidae). 39-54ß Northwest Sci. 46: 207-255. ß 1974. The avian skeletomuscularsystemß Pp. KLEMM,R. D. 1969. Comparativeroyology of the 119-257 in Avian biology, vol. 4 (D. S. Farner hind limb of procellariiformbirds. Southern Il- & J. R. King, Eds.). New York, AcademicPress. linois Univ. Monogr., Sci. Series2: 1-269. BREAZILE,J. E., & M. YASUDA.1979. Systema ner- McGowAN, C. 1979. The hind limb musculature of vosum peripherale. Pp. 473-503 in nomina an- the Brown Kiwi, Apteryx australismantelli. J. atomica avium (J. J. Baumel, A. S. King, A.M. Morphol. 160: 33-74ß Lucas, J. E. Breazile, & H. E. Evans, Eds.). Lon- MITCHELL, P. C. 1913. The peroneal muscles in don, Academic Pressß birdsß Proc. Zool. Soc. London 1913: 1039-1072. C•,ACI•FT, J. 1971. The functionalmorphology of OWEN,R. 1842. Monograph on Apteryx australis,in- the hind limb of the domesticpigeon, Columba cludingmyology. Proc. Zool. Soc.London 1842: livia. Bull. Amer. Mus. Nat. Histß 144: 171-268. 22-41. --. 1974. Phylogenyand evolutionof the ratite --. 1849. XI. On the anatomyof the Apteryxaus- birdsß Ibis 111: 494--521. tralis Shaw, part 2 (Myology). TransßZool. Soc. FISHER,H. I. 1957ßThe function of M. depressor London, III, Part 4: 277-301. caudaeand M. caudofemoralisin pigeonsßAuk --. 1879. Memoirs on the ExtinctWingless Birds 74: 479-486. of New Zealandß London, Van Voorst. FREWEIN,J. 1967. Die Gelenkr3ume, Schleimbeutel OWRE,O. T. 1967. Adaptations for locqmotion and und Sehnenscheiden an den Zehen des Hau- feeding in the Anhinga and the Double-crested shuhnes. Zbl. Veterin3rmed. 14-A: 129-136. Cormorant. Ornithol. Monogr. 6: 1-138. GADOW,H. 1880. Zur vergleichendenAnatomie der PARKER,T. J. 1891. Observations on the anatomy Muskulatur des Beckens und der hinteren Glied- and developmentof Apteryx. Phil. TransßRoyal masse der Ratiten. Jena, G. Fischerß Soc. 182: 25-134. --, & E. SELENKA.1891. Vogel: I. Anatomischer PYCI•ET, W. P. 1900. Part II. On the morphology Theil. In Bronn's Klassen and Ordnungen des and phylogenyof the Palaeognathae(Ratitae and Thier-Reichs, Bd. 6(4)ßWinterß Leipzig, C. F. Crypturi) and Neognathae (Carinatae). Transß GARROD,A. H. 1873. On certainmuscles of the thigh Zool. Soc. London 15: 149-290. of birds and their value in classification,part 1. RAIKOW,R. J. 1975. The evolutionary reappearance Proc. Zool. Soc. London 1873: 626-644. of ancestralmuscles as developmentalanomalies --. 1875. On the disposition of the deep plantar in two speciesof birdsßCondor 77: 514-517. tendons in different birdsß Proc. Zool. Soc. Lon- 1978. Appendicularmyology and relation- don 1875: 339-348ß shipsof the New World nine-primariedOscines April 1982] KiwiMyology 315

(Aves: Passeriformes).Bull. CarnegieMus. No. son and Grossman'sThe anatomy of the domes- 7. tic animals5th ed. (R. Getty, Ed.), Philadelphia, --, S. R. BORECKY,& S. L. BEllMAN. 1979. The W. B. Saunders Co. evolutionary re-establishment of a lost ancestral --. 1976. M. iliotibialis medialis and a review muscle in the bowerbird assemblage.Condor of the M. iliotibialis complexin flamingos. Auk 81: 203-206. 93: 429-433. ROMER,A. S. 1927. The development of the thigh --. 1979. Myologia. Pp. 175-219 in Nomina an- musculatureof the chick. J. Morphol. Physiol. atomica avium (J. J. Baumel, A. S. King, A.M. 43: 347-385. Lucas, J. E. Breazile, & H. E. Evans, Eds.). Lon- VANDENBERCE, J. C. 1970. A comparativestudy of don, Academic Press. the appendicular musculature of the Order Ci- ZusL R. L., & G. D. BENTZ.1978. The appendicular coniiformes. Amer. Midi. Natur. 84: 289-364. myologyof the LabradorDuck (Camptorhynchus --. 1975. Aves: Myology. Pp. 1802-1848 in Sis- labradorius). Condor 80: 407-418.

(continuedfrom p. 291)

Mock*, Timothy C. Moermond, Burt L. Monroe, Jr.*, Paul Sherman, Thomas W. Sherry, Gerald F. William A. Montevecchi*, Robert D. Montgomerie, Shields*,Lester L. Short*,Gary W. Shugart,Douglas DouglassH. Morse, Eugene S. Morton*, Martin L. Siegel-Causey, Walter R. Siegfried*, R. D. Slack, Morton, James A. Mosher, Helmut C. Mueller*, Ron- James N.M. Smith, Kimberly G. Smith*, W. John ald Mumme, Edward C. Murphy, William Murdoch, Smith, BarbaraD. Snapp, David W. Snow*, Noel F. Bertram G. Murray, Jr., Pete Myers*, Bryan Nelson, R. Snyder*, Paul R. Sotherland*, Michael SoulS, Lin- Val Nolan, Jr., Thomas D. Nudds, Raymond J. da K. Southern, William E. Southern*, David W. O'Connor*, Hans Oelke, Harry M. Ohlendorf, Storrs Steadman*,Peter Stettenheim,F. Gary Stiles*, Chris L. Olson*, John P. O'Neill*, Lewis W. Oring*, Gary Stinson, Robert W. Storer, JosephG. Strauch, Jr.*, C. Packard, Kenneth C. Parkes, John Paton, Samuel Ian R. Swingland, J. B. Tatum, Fritz Taylor, Stanley M. Patten, Jr., Ian J. Patterson *, James L. Patton, A. Temple, Charles F. Thompson*, William L. Robert B. Payne*, David L. Pearson,C. M. Perrins*, Thompson*, A. van Tienhoven, Harrison B. Tordoff, Raymond Pierotti, Stuart L. Pimm, Benedict C. Pin- C. Richard Tracy, Elliot J. Tramer, JosephTravis, kowski, A. Poole, Sergej Postupalsky, Dennis M. Melvin A. Traylor*, Wayne Trivelpiece, M.D. F. Power, Harry W. Power, III*, Harold H. Prince*, H. Udvardy, R. A. Vgisgnen, Stephen B. Vander Wall, Graham Purchase, Kerry N. Rabenold, Robert Rai- Beatrice Van Horne, Leigh Van Valen, Sandra L. kow*, Ralph J. Raitt, Fred Ramsey*, Dennis G. Rav- Vehrencamp*, R. D. Vessel, A. Village, Carol Vleck, eling*, Harry Recher, Jan Reese, Richard Reynolds*, Glenn Walsberg,Jeff Walters, John Warham, George Bill Rice, Jake Rice, Terrell D. Rich, Douglas Rich- E. Watson*, Patrick J. Weatherhead, Wesley W. ards, W. John Richardson, Robert E. Ricklefs*, Rob- Weathers, Milton W. Weller*, David F. Werschkul, ert W. Risebrough, Chandler S. Robbins, Robert J. Clayton M. White*, Robert Whitmore*, G. Causey Robel, William B. Robertson, Jr., JamesA. Rodgers, Whittow, Sanford R. Wilbur, David E. Willard, John G. Rogers, Jr.*, Howard A. Ross, John T. Ro- George C. Williams, Timothy C. Williams, Fred E. tenberry*, Stephen I. Rothstein*, Ian C. R. Rowley, Wilson, Rick Wishart, JamesF. Wittenberger*,Larry Lynda Rummell, John P. Ryder*, Paul Samollow*, L. Wolf*, Glen E. Woolfenden*,Ken Yasukawa,Terry ThomasSchoener, Ralph W. Schreiber*,D. M. Scott, Yates*,Jerrold H. Zar*, and Richard L. Zusi*. Special J. Michael Scott, Norm Scott, William A. Searcy*, thanks are due Stuart Houston. MOVEMENTS AND ACTIVITIES OF RADIO-TRACKED BROWN-HEADED COWBIRDS

ALFREDM. DUFTY, JR.• Departmentof BiologicalSciences, State Universityof New York at Binghamton, Binghamton,New York 13901USA

ABSTR•CT.--Thesocial organization of a color-markedBrown-headed Cowbird (Molothrus ater) population was investigatedduring the 1978-1980breeding seasons.Several members of the population were radio tracked, some in more than one season.Monogamous asso- ciations were establishedbetween males and females, and each pair maintained a large nonfeedingrange. Femalesappeared to defend their nonfeeding rangesfrom other females, and malesguarded females from other males.Most cowbird pairs that returnedin successive years exhibited mate fidelity and site fidelity, although some evidence indicatesthat the former is a coincidentalresult of the latter. I proposethat variations in the mating system of this speciesare primarily due to habitat differencesand differencesin the relative abun- danceof the cowbird and its hosts.Received 8 May 1981, accepted2 September1981.

THE Brown-headed Cowbird (Molothrus promiscuity (Friedmann 1929, Nice 1937, Las- ater) is the most abundant cowbird in the key 1950, Darley 1968, Rothstein 1972, Elliott United States and the best-studied member of 1980). Confounding the problem further is the its genus. Most investigations of this species likelihood that the mating systemvaries be- have dealt with rates of brood parasitism and tween geographic areas (Friedmann 1929, El- effects of parasitism on the reproductive suc- liott 1980). cess of hosts (e.g. Norris 1947; Walkinshaw This paper reports on a study of the move- 1949, 1961;Berger 1951; Wiens 1963;Rothstein ments and behavior of cowbirds, using color- 1976; Elliott 1978). In contrast, much less is marking and radiotelemetry. Radio tracking known about the social relationships of cow- has been successfullyemployed in a number birds. The reasons for this relative dearth of of investigationsof passerines(e.g. Cochranet information are twofold. First, cowbirds do not al. 1967; Bray et al. 1975a, 1979), and it is par- constructnests, so a convenientfocal point for ticularly suitable for use with a species as the observationof socialactivity is absent.Sec- wide-ranging as the Brown-headed Cowbird. ond, cowbirds range over large areas (Fried- mann 1929, Nice 1937, Darley 1968), making METHODS continuous observation of individual birds Capture and banding.--Most cowbirds were cap- quite difficult. Nice (1937),Laskey (1950), and tured in mist nets or Potter traps placedaround each Darley (1968) observedcolor-marked individ- of two artificial feeders. The feeders were 0.16 km uals, but only Darley followed a large number apart and were located approximately 0.8 km from of birds over a broad area. Color-marking al- the Binghamton campus of the State University of leviates the problem of individual recognition New York. In 1978, trapping began on 22 April and but does not reduce the difficulties associated continuedthrough the end of June.In 1979,trapping startedon 30 March and continued almostdaily until with locating and following cowbirds. This 29 May. In 1980, trapping was conducted from 20 problem is especially acute with female cow- March through 16 June. Trapping was less intense birds, which are more secretive than males. in 1979 and 1980 than in 1978. All captured birds These difficulties may have contributed to were individually marked with colored leg bands contrastingdescriptions of socialorganization and were fitted with colored leg streamers2.54 cm for this species, ranging from monogamy to in length. Yearling and adult male cowbirds were distinguished on the basis of retention of juvenal underwing covertsin the former group (Baird 1958, Selanderand Giller 1960). Yearling and adult females • Presentaddress: Rockefeller University Field Re- were not differentiated. searchCenter, Tyrrel Road, Millbrook, New York Radio tracking.--Field observationsof the color- 12545 USA. marked population revealed that birds that were lo-

316 The Auk 99:316-327. April 1982 April 1982] CowbirdMovements and Activities 317 cal residentsand appearedto be paired. If thesewere sulted from birds leaving their ranges for short recaptured,they were fitted with transmitters(model periods of time, and such movements (if they oc- SM1, AVM Instrument Co., Champaign, Illinois) curred) are not plotted. Each nonfeedingrange was with a Mallory Duracell RM-312T2 mercury battery determined by connectingthe remaining outermost power source.The birds were then released. locationsof a bird. The area of each range was cal- The radio transmitters were attached to the cow- culated using a polar planimeter. birds by glueingthem to the interscapularregion of Birds from a particular range were assignedthe the spinal tract, just anterior to the points of articu- letter M or F, which designatedthem as male or fe- lation of the humeri. The birds were held in one male, and a number correspondingto their range. hand in the manner illustratedby Raim (1978, Fig. Birds radio-tracked in more than one year retained 2). All the feathers in an area the size of the trans- the same number. Primed (') numbers represent re- mitter unit were cut to within 1-2 mm of the skin. placementbirds. The featherswere cut rather than pulled out, as the Both residentand nonresidentfemales were pres- latter method reduces the retention time of the trans- ent in the study area. Resident females maintained mitter. Epoxy glue (Devcon 5 Minute Epoxy) was nonfeedingranges that overlappedthose of neigh- applied to the ventral surface of the transmitter, boring femalesonly slightly. These femaleswere ob- which was carefully placed onto the exposed area servedfrequently. Nonresidentfemales were rarely and held firmly for five minutes(longer if excessglue observed and never exhibited aggressivebehavior was applied). A few uncut feathers,just anterior and in the study area. They were seenprimarily at feed- lateral to the transmitter, were glued to it to secure ing areas. There was no evidence of nonterritorial the unit further. Each transmitter-battery unit residentfemales in my population, althoughDarley weighed approximately2.5-3.0 g. (1968) noted such females in his. Male cowbirds were All of the cowbirdspulled at the transmittersini- designatedas mated or unmated accordingto their tially, but most ignored them after a few hours. A relationshipswith the residentfemales. No attempt few birds never acceptedthe transmittersand con- was made to catalog systematicallyall the resident tinued to peck at them until they becamedetached. unmated males, and a number of males remained Raim (1978) reports similar behavior. Many of the unbanded throughout the study period. cowbirds had deeper wingbeats following attach- ment of the transmitters,but reactiontime and flight RESULTS speed did not appear to be affected. Transmitters Bandingand sex ratio.--In 1978,50 malesand remained on thosebirds that acceptedthem for 3-21 38 females were banded, a male: female ratio days, at which time they either fell off or were re- of 1.32:1. In 1979, 38 males and 27 females were moved by me. Most tracking was done in the morning and late captured, a ratio of 1.41:1. These values are not afternoon when the birds were most active. Trans- significantlydifferent from unity for the two mitter signals were received by a hand-held AVM years(X 2 = 1.65 and 1.88, respectively;P ) 0.05 Yagi antenna coupled to an AVM model LA 12 por- for both). In 1980, 45 males and 21 females were table receiver.Tracking and behavioralobservations trapped. This male:female ratio of 2.14:1 is were tape recordedin the field and later transcribed. significantlydifferent from unity (X2= 8.74, Tracking locationswere plotted on an aerial photo- P (0.005), as is the combined data for the graph of the study area. Only the nonfeedingloca- three years(X • = 10.09, P (0.005), which give tions of the birds were consideredwhen plotting an overall male: female ratio of 1.55:1. Thirteen ranges. Movements to midday resting areas, where cowbirdsfrom severalranges congregated, were ex- yearling and 36 adult males were captured in cluded from consideration. One such resting area 1978 (one male was not aged). In 1979 and 1980 was locatednear my primary trapping site (Fig. 1). these age classeshad 11 yearlingsand 27 adults In 1979 birds from areas 1, 2, 4, 5, and 6 were known and 13 yearlings and 32 adults, respectively, to rest there, as did several unidentified cowbirds. for a total of 37 yearling and 95 adult cowbirds. Likewise excluded were the locationsof aggressive Radiotracking.--Seven cowbirds were radio encounters in which birds were driven from the tracked in 1978, four females and three males. range of another pair. On occasionthe signal of a The nonfeeding ranges of these birds are tracked bird would disappear, reappearing a few minutes later. Most instances of this nature could be shown in Fig. 1. Areas 1, 2, and 4 of Fig. 1 represent ranges that were identical for both attributedto the signalbeing blockedby vegetation, a male and a female cowbird. These birds were the bird flying to the ground (therebyreducing the effective range of the transmitter), interferencefrom consideredto be paired. The fourth female also citizens' band radios, and solar interference(espe- appeared to be paired, but, because her mate cially near midday). I cannoteliminate the possibil- was unbanded, her status could not be deter- ity, however, that some of thesedisappearances re- mined with certainty. 318 ALFREDM. Dv•r¾, JR. [Auk, Vol. 99

because few data were obtained at these loca- tions. Three birds are known to have been killed in 1979.F5 was taken by a predatoron 21 April. Her mate was seen with a replacementfemale on 23 April, and this female (F5') was subse- quently capturedand radio tracked.Her range appeared to be the same as that of F5. M6 and 200m M9 were found dead on 11 May and 22 May, respectively.Each had been shot with a pellet o gun. A secondmale (M6') paired with F6; he <... was radio tracked and maintained a nonfeed- ing range identical to that of M6. Female F9 Fig. 1. Nonfeeding ranges of Brown-headed was subsequentlyguarded by an unbanded Cowbirds in 1978. X = trapping site. Triangle = male whose range was not determined. midday restingarea for cowbirdsof severalranges. M8 was trackedfor 71 rain over2 daysbefore his transmitter malfunctioned. Observational data were used to supplementthe tracking data Elevenbirds were radio trackedin 1979(Fig. in determining his range. 2). Pairs of birds were tracked in areas 2, 4, 5, Two pairs of cowbirds tracked in 1979 had and 6. Male membersof pairs were followed also been tracked in 1978. These birds settled in areas 8 and 9. A mated female was tracked in much the same areas as in 1978 and once in area 7. The male in area i was not radio again had identical ranges(Figs. 3 and 4). The tracked in 1979. Area 1 had few large trees to male from the remaining pair (M1) returned obstructmy vision, and I was able to follow and mated with a different female. His 1978 his movements over most of his range. The mate was not seen in the study area in 1979. areas where observations of this male were The 1979 pair establishedthemselves near the difficult are labeled as "proposedboundary," male's 1978 area (Fig. 5).

COWBIRD TERRITORIES, 1979 -----proposed boundary

200m 1000m

Fig. 2. Nonfeedingranges of Brown-headedCowbirds in 1979."Proposed boundary" refers to areasin rangesi and 8 where data were sparse. April1982] CowbirdMovements and Activities 319

-1978 ...... 1979 1978 r200 m 19791980 mole I lOOm ...... 1980 femole •-0

r2OOmloOOm._jrr...... Fig. 3. Superimposednonfeeding ranges of a cowbird pair (number2 in Figs. 1 and 2) in 1978and 1979.

Tracking data for 1980 are shown in Fig. 6. Females were followed in areas 4 and 11. Both Fig. 4. Superimposednonfeeding ranges of a cowbird pair (number 4 in Figs. 1, 2, and 6) in birds in areas 6 and 10 were tracked, but the 1978-1980. In 1980 the male and female paired with data for M10 were lost due to equipment fail- different mates. ure. Nonetheless, his range coincided with that of F10. Two males and two females tracked in 1979 The close associationbetween pair members, returned in 1980. One pair (M4-F4) returned resultingprimarily from malesassiduously fol- for the third consecutiveyear. Instead of pair- lowing females (Darley 1968, Dufty 1981; see ing and assuming identical home ranges, as also below), accountsfor the uniformity of they had done in 1978and 1979,F4 settledwith range boundaries for males and females of a new male in an area adjacent to the one in pairs. Males accompainiedfemales throughout which M4 was located with a new female. Each much of the day, although there were extended of their 1980 rangespartially overlappedtheir periods in the morningswhen femaleswere 1979 ranges (Fig. 4). The range sizes for all stationaryand alone,presumably searching for birds, 1978-1980, are presentedin Table 1. nests. For example, F4 was tracked for 1,371 Of the 12 malesthat were trackedduring the min during the mornings of 13-18 June 1978, courseof this investigation,five were yearling a time when both membersof pair 4 had active birds and the rest were adults. transmitters. She was alone and relatively in- Pairing.--The evidence suggestsmonoga- active on 25 occasions, for a total of 457 min mous associations between male and female (x = 18.28 + 10.18 min, range = 7-48 min) or cowbirds. Figures 1, 2, and 6 show that cow- 33.3% of the time she was tracked. birds limited their nonfeeding activities to The coincidentalmovement of the signalsof large yet well-defined areas.The fact that cer- a male and female does not preclude the pos- tain males and females exhibited identical sibility that birds without transmittersalso ac- ranges implies that these birds were paired. companiedthem, but when both membersof On nine occasionsboth members of presump- a tracked pair were visible there was no evi- tive pairs had simultaneously active transmit- dence of additional cowbirds of either sex in ters. Both signals were often found together regularattendance. When a pair was not visi- and moved together throughout the ranges. ble, I couldusually detect the presenceof other 320 ^LFaEDM. DvFr¾,Ja. [Auk,Vol. 99

1978 ...... 1979 TABLE1. Sizes of the nonfeedingranges of Brown------proposed boundary 1979 headed Cowbirds, 1978-1980.

Size (ha) Range number 1978 1979 1980

1 25.5 22.7 22.8 21.1 200m 23 29.3 20.5 29.7 100m 44a 9.9 Lø 4b 14.1 5 19.3 15.6 33.2 • 7 18.1 • •-----• 8 17.5 • ------•' 10 11.0 '•------'-----'-----'-- • 11 15.3

Fig. 5. Superimposednonfeeding ranges of a Rangenumbers correspond to thosefound in Figs. 1, 2, and 6. cowbird pair (number 1 in Figs. 1 and 2) in 1978and 1979. In 1979 the male paired with a different (un- tracked) female. He was tracked visually in 1979. "Proposed bounda•" refers to areas where data had active transmitters, which could have w•r• sparse. made them easier to locate than unmarked males, thereby biasing the data. To avoid this cowbirds by differencesin their vocalizations. bias, data are included only if the birds were These instancesalso occurredirregularly. visible, thus obviating the need for radio track- Further evidence that males and females ing in locatingbirds. Eachfemale cowbird was were paired is presented in Tables 2 and 3. seen more often with her mate only than with Table 2 shows the amount of time each female other males [Wilcoxon matched pairs signed- was seen with her presumptive mate only, rank test (Siegel 1956), T = 0, P < 0.01]. Data with her mate and other males, and with other for 1980 show a similar trend but are not pre- males only, in 1978 and 1979. Instances are sented, because many of the tracked birds' consideredonly if the identities of both mem- mateswere not bandedat the time of tracking, bers of a pair were known. Some male mates so sample sizes are small.

COWBIRDTERRITORIES, 1980

200m IoOOm / 4b •

Fig.6. Nonfeedingranges of Brown-headed Cowbirds in 1980. The male (a) and female (b) •rom area 4 returned to their range of the previous2 yr but had different mates. April 1982] CowbirdMovements and Activities 321

TABLE2. Time (min) that each female cowbird was TABLE 3. Time (min) that each male cowbird was seen in the presenceof her mate, her mate and seen in the presenceof his mate, his mate and other males, and other males. other females, and other females.

Time seen with: Time seen with

Mate and Mate and other Other other Other Cowbird Mate males males Cowbird Mate females females Year (n)a (tl)b (tl)b (tl)b Year (n)a (n)b (n)b (n)b 1978 F1 (7) 53 (11) 7 (3) 20 (5) 1978 M1 (7) 54 (14) 6 (1) 0 F2 (7) 530 (66) 295 (36) 15 (4) M2 (6) 545 (71) 280 (27) 2 (1) F4 (13) 163 (33) 53 (12) 1 (1) M4 (14) 174 (39) 42 (6) 8 (3) 1979 F2 (10) 125 (23) 147 (24) 8 (4) 1979 M2 (9) 151 (27) 71 (14) 3 (1) F4 (3) 25 (6) 11 (3) 24 (4) M4 (3) 25 (6) 11 (2) 0 F5 (1) 29 (3) 1 (1) 0 M5 (8) 134 (13) 41 (3) 0 F5' (2) 28 (4) 5 (1) 0 M6 (4) 19 (6) 18 (3) 15 (3) F6 (2) 15 (4) 2 (2) 3 (1) M6' (8) 28 (6) 15 (1) 7 (1)

Number of days of observation. Number of days of observation. Number of observation sessions. Number of observation sessions.

The analagousdata for males are presented Aggressivebehavior.--To determine whether in Table 3. As with females, each male was or not male cowbirds defend their mates from seen more often with his mate only than with other males,I observedthe outcomeof display all other females combined (T = 0, P • 0.01). bouts between guarding and intruding males A male accompaniedhis mate and other fe- in the presence of resident females. The de- males on feeding grounds and during territo- parture of a male in advanceof the female was rial interactions between the females. If the scored as a defeat for that male. I scored 74 mate of the other female was also present, the instances, involving 9 guarding males (5 = maleswould sing and display to each other. If 8.2 + 7.1 instances/male, range = 1-22) and the other female's mate was not present, the 101 intruders. The guarding males won 64/74 male would sing to the femalesor sit quietly bouts (86%); thus, mated males were signifi- nearby while the femalesinteracted. When a cantlymore likely to win suchbouts than were paired male was with a female other than his intruders (X2 = 39.42, P < 0.005). Of the 10 mate, he would sing to her (if she was alone) casesin which guarding males flew off, four or engagein a display bout with her mate. occurredearly in the season,when guarding "Following"behavior of males.--I have indi- ability may not have been completelyestab- cated above that mated males attempt to stay lished. Two others occurred the day after a near their females. If such is the case, then male (M4) had been fitted with a transmitter, males should follow their females when the lat- perhapswhile he was still adjustingto its pres- ter take flight, while the converseshould not ence. necessarilyoccur. In testing this prediction, Aggressivebehavior between females was flights of pairs were recorded.When one mem- less regularlyobserved. I never saw a female ber of a pair flew, the other member was guard her mate from other females. In the watched to determine if he or she followed the mornings, most female-female encounters first member within 10 s. Of 314 such flights, were at territorial boundaries (n = 51), with females flew first 239 times. Males followed fe- fewer (16) occurringdeep within a female'ster- malesin 206 (86%) of theseflights. Severalin- ritory. In the afternoonscowbirds fed com- stances in which male cowbirds did not follow munally, either on lawns or elsewhere (e.g. females occurred when females made short horse pastures). Feeding sites were not de- flights into vegetation, possiblyto begin nest fended, even thoselying entirely within a fe- searching.Males flew first 75 times, with fe- male's range, and little aggressivebehavior males following on only 12 (16%) occasions. was recorded at these locations. The differencebetween the tendency of males As might be expectedin encountersbetween to follow females and females to follow males neighboringterritorial females, there was often is significant(X • = 129.22,P < 0.005). no clearly observable victor. The females 322 ALFREDM. Dvrr¾, JR. [Auk, Vol. 99 would chatterand give aggressivebill-up dis- a whole (X2= 0.44, P < 0.05). Payne (1965) plays to eachother as they hoppedfrom branch and Darley (1968) also reported that yearling to branch. Eventually they would become sep- males breed successfully.This contrastswith arated, each toward her own territory, and the yearling males of polygynousicterid species, interaction would end. At other times the fe- such as the Yellow-headed Blackbird (Xantho- maleswould take flight in oppositedirections cephalusxanthocephalus) and the Red-winged or would fly togetherout of sight. Raim (1979) Blackbird(Agelaius phoeniceus), where juvenile reports similar activities. Whereas the social males either do not breed or have reduced re- relationshipsbetween males were readily dis- productivesuccess (Willson 1966,Searcy 1979). cernable, such relationships between females The cowbirds that were radio tracked were less obvious and may have operated at a showednonfeeding ranges that averaged20.4 more subtle level. ha in size.These ranges were identicalfor both Removal of females.--Females from seven membersof a pair. Darley (1968) mapped the pairs were temporarilyremoved from the pop- nonfeeding ranges of color-markedcowbirds ulation in May 1979and May 1980, in addition and found that the male member of a pair usu- to the permanent removal of F5 by a predator ally had a largerrange, which overlappedthat in April 1979. These femaleswere kept in iso- of the female. The nonfeeding rangesaveraged lation for 3-8 days. The male member of each 4.5 ha for his females and 6.6 ha for his mated pair confinedhis movementsto the established males, considerablysmaller than the rangesin nonfeeding range, and none was seen in as- my population. It is possiblethat Darley was sociation with a lone female up to the time not able to trackhis birds visuallyover all parts their females were returned. On those infre- of their ranges, resulting in smaller apparent quent occasionswhen a male was seen with a ranges than would have been found by radio female, she was a neighbor in the companyof tracking. Alternatively, the smaller ranges of her mate, and the encounters occurred at the the cowbirds studied by Darley may have re- boundary of the nonfeedingrange or on com- flecteda higher populationdensity in his study munal feeding grounds. The females all re- area. This would increase the cost of defense turned to their respective territories upon re- of the ranges(Brown 1964) and reducethe size lease, and their mates resumed guarding of eachrange. Changes in territory size in pop- behavior. ulations with different densities have been found in several species, such as the Song DISCUSSION Sparrow (Melospizamelodia), Dickcissel (Spiza americana),and GreatTit (Parusmajor) (Tompa The skewed sex ratio of the captured cow- 1962, Zimmerman 1971, Krebs 1971). birds (1.55:1) is similar to the 1.5:1 ratio ob- In addition to having identical nonfeeding served by Friedmann (1929) and Darley (1971). ranges, paired cowbirds spent significantly The imbalance in the sex ratio may be an ar- more time in the presenceof their mates than tifact, because the male:female ratio of im- in the presenceof other cowbirds of the op- mature cowbirds is 1:1 (Hill 1976) and females posite sex. This was effectedprimarily by the are more secretive in their movements than male, who attended the female diligently, fol- males. Fankhauser (1971) and Searcy and Ya- lowing her as she moved throughher territory sukawa (1981), however, have found that male and defending her from the approachof other cowbirds have a higher survival rate than do males.Guarding behavior may reducethe risk females.Furthermore, Darley (1971) and Burtt that a female will engage in extra-pair copu- and Giltz (1976) have shown that female cow- lations, for solitary male cowbirds presented birds are retrapped significantly more often with a model of a female and tape-recorded than are males,suggesting that the proportion female vocalizations rapidly approach the of males in the population may actually be model and occasionallyattempt to copulate underestimated. with it (Dufty 1981). Energetically, it may be Despite my bias of deliberately selecting more feasiblefor a male to guard his mate than males that had been trackedin previous years, to attempt to patrol her entire nonfeeding almost half (s/12)of the paired radio-tracked range (Conder 1949, Brown 1964), especially males were yearling birds, breeding for the becausethe skewedsex ratio of breeding cow- first time. This is not significantly different birds suggeststhat malesmust compete for ac- from their representationin the population as cessto females (cf. Emlen and Oring 1977). The April 1982] CowbirdMovements and Activities 323 severity of the intermale competitionfor fe- because cowbirds perform no parental care. malesis demonstratedby the resultsof the fe- Fledgingsuccess in cowbirdsis determinedby male removal experiment. None of the males the timing of egg deposition,the reactionof whose mate had been removed was able to se- the host to the cowbird egg, and any subse- cure a second female within the time con- quent nestling-fosterparent interactions,fac- straintsof the experiment.A male whose mate tors that would appear to be little affectedby was taken by a predator did obtain a second the presenceor absenceof a lastingpair bond. female, although this occurredearly in the sea- Breeding-sitefidelity may play a more prom- son when some females may still have been inent role in the determinationof reproductive searchingfor territories. success in individual cowbirds. Males form Females,in contrastto males, do appear to hierarchies, with the dominant male pairing defend their nonfeeding ranges from other with the female of that particulararea (Laskey same-sexedconspecifics, resulting in minimal 1950, Darley 1968). Copulation with a female overlap of territoriesof neighboringfemales. may be largelyrestricted to the dominantmale Spacing-outof adjacentfemales is not simply (Westet al. 1981),although early in the season, due to mutual avoidance, for female-female when the hierarchy may not be firmly estab- encounters were characterized by aggressive lished, other males may copulate with her rather than avoidance behavior. The aggres- (Darley 1968).Therefore, dominance in an area sive responseof these femalesto simulated fe- probably translatesinto high reproductivesuc- male intrudersfurther supportsthis hypothe- cess relative to subordinate male cowbirds. sis (Dufty 1981). Darley (1968) has suggested Fidelity to an area where a male has suc- that female cowbirds defend access to host cessfullybred in previous years may facilitate nests. Such territorial behavior may increase acquisitionof a dominant positionin succeed- the reproductivesuccess of individual female ing seasons. In the Red-winged Blackbird, cowbirds by reducing the amount of multiple maleswho have had prior reproductiveexpe- parasitism, thereby reducing competition be- rience in a territory return to that territory the tween their nestlings and those of other fe- following year and readily establish them- males. selves therein (Nero 1956, Searcy1979). Simi- The integrity of the pair bond between cow- larly, site fidelity in male cowbirdsmay have birds was maintained in successivebreeding evolved as a result of intrasexual selection seasons, with pairs exhibiting site fidelity. pressure, and a male's faithfulnessmay be to Friedmann (1929) felt that his birds returned his breeding area and not to his female. Three to the same territories each spring, although observationsare consistentwith this hypoth- he worked with unmarked birds. Nice (1937) esis. Male M1, whose mate did not return in reported faithfulnessto breeding sites in her 1979, paired with a new female, yet he main- color-bandedcowbirds. Darley (1968)also not- tained a nonfeeding range juxtaposedto his ed that pairs occupythe same(or overlapping) 1979 range. Furthermore,when females were nonfeeding areas in successiveyears. Addi- temporarily removed from the population, tional reports of site fidelity can be found in none of the males abandoned their ranges. Fi- Laskey (1950), Hunt (1977), and Kennard nally, the membersof a pair (M4-F4) that had (1978). bred togetherfor two consecutiveyears on the The significanceof a durable pair bond in same territory consortedwith different mates cowbirds is unclear. In long-lived species, the third year, each on a different part of the birds that have paired in previous years have old territory, indicating that site fidelity may higher reproductivesuccess than newly formed take precedenceover mate fidelity. pairs. Coulson (1966) found that Black-legged Data from Darley (1968) also support the Kittiwakes (Rissatridactyla) are able to raise view that successfulbreeding facilitates site more young if they have bred togetherprevi- fidelity. Ten of 12 of his paired malesreturned ously. Likewise, both members of Song Spar- to the study area in the secondyear; only 2 of row, Skylark (Alauda arvensis),and Great Tit 9 unmated resident males did so. pairs will return to their previousnesting area The apparent mate fidelity resultsfrom co- if they successfullyfledged young (Nice 1937, incidental site fidelity on the part of the fe- Delius 1965, Harvey et al. 1979). It is unlikely males. Females defend their territories, as that cowbirdsare affordeda similar advantage, shown by the resultsof playbacksof cowbird 324 ALFREDM. Duffy, JR. [Auk, Vol. 99 vocalizations to resident females (Dufty 1981). The mating system of the Brown-headed They do not desert their territories if their Cowbird is not identical throughoutits range. mates are killed; females F6 and F9 both re- The system I have described is similar to that mained in their territories and paired again reportedby Friedmann(1929) in New York and after their mates died. This is consistent with by Darley (1968) in Ontario, yet Elliott (1980) Darley's (1968) data. As with males, prior has shown that cowbirds are promiscuousin breeding experiencein an area may ease a fe- prairie habitats. Differencesin population size male's subsequentestablishment in that area. and habitat may be crucialfactors in explaining Additionally, because females spend much these contrastingbreeding systems. time searchingfor nests (Friedmann1929), fa- Cowbirds apparently entered the United miliarity with territories could enhancethe ef- States through Texas and the Prairie States ficiency of their nest-searchingactivities, al- (Friedmann 1929) and extended their range lowing them to find more nests or to spend eastward with the removal of large tracts of more time in other activities such as territorial woodland that had served as a barrier to dis- defense or feeding. Therefore, intrasexual persal (Mayfield 1965). The Prairie States still pressure on males to obtain mates and on fe- have the highest densities of cowbirds. Sur- males to obtain breeding territories probably veys show that the population densities of resultsin site fidelity in both sexesand, sec- cowbirds in Oklahoma, Kansas, and Nebraska ondarily, in mate fidelity. Nolan (1978) simi- are 2-3.5 times higher than thosein New York, larly concluded that mate fidelity in Prairie and the difference is increasing(Van Velzen Warblers (Dendroicadiscolor) occurs coinciden- 1972, Dolbeer and Stehn 1979). In contrast, tally, a result of site fidelity on the part of both Wiens (1973) has shown that overall species members of a pair. diversity and population density in the open While the data strongly suggesta monoga- grasslandsof the Prairie Statesis significantly mous relationship between male and female lower than in the shrub/forestecosystems of Brown-headed Cowbirds in my population, the Northeast. Thus, there is a higher density the suppositionof monogamy must be tem- of breeding cowbirds competing for relatively pered by the fact that no copulationswere ob- fewer host speciesin the grasslandsof the Prai- servedduring this study, so the actualmating rie States than in the shrub/forests of New pattern is unknown. Because other species York. Competition in the grasslandswould be thought to be monogamous or polygynous further increasedby the fact that many of the have been shown to be more promiscuousthan parasitized nestsare locatedon the edge of the previously suspected(Bray et al. 1975b, Burns grasslandsin shrubs or thickets, while those et al. 1980), a final conclusionregarding mo- in the open expanseof the grasslandsare rel- nogamy in these cowbirds must await infor- atively free from parasitism (Wiens 1963). Sim- mation regardingmating patternsin the wild. ilar occurrenceshave been noted in open fields of other localities (Berger 1951; Best 1978, Fig. MATING SYSTEMS IN THE 4; Gates and Gysel 1978;Gochfeld 1979). BROWN-HEADED COWBIRD The relatively low cowbird population in the Territoriality is thought to have evolved Northeast, coupled with the more abundant through competition for a limiting resource supply of host nests, could make defense of (Brown 1964). If the resourceis economically these nests profitable for female cowbirds in defensible, then territorial behavior may de- terms of reproductive success.McGeen (1972) velop, that is, the cost to the bird of securing and Elliott (1978) have shown that high levels the resourcemust not outweigh the benefits. of multiple parasitismreduce cowbird fiedging Competition for resourcesvaries directly with success,so territoriality in female cowbirds the density of the population and inversely could act to reduce competition between nest- with the abundance of the defended resource. ling cowbirds. In prairies, where competition Emlen and Oring (1977) have applied Brown's from other femalesmay be keener and the sup- ideas to the evolutionof mating systems.They ply of nests availablefor parasitismlower, fe- feel (p. 222) that "considerablelability in mat- males may undergo a kind of "scramble" com- ing systemsis . . . expectedbetween different petition. That is, they may not be able to populations of a given speciesin different en- defend successfullyan area from all other fe- vironmental or density situations." malesand still locateenough nests in which to April 1982] CowbirdMovements and Activities 325 lay eggs. Under these circumstances they Multiform mating systems have been re- would be expected to abandon territoriality ported in a number of avian species (Arm- and simply lay in whatevernests they find. A strong 1955, Case and Hewitt 1963, Verner similar situation occurs in other species, in 1964, Carey and Nolan 1975, Balfourand Cad- which territorial behavior is reduced or aban- bury 1979).Such variability may promotethe doned when resourcesare not economically efficient use of limited resources,such as mates defensible, yet exclusiveterritories are main- (for male cowbirds) and host nests (for female tained when the cost of such maintenance is cowbirds), and it provides these organisms low (Ewald and Carpenter1978, Myers et al. with a certain amount of flexibility in their re- 1979, Ewald et al. 1980). sponsesto differing demographicand/or en- Females in the Northeast are able to confine vironmental circumstances. their nonfeeding activities to areas that are ACKNOWLEDGMENTS small enough that a single male cowbird is dominant in each.This resultsin monogamous I thank J. J. Christian for his guidancethroughout associations between the birds. In contrast, this project,J. Comenofor his assistancein the field, Elliott (1980) reports that males in the prairies R. Schumannfor allowing me to work on his land, D. Madison for the generous use of his time and do not guard females and do copulate with radiotelemetryequipment, and C. Volter and M. more than one female. As argued above, fe- Searcyfor drawing the figures.I am gratefulto P. F. males in prairie habitats may travel more Elliott, P. D. McArthur, S. I. Rothstein,W. A. Searcy, widely than thosein the Northeast,and a male and K. Yasukawa for critical comments on an earlier cowbird may not be dominant in all parts of draft of this manuscript. This researchwas support- any given female'srange. Becausemales may ed by NIMH grantsMH07218 and MH08458. not be able to defend females from other males LITERATURE CITED under these circumstances,males may maxi- mize their reproductiveoutput by adoptingthe ARMSTRONG,E. A. 1955. The wren. London, Collins. alternatebehavior of copulatingwith those fe- BAIRD, J. 1958. The postjuvenal molt of the male malesthat enterthe areawhere they are dom- Brown-headed Cowbird (Molothrus ater). Bird- inant. According to this interpretation, male Banding 29: 224-228. cowbirds in the Prairie States should have BALFOURßE., & C. J. CADBURY.1979. Polygyny, spacingand sexratio amongHen HarriersCircus smallerranges than females. cyaneusin Orkney, Scotland.Omis Scandina- As a consequenceof thesepresumed differ- vica 10: 133-141. encesin cowbird spatial patterns,certain pre- BERGER,A. J. 1951. The cowbird and certain host dictions can be made regarding the deposition speciesin Michigan. Wilson Bull. 63: 26-34. of cowbird eggsin the Northeastand Prairie BEST,L. B. 1978. Field Sparrow reproductivesuccess States: (1) If females exclude same-sexed con- and nestingecology. Auk 95: 9-22. specificsfrom breeding areasin the Northeast, BRAY, O. E., K. H. LARSEN, & D. F. MOTT. 1975a. then the cowbird eggs found in any one ter- Winter movements and activities of radio- ritory shouldbe similar to eachother (indicat- equipped Starlings.J. Wildl. Mgmt. 39: 795-801. ing that they were laid by the same female), -, J. J. KENNELLY,& J. L. GUARINO. 1975b. Fer- tility of eggs produced on territories of vasec- yet different from eggs found in adjacentter- tomized Red-winged Blackbirds. Wilson Bull. ritories. Remarks by Friedmann (1929:176) 87: 187-195. support this prediction, although no data are ß W. C. ROYALL,JR., J. L. GUARINO, & R. E. presented.Exceptions should occurat areasof JOHNSON. 1979. Activities of radio-equipped territorialoverlap, where eggsfrom neighbor- CommonGrackles during fall migration.Wilson ing birds may be found. In contrast,(2) local, Bull. 91: 78-87. parasitized nests in the Prairie Statesshould BROWN,J. L. 1964. The evolution of diversity in contain dissimilar cowbird eggs, reflecting a avian territorial systems. Wilson Bull. 76: lack of territorial behavior in females. Finally, 160-169. BURNS,J. T., K. M. CHENG, •: F. MCKINNEY. 1980. (3) if territoriality functionsto increasea fe- Forced copulation in captive Mallards I. fertil- male's reproductivesuccess by limiting mul- ization of eggs.Auk 97: 875•79. tiple parasitism, there should be relatively BURTTßH. E., & M. L. GILTZ. 1976. Sex differences fewer cases of multiple parasitism involving in the tendency for Brown-headed Cowbirds two or more females in the Northeast than in and Red-wingedBlackbirds to re-entera decoy the Prairie States. trap. Ohio J. Sci. 76: 264-267. 326 ALVREDM. DvvrY,JR. [Auk, Vol. 99

CAREY,M., & V. NOLAN, JR. 1975ßPolygyny in In- HARVEY, P. H., J. GREENWOOD,& C. M. PERRINS. digo Buntings: a hypothesistestedß Science 190: 1979ßBreeding area fidelity of GreatTits (Parus 1296-1297ß major)ßJ. Anim. Ecol. 48: 305-313ß CASE,N. A., & O. H. HEWITT. 1963. Nesting and HILL, R. A. 1976ß Sex ratio and sex determination productivityof the Red-wingedBlackbird in re- of immature Brown-headed Cowbirdsß Bird- lation to habitatßLiving Bird 2: 7-20. Banding47: 112-114ß COCHP,•N,W. W., G. G. MONTGOMERY,•: R. R. GRA- HUNT, L. B. 1977ß Brown-headed Cowbirds form a •3ER. 1967. Migratory flights of Hylocichla pair bond for threeseasonsß Inland Bird Banding thrushesin spring: a radiotelemetrystudy. Liv- News 49: 223-224. ing Bird 6: 213-225ß KENNARD, J. H. 1978. A method of local limitation CONDER, P. J. 1949. Individual distanceß Ibis 91: of broodparasitism by the Brown-headedCow- 649•555. birdß North Amer. Bird Bander 3: 100-101. COVLSON,J. C. 1966. The influenceof the pair-bond KRœ•3s,J. R. 1971ßTerritory and breedingdensity in andage on the breedingbiology of the Kittiwake the GreatTit, Parusmajor L. Ecology52: 2-22ß Gull Rissatridactyla. J. Anim. Ecol. 35: 269-279ß LASKEY, A. 1950ß Cowbird behaviorß Wilson Bullß DARLEY,J. A. 1968. The social organization of 62: 157-174ß breeding Brown-headed Cowbirdsß Unpub- MAYFIELD, H. 1965ß The Brown-headed Cowbird, lished Ph.D. dissertationß London, Ontario, with old and new hostsßLiving Bird 8: 13-28ß Univ. Western Ontarioß McGEEN,D. S. 1972. Cowbird-hostrelationshipsß ß 1971. Sex ratio and mortality in the Brown- Auk 89: 84-93ß headed Cowbirdß Auk 88: 560-566ß MYERS,J. P., P. G. CONNORS,& F. A. PITELtrA. 1979ß DELIUS,J. D. 1965. A population study of Skylarks, Alauda arvensis. Ibis 107: 466-492. Territory size in wintering Sanderlings:the ef- fectsof prey abundanceand intruder densityß DOL•3EER,R. A., & R. A. STœHN.1979. Population Auk 96: 551-561ß trends of blackbirds and Starlings in North America, 1966-1976ßU.S. Fish and Wildl. Serv. NERO,R. W. 1956ßA behavior study of the Red- winged Blackbird1. Mating and nestingactivi- SpecialSci. Rept., Wildl. No. 214. ties. Wilson Bullß 68: 5-37ß DOFTY,A. M., JR. 1981. Socialorganization of the Brown-headed Cowbird, Molothrus ater, in New N•CE, M. M. 1937ßStudies in the life history of the York Stateß Unpublished Ph.D. dissertationß Song Sparrow, I. Transß Linnaean Soc. New York 4: 1-247ß Binghamton,New York, StateUniv. New York. ELLIOTT,P. F. 1978. Cowbird parasitismin the Kan- NOLAN,V., JR. 1978ßThe ecologyand behavior of the Prairie Warbler Dendroica discolorß Ornithol. sastallgrass prairie. Auk 95: 161-167ß 1980. Evolution of promiscuity in the Monogr. No. 26. Brown-headed Cowbirdß Condoi 82: 138-141ß NORRIS, R. T. 1947ß The cowbirds of Preston Firth. Wilson Bull. 59: 83-103ß EMLEN,S. T., & L. W. ORING. 1977. Ecology, sexual selection,and the evolution of mating systemsß PAYNE,R. B. 1965ßClutch size and numbersof eggs Science 197: 215-223ß laid by Brown-headed Cowbirdsß Condor 67: 44-60ß EwALD, P. W., & F. L. CARPENTERß1978. Territorial RAIM, A. 1978ß A radio transmitter attachment for responsesto energymanipulation in the Anna HummingbirdßOecologia 31: 277-292ß smallpasserine birdsß Bird-Banding 49: 326-332ß 1979ß Territoriality in the female Brown- , G. L. HUNT, JR., & M. WARNER. 1980. Ter- headed Cowbird (Molothrusater). Paper pre- ritory size in Western Gulls: importanceof in- sentedat the 97thStated Meeting of the A.O.U., trusion pressure, defense investments, and CollegeStation, Texasß vegetationstructure. Ecology 61: 80-87ß ROT•STEIN,S. I. 1972ßTerritoriality and mating sys- FANKHAUSER,D. F. 1971. Annual adult survival rates tem in the parasitic Brown-headed Cowbird of blackbirds and StarlingsßBird-Banding 42: (Molothrus ater) as determined from captive 36-42. birdsß Amer. Zool. 12:659 (abstract)ß FRIEDMANN,H. 1929. The cowbirds:a study in the ß 1976ß Cowbird parasitism of the Cedar biologyof socialparasitismß Springfield, Illinois, Waxwing and its evolutionaryimplicationsß Auk C. C. Thomasß 93: 498-509ß GATES, G. F., & L. W. GYSEL. 1978. Avian nest dis- SEARCY,W. A. 1979ßMale characteristicsand pair- persionand fledgingsuccess in field-foresteco- ing successin Red-winged BlackbirdsßAuk 96: tonesßEcology 59: 871-883. 353-363ß GOCHFELD,M. 1979. Broodparasite and host coevo- , & K. YASUKAWA. 1981ß Sexual size dimor- lution: interactions between Shiny Cowbirds phism and survival of male and female black- and two speciesof meadowlarksßAmer. Natur. birds (Icteridae). Auk 98: 457-465. 113: 8554376. SELANDER,R. K., & D. R. GILLER.1960ß First-year April 1982] CowbirdMovements and Activities 327

plumages of the Brown-headedCowbird and headed Cowbird on Empidonaxflycatchers in Red-winged Blackbird. Condor 62: 202-214. Michigan. Auk 78: 266-268. SIEGEL,S. 1956. Nonparametricstatistics for the be- WEST, M. J., A. P. KING, &: D. H. EASTZER.1981. havioral sciences. New York, McGraw-Hill. The cowbird: reflectionson developmentfrom TOMPA, F. S. 1962. Territorial behavior: the main an unlikely source.Amer. Sci. 69: 56-66. controllingfactor of a local Song Sparrow pop- WIENS,J. A. 1963. Aspectsof cowbirdparasitism in ulation. Auk 79: 687-697. southern Oklahoma. Wilson Bull. 94: 130-139. VAN V•LZ•N, W. T. 1972. Distribution and abun- --. 1973. Pattern and processin grasslandbird dance of the Brown-headedCowbird. Jack-Pine communities. Ecol. Monogr. 75: 237-270. Warbler 50: 110-113. WILLSON,M. F. 1966. Breeding ecologyof the Yel- V•RNV.R, J. 1964.Evolution of polygamyin the Long- low-headed Blackbird. Ecol. Monogr. 36: 51-77. billed Marsh Wren. Evolution 18: 252-261. ZIMMV.mWAN, J. L. 1971. The territory and its density WALKINSHAW,L. H. 1949. Twenty-five eggs appar- dependent effect in Spiza americana.Auk 88: ently laid by a cowbird.Wilson Bull. 61: 82-85. 591-612. ß 1961. The effect of parasitismby the Brown-

An International Symposiumon the Tamaulipan Biotic Provincewill be held at La Quinta Royale Motor Inn, CorpusChristi, Texas,on 28-30 October1982. The symposiumwill emphasizeall ecologicalaspects of the biome: vegetation,invertebrates, vertebrates, ecological structure and function,biological resources (use and effects),and management.Abstracts of contributedpapers are due 1 August 1982;abstracts for invited papers must be receivedby 1 September1982. For information on the program, contactGene W. Blacklock, Curator, Welder Wildlife Foundation, P.O. Drawer 1400, Sinton, Texas 78387 or David Riskind, Head, ResourceManagement Section, Texas Parks and Wildlife Department, 4200 Smith School Road, Austin, Texas 78744.For informationon registration,contact Jimmie R. Picquet, Director, John E. Conner Museum, Texas A&I University, P.O. Box 2172, Kingsville, Texas 78363.

The joint 8th International Conference on Bird Census Work and the 6th Meeting of the European Ornithological Atlas Committee will be held at Newland Park College, Chalfont St. Giles, Buckingham- shire, United Kingdom, on 5-9 September1983. The conferencewill include papers on validation and testingof censusmethods, application of censustechniques to bird communitystudies within and between habitats,description of vegetationand habitats for use in ornithologicalstudies, atlas studies,and data processingmethods. For further information, contact R. J. Fuller, British Trust for Ornithology, Beech Grove, Tring, Hertfordshire, HP23 5NR, United Kingdom. EFFECT OF HABITAT DECIMATION ON RING-BILLED GULL COLONY- AND NEST-SITE TENACITY

LINDA K. SOUTHERN AND WILLIAM E. SOUTHERN Departmentof BiologicalSciences, Northern Illinois University, DeKalb, Illinois 60115 USA

ABSTP,•CT.---Co1ony-sitetenacity and nest-sitetenacity have been documentedin several larids,but the proximatefactors affecting fidelity remainpoorly understood. We examined the effectof severebreeding-habitat alterations (bulldozing) on site tenacityin Ring-billed Gulls (Larusdelawarensis). Return ratesof wing-taggedadults were similar in bulldozedand unchangedparts of the colonysite. In bulldozedareas, however, most ring-bills abstained from nesting;those that did breedshowed an increasedtendency to moveto a differentnest site. Received14 April 1981,accepted 31 August1981.

FIDELITYto a previously used nesting area ing on the Great Lakes, Ludwig (1974) sug- has been demonstrated or hypothesized for gestedthat little, if any, attachmentwas shown many bird species (see Greenwood 1980 for to colony sites of various stabilities.Southern partial review). In coloniallynesting birds, the (1977)documented the existenceof strongcol- areato which a bird returnsfrom one breeding ony-site tenacity in this speciesat the stable seasonto the next may be as large, and rela- RogersCity Calcite site. Recently,we reported tively imprecise, as a colony site (colony-site that 96.7% of 152 adult Ring-billed Gulls re- tenacity) or as precise as a specificnest site turning to the RogersCity site were faithful to (nest-site tenacity). the same colony subdivision, and most birds Severallarids have been subjectsof investi- bred within 3 m of their previous year's nest gations of colony-sitetenacity and/or nest-site site (at least 69.5%, n = 59, Southern and tenacity. Austin (1940) documenteda well-de- Southern1979). Blokpoel and Courtney (1980) veloped tendency toward colony-sitetenacity reported similar results at another Great Lakes in Common Terns (Sterna hitundo), Arctic colony site. Of the 83 banded ring-bills they Terns (S. paradisaea),and Roseate Terns (S. resighted, 71 (85.5%) nested in the same col- dougallii). He subsequently reported (Austin ony sub-areafor 2 yr. In this case,at leastsome 1949) that 86.2% of 115 Common Terns bred of the birds that changedareas were forcedto within about 9 m of their former nest site. move, becausetheir previous site had been in- Coulson and White (1958) noted that breeding undated. Black-legged Kittiwakes (Rissa tridactyla) Although the existenceof colony- and nest- showedstrong attachment to the colonysite at site tenacity has been demonstratedin several which they had nested previously, but that larid species,investigations of the proximate 39.5% changed nest sites between years. Ver- factorsinfluencing tenacity have been limited. meer (1963) found that most (81.0%) Glaucous- Chronological age and breeding experience winged Gulls (Larusglaucescens) nested within have been convincingly shown to correlate about 5 m of their previoussite. Herring Gulls with increasing degrees of site attachment (L. argentatus)reportedly show a strong ten- (Austin 1940, Coulson and White 1958, Blok- dency to return to the vicinity of previously poel and Courtney 1980). Other factors, such used nest sites (84% to same sub-colony,Cha- as predation, low reproductive success,and brzyk and Coulson 1976). Franklin's Gulls (L. changesin habitat, have been implicated in pipixcan)were reported to visit the colony site causing breakdowns of tenacity (e.g. Mc- where they had nested in previous years, but Nicholl 1975, Erwin 1977, Conover and Miller after obtaining mates they often moved to a 1978).In spite of the obviousnegative impacts different site (Burger 1974). Nest-site tenacity of these factors, however, there are several ac- also was suggestedfor this species(op. cit.). countsof continuedre-use of adverselyaffect- For Ring-billed Gulls (L. delawarensis)nest- ed sites (e.g. Austin 1940, 1949; Southern and

328 TheAuk 99: 328-331. April 1982 April 1982] Ring-billedGull Site Tenacity 329

14 65

[] Bulldozed 57 40 A. B8 • 5 BO 84[] u.o.eed rr 71tB ,OO FC. _• •o • 60

0 • 1978 1979 197B 1979 1978 1979

YEAR OF MARKING Fig. 1. The numerals above histogram columns indicate the number of birds each represents.A. Pro- portion of marked birds from bulldozed and unaltered areas that were resighted at the Calcite colony in 1980. B. Proportion of resighted birds, from bulldozed and unaltered areas, that did not nest in 1980. C. Proportion of birds resightedas nestersin 1980, from bulldozed and unaltered areas,that nested in the same subdivision as in previous years.

Southern1978; Southern et al. 1979;Buckley bird was outfitted with a standard U.S. Fish and and Buckley 1980; Petersen1980). Wildlife Serviceband and a patagialtag carryinga In 1980, we addressedthe question: does unique code (Southern1971). We tagged 313 birds in 1978 and 395 in 1979. decimationof the breeding habitat affectnest- site tenacity in Ring-billed Gulls and, if so, in We designated and marked subdivisions of the colonyon the basisof permanentphysical structures, what way?Our experimentalsetup was pro- e.g. harbormarkers. Subdivisions varied widely in vided for us by the unexpectedbulldozing of shapeand size, averagingabout 600 m2 (mappedin partsof our studysite. By observingthe reac- Southern and Southern in press). We observed tions of Ring-billedGulls encounteringthese markedbirds from a car, servingas a blind, parked unusualhabitat changes,we hopedto gain a along gravel roads,which run throughoutthe col- better understandingof the influenceof such ony, and noted the subdivisions in which marked changeson nest-sitetenacity in this species. birds were sighted and/or nested.Observations were The changescaused by heavy equipment made on a near daily basisduring 13 May-13 June might simulatethe type of habitat alterations 1978, 8-23 May 1979, and 10 May-27 June1980. that ring-bills could experienceas a result of During the fall of 1979,when breedingbirds were not present,portions of the colonysite were covered ice or wavesscouring low-lying colonysites. with 15-30 cm of fill and graded. These alterations affectedsome areas that had been usedfor nesting STUDY AREA AND METHODS by birds wing-markedin 1978and 1979.All vege- tation was eliminated in these areas and a hard- This study was conductedat the Calcite colony packed, bare limestone substrate resulted. In con- near Rogers City, Michigan (Presque Isle County, trast, unaltered areas supportedground cover and 45øN, 83øW).A manmade peninsula that servesas a weedy vegetationduring the summer(e.g. Capsella, breakwall for the harbor area of U.S. Steel's calcite Chenopodium). plant hasalso served as a colonysite for nestinggulls for several decades. RESULTS In 1978-1979we captured adult Ring-billed Gulls during mid- to late-incubation by means of a can- Marked birds that had nestedpreviously in non-net. Cannon-netting was conductedonly in fa- bulldozed and unchangedareas were sighted vorable weather and with a minimum of five persons at the Calcite colony with similar frequencies present to expedite the procedure. Each captured in 1980 (Fig. 1A; marked 1978, X2 = 0.98, • = 330 SOUTHERNAND SOUTI-I•RN [Auk, Vol. 99

TABbE1. Comparison of 1979 and 1980nest censuses again in bulldozed areas than did so in unal- and mean clutch sizes in bulldozed and unaltered tered areas (Fig. 1C; marked 1978, X2= 8.32, areas. p = 1, P < 0.01; marked 1979, X2 = 45.92, p = Un01tered Bulldozed 1, P < 0.01). Nest-census and clutch-size data for 1979 Number of nests and 1980 are comparedbetween bulldozed and 1979 5,625 1,415 unaltered portions of the colony site in Table 1980 6,495 1,242 1. Although the total number of nests in bull- Change +15.5% -12.2% dozed areasdeclined between years, more than Mean clutch size 1,200 pairs bred in these areasin 1980. Along 1979 2.68 2.64 with the unchangedclutch size betweenyears, 1980 2.61 2.64 this would seem to indicate that the altered Change -2.6% 0 habitat was suitablefor nesting. We have very little information on the effect of habitat changeson the pair bond. Of five marked pairs identified in 1979 from unaltered 1, P = 0.32; marked 1979, X2= 0.25, •, = 1, areas,four remained togetherin 1980.One out P = 0.62). Overall resighting rateswere low in of three pairs from changedareas did not re- 1980 (marked 1978, 31.9%; marked 1979, unite in 1980. These samplesare not adequate 33.9%). We suspecthuman activities, partic- for drawing conclusions. ularly harassment and killing of birds away Male and female Ring-billed Gulls showed from the colony site, were partially responsi- similar degrees of site tenacity. Of 24 birds ble. A local outbreak of histoplasmosis, asso- identified as females(Southern 1981), 3 (12.5%) ciated with the colony site, had evoked sharp changed subdivisions in 1980; 2 of the 3 were public reactionagainst the gulls. It is unlikely, from bulldozed areas. Similarly, 13.3% of 45 however, that any one group of birds was af- males changedsubdivisions; 3 of these6 birds fected by thesefactors to a greaterextent than others. were from disturbed areas. Our sample sizes are not sufficientfor determining whether or Of thosebirds we recordedat the colony site not one sex more readily deserted bulldozed in 1980, significantly more from bulldozed areas than the other. areasdid not breed in the 1980 season(Fig. lB;

marked 1978, X2= 6.93, •,= 1, P < 0.01; DISCUSSION marked 1979, X2 = 13.92, •, = 1, P < 0.01). We designateda bird as a nonbreederif we never Severe breeding-habitat alterations (bull- saw it copulating, attending a nest or chicks, dozing) affected Ring-billed Gulls in the fol- or repeatedly defending a territory. Because lowing way. Birds that had nestedpreviously our sample of tagged birds was distributed in areas bulldozed in fall 1979 were seen at the throughout the colony, we had ample oppor- colony in 1980as frequently as thosefrom un- tunity to sight birds in all subdivisions. changedareas, but their tendencyto breed was Nonbreeding birds from bulldozed and un- reduced significantly.Among thosebirds from altered areas seemed to differ in their move- disturbed areas that did breed, relocation to a ments at the colony, althoughstatistically sig- different colony subdivision occurred with nificant differenceswere not shown (X2 = 3.19, greater than expected frequency. Ring-bills •, = 1, P = 0.07). Nonbreeders from un- that encountered a drastically changed local changed areas were sighted frequently (33 environment upon arrival at the colonyand did times) in the vicinity of their old nest site (i.e. not breed in 1980 were equally likely to be same subdivision), but not uncommonly (14 sighted near their old site or in other parts of times) elsewhere in the colony. Birds from the colony. bulldozed areaswere seennear and away from It appears that bulldozing somewhat re- their old sites equally often (15 times each). duced the attractiveness of the habitat to re~ Figure 1C illustrates the tendency of those turning gulls. The number of nests in bull- birds that did breed in 1980 to nest in the same dozed areas decreased between 1979 and 1980, subdivision they had used the precedingyear. while increasingin unaltered areas. Bulldozed Lower proportions of breeding birds nested areas were more open than unaltered areas, April1982] Ring-billedGull Site Tenacity 331

and this may have contributed to an unfavor- Co•ow•, M. R., & D. E. M•Lr•. 1978. Reaction of able reactionon the part of somebirds. Ring-billedGulls to predatorsand human dis- Over 1,200 pairs did nest in bulldozed areas, turbancesat their breeding colonies.Proc. 1978 but most marked birds familiar with those Conf. Colonial Waterbird Group: 41-47. CouLso•, J. C., & E. W•tE. 1958. The effectof age areas moved or did not nest. This implies that on the breedingbiology of the Kittiwake. Ibis many birds breeding in disturbed areas may 100: 40-51. have been immigrants. Somefaithful, but non- ERw•, R. M. 1977. Population and colony site dy- breeding, marked birds remained near their namics in selected Massachusetts waterbirdsß prior sites throughout the season,and some Proc. 1977 Confß Colonial Waterbird Group: occasionally defended territories. By their 19-25. presence,these tenacious ring-bills might have GRI•EiX/WOOD,P. J. 1980. Mating systemsßphilopatry attractednewcomers that perceived no change and dispersalin birds and mammalsßAnim. Be- in the habitat and proceededto breed. hav. 28: 1140-1162. Luvw•c, J. P. 1974. Recent changesin the Ring- ACKNOWLEDGMENTS billed Gull population and biology in the Lau- rentjan Great Lakesß Auk 91: 575-594ß We appreciatethe financial support of the McNIcHorr, M. K. 1975. Larid site tenacity and Northeastern Bird-Banding Association, group adherencein relationto habitatßAuk 92: through the E. Alexander BergstromAward to 98-104. the seniorauthor. J. P. Ryder, P. L. Ryder, and Pst•s•, M. R. 1980. Red fox predationof seabirds at Shaiak IslandßAlaskaß Pacific Seabird Group an anonymousreviewer provided helpful com- Bullß 7: 54-55. (Abstractß) ments on the manuscript. Advice on statistical SoutHErn, L. K. 1981. Sex-related differences in ter- procedureswas generouslyfurnished by J. H. ritorial aggressionby Ring-billedGulls. Auk 98: Zar, and illustrationswere prepared by Jane 179-181. Glaser.Access to the colonysite, for which we ß & W. E. SouT•s•. 1978. Absence of noc- are grateful, was grantedby U.S. Steel. turnal predatordefense mechanisms in breeding gullsß Proc. 1978 Conf. Colonial Waterbird L•TS•,•TURE Group: 157-162. ß & --. 1979. Philopatry in Ring-billed Aust•, O. L. 1940. Some aspectsof individual dis- Gullsß Proc. 1979 Confß Colonial Waterbird tribution in the Cape Cod tern colonies. Bird- Group, vol. 3: 27-32. Banding 11: 155-169. SOu•ER•, W. E. 1971. Evaluationof a plasticwing- ß 1949. Site tenacity, a behaviourtrait of the markerfor gull studiesßBird-Banding 42: 88-91. Common Tern (Sterna hirundo Linn.). Bird- ß 1977. Colony selectionand colony site te- Banding 20: 1-39. nacity in Ring-billed Gulls at a stablecolonyß BLOKPOEL,H., & P. COURTNEY.1980. Site tenacity Auk 94: 469-478. in a new Ring-billed Gull colony. J. Field Or- ß & L. K. Sou•. In pressßColony census nithol. 51: 1-5. results as indicatorsof pre-hatchingperturba- BUCKLEY, F. G., & P. A. BUCKLEY. 1980. Habitat tions. Colonial Waterbirdsß selection and marine birds. Pp. 69-112 in Be- ß S. R. P^tto•, & L. A. H^•s. 1979. Dif- havior of marine animals (J. Burger, B. L. Olla, ferentialresponse of Ring-billedGulls and Her- and H. E. Winn, Eds.). New Yorkß Plenum ring Gulls to fox predation.Proc. 1979 Conf. Press. ColonialWaterbird Group, vol. 3: 119-127. BuRc•, J. 1974. Breedingadaptations of Franklin's V•M•E•, K. 1963. The breeding ecology of the Gull (Laruspipixcan) to a marshhabitatß Anim. Glaucous-winged Gull (Larus glaucescens)on Behav. 22: 521-567. Mandarte Island, BßCßOcc. Papers, British Co- C•^B•z•ic, G., & J. C. CouLso•. 1976ß Survival and lumbia Provincial Mus., No. 13. recruitmentin the Herring Gull Larusargentatus. J. Anim. Ecol. 45: 187-203. PARENTAL FEEDING OF NESTLING YELLOW WARBLERS IN RELATION TO BROOD SIZE AND PREY AVAILABILITY

GLORIA C. BIERMANN AND SPENCER G. SEALY Departmentof Zoology,University of Manitoba,Winnipeg, Manitoba R3T 2N2, Canada

ABSTRACT.--Thefeeding by adult Yellow Warblers (Dendroicapetechia) of nestlingsin broodsof 3, 4, or 5 young was studied during two breeding seasonson the foresteddune ridge near Delta, Manitoba. Broodsof 2-day-old youngwere selectivelyfed geometridlarvae by the adults, and broodsof 8-day-oldyoung were selectivelyfed chironomidsand geo- metrid and other Lepidoptera larvae. The diet varied among broodsof 3, 4, and 5 young at both ages,but this variationdid not appear to affectthe growth ratesof the young. The total weight of items brought per half hour increasedwith brood size among broods of 2-day-old young. Broodsof four 8-day-old youngwere fed the greatestamount of food. The diet of the nestlingsis contrastedwith the diet of adult Yellow Warblers observedin the same locality. Received15 October1980, accepted19 October1981.

THE Yellow Warbler (Dendroicapetechia), 1978and 1979on a portion of the foresteddune ridge along with six other speciesof primarily in- that separatesLake Manitoba from the Delta Marsh, sectivorouspasserines, nests densely on the about 5 km west of Delta (50ø11'N, 98ø19'W), Mani- foresteddune ridge that separatesLake Man- toba. Willows (Salix amygdaloidesand S. interior), itoba from the Delta Marsh, Manitoba. Both Manitoba maple (Acernegundo), and greenash (Frax- the adults and nestlingsof the speciesstudied inus pennsylvanica)account for over 95% of the tree flora and provide a forest canopy about 10-14 m in so far in this areaare apparentlysustained by height (see MacKenzie 1979). The prominent shrubs a superabundantfood resource,adult midges are elderberry(Sambucus pubens), dogwood (Cornus (Diptera: Chironomidae) (Busby and Sealy stolonifera),and raspberry (Rubusideaus) (see Goos- 1979, Biermann 1980, Sealy 1980).The greatest sen 1978, Busbyand Sealy 1979). demand on the food resourcesin the ridge We located 139 active Yellow Warbler nests and communityprobably occursduring the period recordedtheir contentseach day during egg-laying when the young are being reared.It is impor- and every 3-4 days thereafter. Observations were tant, therefore, to determine the diets of nest- made at 19 nestswith broodsof 3, 4, or 5 young on lings of the speciesin this communityin re- day 2 (young young, YY) and day 8 (old young, OY) after the first chick in each brood had hatched. At 10 lation to the food resources that are available nests at least one parent was color-banded, at 5 nests at this time. We did this for the Yellow War- neither parent was banded, and at the other 4 nests bler, and we examined the rates at which at least one parent was aluminum banded. Adults at adults fed their young in broodsof 3, 4, and each nest were separableby plumage differences. 5 at two ages.To determine the effectsof any Four nests of each of the three brood sizes at each feeding differencesthat might have been ob- age were observed(except that 3 nestswith broods served, we also measuredthe growth rate of of five 8-day-oldyoung and 7 nestswith broodsof the young. The diet of the nestlingsrelative to three 2-day-old young were observed).Most of the the prey available was also contrastedto that nestsobserved were lessthan 2 m abovethe ground. of adult Yellow Warblers studied on the same Observationswere made from a blind placed 1-2 m area by Busbyand Sealy(1979). As the diet of from the nest the day before observationswere be- gun, at four timesof the day: 0630-0830,1000-1200, nestlings of most parulid speciesis poorly 1530-1730, and 1830-2030 (CDT). Observations known, beyond the general knowledge that spanned 12 June to 1 July in 1978 and 20 June to 19 they are insectivorous,our studyfills in a gap July in 1979. Parental activities at or near the nest for the Yellow Warbler in one area. and the number and identity of prey items fed to the young were recorded. Large insects that protruded METHODS from the parents' bills could be identified, but small items could not. Therefore, the prey we identified We studied parental care in a denselynesting pop- were biased toward larger sizes. Items brought were ulation of Yellow Warblers during the summersof unidentifiablein 347of 1,211feeding trips observed,

332 TheAuk 99:332-341. April 1982 April 1982] YellowWarbler Parental Feeding 333

T^BLE1. Feedingrates and food loads(mean + SE) of Yellow Warbler parents.

Trips/half hour Items/trips Age and number of young Males Females Males Females 3YYa (48)b 1.9 + 0.2c 1.0 + 0.2 1.2 + 0.1 1.2 + 0.1 4YY (32) 2.6 + 0.2c 1.6 + 0.2 1.3 & 0.1 1.2 + 0.1 5YY (36) 3.1 + 0.3d 0.9 + 0.3 1.2 & 0.1 a 1.0 +0 3OY (32) 3.8 + 0.5 3.4 + 0.2 1.9 + 0.1 1.5 + 0.1 4OY (31) 5.2 + 0.5ø 2.7 + 0.4 2.6 + 0.1 2.2 + 0.2 5OY (32) 4.2 + 0.3d 5.6 + 0.5 1.5 & 0.1 ½ 1.2 + 0.1 YY (youngyoung) refers to 2-day-oldyoung; OY (old young)refers to 8-day-oldyoung. Number of half-hour observationperiods in parentheses. P -c 0.01 between males and females. P -c 0.025 between males and females.

due in part to their smallersize, but alsobecause the which they jumped out of the nest following han- parents'positions while feedingthe youngobscured dling. our view. Statistical tests used include analysis of variance In 1979,27 8-day-oldnestlings were collected,and (ANOVA), linear regression,Chi-square, and Wil- 70% ethanol was injected immediately into their coxonsign rank test. The level of significanceused esophagi.The stomachcontents were removed with- was P • 0.05. in an hour and stored in ethanol. Food items were identified later in the laboratoryusing a variable- RESULTS powermicroscope. Intact prey itemswere few in the samples,but head capsules,thoraxes, wings, occip- Feedingrates and food loads.---Timeof day ital rings, and mandiblescould be identified to order had no effect on the feeding rate of adult Yel- and often to family. This method probably biases low Warblerswith young of either age. Broods resultstoward insectswith thick hard coverings.The of five 2-day-old young were fed significantly prey itemsobserved being fed werecompared to the more often by malesthan were smallerbroods prey items in stomachsto assesspotential bias. (Table 1). Males with 8-day-old young fed The foliage arthropodfauna of a 100-m line tran- sectof a portion of the study areawas sampledusing broodsof 4 young more often than broodsof a standardinsect sweep net, in the mannerdescribed 5 or 3. Among females,the only significantre- by Busby and Sealy (1979). They found that this sult was that broods of five 8-day-old young method yielded representativesamples when as- were fed more often than were smaller broods sessingthe availabilityof arthropodsto adultYellow [(two-way ANOVA: maleswith YY-brood size, Warblerson this studyarea. The faunawas sampled F(2, 104) = 6.16; time of day, F(3, 104) = 2.47; in the morning and eveningof everyfifth day during males with OY-brood size, F(2, 83)= 2.94; the observationperiods. The sampleswere sorted time of day, F(3, 83) = 0.71; femaleswith YY- and identified to order or family and grouped into brood size, F(2, 104) = 2.52; time of day, F(3, five categoriesto permit comparisonwith the prey 104) = 1.40; femaleswith OY-brood size, F(2, being fed to the nestling warblers. These groups were: Chironomidae and Culicidae (midges and 83) = 11.04; time of day, F(3, 83) = 2.30]. mosquitoes),Geome•ridae larvae (inchworms), all The number of items brought per trip by other Lepidopteralarvae, all other Diptera, and all males and females with 2-day-old young was other insectsand arthropods. constantthroughout the day and did not vary The dry weights(to the nearest0.1 mg) of the sep- amongthe differentbrood sizes [Table 1; two- arate samplesof midges, mosquitoes,and larvae way ANOVA: males--brood size, F(2, 229) = were determinedafter being oven-dried at 60øCfor 0.77; time of day, FO, 229) = 0.76; females-- one week. The two groupsof larvaewere then com- brood size, F(2, 76) = 1.43; time of day, F(3, bined, and the protein contentof the larvaeand the 76) = 1.50]. Males feeding 8-day-old young chironomid and culicid sampleswas determined by the Manitoba Departmentof Agriculture. broughtthe largestloads to broodsof 4 (Table Yellow Warbler nestlings in randomly chosen 1) and generallybrought larger loads in the nestswere weighedto the nearest0.1 g between2030 early morning and smaller loads in the late and 2130 (CDT) on a triple-beambalance. Nestlings afternoon[two-way ANOVA: brood size, F(2, could be weighed until day 6 post-hatching,after 273) = 30.04; time of day, F(3, 273) = 3.51]. 334 BIERMANNA•D SEALY [Auk, Vol. 99

TAI31•E2. Number of insectsin each group observedbeing fed by adult Yellow Warblersto young in three brood sizes at two ages.

Broodsize and age 3YY 4YY 5YY 3OY 4OY 5OY

Insectgroup (48)a (91) (52) (83) (32) (111)(108) (121) (82) (159) (178)(134) Chironomids and culicids (I) 22 28 26 29 1 34 68 110 117 247 95 73 Geometrid larvae (II) 8 48 15 42 18 91 10 37 6 17 34 49 Other larvae (III) 1 2 0 0 0 2 2 8 1 2 1 3 Other Diptera (IV) 1 0 1 1 3 5 5 1 5 1 0 1 All other insects (V) 1 0 1 1 0 5 5 6 13 51 15 5 Feeding trips with unidentified prey 19 16 15 21 10 7 42 43 21 42 55 44 Total number of feeding trips observed.

Femalesfed broodsof four 8-day-oldyoung the When the young were 2 days old, more food largestnumber of itemsper trip (Table1) and items were brought by malesthan by females. did not vary their rate with time of day [two- Overall, 2~day-oldbroods of 4 were fed more way ANOVA: brood size, F(2, 238)= 25.91; food items than broods of 5 or 3. When the time of day, F(2, 238) = 1.74]. young were 8 daysold, more food items were Males fed 2-day-old young more often than brought by malesto broodsof 3 and 4 than by females regardlessof brood size [ANOVA: females, while the reverse was true for broods broodsof 3, F(1, 94) = 8.54; broodsof 4, F(1, of 5. Overall, broods of 4 were fed more items 62) = 7.97;broods of 5, F(1, 70) = 29.95].With than broods of 3 or 5 at this age. 8-day-oldyoung, males and females fed broods Fooditems available and used.--As time of day of 4 at equivalentrates [F(1, 62) = 0.68], males did not affect the type of prey the adults fed broods of 3 more often than their mates did brought to the young (Chi-square,P • 0.05 in [F(1, 60) = 14.3], and femalesfed broods of 5 10 of the 12 cases), the observations from all more often than males did [F(1, 62) = 5.33]. time periodswere combined.At eachage, sig- Both sexes fed 8-day-old young more often nificant differencesin the proportion of each than 2-day-old young [females---broodsof 3, of the five insect groups in the diet brought by F(1, 78) = 61.9; broods of 4, F(1, 61) = 5.04; femaleswere observedamong brood sizes (Ta- broodsof 5, F(1, 66) = 69.0; males--broodsof ble 2; YY: X2 = 30.6, 8 df; OY: X2 = 36.3). At 3: F(1, 78) = 15.6; broodsof 4, F(1, 61) = 22.3; both ages, femaleswith 5 young brought more broods of 5, F(1, 66) = 6.9]. geometrid larvae and fewer chironomids and Males with 5 young brought a larger food culicids than did females with broods of 3 and load per trip than did femalesat both agesof 4 young. There were also significant differ- nestlingsstudied. No other significantdiffer- encesbetween 2- and 8-day-old young in the encesbetween parents were observed[ANO- proportion of items brought by females with VA: 3YY, F(1, 102)= 0.63; 4YY, F(1, 97)= broodsof 4 and 5 young (broodsof 3: X• = 4.2, 1.02; 5YY, F(1, 124)= 5.39; 3OY, F(1, 142)= 4 df; broods of 4: X• = 29.9, broods of 5: X• = 0.68; 4OY, F(1, 176) = 3.40; 5OY, F(1, 211) = 38.1). A larger proportion of chironomidsand 10.3]. culicids and a smallerproportion of geometrid All parents except females with 5 young larvae were fed to the older young in broods brought greaterload sizeswhen feedingolder of 4 and 5, but females that were feeding young[females--broods of 3, F(1, 93) = 8.27; broods of 3 maintained a relatively constant broods of 4, F(1, 96) = 17.09; broods of 5, F(1, diet between the two ages. 143) = 3.52; males---broods of 3, F(1, 151) = The proportion of insectsin each group ob- 53.1; broods of 4, F(1, 177) = 52.5; broods of served being fed by males (Table 2) differed 5, F(1, 192) = 7.94]. significantly among brood sizes at each age April 1982] YellowWarbler Parental Feeding 335

TABLE3. Total number of insects in each group Yellow Warblers selected geometrid larvae found in the stomachsof nine 8-day-oldwarbler (P = 0.016) and avoided feeding all other lar- nestlings in each brood size. vae (P = 0.031) and all other insects (P =

Brood size 0.016) when feeding 2-day-old young. Adults Insect feeding8-day-old young selectedchironomids groupsa 3 4 5 and culicids (P = 0.008), geometrid larvae I 76 109 71 (P = 0.008), and other larvae (P = 0.063) and II 28 17 21 avoided all other Diptera (P = 0.016) and all III 2 4 4 other insects (P = 0.008). Thus, adult Yellow IV 14 5 10 V 13 15 16 Warblers selectedparticular prey items as food for their nestlings. See Table 2. Weightand protein content of nestlingfood.- The averagedry weights of insectsin groups I, II, and III were 3.1, 6.0, and 8.5 rag, respectively (YY: X2= 20.5, 8 df; OY: X2= 103.4). Again, (Biermann 1980). The weight of an "average" broods of 3 and 4 were fed more chironomids insect fed to a nestlingwas calculatedfor each and culicids and fewer larvae than broods of parent with each brood size at each age by 5. The proportion of items in each group multiplying the number of individual prey brought by males also differed significantly items of each group being fed (Table 2) by the betweenthe two ages(broods of 3: X2 = 34.3, averageweight of insectsin that group, sum- 4 dr; broods of 4: X2 = 130.5; broods of 5: X2 = ming for groupsI, II, and III and dividing by 30.3) and followed a trend similar to that of the the total number of insectsin the three groups. females. This estimate, multiplied by the total number Intersexual differences in the proportion of of insectsbrought per half hour (Table1) yields items observedbeing fed to the nestlingswere an estimateof the averageweight of food items also significant,except between parents feed- brought per half hour (see Table 4). Because ing four 2-day-oldyoung (3YY: X 2 = 12.7;4YY: the items observedbeing fed were biased to- X•= 5.5; 5YY: X2= 10.3; 3OY: X •= ward larger prey, the true weight of food 13.7; 4OY: X• = 10.8; 5OY: X• = 11.2; all have brought per half hour was smaller. As the 4 df). Males generallybrought more geometrid number of feeding trips in which prey were larvae than did females.Females brought more unidentified was relatively constant (usually chironomids and culicids. about one-quarterto one-third of the feeding The proportionsof items observedbeing fed trips, Table 2), however, the values calculated to nestlingsand those in the stomachs(Table are useful in comparing the feeding rates. 3) differed significantly in each brood size, Generally,males brought heavier prey items mainly due to differencesin the proportionof than did females.Heavier items were brought insectsin groupsIV and V. Theseinsects were to 2-day-old young than to 8-day-old young, usually small and were underestimatedin the becausea larger proportion of the diet of 2- field. The stomachcontents of the young in the day-old young was Lepidopteralarvae. Males three brood sizeswere not significantlydiffer- feeding2-day-old young broughtmore food to ent (Xs = 13.5, 8 df), although the trend found largerbroods. Females fed 2-day-oldyoung at is similar to the field-observed diet in that the a rate unrelated to brood size. Femalesbrought largest proportion of chironomids and the more food to larger broodsof 8-day-old young smallest proportion of larvae were fed to than to smaller broods. Males did not. Overall, broods of four. at 8 days, broodsof 4 were fed more food by The prey items fed are compared in Figs. 1 weight than were broodsof 5 or 3. and 2 to prey items available in each sampling Other factorsapparently are also important. period. Although the total number of insects The fat, ash, protein, and water content of in- varies, the relative importance of each group sects,as well as their digestibility, may affect remained fairly constantthroughout the ob- the proportionof the dry weight that is usable servation periods. A Wilcoxon sign test was by the nestling.The proportionof dry weight used to compare the proportion available in that is protein was 59.8% and 62.6% for the each group to the proportion(by number) of chironomids and larvae, respectively. Protein that group observedto have been fed. Adult contentis likely to be the most important com- 336 BIERMANN^NV SE^L¾ [Auk, Vol. 99

Available Used 2:51I 2 42 II A I ? []-yy $ 129 [] -oy 117' 0 I I I I 60 40 20 0 20 40 60 80

25 I 57 $1 I[ 12 B 24 IO I I 1• 43 2 $ 0 0 6O 40 20 0 20 40 60 80 , 1 66 I ]258 •//////////////////////////• 216 Tr 6 [ 40

rrr 2 c • 39 iv 59 o o •z 134 I I 6O 40 20 0 20 40 60 80

•////////////////////////////• 193

II D IO

6

I I I I 14 40 20 0 20 40 60 80 Proportion in Sample % Proportion in Diet % Fig. 1. Proportion, by number, of insectsin groupsI-V in sweep net samples,and the proportion of insectsin eachgroup observedbeing fed to 2- and 8-day-oldYellow Warblersin 1978.A, 12-16 June;B, 17-22 June;C, 23-27 June;D, 28 June-1 July. Absolutenumbers of insectsare given on the sides.

ponent of a nestling'sdiet, becauserapid body is protein to determine the amount of protein growth occurs during this stage of develop- in the insects.The total weight of protein fed ment (Kear 1972). by the parents of the nestlingsat each age and The averagedry weight of the insectswas in each brood size was determined using cal- multiplied by the proportionof the weight that culations described above (Table 4). Compar- April 1982] YellowWarbler Parental Feeding 337

Available Used i I 588[ II 4 'rrr 17 A [--I-yy • 81 E3-oy •' 62 I 8O 60 • 20 0 20 40 60 T 416 J I 93 • 8 B 2 7 iv 86 6 '•T 106• • 8O 60 • 20 •0 20• I40 • 80 5• • •9

rr 4 C 60

]]• 6 5

1•' 159 2

i i i 16 80 60 40 20 0 20 • 60 I 194 • 124 D H 4

'rrr 5

2Z :550 'v 202

8O 6O 40 20 0 20 • 60 80

I 104 [ •//////•,y////////)/////•/ 28

II 2 2 m' 2

'•7 115 0 • 65• 8O 60 40 20 0 20 40 60 8•0 I 338 II 3 F ]]I 2 i • 220 o '•T 140• 8O 60 40 20 o 20 • 60 80 Proportionin Sample % Proportion in Diet % Fig. 2. Proportion,by number,of insectsin groupsI-V in sweepnet samples,and the proportionof insectsin eachgroup observedbeing fed to 2- and 8-day-oldYellow Warblersin 1979.A, 20-24 June;B, 25-29June; C, 30 June-4July; D, 5-9 July;E, 10-14July; F, 15-19July. Absolute numbers of insectsare given on the sides.

isons among brood sizes are similar to those the growth rates of the young. Fig. 3 shows describedfor the dry weight of food brought the growth rate, measuredas changein body by the parents, but the differences in the weight, of nestlings in broods of 3, 4, and 5. amount fed among the brood sizes are gener- Becausethe same nestlingswere not weighed ally decreased. every day and the weights shown are means Growth rates.---As broods of 3, 4, and 5 of all nestlingsweighed, the curvesdo not con- young are fed different proportions of food form exactly in shape to standard growth items at different rateson a per nestlingbasis, curves. The overall growth rates of young in these differencesmight be expectedto affect the brood sizes can be compared using the 338 BIERMANNAND SEALY [Auk,Vol. 99

TABLE4. Average dry weight (mR) and protein content (mR) of insectsfed by male and female parent Yellow Warblers to broods of 3, 4, and 5 young at two ages.

Averagedry Averagedry Mean dry weight Averagedry Brood weightof average weightof food ofprotein per weightof protein size insectbrought broughtper half-hour averageitem broughtper half-hour and age Males Females Males Females Males Females Males Females

3 YY 5.0 4.0 11.6 4.8 3.2 2.5 7.4 3.0 4 YY 4.8 4.2 16.4 7.9 3.0 2.2 10.2 4.2 5 YY 5.3 5.9 19.5 5.3 3.3 3.7 12.2 3.3 30Y 4.3 3.6 30.8 18.4 2.5 2.2 18.0 11.2 40Y 3.3 3.3 45.8 19.4 2.0 2.0 27.0 11.8 50Y 4.4 3.9 27.5 26.1 2.7 2.4 17.0 16.1

regressionequations provided in Fig. 3. A test (Moreau 1947, Lack and Silva 1949, Lack and for the equality of slopes(Sokal and Rohlf 1969) Lack 1951, Royama 1966, Morehouse and showed no significantdifference in the growth Brewer 1968, Hussell 1972, Best 1977, Walsh rate of the young in the three brood sizes [AN- 1978). The feeding rate of adult Yellow Warb- OVA: F(2, 11) = 0.65]. Using a graphicaltech- lers was greatestin the largestbrood size when nique to convertthe growth curve to a straight broods were 8 days old. Food loads, however, line (Ricklefs 1967), we calculatedthe specific were not the largest(Table 1). At 2 daysof age, growth rate constant (K) for young in each both feeding rate and food-loadsize were great- brood size, using a weight asymptoteof 11.0 est for broods of 4. g. The rate constants(0.546, 0.551, and 0.541 Other workers have also found that feeding for broods of 3, 4, and 5 young, respectively) rate is not directly related to brood size. Pin- again suggestedsimilar growth ratesof young kowski (1978) found no positive relation be- in the three brood sizes. tween brood size and feeding rate by either male or female Eastern Bluebirds (Sialia sialis). DISCUSSION Seel (1969) found that the feeding rates for FEEDING RATES AND FOOD LOADS broods of 3, 4, and 5 House Sparrows (Passer domesticus)did not differ significantly. Time of day.--Time of day did not usually Feeding rates were shown in our study and affect the feeding rate and load size of adult by several others to increase as the nestlings Yellow Warblers feeding their young. In many grew older (Gibb and Betts1963, Royama 1966, species,most feeding usually occursjust after Seel 1969, Nolan 1978, Pinkowski 1978, Walsh dawn and at dusk, and the least occurs at mid- 1978). Larger young generally require more day or mid-afternoon (Best 1977, Nolan 1978, food to supply sufficientenergy for thermoreg- Pinkowski 1978). These variations are usually ulation and developmentalprocesses. related to differences in the activities of the The role of males relative to females in feed- prey species,as many insectsare most active ing the young varies among species. Nolan in the coolestparts of the day, or are associated (1978) found that female Prairie Warblers (D. with the need of the birds to feed heavily just discolor)assumed a slightly larger proportion before and after the overnightfast. The diet of of the feeding duties, whereasField Sparrows some bird speciesmay also vary with time of (Spizellapusilla) share this duty nearly equally day (Walsh 1978). In the presentstudy, how- (Best 1977). On our study area, male Yellow ever, early-morning sampling was not done Warblers usually assumed the larger propor- just after dawn, nor was evening sampling tion of feeding duties. done just before sunset. The lack of dietary Gibb and Betts (1963), Royama (1966), and variation during the day indicates that insect Morehouse and Brewer (1968) noted that the activities did not greatly affect the foraging feeding rate and the size of items brought were success of the birds. inverselyrelated. Nolan (1978)found that adult Broodsize and age.--Generally, adult feeding Prairie Warblers brought similar sized food rate increases with an increase in brood size items, but, as the young grew older, both the April 1982] YellowWarbler Parental Feeding 339

15 feeding rate and the size of items fed in- 10 creased. Thus, he observed no inverse corre- lation between feeding rate and the size of

8 items fed. In the Yellow Warbler, males brought heavier items and fed the young more often. Within each sex, however, the inverse relation between feeding rate and food size held.

DIET OF NESTLINGS

Busby and Sealy (1979) found the diet of

I I I adult Yellow Warblers during the breeding 2 3 4 seasonat Delta to be composedmainly of Dip- Age ( Days tera, particularly chironomids. Studies of the diet of adult Yellow Warblers in other areas

10 have found Hymenoptera or Homoptera and Coleopterato be the major componentsof the

8 diet (Kendeigh 1947, Frydendall 1967). Adult Yellow Warblersapparently are flexiblein their choice of food and forage opportunistically (Busby and Sealy 1979). Adult chironomids and culicids and larval geometridswere the main componentsof the diet of nestling Yellow Warblers. The propor- tion of these items in the diet changed with the nestlings' age. Two-day-old young were fed more geometridlarvae than were 8-day-old I [ I I I 2 5 4 5 young, with a corresponding change in the Age (Days) proportionof chironomidsand culicids.Stom- ach analysesconfirmed, to an extent, feeding observations and indicated that chironomids, 5 not culicids, were the most important insect in group I. Apparently, the abundant chirono- 8 raids are an important source of nourishment for both adult and nestling Yellow Warblers during the breeding seasonat Delta. Diet and prey availability.-•Generally, adult Yellow Warblers did not feed their nestlings prey in proportion to its availability. They were selective.At both nestling ages, geome- trid larvaewere fed in proportionsgreater than were available. Other studies of passerine I I t I I nestling feeding have found Lepidoptera lar- I 2 • 4 5 vae to be important componentsof the diet Age (Days) (Best 1977, Pinkowski 1978). In the present Fig. 3. Growth curves of Yellow Warblers in study, all other larvae and all other insectswere broods of 3 (A), 4 (B) and 5 (C) young. The hori- under-represented in the diet of 2-day-old zontal lines are the mean weights, the vertical lines the ranges, and the solid bars denote 1 SE on each young, comparedwith what was available. side of the mean. The numbers above each vertical When 2-day-old young were fed a geometrid line are samplesizes. Regression equations: 3 young, larva, the parents often passedit back and forth y = 1.38 + 1.31x; 4 young, y = 1.56 + 1.39x; 5 betweenthemselves and pulled on it, breaking young, y = 1.30 + 1.20x. The slopesof all regression it into smaller, more manageablepieces that equationsare significantlydifferent from zero. were then fed to the nestlings.Attempts to feed 340 BIERMANNAND SEALY [Auk, Vol. 99 a larva whole to the nestlingsusually failed. ACKNOWLEDGMENTS Becauseother larvae are usually wider and We thank Dr. J. M. Shayand the staffof the Uni- heavier than geometrid larvae, they would versity of Manitoba Field Station (Delta Marsh) for have required even more processingby the providing pleasantsurroundings and help during parentsbefore they couldbe fed to the young. our field studies.The PortageCountry Club kindly Probably more effort is required than is prac- permitted us to conductpart of this work on their tical. Best (1977) found that girth rather than property.B. D. J. Batt, L. B. Best,T. D. Galloway,H. lengthlimited the size of food that youngField B. ToLdoff,T. J. Wiens, and two anonymousreview- Sparrows could ingest. Several studies have ers commentedon earlier drafts of the manuscript, shown that the size of food items increases and their criticalsuggestions were valuable. We ex- tend our appreciationto H. den Hann, J. L. Morgan, with nestling age, suggestingthat the relative J. M. Porter, and G. D. Sutherland, who assisted in size of the nestlingsand the food may be im- aspectsof the field work. T. D. Galloway kindly portant (Gibb and Betts 1963, Best 1977, Nolan identified and confirmed our identifications of the 1978, Pinkowski 1978, Walsh 1978, Johnson et arthropodsfrom the stomachsamples. D. Waddell, al. 1980). All other insects collected in the in- Manitoba Departmentof Agriculture,arranged for sect samplesincluded many hard bodied in- the analysisof the protein contentof the insectfood, sectssuch as Coleoptera,Hymenoptera, and for which we are grateful. Hemiptera, all of which may have been un- This work was funded by grantsfrom the Natural suitablefood for young birds. Sciencesand EngineeringResearch Council of Can- Chironomids and culicids and all other lar- ada (A9556)and the Universityof Manitoba Research Boardto Sealy.Biermann received personal support vae were selectedpreferentially for 8-day-old from a University of Manitoba Fellowship for young, as were geometrid larvae. At 8 days 1979-1980.This paper is contributionnumber 70 of of age, the nestlingsare nearly adult size and the University of Manitoba Field Station (Delta can swallow larger larvae whole. Chironomids Marsh). and culicidshave soft abdomensthat are easily LITERATURE CITED digested,and, becausethey move slowly, sev- eral can be broughtto the nestin one trip. Part BEST,L. B. 1977. Nestlingbiology of the Field Spar- of the body, however, is enclosedin a chitin- row. Auk 94: 308-319. ous exoskeletonthat is lesseasily digested than BIeRMANN,G. C. 1980. Parentalcare, growth rates soft-bodied insects(Boprot et al. 1976). The ex- and pre-fledgingcondition of Yellow Warbler tent of the occurrenceof a gastricchitinolytic nestlingsin differentbrood sizes. Unpublished M.Sc. thesis,Winnipeg, Manitoba, Univ. Mani- system in birds is unknown (see Jeuniaux toba. 1961), but it has been found in most of the few BORROR,D. J., D. M. DELONG,& C. A. TRIPLEHORN. species studied (Ziswiler and Famer 1972). It 1976. An introductionto the study of insects, is probably reasonableto assumethat not all 4th ed. New York, Holt, Rinehart and Winston. of the proteinof a hard-bodiedinsect is digest- BUSBY,D. G., & S. G. SEALY.1979. Feeding ecology ible by nestlings. Thus, the difference in the of a populationof nestingYellow Warblers. Can. total digestible protein brought per half hour J. Zool. 57: 1670-1681. FRYDENDALL,M. J. 1967. Feeding ecologyand ter- to broodsof four and five 8-day-oldyoung is ritorial behaviorof the Yellow Warbler. Unpub- probably smaller than the difference between lished Ph.D. dissertation,Logan, Utah, Utah the values calculated in Table 4. State Univ. The similarity in growth ratesof the young GIBB,J., & M. M. BETIS.1963. Food and foodsupply in the three brood sizes (Fig. 3) indicatesthat of nestling tits (Paridae) in Brecklandpine. J. differencesin feedingrates and food itemsfed Anita. Ecol. 32: 489-533. do not affectthe growth of the young. Thus, GOoSSEN,J.P. 1978. Breedingbiology and repro- the larger number of insectsbrought by par- ductive success of the Yellow Warbler on the ents with four young may have been of lower Delta Beach Ridge, Manitoba. Unpublished M.Sc. thesis, Winnipeg, Manitoba, Univ. Man- quality, as suggested,or, alternatively, the itoba. young in broodsof four may have been over- HUSSELL,D. J. T. 1972. Factorsaffecting clutch size fed, in that they may not have been able to in Arctic passerines.Ecol. Monogr. 42: 317-364. assimilate all the nutrients from their food. JEUN•AUX,C. 1961. Chitinase: an addition to the list Overfeedingin birds has never been reported, of hydrolasesin the digestive tract of verte- however, and would be contrary to expecta- brates. Nature 192: 135-136. tions of optimal parental behavior. JOHNSON,E. J., L. g. BEST, & P. A. HEAGY. 1980. April1982] YellowWarbler Parental Feeding 341

Food sampling biases associatedwith the "Lig- the Prairie Warbler Dendroica discolor. Ornithol. ature method." Condor 82: 186-192. Monogr. 26: 1-595. KEAR,J. 1972. Feeding habits of birds. Pp. 471-504 Pt•KOWSKI,B.C. 1978. Feeding of nestling and in Biology of nutrition (R. N. T. W. Fiennes, fledgling Eastern Bluebirds. Wilson Bull. 90: Ed.). Oxford, Pergamon. 84-98. KENDEIGH,S.C. 1947. Bird population studies in RICKLEts,R. E. 1967. A graphicalmethod of fitting the coniferousforest biome during a sprucebud- equationsto growthcurves. Ecology 48: 978-983. worm outbreak.Biol. Bull. Dept. LandsForests 1: ROYAMA,T. 1966. Factorsgoverning feeding rate, 1-100. food requirement and brood size of nestling LACK,D., & E. LACK. 1951. The breeding biology of Great Tits Parusmajor. Ibis 108:313-347. the swift Apusapus. Ibis 93: 501-546. S•^L¾,S. G. 1980.Reproductive responses of North- , & E. T. SILVA. 1949. The weight of nestling em Orioles to a changingfood supply. Can. J. robins. Ibis 91: 64-88. Zool. 58: 221-227. MAcKEnzIE, D. I. 1979. Nest site selection and co- S•L, D. C. 1969. Food, feeding ratesand body tem- existence in Eastern and Western Kingbirds at perature in the nestling House Sparrow Passer Delta Marsh, Manitoba. Unpublished M.Sc. the- domesticus at Oxford. Ibis 111: 36-47. sis, Winnipeg, Manitoba, Univ. Manitoba. SoK^L, R. R., & F. J. ROHLF. 1969. Biometry. San MOREAU, R. E. 1947. Relations between number in Francisco, W. H. Freeman and Co. brood, feeding rate and nestling period in 9 WALSH,H. 1978. Food of nestling Purple Martins. species of birds in Tanganyika Territory. J. Wilson Bull. 90: 248-260. Anita. Ecol. 16: 205-209. ZISWILER,V., & D. S. FARNER.1972. Digestion and MOREHOUSE,E. L., & R. BREWER.1968. Feeding of the digestivesystem. Pp. 343-430in Avian biol- nestling and fledgling Eastern Kingbirds. Auk ogy, vol. 2 (D. S. Farnerand J. R. King, Eds.). 85: 44-54. New York, Academic Press. NOLAN, V., JR. 1978. The ecologyand behavior of SEASONAL VARIATION AND FUNCTION OF MOBBING AND RELATED ANTIPREDATOR BEHAVIORS OF THE AMERICAN ROBIN (TURD US MIGRATORIUS)

DOUGLAS H. SHEDD 1 Sectionof Ecologyand Systematics,Cornell University, Ithaca, New York 14850 USA

ABSTR•CT.--AmericanRobins (Turdusmigratorius) displayed four classesof approachto a ScreechOwl (Otus asio):silent approach,vocal approach,mobbing, and attacking.Mob- bing and attackingwere confinedto the breeding season.Mobbing first appearedin March and reacheda maximum duration per mobbing bird in May and a maximum frequency in June. Mobbing and attacking were most often performed by pairs of robins or by single birds. These responseswere directedat a ScreechOwl locatedon the territory of the pair or individual. Silent approachand vocal approachoccurred at all times of the year. During the breeding season,birds approacheda ScreechOwl locatedon a territory other than their own, observedthe owl being mobbed, but joined in mobbing only if the owl flew. Outside the breeding season,robins gatheredin large mobile flocks. Birds in suchflocks sometimes approached a ScreechOwl but did not mob. I suggestthat the contextsin which these responsesoccurred indicate that mobbing and attackingare beneficialfor robinsonly when birds areconfined to territoriesor homeranges. The presenceof young on territoriesfurther increasesthe amount of parental mobbing. Received20 August1980, accepted4 November1981.

MANY speciesof birds, in diverseorders, are ators during the breeding season, possibly known to exhibit mobbing behavior. The even serving mainly as a nest-defensemecha- widespread occurrence of this behavior is nism. somewhatsurprising, for it often involves an The objectivesof this study were (1) to test animal seeking out a potentially dangerous for a breeding season-mobbingassociation in predatorand performingconspicuous displays the AmericanRobin (Turdusrnigratorius), and near it. That mobbing may be risky is con- (2) to investigatethe adaptive significanceof firmed by reportsof mobbingbirds being cap- this association, if it were found. tured by predators (Hartley 1950, Cade 1967, METHODS Denson 1979).As a result, the adaptive signif- icance of mobbing has remained largely an From 1 June 1974 through 31 August 1976, field enigma, although many hypotheseshave been trials were performedon a regularweekly basis.The advanced(Marler 1956;Humphries and Driver procedureduring these trials was planned to simu- 1967; Curio 1975, 1978; Cully and Ligon 1976). late a natural mobbing episode.A mounted Screech Owl (Otus asio)was placed in a conspicuouslocation Certain aspectsof the behavior, however, do 2 m from the ground. A recordedScreech Owl call suggestpossible functions. An associationbe- was then played by means of a cassettetape recorder tween the breeding season and an increased with the speaker located 10 cm from the mounted tendency to mob has been reported in a num- specimen. The taped calls were played on either a ber of species(Bolles 1890, Altmann 1956, Cu- Norelco Tape CassetteCarry-Corder "150" (Model rio 1975). In some casesactual possessionof a E13302P)or a PanasonicAC/Battery Portable Cassette nest is necessaryfor mobbing to occur (Horn Recorder RQ-209AS. All trials were observed from 1968).This relationshipbetween mobbing and concealedpositions 10 m from the mount. reproduction indicates that mobbing may be Comparative tests showed no significantdiffer- a particularly advantageous response to pred- encebetween the responseto a stuffedspecimen and taped call and the responseto a captivelive owl and taped call. Observationsof birds mobbingwild owls further confirmedthe adequacyof the test situation. • Present address:Department of Biology, Ran- Using a similar experimentaldesign, Curio (1975) dolph-MaconWoman's College, .Lynchburg, Virgin- found no difference between the responsedirected ia 24503 USA. at stuffed specimensand live predators.

342 The Auk 99: 342-346.April 1982 April1982] SeasonalVariation in MobbingBehavior 343

TABLE1. Percentageof successfultrials, averageminutes of mobbing/bird (AMM/B), and averagenumber of birds/successfultrial for eachmonth (Ithaca,New York). Robinswere presenton studysites in signif- icant numbersfrom 1 March through15 November;a few individualswere presentthroughout the winter.

Average Percentage number of Successful of successful birds/success- Month Total trials trials Total birds trials AMM/B ful trial

January 35 0 0 0.0 0.0 0.0 February 18 0 0 0.0 0.0 0.0 March 37 2 3 5.4 1.0 1.5 April 59 7 11 11.9 3.2 1.6 May 103 24 36 23.3 5.9 1.5 June 134 44 67 32.8 5.0 1.5 July 151 36 65 23.8 4.1 1.8 August 172 15 15 8.7 2.7 1.0 September 91 0 0 0.0 0.0 0.0 October 95 0 0 0.0 0.0 0.0 November 54 0 0 0.0 0.0 0.0 December 29 0 0 0.0 0.0 0.0

Data for many variableswere collectedduring each (Sabalpalmetto) thickets. Robins were abundant in trial; variables that proved useful include the num- both localities, drawn in part by the fruiting pal- ber of individualsmobbing, the presenceor absence metto. Where flockswere present, densities exceed- of vocalizationsand displays,the number of attacks ed 1,000 birds/ha. Screech Owls were found at both on the predator mount, and the total minutes of Florida study sites. mobbing performed by each bird. Mobbing behavior is only one phasein a contin- Data from the study period were totaled and then uum of behavioralresponses to a predatorthat range treatedas a single12-month series. For eachmonth, from fleeingto attacking(Simmons 1952). Although the averagenumber of minutesof mobbing per bird responsesintergrade continuously, it is usefulto cat- (AMM/B) was calculatedby summingthe minutesof egorizethem for purposesof data collection.In this mobbingand dividing by the total numberof mob- study positive responses (i.e. responses involving bing birds. Also calculatedfor eachmonth were the movement toward the predator) were divided into percentageof successfultrials (trials that resultedin four classes: mobbing)and the averagenumber of mobbingbirds Silentapproach .--In many trialsin which either the per successfultrial. mounted owl or a captive owl was used, birds were Means for each month were tested for statistical observedlocating the owl and then silently observ- probability of equality using a singleclassification ing it for severalminutes without vocalizing.Even- ANOVA, with meansweighted accordingto the re- tually, these birds flew off, usually leaving the vicin- ciprocal of the variance of the samplesfrom which ity of the owl. they were drawn. This adjustmentresults in an ap- Vocal approach.--Birds vocally approaching a proximate test of equality of means when variances predator gave alarm calls typical of mobbing birds are heterogeneous(Snedecor 1956, Sokal and Rohlf but remained several meters away and exhibited 1969). few, if any, of the visualdisplays typical of mobbing. The fieldwork for this research was carried out in Suchbirds seemedto experiencelittle risk of capture. the vicinity of Ithaca, New York. During December Mobbing.--Mobbingbirds approacheda predator and January1976, trials were also performed at var- closely (within 1 m) and performed a combination ious locationsin Florida. The New York study sites of visual displays and vocalizations. Visual displays consisted mostly of deciduous woodland where typical of mobbing consistof intention movements beech (Fagusgrandifolia) and sugarmaple (Acer sac- to approachor flee, such as wing-flicking, tail-wag- charum) predominate. Floodplains dominated by ging, and nervous hopping. Mobbing vocalizations Fraxinusspecies and large areasof old field habitat are loud and high-pitched. Generally calls are of were also sampled. Robins were common in all of 0.1 s or less in duration and of a narrow range of these areasfrom early March until late October, with frequencies;this combinationpresumably makes the breeding densitiesaveraging 5-8 pairs/ha. callseasy to locate(Marler 1959).Both visual displays Testing in Florida was done at Merritt Island Na- and vocalizations are frequently repeated during tional Wildlife Refuge and on Sanibel Island. Both mobbing. areascontained extensive regions of coastalpalmetto Attacking.--Mobbing birds sometimes suddenly 344 DOUGLASH. SHEDD [Auk,Vol. 99

TABLE2. Attacksby Robins at study sitesin Ithaca, TABLE3. Averagenumber of Robins/trialdisplaying New York. Attacks occurred only from April silent or vocal approach(SVA/ST) for each month. through August. Totaltrials: TT; trialswith silentor vocalapproach: ST. Average number Immature- Trials of attacks/ Adult-plumaged plumaged Total with Total trials with birds birds Month trials attacks attacks attacks Month TT ST SVA/ST ST SVA/ST April 59 3 5 1.7 May 103 6 81 13.5 New York June 134 7 14 2.0 January 35 0 0.0 July 151 5 8 1.6 February 18 0 0.0 August 172 6 12 2.0 March 37 7 2.9 April 59 12 1.8 May 103 17 1.5 June 134 18 1.9 2 1.0 July 151 32 3.1 35 3.5 ceasedmobbing and pressedan attack directly on August 172 40 3.4 45 8.0 the predator.Such attacks consisted of shortdashing September 91 12 3.6 12 5.3 October 95 24 8.3 flights, culminatingin strikeson the predator,usu- November 54 1 1.0 ally about the head or neck. December 29 0 0.0

Florida RESULTS December 57 15 17.9 January 35 11 35.2 Robins were present in Ithaca in significant numbers from early March until middle No- vember (with isolated individuals lingering through the winter), but mobbing was con- corporatedinto the generalmobbing activities fined to the sampling period extendingfrom of the birds and consistedof swoopingflights March throughAugust (with a singleexception towardthe mountedowl (usuallydirectly at the noted below). Monthly averagesfor the AMM/ head),followed by an abruptretreat. Only one B index rose during this time from a minimum attackwas observedthat was not precededby of 1.0 min of mobbing/birdin March to a max- preliminary displays of mobbing. Robins imum of 5.9 in May and then declined to 2.7 mobbing wild owls attackedthem in a similar in August (P < 0.05, F = 14.759).The average fashion. amount of mobbing performed by each bird The behaviorsof silent and vocalapproach increased with the advance of the breeding were more common in Ithaca outside the seasonuntil May and then declined(Table 1). breedingseason (Table 3). At theFlorida study The percentageof trials that was successful sites,large numbers of robinswere exposedto in producingmobbing amongrobins rose from the owl mount and taped call. None of these a low of 5.4% in March to a peak in June of birds was observedto mob, althoughsome 32.8%. The percentagethen declined to 8.7% displayedsilent or vocalapproach. in August (Table 1). Silentapproach was typicalof immaturerob- During the 6 monthsin which mobbing was ins, which were first observedresponding in observed, there was no significantchange in June(Table 3). Thesebirds alwaysarrived after the number of birds respondingduring each adult birds were mobbingand may have been successfultrial (Table 1). Trialsin Julyaveraged attractedby the adults' alarm calls rather than 1.8 birds/successfultrial for a maximum, while by the owl call.By Julymany immatures were in August a minimum of 1.0 was reached attracted,and mostcontinued to displaysilent (P < 0.05, F = 0.439). approach. In August the number of immature Other responsesthat were closelyrelated to robins attracted outnumbered adults. At this mobbing also varied seasonally.Robins were time immaturesresponded directly to the owl observedattacking the owl mount only during call and displayedvocal approach.Immature the breeding season(Table 2). The number of robinswere observedto mob on only two oc- attacks was highest in May. Attacks were in- casions(22 July 1974, 11 August 1976). April 1982] SeasonalVariation in MobbingBehavior 345

DISCUSSION of developmentduring the breeding season similar to that observed in antipredator behav- Seasonalvariation in mobbing.--Severalstud- iors such as distraction displays and reactivity ies of mobbing behavior have noted an asso- to distress calls (Stefanski and Falls 1972, Barash ciation between this behavior and reproduc- 1975). tive activity (Edwards et al. 1949, Root 1969, The manner in which mobbing may function Curio 1975). In the present study, a peak in to reduce predation remains open to debate. mobbing activity occurred in May and June Becausemobbing by robins is sometimesin- (Table 1), markedly overlappingwith the in- terspersedwith direct attackson the predator terval of 2 May through 23 Junewhen 70% of (Table 2), and becausemobbing displays are the hatching and fledging of young could be often virtually identical to displaysused dur- expected (Howell 1942). The willingness of ing territorial encounterswith conspecifics,it robins to mob during the breeding season is likely that mobbing functionsin this species probably stems from at least two factors: (1) to drive predatorsaway. Certainly the advan- birds are territorial at this time and lack the tage inherent in driving a ScreechOwl away option of easy relocationaway from predators from a territory is evident from a consideration on their territories, and (2) birds are likely to of the owl's diet, which during the spring may have eggs or young on their territories. consist predominately of small and medium- Restrictionto a smallhome range or territory sized birds (Allen 1924, Stewart 1969, Van may be an important factor influencing the Camp and Henny 1975). performanceof mobbing. The single instance Silentand vocal approaches.--Silent and vocal of postreproductivemobbing by robins ob- approaches seem to occur in situations in served in this study occurredon one of the which the risk of mobbing is greaterthan any New York study sites in January.Two robins, benefits that might result. Silent approach is confinedto a sumac(Rhus typhina) thicket by a low-risk behavior allowing birds to locate heavy snows, mobbed vigorously when ex- and then avoid a predator.Vocal approach may posed to the mounted owl and taped call. This discourage a predator from further hunting was in marked contrastto the responsetypical without the risk of the closeapproach necessary of robins outsidethe breeding season(see be- for mobbing (see "Perception Advertisement" low). in Curio 1978). The relationship between mobbing and ter- Silent and vocal approachesappear in two ritoriality is further demonstratedby a consid- contexts: (1) Breeding birds show these re- eration of Table 1, which indicates that, al- sponseswhen locating predatorsnot on their though the percentage of trials successfulin own territories. Observationsof the mobbing stimulating mobbing increased greatly from of live owls revealed that birds from distant March until June, the number of birds mob- territories were often attracted to a mobbing bing during a successfultrial did not vary sig- pair but joined in mobbing only if the owl was nificantlyduring this period, alwaysbeing less forced into flight by the mobbers. A similar than an average of two birds. The territorial observationhas been made of coloniallynest- distribution of the mobbing birds made it ap- ing gulls by Kruuk (1976). (2) Birds show these parent that they were individuals maintaining responseswhen they abandon territoriesat the territories where the Screech Owl was located. closeof the breeding season.All of the birds Just as maintenance of a territory seems to attracted to the mounted owl and taped call have the effect of increasing the intensity of from Septemberuntil November in New York mobbing, a breeding bird's geneticinvestment were such birds (with the single exception of in young is likely to have a similar effect. The the mobbing pair noted above). Similar re- risk entailed by a mobbing bird would be sponseswere obtained in Florida during De- counterbalanced by an increase in fitness cember and January. The factor most respon- throughkin selectionif mobbingreduced mor- sible for the cessationof mobbing at the end tality in that bird's young. If this were true, of the breeding seasonand the substitution of mobbing intensity would increaseas the egg- silent or vocal approachesappears to be the nest unit grew older and parental investment movement of territorial birds into highly mo- increased.This was observedto occurby How- bile nonbreedingflocks. Robins in suchflocks ell (1942)and is suggestedby the AMM/B index would seemto have little to gain by mobbing; (Table 1). Mobbing seemsto follow a pattern it is presumablymore adaptivefor them to lo- 346 DOUGLASH. SUEDD [Auk, Vol. 99 cate a predator and then avoid further contact CurLY, J. F., & J. D. LIGON. 1976. Comparativemob- with it. bing behavior of Scrub and Mexican jays. Auk The numeroustheories currentlybeing pro- 93: 116-125ß posedas to the function of mobbing have been CURIO,E. 1975. The functional organizationof anti- synthesizedby Curio (1978)into nine hypoth- predator behaviour in the Pied Flycatcher: a study of avian visual perceptionßAnim. Behav. eses with resultant predictions of prey and 23: 1-115. predator behavior. Of these hypotheses,some ß 1978. The adaptive significance of avian variationof Curio's"Moving on" modelseems mobbingßI. Teleonomichypotheses and predic- most consistentwith my resultsßOther factors, tions. Z. Tierpsychol.47: 175-183ß however, may be operating concurrently.For ß U. ERNST,& W. VIETH. 1978. The adaptive example, Curio et al. (1978) suggestthe pos- significance of avian mobbing. II. Cultural sibility of cultural transmissionof information transmission of enemy recognition in black- about predators. Such an hypothesis is sup- birds: effectiveness and some constraints. Z. ported by observationsin this studyindicating Tierpsychol.48: 184-202. that immature robins respondinitially to calls DENSON,R. D. 1979. Owl predation on a mobbing' Crow. Wilson Bull. 91: 133. of mobbing adults and closelyobserve these EDWARDS, G., E. HOSKING, & S. SMITH. 1949. Re- adultsßBecause many factorsmay be operating actionsof somepasserine birds to a stuffed Cuck- to causemobbing, it is usefulto keep in mind oo. Brit. Birds 42: 13-19. that many of the proposedhypotheses are not HARTLEY,P. H. T. 1950. An experimental analysis mutually exclusive. Also, given the frequent of interspecificrecognition. Symp. Soc. Exp. occurrenceof mobbing among groups of un- Biol. 4: 313-336. related birds (and mammals), it would be un- HORN, H. S. 1968. The adaptive significanceof co- realisticto hope for an explanationof the func- lonial nesting in Brewer'sBlackbirds (Euphagus tion of mobbingthat would apply to all species cyanocephalus).Ecology 49: 682494. in all situationsß HOwEkE,J. C. 1942. Notes on the nesting habits of the American Robin. Amer. Midl. Natur. 28: ACKNOWrEDGMENTS 529403. HUMPHRIES, D. A., & P.M. DRIVER. 1967ß Erratic This paper is based on a portion of a PhßDßdis- displays as a device against predators. Science sertation submitted to Cornell University. Financial 156: 1767-1768. supportfor this researchwas provided by the Frank KRUUK,H. 1976. The biological function of gulls' M. Chapman Fundß the Section of Ecology and Sys- attraction toward predators. Anim. Behav. 24: tematicsßthe Peregrine Fund, and a Randolph-Ma- 146-153. con Woman's College Humanities Grant. I owe a MARKER, P. 1956. Behaviour of the Chaffinchß Frin- specialdebt of gratitude to T. J. Cade for suggesting gilla coelebs.Behav. Suppl. 5: 1-184. this researchtopic and for providing equipment, ad- ß 1959. Developments in the study of animal vice, financial supportßand editorial assistanceßP. communicationßPp. 150-206in Darwin's biolog- Marks and D. Q. Thompson were also very helpful ical work: some aspectsreconsidered (P. R. Bell, in many aspectsof my research. J. A. Wiens, T. Ed.). Cambridge,Cambridge Univ. Pressß Michalik, W. A. Coulterßand several anonymous re- RooT, R. 1969. The behavior and reproductive suc- viewers made many valuable commentsßAbove all, cess of the Blue-gray Gnatcatcher.Condor 71: I thank my wife Barbara for her unceasingencour- 16-31. agement and supportß SIMMONS,K. E. L. 1952. The nature of predator re- actionsin breeding birds. Behaviour4: 161-171. LITERATURE CITED SNEDECOR,G. W. 1956. Statistical methodsß Ames, ArrEN, A. A. 1924. A contribution to the life history Iowa, Iowa State College Pressß and economic status of the Screech Owl. Auk 41: So•

NICOLAAS A.M. VERBELK Departmentof Biology,Simon Fraser University, Burnaby, British Columbia V5A 1S6,Canada

ABSTRACT.--Predationby NorthwesternCrows (Corvus caurinus) on the eggsof Double- crested Cormorants(Phalacrocorax auritus), Pelagic Cormorants(P. pelagicus),Glaucous- wingedGulls (Larus glaucescens), and PigeonGuillemots (Cepphus columba) was studiedin the summersof 1976-1980.Crows took eggsto specificsites on their territorieswhere they ate them and where we counted the shells. In 1980, the crows took 1,167 cormoranteggs, an estimated22% of all eggsin first clutches.When Bald Eagles(Haliaeetus leucocephalus) were present,more cormoranteggs were lost to crows than on days when eagleswere absent.More eagleswere presenton weekendsthan on weekdays,presumably because the localeagle population was kept inadvertentlyon the moveby peopleseeking recreation on the water. Received10 November1980, accepted I September1981.

CROWS(Corvus spp.) are notoriouspredators them in places to which the crows had no access on the eggsand young of other birds. This pa- (1976-1979)or took them away in a plasticbag (1980). per documentsthe predationof an island pop- The crow territory on which the eggs were found ulation of Northwestern Crows (Corvus cauri- was recorded. In 1978 and 1980 we distinguished nus)on the eggsof Double-crestedCormorants betweenfresh eggs (taken on the day on which they were found) and old eggs(taken the previousday or (Phalacrocoraxauritus), Pelagic Cormorants (P. earlier). We did not attempt to distinguish between pelagicus),Glaucous-winged Gulls (Larusglau- the eggsof the two speciesof cormorant,because the cescens),and Pigeon Guillemots(Cepphus co- eggsare similarlycolored and shapedand overlapin lumba).The study was conductedon Mandarte size [Double-crestedCormorant: length 62.65 + 1.85 Island (48ø38'N, 123ø17'W), British Columbia. mm, width 39.27 + 1.16, n = 44 (Verbeek MS); Pe- The cliffs on the southwest side of the island lagic Cormorant: length 58.85 4- 3.83 mm, width rise abruptly to a height of 10-30 m, and the 37.37 4- 1.23, n = 20 (Palmer 1976)], and many of the island slopesdownward to a height of 1-5 m eggswere too fragmentedto measure.The presence along the northeastshore (Fig. 1). PelagicCor- or absenceof Bald Eagleswas recordeddaily in all morants nest singly or in small, scattered 5 yr. In 1980we alsorecorded how many timeseagles flew over the island each day and whether or not groupson rocky ledgesalong the steepcliffs, they disturbedthe gulls and the cormorants.The re- while Double-crestedCormorants nest in tight searcherson Mandarte Island carefully avoided dis- colonies at the top of the cliffs. Gulls nest in turbing the cormorantsthroughout this study. the grasson either side of a strip of shrubsand trees that grow along the long, centralaxis of RESULTS the island, and the guillemotsnest in burrows under rocks and in crevices in the meadows Althoughthe total numberof cormoranteggs and the cliffs, respectively.The crows nest in found varied among years, most were found the shrubs and trees (Fig. 1). in June each year (Table 1). Similarly, eggsof Glaucous-wingedGulls were found mostly in METHODS June (Table 2). The crows took few eggsof Pi- geon Guillemots, presumably because their Starting about 1 h before sunset,my colleagues nests were concealed and difficult to enter. We and I searchedthe island daily in May, June, and may have misidentified some fragmentedPi- July1978, 1979,1980 and about three times per week in 1976 and 1977 for seabird eggs that had been geon Guillemot eggsas gull eggs. preyedupon. In most caseswe had no difficultyin A crow transportsa whole cormorantegg in decidinghow many whole eggswere representedby its bill (lengthwise)and takes it to a secluded the partial shells and fragments found. When in spot on its territory to be stored or eaten (see doubt, we guessedconservatively. We removed all also Montevecchi1976). All storagesites were the eggshellsand shell fragments and 'either cached located within or at the outer edge of the

347 The Auk 99: 347-352. April 1982 348 NICOLAASA.M. VœRBœœ•C [Auk, Vol. 99

.'-..::.:•ß SH R U B S 100m •GRASS •k,N ' ' I•JCLIFF Fig. 1. MandarteIsland. The blackdots represent the locationof crownests in 1980.

shrubsand trees.A pair of crowstypically used on neighboring islands, and other nonbreed- the same site(s) throughout the season and ing eaglesvisited Mandarte Island. Bald Eagles usually the same sites were used each year. prey on the adult gulls (pers. obs.) and their Sites were variable in nature, but all had some chicks (Hayward et al. 1977, pers. obs.), and object that offered visual concealment from they have been reported to take Pelagic Cor- above. Eggswere hidden under a shrub, under morants as well (Campbell 1969). a fallen tree trunk, among exposedtree roots, At the arrival of a Bald Eagle, gulls were the at the edge of small earth banks, under over- first birds to respond. If the eagleflew low over hanging rocks,and inside an unused shed. The the island, the gulls would fly up, and, as the crowsnever took eggsto sitessuch as an open eagle continued to approachthe birds, a wave meadow where they would be in full view. A of anxiety spread down the length of the is- pair of crowsthus couldaccumulate many eggs land. At the first indication of an approaching at the same site(s) in the course of a summer. eagle, the crowsappeared in the topsof bushes It appearedthat each pair of crows (and pre- and trees. As the excitement increased, the sumably each individual) had its own consis- crows flew high into the air, which is a typical, tent way of opening an egg; somemade a hole direct responseof the genus Corvusto poten- in the side of the egg and ate the contents tially dangerousavian predators.It gave them through it; others fragmented the shell. Al- the direct advantageof being above the pred- though some of the opened eggs contained ator and the indirect advantageof an overview some albumen when we found them, none of the island. Within seconds, crows from all containedyolk, and we rarely found cormorant over the island converged on the cormorant eggs in advanced stagesof incubation (blood cliffs, where they soaredin updrafts over the inside the shell). nests, dropping down swiftly to take an egg or Cormorant eggs and young were taken from young when the opportunity arose. We have the nest by crows either during nest change- seenincubating, female crowsleave their nests overs, when one incubating parent left the nest during suchdisturbances to take an egg quick- to be replacedby its partner (Drent et al. 1964, ly, eat it on the territory, and return to incu- pers. obs.) or during disturbanceswhen cor- bate. During some disturbances,such as on 28 morants left their nests. Disturbances often oc- May 1976, we saw a crow bring an egg to its curred when adult Bald Eagles, which nested usual hiding place, deposit it there, then head

TABLE1. The number of cormoranteggs found in NorthwesternCrow territoriesby month and year.

1976 1977 1978 1979 1980

Month Number % Number % Number % Number % Number %

May 102 35 33 12 154 37 139 10 207 18 June 170 58 203 76 176 43 827 60 533 46 July 22 7 31 12 82 20 422 30 427 37 Tot• 294 267 412 1,388 1,167 April 1982] EggPredation by NorthwesternCrows 349

TABLE 2. The number of Glaucous-winged Gull TABLE4. Frequencyof freshly taken cormoranteggs (GWG) and Pigeon Guillemot (PG) eggsfound in found on weekends(Friday to Sunday)and week- Northwestern Crow territories in 1978, 1979, and days (Monday to Thursday)? 1980. a Weekdays 1978 1979 1980 Weekends (3 days) (4 days)

Time GWG PG GWG PG GWG PG Number Per- Number Per- Year of eggs centage of eggs centage 15-31May 2 0 0 0 3 0 1-15 June 15 I 14 0 22 1 1978 194 b 56 152 44 16-30 June 26 2 28 I 40 0 1980 525 59 373 41 1-21 July 25 0 25 1 32 0 a Only in 1978 and 1980 did we distinguish daily between freshly Comparable data were not collectedin 1976 and 1977. taken cormoranteggs and thosetaken the previousday or earlierbut not found by us. •' In both years significantlymore cormoranteggs were taken by crowson weekendsthan on weekdays(Chi squaretest, P < 0.001). back immediately to get another one. After a single prolonged eagle disturbanceon 21 June 1980, we found 17 recently laid eggs on one in 1980 eagleswere present at about equal fre- crow territory. There is little doubt that the quency throughout the season(Table 3). Visits eggs were brought in by a pair of crows, one of the eagles to the island were apparently after the other and eaten later when the dis- more common during this study than they turbance was over. Some uneaten eggs were were in 1959 and 1960 (Drent et al. 1964). cached. When such eggs were left by us, they Another source of disturbance was visits of were eaten invariably within the next 2 days. small boats (sailing vessels,fishing boats, and Not all Bald Eagles that visited the island sightseers)too close to the cormorant cliffs. In caused disturbances. Some merely flew over 1978 and 1980, when we distinguished daily the island en route elsewhere. Eagles, espe- between fresh eggs and old ones (see Meth- cially immatures,would rest on the island oc- ods), significantlymore fresh eggswere found casionally. They typically did this in the few on weekends (including Fridays, as many trees on the island, but sometimesthey rested small vessel operators appeared to start their near the top of the bare cliffs where the Dou- weekends early) than on the remaining 4 days ble-crested Cormorants nested. In the latter of the week (Table 4). The daily take of fresh case, the cormorants left their nests and con- cormorant eggs in 1980 is shown in Fig. 2. In tinued to fly back and forth alongthe cliffs un- addition, significantlymore eagle scaresalso til the eagles departed. During such distur- occurredon weekends than on weekdays, and bances, the crows had unhindered accessto the many more cormoranteggs were taken on days cormorant nests. The number of days in each with eaglescares than on days without scares, month on which eagleswere present on the regardlessof the time of the week (Table 5). In island over the 5 yr is shown in Table 3. In the absenceof eagle scares,no significant dif- 1976, 1977, and 1978, the percentageof days in ferencesexisted between the number of eggs each month that eagleswere present declined taken on weekendsand on weekdays(Table 5). from April to July, but in 1979 and especially The breeding populations of the two cor-

TABLE3. The number of days peoplespent on Mandarte Island eachmonth and the percentageof those days during which Bald Eagleswere present.

March April May June July Year Days % Days % Days % Days % Days % 1976 10 70 21 71 31 45 30 20 29 7 1977 4 75 19 75 31 55 30 47 29 41 1978 0 0 23 61 31 45 30 30 31 23 1979 6 67 29 62 31 65 30 70 22 23 1980 0 0 7 100 31 68 30 80 21 71 350 NICOLAASA.M. VERBEEK [Auk, Vol. 99

70'

w co

7

5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 MAY JUNE JULY Fig. 2. The numberof freshlytaken cormorant eggs found in 1980.The shadedbars represent weekends and the unshadedbars representweekdays as defined in Table 4.

morant species on Mandarte Island have in- mean clutch size for Pelagic Cormorants (3.83 creased since Drent et al. (1964) reported on eggs, n = 52) doesnot differ significantlyfrom them. In 1980 we counted 545 nests of Pelagic that of Double-crested Cormorants (3.94, n = Cormorants and 794 nests of Double-crested 109, Robertson 1971). Thus, the cormorants on Cormorants. These counts were made from a Mandarte Island potentially could have laid boat, and the count of Double-crested Cor- 5,200 eggs in 1980 (not counting replacement morant nests may be an underestimate. The clutches), of which the crows took 1,167, or

TABLE5. Relationshipbetween presence or absenceof Bald Eaglescares, time of the week, and the number of freshly taken cormoranteggs found in 1980.

Weekendsa Weekdays Number Number Eggs Number Number Eggs of days of scares taken of days of scares taken With scares 23 68 b 451 ½ With scares 24 46 270 Without scares d 13 0 74 e Without scares 22 0 103

Weekends as defined in Table 4. Significantlymore eagle scaresoccurred on weekendsthan on weekdays(Chi-square test, P < 0.05). Significantlymore eggswere taken on weekendsthan on weekdays,(Chi-square test, P < 0.001), in the presenceof eaglescares. Eaglesabsent all day or present but flew over high or otherwisedid not disturb the cormorants. No significantdifference exists between eggs taken on weekendsand weekdaysin the absenceof eaglescares (Chi-square test, P > 0.05). April 1982] EggPredation by Northwestern Crows 351

I N;70

;94

N=1224

N =115 I I

20 30 10 20 20 30 APRIL MAY JUNE Fig. 3. The mean and standard deviation of clutch commencementof four avian speciesnesting on Mandarte Island. Only the range is shown for Double-crestedCormorants, because insufficient data were available.The data for cormorantsand Pigeon Guillemotsare basedon Drent et al. (1964);those for crows and gulls are basedon this study. about 22%. About 2,200 pairs of Glaucous- ly May. Probably not more than 5-10 yearlings winged Gulls nest on the island, the same as and visitors were involved in 1980. The mean in 1962 (Vermeer 1963). The mean clutch size datesand rangesof egglaying of the crowsand was 2.74 (n = 1,027, Verbeek MS). The crows the other speciesare shown in Fig. 3. The first took few of the eggs (Table 2). Drent et al. cormorant eggs taken by crows on Mandarte (1964) estimated the breeding population of Island were found on 30 April 1977, 25 April Pigeon Guillemots at approximately100 pairs, 1978, 29 April 1979, and 28 April 1980. and they thought that the colony had reached maximum size. The mean clutch size was 1.91 DISCUSSION (n = 162, Drent et al. 1964), so that a total of 191 eggsin first clutcheswas laid. The number The Northwestern Crow on Mandarte Island of crowsnesting on MandaRe Island increased nests early, well ahead of the seabirds whose from 13 pairs in 1976 to 25 pairs in 1980. Be- eggs it eats (Fig. 3). We do not wish to imply, sidesthese breeding pairs, somenonbreeding however, that early nesting in the crow has yearlings (young of the previous year) were evolved in responseto the nesting of the other present, and some birds occasionallyvisited species,because the Northwestern Crow nests from neighboringislands. The number of year- early outside seabird colonies as well. Never- lings and visitorswas highest in April and ear- theless, on Mandarte Island, the crow has cor- 352 NICOLAASA.M. VERBEEOC [Auk, Vol. 99 morant eggs and those of other seabirdsavail- without eagle scares. In addition, more eggs able to it from late April to late July (Fig. 3). are lost on weekends (Table 4, Fig. 2) than on The crow is mainly a predator of cormorant weekdays(Table 5). It appearsthat the week- eggs, and I suspectmainly of PelagicCormo- end activityof peopleseeking recreation on the rants. The Pelagic Cormorant nests tend to be water and along the shores of islands sur- isolated from one another, in contrast to those rounding Mandarte Island inadvertently stirs of Double-crested Cormorants, which nest in up the eagle population. People, then, are di- tight groups.The latter's coloniesmay be more rectly (sailing too closeto the cormorantcol- difficult for a crow to enter, and perhaps the ony) and indirectly (stirringup the eagles)re- birds are less easily panicked. Additionally, sponsiblefor part of the egg loss suffered by Montevecchi (1976) has shown that crows pre- the cormorants to Northwestern Crows. fer smallereggs over larger ones. The apparent freshnessof the eggs that were taken, even late ACKNOWLEDGMENTS in the season, suggeststhat perhaps a large I thank JoanMorgan, Linda Graf, Loic Legendre, proportion of the eggs were obtained from the JamieSmith, and Alan Stuart for their help. Financial samevictims, which may have lost first as well support was provided by the Natural Sciencesand as possiblereplacement clutches. It is possible EngineeringResearch Council of Canada.The Tsey- that the cormorantsflush easier from newly cum and Tsawout Indian Bands gave us accessto laid eggs than from those that have been in- their island. I appreciatethe helpful commentsI re- cubated for some time. ceived from M. A. Howe, W. A. Montevecchi, C. D. Very few eggs of Pigeon Guillemots were Orr, and H. Richardson. taken (Table 2), as was shown also by Drent et LITERATURE CITED al. (1964). Glaucous-wingedGulls also lost few eggs to Northwestern Crows. When an eagle CAMPBELL,R. W. 1969. Bald Eagle swimming in caused the gulls to leave their nests temporar- oceanwith prey. Auk 86: 561. ily, the crowson the island did not fly to the DRENT, R. H., G. F. VAN TETS, F. TOMPA, & K. VER- MEER.1964. The breeding birds of Mandarte Is- gull nests but to those of the cormorants.Gull land, British Columbia. Can. Field-Natur. 78: eggs are not available otherwise, as the eggs 208-263. are coveredalmost constantly.Most gull eggs HAYWARD, J. L., JR., G. W. HUMPHREY,C. J. AM- that were taken by the crows were probably LANER,& J. F. STOUT.1977. Predation on gulls those that were abandoned after the others in by Bald Eaglesin Washington.Auk 94: 375. the clutch had hatched. Table 2 shows that MONTEVECCHI,W. A. 1976. Egg size and the egg most of the gull eggs appearedon crow terri- predatory behaviour of crows. Behaviour 57: tories in the secondhalf of June and into July, 307-320. and we noticed that these eggs consistently PALMER,R. S. 1976. Handbook of North American birds. Vol. I. New Haven, Connecticut, Yale smelledputrid, indicating that they had been Univ. Press. spoiled for some time. I believe that most of ROBERTSON,I. 1971. The influence of brood-size on the egg loss suffered by Glaucous-winged reproductivesuccess in two speciesof cormo- Gulls is due to intraspecific predation. rant, Phalacrocoraxauritus and P. pelagicus,and The increased predation of cormorant eggs its relation to the problem of clutch size. Un- in 1979 and 1980 (Table 1) is undoubtedlydue published M.Sc. thesis, Vancouver,British Co- in part to the increasein the crow population lumbia, Univ. British Columbia. on the island, but more important, to the con- VERMEER,K. 1963. The breeding ecology of the tinued, frequent visits of eagles in June and Glaucous-wingedGull (Larusglaucescens) on July (Table 3). It is clearly shown in Table 5 Mandarte Island, B.C. British Columbia Provin- that cormorants lose more eggs to crow pre- cial Mus., Occ. Pap. 13. dation on days with eagle scaresthan on days TERRITORY SIZE IN MEGACERYLE ALCYON ALONG A STREAM HABITAT

WILLIAM JAMESDAVIS 1 Departmentof Biology,University of Cincinnati, Cincinnati, Ohio 45221 USA

ASSTRACT.--BeltedKingfishers (Megaceryle alcyon) defend both breedingand nonbreed- ing territories.The smallnonbreeding territories serve as feedingterritories, and their size is inverselycorrelated to food abundance.In the breedingseason, when nest sitesare a limitingresource, territory size doesnot significantlycorrelate with foodabundance. The averageweight of fledglings/nest,however, is significantlycorrelated with territorysize. In both seasons,kingfishers appear to use streamriffles as preferredforaging sites, although poolsare more prevalent.Riffles may representenvironmental cues by which individuals assesshabitat quality. Received19 February1981, accepted 30 September1981.

NUMEROUS studies have revealed correla- ritory defended [see Nice (1941)for classifica- tions between territory size and various envi- tion of territories]. For example, breeding ter- ronmental parameters. Most frequently, re- ritories of birds usually possess multiple source density (Pitelka et al. 1955, Stenger resources, such as nest sites and food, while 1958, Gill and Wolf 1975, Salomonson and Bal- non-breedingterritories are often only "feed- da 1977) and/or population density (Myers et ing territories" (Welty 1975); thus, in the al. 1979, 1980; Ewald et al. 1980) are cited, but, breeding season,the number and distribution as is now becoming evident, the proximate of available nest sites may influence the suit- causal mechanismsresponsible for these cor- ability of a habitat (von Haartman 1959). If relationsare not always obvious.For example, nest-site limitations restrict the total number although a negative correlationbetween terri- of breeding birds, territory sizesmay become tory size and food densitymay imply that an less dependent upon the availability of other animal has the ability to assessresource den- resources such as food. sity in relationto its needs,it may actuallyonly Belted Kingfishers(Megaceryle alcyon) are reflect the increased population pressure in overtly aggressive birds that, along small areasthat attractmany animalsbecause of high streams, defend territories with well-defined food density. Alternatively, negative correla- boundaries and maintain exclusive use of the tions might resultfrom an animal'srecognition resources on their territories. Although the of habitat factors that are directly associated strong territorial nature of this bird is well with food (Hild•n 1965, Morse 1976) rather known to naturalists,I am unaware of any pre- than its assessment of food itself. Seastedt and vious empirical studies reporting analysesof Maclean (1979) have shown that the size of the territoriality of this species.During the breeding territories of Lapland Longspurs breeding season(March-June), both the male (Calcariuslapponicus) is inversely related to the and female of a breeding pair jointly defend "expected" food density of the habitat com- the territory on which their nest is located. prised by the territory rather than to the actual Nest sites are chosen before the boundaries of food density in a given year. They conclude breedingterritories are firmly established(Da- that longspursrespond to the habitat compo- vis 1980). In the autumn, birds of both sexes, sition of an area rather than to the food den- including the young of the year, defend indi- sity. vidual territories. The present study investi- The relative importanceof different environ- gatespossible determinants of territorysize in mental parametersas determinantsof territory the BeltedKingfisher in both the breedingand size may be dependentupon the type of ter- nonbreeding seasons.

STUDY AREA AND METHODS Presentaddress: Department of Zoology,Uni- Kingfisherswere observed along a 16.8-kmstretch versityof Texasat Austin,Austin, Texas 78712 USA. of IndianCreek in ruralsouthwestern Ohio. During 353 The Auk 99: 353-362. April 1982 354 WILLIAMJAMES DAVIS [Auk, Vol. 99 the study period, Indian Creek had an averageflow 1965). Prey were classified to family, genus, or rateof 1.9 + 0.4 ma/s,although runoff produced from speciesby noting the morphologicalcharacteristics heavy rains occasionallyincreased the volume of of the respectivetaxonomic groups. Cyprinids were flow (at times by as much as 10x). Creek width and classified into one of three groups: stonerollers depth varied from 5 to 19 m and 0.05 to 1.8 m, re- (Campostomaanomalum); minnows, which included spectively,at normal flow. Turbidity was usuallylow all identifiable cyprinid speciesother than C. an- exceptafter heavy rains. Substratevaried; looserock omalum,or unidentified cyprinids, which included was predominant in riffles, while gravel and sand all specimensthat could not be identified to genus. were the usual substrateof deeper pools. Nesting The second method (1 June-23 June) involved plac- banks suitableto M. alcyonoccur regularly along In- ing velcro collarsaround the necksof nestling king- dian Creek where the water erodes soil from bor- fishers for a period of 2 h from 0700 to 0900. This dering banks. Secondary forest growth dominates procedureprevented the young birds from swallow- much of the creekline. Dominant tree species are ing food brought to them by the adults, and, as a Americansycamore (Plantanus occidentalis), sand bar result, fish accumulated at the entrance of the nest willow ($alix interior), several maple species (Acer chamber. Weight, length, and speciesof these fish spp.), and a variety of oaks (Quercusspp.). were recorded.Additional data on prey size and type The bulk of the datawas collectedbetween Ianuary were collectedas kingfisherswere observedfishing 1978 and October 1979. Kingfishers were banded or when items were dropped into the mist net during with U.S. Fish and Wildlife bands and marked with banding of adults. lead-free paint along the edge of the rectrices,using Estimates of food abundance were obtained for six a combination of four colors. Individual birds could nonbreeding territories (SeptembersOctober1978) easily be identified in the field within a 50-m range and for six breeding territories (June 1979) by sam- using binoculars. pling the entire length of each territory by electro- The territory size of marked individuals was mea- shocking. A shortage of manpower prevented fish sured as length (m) instead of as area, becauseter- sampling on any of the 21 nonbreedingterritories ritories followed the course of the stream. Territory measured in 1979, but in October 1979, 1,200 m of sizes were assessedby recording the movement of Indian Creek were electroshocked,with pools and marked individuals that were "herded" along their riffles sampledseparately, in order to documentthe territories (Wiens 1969). By walking along the relative locationof the different fish species.Electro- stream,I forcedthe residentkingfisher(s) to fly ahead shockingwas accomplishedby suspendingtwo elec- until the end of the territory was reached. At this trodes, positioned at opposite ends of 1.3-m pole, in point the kingfisher(s)would reversedirection to fly the water while moving upstream at a constant back into its territory. The locationsof the up- and speed. Current was supplied to the electrodesby a downstream boundaries were determined a mini- 110-volt AC (1,500 watt) generator. Shocked fish mum of five times by this method. The site of ag- were collectedin nets by two personsfollowing a gressive confrontation between adjacent territory third person carrying the electrodes.Fish collected holders was used to confirm boundaries between ter- were counted and classified accordingto size and ritories. Lateral boundaries of territories were de- speciesbefore they were retumed to the stream. Fish marcared by the treeline along the edge of the mortalitywas reducedto lessthan 1% by supplying stream. The following information was later trans- oxygen to the collectingbuckets via air-stone and ferred to detailed maps: length and width of stream, portable air pump. depth measurementsof stream, lengths of pools and Electroshockingwas the method of choicefor sev- riffles, lengths of exposed banks, location of nests, eral reasons:(1) it was more practical for sampling and number of perchesalong the stream'sborders. long stretchesof streams(seining was the altemative Riffles were defined as turbulent flow (5-15 cm in method); (2) sampling of only surfacefish was de- depth) connectingpools. Six nonbreedingterritories sired; and (3) although small fish are generally less were measuredin the fall of 1978, 11 breeding ter- sensitive to sampling by this method than larger ritories in the spring of 1979, and 21 nonbreeding fish, the size of fish vulnerableto predationby king- territories in the late summer and fall of 1979. fishers(4-14 cm) appearseasily sampledby electro- Two methods were used to determine the size and shocking (McCormack 1962; results of preliminary speciesof prey selectedby Megacerglealcyon. In the samplingby the author).Fish were assignedinto five first (24 May-23 Iune), I assessedthe speciesand size size classes: 4-5.9 cm, 6-7.9 cm, 8-9.9 cm, 10-11.9 of prey brought to the nest by observingadult feed- cm, and 12-14 cm. Fish smallerthan 4 cm and larger ing activity with the aid of blinds and a 20x tele- than 14 cm were excluded from the samples, as these scope.Because kingfishers carry prey in their beaks, extreme sizeswere never seen taken by kingfishers. I could estimatefish size by comparingthe length of For each fish species, 15 specimensin each size the fish with the bird's bill (culmenlength of indi- classwere weighed to obtain an averageweight for vidual birds was recorded during banding; Drent that size class.These averagedweights were multi- April 1982] KingfisherTerritory Size 355

B

CRAY FISH 133%

NON- 60 M INNOWS 10,2% •u• 50 36"/. M INNOWS 12.7% LL. •.0 29 LL C) STONEROLLERS • 3ø• % 20

UN IDENTIFIED 10, CYPRINIDS •X\\\\\\\\\\\\\\\I 2•1%

1•3 ' 3'0 ß 5"0 ' /*-6 6-8 8-1010-12 12-14 NUMBER OF PREY SIZE CLASS (CM) Fig. 1. Foodcapture data. Shown are the speciescomposition (graph A) and size (graphB, hatchedbars) of prey items taken by kingfishers.Data are from observationsmade on 22 dates and from velcro collar method (seetext for details). Open bars in graph B representthe expectednumber of fish in each size class that should be taken by kingfishers, calculatedfrom the percentageof each size classpresent in electro- shockingdata (breeding season1979, total number of fish sampled = 3,678).

plied by the number of individuals sampled within deserted, but in each case new, successful nests were each size class; then, all size classes were added to- made within the original territory. gether to obtain the total gramsof fish per species. The number of grams of each specieswere added RESULTS together and divided by the length of stream sam- pled to obtain the total gramsper meter of stream. Prey andforaging behavior.--Cyprinid species Relative food abundanceper territory was calculated comprisethe majority of the kingfisher'sdiet from thesedata (grams.meter-•'territory •). (Fig. 1A); stonerollers(C. anomalum)were ob- Preferredforaging sites of individual birds were servedmost frequently. If, as suspected,a sub- identified. Each time a bird was observedforaging, stantial proportion of the unidentified cypri- selectedstream parameters of the immediatehabitat nids in Fig. 1A are C. anomalum,this species were recorded,e.g. whether the individual was seen would be under-representedin the figure. In fishing in a shallowpool (depth < 30 cm), a riffle addition, the percentageof crayfish(Cambarus (as previously defined), or deep pool (depth > 30 cm). Both sexes were included, and data were col- spp.) may not reflect the importanceof this lectedon 15 separatedates during the fall of 1979. species;19 of the 21 observationsof kingfishers BetweenApril and June 1979, data were collected bringing crayfishto the nest occurredduring from 14 nests on Indian Creek. Recorded were: (1) periodsof high water and high turbidity when number of eggs laid, (2) number of eggs hatched, fish were difficult to catch. (3) weight of nestlings (g), (4) incubation period The size range of fish taken by kingfishers (days), and (5) number of young fledged.To sample is shown in Fig. lB. Also shown is the "ex- eachnest, I dug an entrancefrom the top of the bank pected" distribution of prey, calculatedfrom down to the backsideof the next chamber. A pre-cut the proportion of each size classrepresented plywood door assemblywas used to resealthe nest chamber between visits. Nests were checked an av- in the electroshockingdata from Indian Creek erage of once a week; near fledging, nests were (data are pooled from all territories). A Chi- checkedevery other day. Data collectionis more like- squaretest showedno differencebetween the ly to causenest desertionearly in the nesting cycle observedand expectedsizes taken (X• = 5.08; (i.e. before incubation) than after the eggs have 0.3 > P > 0.2; df = 4), indicating that king- hatched. In very early spring (1979), two nests were fishers take prey of different sizes in propor- 356 W•LL•AMJAMES DAVIS [Auk, Vol. 99

data from riffles and pools were combined. 59 NO. OF Although it was evident in the field that sig- FISH nificantlymore fish were taken from riffles, the initial data couldnot be analyzedto determine the differencein the number of fish occurring between these areas. Such data were available, • n 3119 however, for the 1,200 m of Indian Creek elec- troshockedin October 1979. Using these data, the distribution of prey species (fish) was found to be significantlygreater in riffles than in pools (n,oo•= 12, •, = 48.65 g/m; nr•fae.•= 12, ir = 195.46 g/m; t = -3.11, P < 0.01, df = 22; see example in Fig. 2). Stonerollerswere most commonly found in the riffles (X2= < 400m234, P • 0.001, df = 1), while other cyprinids, Fig. 2. Distribution of fish in pools and riffles of taken as a group, showedno preferencefor rif- a typical nonbreeding territory. Hatched bars rep- fles or pools (X2 = 1.29, P > 0.2, df= 1). resent numbers of fish caught in riffle sections;open Campostomaanomalum is known to be a riffle bars representnumbers of fish caught in pools. species(McClane 1978). In the nonbreeding season,kingfishers use streamhabitats differentially: during 80 obser- tion to the relative abundance of each size. vations, kingfishers were found fishing long Appropriate data are not available to test ad- riffles 71.3% of the time, along shallowpools equately whether or not kingfishersselect in- 20% of the time, and at deep pools8.7% of the creasinglylarger prey acrossthe nestingseason time, despite the fact that riffles comprised to feed their progressivelylarger young. The only 36% of the stream's length (X•= 35, size distribution of the fish electroshocked in P < 0.001, df = 2). Unfortunately, foraging the nonbreeding season suggeststhat small data were not collectedfor the breeding sea- fish are more abundant than would be pre- son, but my observationsindicate that king- dictedfrom the springdata (X2 = 79, P >>0.01). fishers behave similarly in the spring. More- It is unknown whether or not kingfisherstake over, the predominant use of C. anomalumas on-the-averagesmaller prey in the non-breed- food in the breeding seasonsuggests that they ing season,but field observationconfirms that fish primarily along riffles. they continue to feed mainly on fish (Davis Territories.--Territories defended in the 1979 1980). breeding seasonwere more than twice as large In both seasons, stonerollers are the most as those defended in the 1979 nonbreeding abundant prey species,comprising 70% of the season (breeding, n = 6, i = 1,030 + 219 m; total fish sampled by electroshockingin the nonbreeding,n = 21, i = 389.29 + 92.63; t = nonbreeding season (data pooled from 1978 3.55, P • 0.005, df = 31). The data are incon- and 1979) and 72% in the breeding season; clusive as to whether or not food is a causal cyprinidsother than stonerollerscomprise 20% factor affecting territory size. For example, the and 21%, respectively.There is no statistical size of nonbreedingterritories is significantly difference in the proportion of stonerollersto inversely related to food density (g/m; r •= "minnows" between seasons (X2= 0.109, -0.98, n = 6, P < 0.01), while the size of P > 0.7, df = 1; total fish sampledin the non- breeding territories is not significantlycorre- breeding season= 3,578, in the breeding sea- lated (r = -0.71, n = 6, P > 0.05). Care must son = 3,678), although noncyprinid species be exercisedwhen interpreting causalrelation- are more abundantin the non-breedingseason ships from simple correlations: in both sea- (X2 = 32.68, P < 0.001, df = 1). This relatively sons, the negative correlationsmay result from large proportion of stonerollersexplains the the indirect effect of intruder pressure, i.e. large proportion of this speciesin the king- more birds are attracted to higher food densi- ilsher's diet (Fig. 1A). Eipper (1956) also re- ties (Myers et al. 1980). ports that stonerollersare a favorite prey of As mentionedpreviously, riffles containsig- kingfishers. nificantly more fish than do pools, and king- During the initial electroshockingruns, the fishersforage more often at riffles. If riffles are April 1982] KingfisherTerritory Size 357

Fig. 3. Riffle length versus size of nonbreeding territories on Indian Creek, 1978. Riffle length and territory size are measuredin meters. Data points fit an exponentialequation (representedby the dashed line) with r s = 0.75, n = 19 and P ( 0.01. an important cue, one would expect territory Starting in March, confrontations between size to be inverselyrelated to total riffle length, males along sections of streams containing as is the case(shown in Fig. 3) for nonbreeding nesting banks were a commonsight, suggest- territories (n = 19, r 2= 0.75, P < 0.01). Al- ing the presenceof more males than nesting thoughriffle length and breedingterritory size banks. In addition, females seemedin surplus are not significantly correlated (n = 8, r = as well, because, on two occasions when the -0.45, P > 0.10), breeding territories that female of a pair disappeared, a new female have a greater concentrationof riffles around took her place with the resident male. White the nestare significantlysmaller than otherter- (1953)and Hamas (1974)also reported nest-site ritories, as is shown in Fig. 4 (.• = 827.5, n = limitations for the Belted Kingfisher. An esti- 4; ß = 1,213.75, n = 4; t = -3.76, P •0.01, mate of the number of kingfisherscompeting df = 6). These resultsmight be explainedby for nests is not available due to the fact that postulating different functions for breeding birds were not banded until after territories and nonbreedingterritories: nonbreeding ter- were established. ritories appear to be essentiallyfeeding terri- Relationships of territory size and food tories, while breeding territories contain two abundance with reproductive successare sum- important resources, food and a suitable nest marized in Table 1. Significantcorrelations oc- site. cur between the following setsof parameters: Nest-sitelimitation and reproductivesuccess.- (1) territory size vs. averageweight of fledg- Suitable nesting banks appear to be in short lings (r = 0.93), and (2) food density (g/m) vs. supply.Of 24 banksalong a 10.6-kmsection of number of birds fledged/nest(r = 0.96). The Indian Creek, 13 banks were occupiedby nest- fact that food density (g/m) is significantlycor- ing pairs, 6 banks were unsuitable for oc- related with number of fledged/nestwhile ter- cupation due to substrate characteristics(com- ritory size is not (r = -0.45) suggeststhat the posedof a high percentageof clayand/or stone) size of the territory is less important than food or location near areas receiving much use by density for production of offspring. humans (e.g. dredging operations), and 5 banks remained unoccupied due to competi- DISCUSSION tive exclusionby territorial individuals. Belted Kingfishers are not colonial nesters and will Nest sites,food, and territorysize.--Both the defend unoccupied nesting banks from con- scarcityand location of nest sites in relation to specificsif they are within their territories. riffles appear to influence territory establish- 358 W•LL•AMJAMES DAWS [Auk,Vol. 99

LU

c•500t ,• 40•,(1037) oo-I 363(765) 362(757) z

• 300-

w• • • • ,' ' .... ,.;"•d223 ...... (,0• 2 273(1282)

• 100' z• ,' •..

100 300 500 700 900 1100 1300 DISTANCE FRON NEST Fig. 4. Cumulativeriffle length versus distance &om nest. On the ordinateis the cumulativelength of shallowwater (depth • 15cm) as the distance from the nest increases symmetric•ly by 25m onthe abscissa. Datawere obtained from detailed maps made of eachterrito•, whichincluded location and length of riffles andpools along the stream. Adjacent numbers represent total length of sh•lowwater (primarily riffles), and numbersin parenthesesrepresent lengths of territories. mentby Megacerylealcyon. Theoretically, nest- mineswhether or not a particularpair of king- sitelimitations can suppress the breedingpop- fisherswill defenda large or smallterritory. ulation density and thus lead to a lower level Evidence is not in favor of a random selection of competitionfor food than would occurif all of nest sites, however. First, males arrive at potentialbreeders were ableto nest.A scarcity least a month before the females to find a suit- of nest sites is often observed for birds (see able nesting bank (somemales in mild winters Temple 1977), and field observations indicate remain the year around). Second,within most competition for available nesting sites for breeding territories along Indian Creek, more kingfishers(Davis 1980). Accordingly, territo- than one suitablenesting bank was available riality in the breedingseason may primarily from which to choose. serve to secure a nest site, while the need to In the event that food is concentrated near defend a food resourceis secondary.This a nestsite, defense of a smallterritory may save study presentsdata that supportthis hypothe- a pair time and energy both in defenseof the sis becauseterritory size and food abundance territory and in the transportof fish backto the (g/m) are not significantlycorrelated. nest. To feed a nest full of youngkingfishers Other data, however,suggest that food dis- may require considerable time and energy. tributiondoes influence the size of breeding Using data from Vessel (1978), I calculatethat territories, in that territory size appearsto be at maximumgrowth rate eachnestling can con- establishedin a consistentrelation to the prox- sume approximately11.2 fish/day.Given 15 h imity of productivefood patchesnear the nest of daylight, eachadult of a pair would need to (Fig. 4), i.e. the smallestterritories enclosethe catch2.6 fish/h in order to feed sevenyoung, richestfood sources.Hole-nesters typically se- not including fish for themselves.During un- lect a nest site beforeestablishing their terri- favorableweather, this may be a difficulttask, tories (von Haartman 1957)•a generalization one that is aggravatedif food is locatedfar from that accuratelydescribes the BeltedKingfisher the nests. As a result, access to an uncontested (Bent 1940, Davis 1980). Unless kingfishers and rich foodsource may be criticalfor raising evaluatefood abundanceduring nest-sitese- a full clutch of young kingfishers.As Seastedt lection, my data imply that an element of and Maclean(1979) have pointed out for Lap- chance enters into the equation that deter- land Longspurs,habitat quality may be better April 1982] KingfisherTerritory Size 359

TABLE1. A summary of correlations(r) and significancelevels between reproductive parameters,territory size, and food abundance. Sample size (n) varies due to the facts that not all territories were sampled for food or nest data were incompletefor some territories.

Food abundance

Territory size Grams/meter Total grams

r n P r n P r n P

Number fledged/nest -0.45 8 NSa 0.96 5 0.02 -0.27 5 NS Averageweight of fledglingsper nest 0.93 7 <0.01 -0.50 5 NS 0.62 7 NS Total weight of fledglingsper nest -0.22 7 NS 0.76 5 NS -0.39 5 NS

• NS = P > 0.05.

represented by food density than by total the question of why kingfishersdefend the quantity. sizes of territory that they do is still not an- When food is not concentrated near the nest, swered. it might be advantageousto defend as large a There are two prevailing hypothesesthat ad- territory as possible, first, becauselarger ter- dress the causal relationship between food ritories may contain greateramounts of food abundance and territory size (Myers et al. (total biomass/territory)and, second,because 1980).The first is the "sufficientresource" hy- larger territories may contain a more reliable pothesis(see Verner 1977), which is derived food supply, i.e. a larger supplyof alternative from the classicalconcept that the basic func- food resources.During fluctuations in water tion of territoriality is to secureadequate re- levelsfollowing heavy or prolongedrains, fish- sources.This hypothesisstates that individu- ing becomesdifficult, and kingfishersare ob- als shoulddefend only thoseresources that are ligated to switch to crayfishas prey, a food required. The secondhypothesis predicts that item not found in riffles. Perhaps as a conse- competition is most severein areasof high re- quenceof both thesefactors, parents with larg- sourcedensity, leading to increasedcosts of er territories that fledge the same number of defense;as a result, territory size is limited by offspringas thosewith smallerterritories may competition for resources(Seastedt and Mac- produceheavier young. Theseincreased fat re- lean 1979; Myers et al. 1979, 1980). Relevant serves could increase the chances of survival data on the BeltedKingfisher are not yet avail- of fledglings(Lack 1966), becausethe first few able to enable a choicebetween thesehypoth- weeks after leaving the nest is a most critical eses; circumstantial evidence, however, favors period for young kingfishers.Early attemptsat the latter. Expansion of breeding territories fishing are most often failures for the young was observed twice, once in 1978 (Davis 1980) birds, becausethey are weak flyersand appar- and oncein 1979, after an adjacentpair of king- ently must learn how to fish (Bent1940, White fishers deserted their territory due to human 1953, Davis 1980);the fact that parentscontin- interference at their nests. These observations ue to feed fledglings suggeststhat young have supportthe hypothesisthat breedingterritory difficultyin obtainingtheir own food. Because size is limited by competition. In a reverse Megacerylealcyon typically raises only one sense, severe nest-site limitations may pro- clutchper year, the survivalof fledglingsmay mote large breeding territories by reducing markedlyaffect the relativefitness of breeding competition. pairs. It appearsthat assessingthe relative val- In the nonbreedingseason, territories of in- ue of large versus small territories in terms of dividuals were remarkably constant in size. reproductivesuccess may depend upon envi- During prolongedrains, nonbreedingterrito- ronmental circumstances. As argued above, ries were often deserted but were reestablished the weight at fledgingmay be important when by the original owners with the original the environment fluctuates,but fledging more boundaries when the water levels returned to but lighter young may increasefitness in a normal. In December, some territories gradu- more constant, productive environment. Both ally increasedin size as kingfishersbegan to strategiesmay be utilized by kingfishers,but disappear,presumably due to their migration 360 WILLIAMJAMES DAVIS [Auk,Vol. 99 south. When and how many birds migrated nest site, breeding birds are able to use habitat appeared to be related to the severity of the cues as important parameters in establishing weather. During cold periods, accumulationof territory size. Only one pair defended a terri- ice along the stream blocked accessto fishing tory larger than expected,when size was pre- areas;the few birds remainingduring the win- dictedby the criterionof the cumulativelength ter months (December-February) seemed to of riffles. This pair initially defendeda smaller usecrayfish more extensively, as gauged by the territory (approximately 830 m) but increased accumulation of their exoskeletons in the win- its size after the adjacentdownstream territory ter roosting nests examined. How dependent was deserted due to human interference at the kingfishersare on crayfishin the winter is still nest. In addition, the relationship between the unknown. Accumulationof exoskeletonsmay size of nonbreedingterritories and riffle length only indicate that these parts are not digestible (Fig. 3) suggeststhat kingfishersmay assess while most parts of fish are digestible (White habitat quality (amount of food available) by 1953). In terms of the proportionof digestible observing the length of fifties present. biomass/preyitem, fish are probably a higher A "new" functionof territoriality.--The view quality food item. adopted in this study and held by most biol- Proximate cues to assess resource abundance.- ogistsis that territoriality evolvedbecause de- Although fish availability may fluctuateduring fense of essentialresources directly enhances the season (Davis 1980), particular habitat pa- the reproductive successof the defender (Nice rametersassociated with food may remain rel- 1941, Lack 1943, Hinde 1956, Brown 1964). But atively constant,e.g. the amountof riffles pres- recently, it has been proposed that territorial ent along a section of stream. Thus, if direct behavior could also be adaptive by decreasing assessmentof food by kingfishers is difficult, the successof one's reproductivecompetitors Megacerylealcyon may instead respondto hab- (Verner 1977). For example, if the cost of de- itat parameters such as riffles. Morse (1976) fense of a resource is minimal, the benefit in believes that warbler speciesrespond to cues terms of relative reproductivesuccess may be that are easily monitored and are correlatedto substantial if an individual can prevent con- important factors. Similarly, data on Sparrow- specificsfrom obtaining a nest or food. A cru- hawks (Accipiternisus) (Newton et al. 1977)and cial assumptionof the hypothesisof "super- Lapland Longspurs (Seastedt and Maclean territories" is that the cost of defense is not 1979) indicate that these speciesrespond to al- prohibitive. Along this line, Rothstein (1979) ternative habitat parameters rather than as- arguesfor the evolution of "inhibitory traits," sessingfood density directly. Other speciesare like aggression,in animals that use specialized known to search for specific prey types by nest sites or utilize highly concentratedfood searchingfor specific sites or patches,rather resources. In the present study, 28% of the than the prey item itself (Royama1970, Parus available nest sites appearedto remain unoc- major; Alcock 1973, Agelaiusphoeniceus; Tin- cupieddue to exclusionof potentialnesters by bergen1976, Sturnus vulgaris). The Little Green extant territory owners. Such aggressivebe- Kingfisher(Chloroceryle americana) also prefers havior, characteristicof territorialityin Mega- to fish along riffles within a stream habitat. cerylealcyon, could be an "inhibitory trait" as Out of 62 observationsmade by the author in describedby Rothstein. If so, territoriality in Costa Rica during the summer of 1981, the this speciesmay be adaptive not only because Green Kingfisherwas found foragingalong rif- it secures resources essential to survival, but fles 84% of the time. Kingfishersmay not in- also becauseit excludescompetitors from re- nately recognize riffles but instead learn to sourcesthat are not utilized by a defender. associatethem with food by positive rein- Several investigators (Getty 1979, Pleasants forcementsubsequent to fishing successalong and Pleasants1979) have voiced strong criti- riffles. Future experiments that manipulate cismof this hypothesis.More empiricalstudies food abundance are needed to determine the are needed to determine whether or not ani- role of habitat parameters(riffles) in the assess- mals do defend territories larger than are ment of habitat quality. neededto fulfill their own requirements.More The general pattern displayed in Fig. 4 sug- extensivedata on reproductivesuccess and ter- gests that in addition to choosinga suitable ritory size are required, and animals that have April 1982] KingfisherTerritory Size 361

specializednest-site or food requirementsthat LACK, D. 1943. The life of the Robin. London, H. F. are easily defended should be critically evalu- & G. Witherby. ated. 1966. Population studies of birds. Oxford, Clarendon Press. MCCLANE,A. J. 1978. McClane'sfield guide to fresh- ACKNOWLEDGMENTS water fishes of North America. New York, Holt, I would like to thank T. C. Kane, my major advi- Rinehart and Winston. sor, and Fred Wasserman, D. R. Osborne, and Linda McCORMACK,J. C. 1962. The food of young trout Mealey for contributing helpful suggestionsduring (Salmo trutta) in two different backs. J. Anim. the first drafts of this paper. I also appreciate the Ecol. 31: 305-316. valuable commentsof J. P. Myers and two anony- MORSE,D. H. 1976. Variables affectingthe density mous reviewers. Mike Bishop, Lisa Aren, and Mas- and territory size of breeding spruce-woods tin Mount made the study possibleby volunteering warblers.Ecology 57: 290-301. their time to collect field data. Financial support MYERS,J.P., P. G. CONNORS,& F. A. PITELKA. 1979. came from Sigma Xi, University of Cincinnati Re- Territory size in wintering Sanderlings:the ef- searchCouncil Summer Fellowships, and University fect of prey abundance and intruder density. of Cincinnati Graduate Student Research Funds. Auk 96: 551-562. This study fulfilled in part the requirementsfor an --, --, & --.. 1980. Optimal territory M.S. degree. size and the Sanderling:compromises in a vari- able environment. Pp. 135-158 in Foraging be- havior: ecological,ethological, and psychologi- LITERATURE CITED cal approaches(A. C. Kamil and T. D. Sargent, ALCOCK,J. 1973.Cues used in searchingfor foodby Eds.). New York, Garland STPM Press. red-wingedblackbirds (Agelaius phoeniceus). Be- NEWTON, I., M. MARQUISS, D. N. WEIR, • D. MOSS. haviour 46: 174-188. 1977. Spacing of Sparrowhawk nesting territo- BENT, A. C. 1940. Life histories of North American ries. J. Anim. Ecol. 46: 425-441. cuckoos,goat-suckers, hummingbirds and their NICE, M. M. 1941. The role of territory in bird life. allies. U.S. Natl. Mus. Bull. 176. Amer. Midl. Natur. 26: 441-487. BROWN,J. L. 1964.The evolutionof diversityin PLEASANTS,J. M., & B. Y. PLEASANTS.1979. The avian territorial systems. Wilson Bull. 76: super-territoryhypothesis: a critique, or why 160-169. there are so few bullies. Amer. Natur. 113: DAVIS,W. J. 1980.The BeltedKingfisher (Megaceryle 609614. alcyon).'its ecologyand territoriality.Unpub- PITELI(A, •F. A., P. W. TOMICH, & G. W. TREICHEL. lished Master's thesis. Cincinnati, Ohio, Univ. 1955. Ecologicalrelations of jaegarsand owls as Cincinnati. lemming predators near Barrow, Alaska. Ecol. DRENT,R. H. 1965.Breeding biology of the Pigeon Monogr. 25: 85-117. Guillemot, Cepphuscolumba. Ardea 53: 99-160. ROYAMA,t. 1970. Factors governing the hunting EIPPER,A. W. 1956. Differencesin vulnerabilityof behaviour and selectionof food by the Great Tit the prey of nestingkingfishers. J. Wildl. Mgmt. (Parusmajor L.). J. Anim. Ecol. 39: 619-668. 20: 177-183. ROTHSTEIN,S. I. 1979. Gene frequenciesand selec- EWALD, P. W., G. L. HUNT, & M. WARNER. 1980. tion for inhibitory traits, with specialemphasis Territory size in western gulls: importanceof on the adaptivenessof territoriality.Amer. Nat- intrusion pressure, defense investments, and ur. 113: 317-331. vegetationstructure. Ecology 61: 80-88. SALOMONSON,g. G., & R. P. BALDA. 1977. Winter GETTY,T. 1979. On the benefitsof aggression:the territorialityof Townsend'sSolitaires (Myadestes adaptivenessof inhibition and superterritories. townsendi)in a pition-juniper-ponderosa pine Amer. Natur. 113: 60•609. ecosystem.Condor 79: 148-161. GILL, F. B., & L. L. WOLF. 1975. Economics of feed- SEASTEDT,t. R. & S. F. MACLEAN.1979. Territory ing territorialityin the Golden-wingedSunbird. size and composition in relation to resource Ecology 56: 333-345. abundance in Lapland Longspurs breeding in HAARTMAN,L. VON. 1959. Adaptationin hole-nest- Arctic Alaska. Auk 96: 131-142. ing birds. Evolution 11: 339-347. STENGER,J. 1958. Food habits and available food of HAMAS,M. J. 1974. Human incursionand nesting Ovenbirdsin relationto territorysize. Auk 75: sitesof the BeltedKingfisher. Auk 91: 835-836. 335-346. HILDf•N, O. 1965. Habitat selection in birds. A re- TEMPLE,S. T. 1977. Endangered birds. Madison, view. Ann. Zool. Fennica 2: 53-74. Wisconsin, Univ. Wisconsin Press. HINDE, R. A. 1956. The biologicalsignificance of TINBERGEN,J. M. 1976. How Starlings(Sturnus vul- territories of birds. Ibis 98: 340-369. garis L.) apportiontheir foragingtime in a vir- 362 WtLLmMJAMœs DAws [Auk,Vol. 99

tual single-prey situation on a meadow. Ardea WErT¾, J. C. 1975. The life of birds, 2nd Ed. Lon- 64: 155-170. don, W. B. SaundersCompany. VœRNœR,J. 1977. On the adaptive significanceof ter- W•rrœ, H. C. 1953. The EasternBelted Kingfisher ritoriality. Amer. Natur. 111: 769-775. in the Maritine Provinces. Fish Res. Board Can- VESSœL,R. D. 1978. Energeticsof the Belted King- ada Bull. 94. fisher (Megacerylealcyon). Unpublished Ph.D. W•œNs,J. A. 1969. An approachto the study of eco- dissertation. Corvallis, Oregon, Oregon State logicalrelationships among grassland birds. Or- Univ. nithol. Monogr. No. 8. SHORT COMMUNICATIONS

EggshellStrength and Cowbird Parasitismof Red-winged Blackbirds

GILBERTW. BLANKESPOOR,JOHN OOLMAN, AND CRAIG UTHE Departmentof Biology,Augustana College, Sioux Falls, South Dakota 57197 USA Recently, there has been much interest in avian We suggestthat cowbirds have evolved an addi- brood parasitism(review by Payne 1977). Its study tional kind of indirect strategyto reducethe number provides a unique opportunity to gain some under- of competing Red-winged Blackbird nestlings. This standingof how two distinct and different lineages strategy involves eggshell strength and egg survi- coevolve (Robertson and Norman 1977). Rothstein vorship.We suggestthat Red-wingedBlackbird eggs (1975) suggeststhat, during the egg-laying period, are structurallyweaker and more likely to be broken host speciesfor the Brown-headedCowbird (Mol- (and subsequentlyremoved) during egg-laying and othrusater) fall into two mostly nonoverlappingcat- incubation than are cowbird eggs. As part of a 1979 egories: some species,"rejecters," exhibit behaviors and 1980study of egg porosityin thesetwo species, that, at leastin part, negatethe effortsof the parasite, we used the scanning electron microscopeto mea- while others, "accepters," appear to tolerate the sureshell thickness in eggscollected before the onset presenceof cowbird eggs. of incubation. The thickness measurement for each In 1975and 1976we monitored cowbird parasitism egg was the mean of five shell fragmentsnear the in populations of an accepter species, the Red- equatorof the egg. Thicknessmeasurements did not winged Blackbird (Agelaiusphoeniceus), in Lincoln include the shell membranes. Thirty Red-winged County, SouthDakota. Rates of parasitismmeasured Blackbird eggs (mean fresh weight = 4.03 g + 0.43 at the beginning of incubation were 14 out of 19 nests SD) had a mean thickness of 41.38 /zm + 3.56 SD, (21 out of 90 eggs)in 1975and 14 out of 34 nests(23 while 30 cowbird eggs (mean fresh weight = 3.26 out of 136 eggs)in 1976. Theseresults confirm other g + 0.24) had a mean thicknessof 58.79/zm + 2.67. published reports that Red-winged Blackbirds are Thus, using thickness as one measure of shell commoncowbird hostson the GreatPlains (Hergen- strength, we concludethat cowbird eggs are clearly rader 1962,Hill 1976,Facemire 1980). The higher rate stronger. The fact that the heavier Red-winged of parasitism in 1975 might have resulted from a Blackbirdeggs had thinner eggshellsruns counter to drought-induced reduction in the number of avail- the usual relationship between egg weight and shell able host nests. thickness in birds (Ar et al. 1974). Apparently, Brown-headed Cowbird nestlings But, in fact, are Red-winged Blackbirdeggs more show few adaptationsfor the parasitichabit. They likely to be crackedand broken?During the porosity do not hatchsignificantly sooner than hostnestlings, study, a number of eggsof both specieshad unex- their growth rates are similar to those of related pectedlyhigh rates of water loss. Upon closeexam- species (Payne 1977), and they do not evict nest ination, these eggswere determined to have cracks. mates. Recently, Eastzeret al. (1980) have shown that Becauseall eggs were wrapped in cotton wadding cowbird nestling survival in Barn Swallow (Hirundo before being transportedfrom the field to the labo- rustica)nests was no better than four of five nonpar- ratory, we are confidentthat mostof theseeggs were asitic species,although a fledglingcowbird was suc- alreadycracked in the nest. The numbersof cracked cessfulin elicitingfeeding behaviorfrom Barn Swal- eggsin parasitizednests were as follows:1979, 23 of low foster parents. 65 (35.4%) Red-wingedBlackbird eggs and 14 of 81 Cowbird nestlings often do enjoy a competitive (17.3%) cowbird eggs; 1980, 35 of 89 (39.3%) Red- advantage,however, becausethey typicallyoccur in wingedBlackbird eggs and 19of 151(12.6%) cowbird the nestsof smallerhost species.Another strategy eggs.Pooling the data for both years,the proportion for gaining a disproportionatelylarge share of food of crackedeggs in Red-winged Blackbirdswas sig- resources involves a reduction in the number of com- nificantlyhigher (X2 = 36.41, df = 1, P < 0.005). peting host nestlings.One way cowbirdsaccomplish Presumably,an individual egg is crackedeither by this is by removinghost eggs. We examinedour 1975 being steppedon by the female or by being jostled and 1976samples of Red-wingedBlackbird nests for against other eggs. In both cases,eggs with thicker, evidence of host-eggremoval (Table 1). If one as- strongershells (i.e. cowbird eggs)would enjoy a dis- sumes a normal clutch size of four, then cowbirds at 10 tinct advantage.With regard to egg jostling, it is ex- of the 37 parasitizednests showed no ejection be- pected that, when a Red-winged Blackbird egg is havior. Becausein the remainingnests a number of jostled against a thicker-shelled cowbird egg, it is Red-winged Blackbird eggs may have disappeared more likely to crack than when it is jostled against for other reasons(see below), we concludethat cow- another Red-winged Blackbirdegg. This idea can be birds quite often fail to eject Red-winged Blackbird tested by comparing the numbers of cracked Red- eggs. winged Blackbird eggs in parasitized and nonpar-

363 364 Short Communications [Auk, Vol. 99

TABLE1. Nest contentsat the beginning of incuba- data across1975 and 1976 and using a one-tailed test, tion or at the time of nest discovery. egg loss was significantly greater in Red-winged Blackbirds (Z = 2.65; P < 0.01). Eighteen of the 27 Number of nests eggslost (66.7%) were in parasitizednests. Nest contents a 1975 1976 We believe this differential egg losswas due to a greater tendencyfor Red-winged Blackbirdeggs to 5 RW -- 1 break and to be subsequentlyejected by female Red- 4 RW 4 17 winged Blackbirds. There are alternative explana- 4 RW, 1 CB 5 3 tions, however, that need to be examined. Because 4 RW, 2 CB -- 1 cowbird eggsare better camouflaged,it could be ar- 4RW, 3 CB i -- 3 RW 3 7 gued that differential egg survivorship was due to 3 RW, 1 CB 4 4 a greater probability of Red-winged Blackbirdeggs 3 RW, 2 CB 4 5 being preyed upon during partial predation.Picman 3 RW, 3 CB -- 1 (1977) demonstrated that the Long-billed Marsh 2 RW -- 2 Wren (Cistothoruspalustris), a common breeding 2 RW, 1 CB 1 2 speciesin the study area, does break and remove 2 RW, 2 CB 1 2 Red-winged Blackbird eggs. His results, however, 2 RW, 4 CB 1 -- indicate that Marsh Wrens respond equally to eggs 1 RW, 1 CB -- 1 of a differentcolor. It may alsobe arguedthat female 1 RW, 4 CB 1 -- cowbirds remove some Red-winged Blackbird eggs • RW = Red-wingedBlackbird eggs; CB = Brown-headedCowbird after incubation has begun. Although suchbehavior eggs. has been observedby Mayfield (1961)in the Kirt- land's Warbler (Dendroicakirtlandii), it apparently occursonly rarely, becausecowbirds usually lay an egg of their own after removing a host egg and such asitized nests. Ideally, only eggs from clutchesof eggslaid after incubation essentiallywould be wast- equal size should be used for suchcomparisons. Be- ed. causeno samplefor a single clutch-sizecategory was From the parasite'spoint of view, it is surprising sufficientlylarge, we included eggsfrom all clutch- that the Red-winged Blackbird is used so often as a size categoriesin our analysis. The numbers of host, becausecowbird nestlingswould be expected crackedeggs were as follows:1979, 35 of 89 (39.3%) to do better in the nests of smaller hosts. The nega- in parasitizednests and 33 of 120 (27.5%) in non- tive effect of large host size, however, can be coun- parasitizednests; 1980, 58 of 154 (37.6%) in parasit- ized nests and 50 of 230 (21.7%) in nonparasitized tered by reducing the number of host eggs (and nests. Pooling the data for both years, the number thereforehost nestlings).Cowbirds often do this by ejecting host eggs. An additional, favorable change of crackedeggs was significantlyhigher in parasit- in the ratio of host to parasite eggs would occur if ized nests (X2 = 11.57, df = 1, P < 0.005). host eggs had lower survival rates during incuba- We have demonstratedthat by the time incubation tion. This study indicatesthat suchis the casewhen begins a significantlylarger number of Red-winged Red-winged Blackbirdsserve as hosts, primarily be- Blackbirdeggs sustains cracks than do cowbird eggs. It remains to be demonstrated that Red-winged cause Red-winged Blackbird eggs are structurally weaker. Blackbirdeggs actually survive lesswell in the field We thank the officials at the Lake Andes National during incubation. Our assumption is that most Wildlife Refugefor permissionto conductthe study cracked eggs subsequentlybreak and are removed at the Atkins Waterfowl Production Area and the from the nest. In 1975 and 1976 during incubation, National ScienceFoundation for financial support in we visited parasitized and nonparasitized nests al- its Undergraduate ResearchPariticipation Program. most every day and were able to compare egg sur-

vivorship in the two species.Eggs that were lost be- LITERATURE CITED causeof whole-nest predation were not included in the comparison. Typically, single eggs would dis- AR, A., C. V. I•AGANELLI, R. B. REEVES, D. G. appear from a nest between one visit and the next. GREENE,& H. RAHN. 1974. The avian egg: water In 1975, 14 Red-winged Blackbirdeggs were lost dur- vapor conductance,shell thickness,and func- ing 580 egg-days,while no cowbird eggswere lost tional pore area. Condor 76: 153-158. during 245 egg-days.In 1976, disregardingtwo cow- EASTZER,D., P. R. CHu, & A. P. KING. 1980. The bird eggs that were laid maladaptively during in- young cowbird: average or optimal nestling? cubation, 13 Red-winged Blackbird eggs were lost Condor 82: 417-425. during 691 egg-dayswhile 1 cowbird egg was lost FACEMIRE,C. F. 1980. Cowbird parasitism of marsh- during181 egg-days. The differencebetween rates nesting Red-winged Blackbirds. Condor 82: of egg loss for the two specieswas tested by using 347-348. a test of proportions(Ferguson 1966: 176). Pooling FERGUSON,G. A. 1966. Statistical analysis in psy- April 1982] ShortCommunications 365

chology and education, 2nd ed. New York, PAYNE,R. B. 1977. The ecologyof brood parasitism McGraw-Hill Book Co. in birds. Ann. Rev. Ecol. Syst. 8: 1-28. HERGENRADER,G. L. 1962. The incidence of nest- PIeMAN, J. 1977. Destruction of eggs by the Long- parasitism by the Brown-headed Cowbird on billed Marsh Wren (Telmatodytespalustris palus- roadside nesting birds in Nebraska. Auk 79: tris). Can J. Zool. 55: 1914-1920. 85-88. ROBERTSON,R. J., & R. F. NORMAN. 1977. The func- HILL, R. A. 1976. Host-parasiterelationships of the tion and evolution of aggressivehost behavior Brown-headed Cowbird in a prairie habitat of towards the Brown-headed Cowbird (Molothrus west-central Kansas. Wilson Bull. 88: 555-565. ater). Can. J. Zool. 55: 508-518. MAYFIELD,H. F. 1961. Vestiges of a proprietary in- ROT•STEIN,S. I. 1975. An experimentaland taleon- terest in nests by the Brown-headedCowbird omic investigation of avian brood parasitism. parasitizing the Kirtland's Warbler. Auk 78: Condor 77: 250-271. 162-166. Received13 July 1981, accepted19 October1981.

Evidenceof a BorealAvifauna in Middle Tennesseeduring the Late Pleistocene

PAUL W. PARMALEE AND WALTER E. KLIPPEL Departmentof Anthropology,University of Tennessee,Knoxville, Tennessee 37916 USA A faunistically significant, stratified cave deposit environment for an extended period of time in this was discoveredduring an archaeologicalsite survey area during the Late Wisconsinan (Parmalee and of the proposedTennessee Valley Authority (TVA) Klippel 1981). Columbia Dam reservoir that will impound approx- The avifauna from Cheek Bend Cave is extremely imately 30 km of the Duck River in middle Tennes- diversified and reflects both open grasslandor sa- see. This small cave, known as Cheek Bend Cave vanna and forest habitat. Remains of aquatic birds (40MU261), is in a limestone bluff 20 m above the (e.g. Pied-billed Grebe, Podilymbuspodiceps, and at river bed, 13 km east-southeastof Columbia, Maury least three speciesof ducks) are rare in the deposit, County. Testing for the presenceof aboriginal oc- although severalspecies associated with riverine or cupation began in September 1978 and, as a result marsh habitats are represented.Most notable among of encounteringhuge quantitiesof animal bone and theseare the Yellow Rail (Coturnicopsnoveboracensis) shell at considerabledepths that appearednot to be and the BlackRail (Laterallusjamaicensis), both prob- attributable to man, was continued until May 1979. ably rare transients in Tennessee,although their Three 1 x 2-m excavationunits placedalong the east modern distribution in the Midsouth is poorly wall revealed stratified deposits of animal remains known. We identified a minimum of approximately to a depth of about 4.5 m. 60 species,representing 28 families, from the avian On the bases of the speciescomposition within sample. Elementsof numerous passerinespecies of and among strata, differing fill zones, and certain the families Parulidae and, particularly, Fringillidae otherstratigraphic features, the depositreflecth two comprised 75-95% of the avian samples from the distinct and well-defined major episodesof fill. The Holocene levels. Except for the PassengerPigeon top 2 m represent the Holocene epoch and contain (Ectopistesmigratorius) and the Ruffed Grouse (Bo- remains of only extant modem species.In contrast, nasa umbellus),all of the identified species repre- the faunal assemblagein the bottom 2 m, although sented in the Holocene levels are still a part of the it includes some speciesthat still inhabit the cave avifauna of middle Tennessee.A comprehensivere- area (presently a deciduous Western Mesophytic port dealingwith the stratigraphyand chronologyof Forest),contains many that are now extirpatedand CheekBend Cave, its archaeologicalcomponent, and reflect a prairie habitat (Prairie Chicken, Tympanu- the total faunal assemblageis being preparedfor the chuscupido; 13-lined ground squirrel, Spermophilus TennesseeValley Authority. tridecemlineatus,pocket gopher, Geomys sp.) or a bo- Several speciesof birds identified from elements real environment (e.g. arctic shrew, Sorexarcticus; recovered in the bottom 2 m of the deposit are es- water shrew, Sorexpalustris; red-backed vole, Cleth- pecially significant,because they not only represent rionomysgapperi ; yellow-cheekedvole, Microtusxan- new stateor regionalrecords but alsobecause they thognathus; red squirrel, Tamiasciurushudsonicus ; suggesta boreal climate and habitat in this region northern flying squirrel, Glaucomyssabrinus; and during the Wisconsinan.Five speciesworthy of spe- heather vole, Phenacomysintermedius). Recovery of cial comment are the Hawk Owl (Surnia ulula), Boreal a partial carapaceof a wood turtle (Clemmysinsculpta) Owl (Aegoliusfunereus), Saw-whet owl (Aegolius also suggests,based on the presentcenter of distri- acadicus),Gray Jay (Perisoreuscanadensis), and Pine bution of the wood turtle, a well-established boreal Grosbeak (Pinicola enucleator). In addition to these 366 Short Communications [Auk, Vol. 99

TABLE1. Avian speciesidentified from the late Pleis- Siskin (Carduelispinus) also suggestextensive conif- tocenestrata of Cheek Bend Cave, Maury County, erous forests in the vicinity of Cheek Bend Cave. Tennessee. Althoughmost species comprising the avian assem- blage reflecta forest habitat, the presenceof mead- Falconidae owlark (Sturnellasp.), Horned Lark (Eremophilaal- AmericanKestrel, Falco sparverius pestris), Prairie Chicken, Sharp-tailed Grouse Tetra onida e (Pedioecetesphasianellus), and somerodents such as Ruffed Grouse, Bonasa umbellus the pocketgopher (Geomyssp.) is indicative of these Sharp-tailedGrouse, Pedioecetes phasianellus a forestedareas having been interspersedwith short- Greater Prairie Chicken, Tympanuchuscupido grassprairie or savanna. The Hawk Owl lives in Canada and Alaska in co- Rallidae niferous forests mixed with birch scrub, tamarack Yellow Rail, Coturnicopsnoveboracensis bogs, and muskegs(Bent 1938). This owl is a rare Scolopacidae and sporadicwinter vagrant in northern Plains and CommonSnipe, Capellagallinago • Midwest states; there are no historic records of it American Woodcock, Philohela minor from Tennessee and no Pleistocene records from Columbidae anywhere. Identification of Surniaulula from Cheek PassengerPigeon, Ectopistesmigratorius Bend Cave is basedon a completephalanx (phalanx I, digit I) and premaxilla(Fig. 1). Distribution of the Strigidae Boreal Owl is similar to that of the Hawk Owl, as is Barred Owl, Strix varia its choiceof habitat. The only fossil(Holocene?) rec- Boreal Owl, Aegoliusfunereus a ord of A. funereusis from ShelterCave, New Mexico Saw-whetOwl, Aegoliusacadicus • Hawk Owl, Surnia ulula a (Brodkorb 1971). Evidence of its occurrence in his- toric times in Tennesseeis lacking. Diagnostic ele- Apodidae ments of at least two individuals of the Boreal Owl Chimney Swift, Chaeturapelagica • from Cheek Bend Cave consistof the proximal two- Picidae thirds of a right and shaft sectionof a left tarsometa- CommonFlicker, Colaptesauratus a tarsus, distal one-third of a right and distal end of Red-belliedWoodpecker, Melanerpes carolinus a left tibiotarsus,distal end of a right femur, head Hairy Woodpecker,Picoides villosus of a left humerus, distal end of a left ulna, proximal Tyrannidae end of a left radius, sectionsof a lower jaw, articular EasternPhoebe, Sayornis phoebe end of a left scapula,coricoidal articular facet and manubrium sectionof a sternum,and 18 phalanges. Alaudidae Although the Saw-whet Owl is not restrictedto Horned Lark, Eremophilaalpestris • the northern coniferous forests, its center of distri- Corvidae bution and greatestabundance lies in the woodland Blue Jay, Cyanocittacristata areas acrosssouthern Canada, the Great Lakes states, Gray Jay, Perisoreuscanadensis • and in the northeastern United States. It is a rare Sittidae winter residentin the Midsouth;there are only four reported sight recordsof this owl in the Nashville Red-breasted Nuthatch, Sitta canadensisa area (ca. 60-80 km north of Cheek Bend Cave) (Rig- Turdidae gins 1977).All remainsof A. acadicuswere recovered Robin, Turdusmigratorius from the same late Wisconsinan levels (300-400 cm Icteridae below the surface) in which the bones of the Hawk Owl and Boreal Owl occurred. Remains of the Meadowlark, Sturnellasp. Screech Owl (Otus asio), a common inhabitant of Fringillidae primarily deciduousforests of easternNorth Amer- PurpleFinch, Carpodacuspurpureus ica, were found in the Holocene levels of Cheek Bend Pine Grosbeak, Pinicola enucleator• Cave but not in the Pleistocene levels. Elements of Pine Siskin, Carduelispinus a the Saw-whet Owl consist of the distal section of a cf. Towhee, Pipilo erythrophthalmus cf. White-throatedSparrow, Zonotrichia albicollis right and left coracoid,articular end of a right scap- ula, proximalone-half of a right and distal one-third Remainspresent only in late Pleistocenestrata. of a left radius, distal end of a left ulna, a complete right tibiotarsus, distal one-third of a right tarso- metatarsus,and sevenphalanges. In easternNorth America both the Gray Jay and five, remainsof 23 other specieswere identifiedfrom Pine Grosbeakoccur throughout the coniferousfor- the late Pleistocene strata (Table 1). In addition to ests of southern Canada, and in the United States the three owls, Gray Jay, and Pine Grosbeak,the they seldom appear farther south than the upper Red-breasted Nuthatch (Sitta canadensis)and Pine Great Lakes and northern New England states. Like April1982] Shor•Communications 367

a c j

a

b b d e

A B

a

e g b c d f

c

Fig. 1. Avianfossils from Cheek Bend Cave, Maury County, Tennessee. A. Surniaulula: a, premaxilla; b, phalange.B. Aegoliusacadicus: a, tibiotarsus; b, radii; c, coracoids; d, ulna; e, tarsometatarsus.C. Aegolius funereus:a, sternum;b, humerus;c, radius;d, ulna;e, femur;f, tarsometatarsi;g, tibiotarsi.

the Hawk Owl and Boreal Owl, these two passerines teau, basedon the analysesof pollenrecovered from are associated with a boreal habitat and seldom wan- sedimentcores from Anderson Pond (White County) der or migrategreat distances. Identification of Pin- and Mingo Pond(Franklin County), Delcourt (1979) icola enucleator was based on sections of the lower was able to show that several major vegetational jaw andleft carpometacarpus (distinct from Pheuticus changeshave taken placeduring the past 25,000 and Hesperiphona)and that of Perisoreuscanadensis years. During the Wisconsinanglacial maximum, on limb elementsincluding the distal ends of two about 19,000-16,300yr BP, borealtaxa of jack pine, left humeri and two complete left carpometacarpi. spruce,and fir weredominant, and "Between16500 Fossilrecords of thesepassetines are lacking(Brod- and12500 yr BP,mixed coniferous deciduous forests korb 1978). coveredthe landscapebetween 34øN and 37øNlati- In a recentstudy of the late Pleistocenevegetation tude" (Delcourt1979: 276). In his discussionof the of the easternHighland Rim and CumberlandPla- vegetationof easternNorth America18,000 yr ago, 368 Short Communications [Auk, Vol. 99

Wright (1981:122) statesthat "Borealforest formed CheekBend Cave (40MU261)and the analysisof the a belt perhaps1000 km broad, starting at or near the faunal materialswere made possible. [Laurentide] ice front in the north and ending south near a line extendingfrom central Georgia around LITERATURE CITED the southernend of the AppalachianHighlands to BEtat, A. C. 1938. Life histories of North American Tennessee.... "Recovery of remains of predomi- birds of prey (part 2). U.S. Natl. Mus. Bull. No. nately boreal small mammals from the lower levels 170. of Cheek Bend Cave and those of four avian species BROVICORB,P. 1971. Catalogueof fossil birds: part whose present primary northern ranges coincide, 4 (Columbiformes through Piciformes). Bull. coupledwith pollen analysesdata, suggestthat the Florida State Mus. 15: 163-266. Nashville Basin was characterizedby climatic con- ditions that included somewhat lower annual mean ß 1978. Catalogueof fossilbirds: part 5 (Pas- seriformes).Bull. Florida State Mus. 23: 139-228. temperaturesthan exist today and a climax vegeta- DErCOURt,H. R. 1979. Late Quaternary vegetation tion dominated by coniferousforest. Although the history of the eastern Highland Rim and adja- possibility exists that the former presenceof the cent Cumberland Plateau of Tennessee. Ecol. Hawk Owl, BorealOwl, Saw-whetOwl, Gray Jay, and Pine Grosbeak at this cave site was the result of Monogr. 49: 255-280. PARMALEE, P. W., & W. E. KLtPPEL. 1981. Remains southward-wanderingvagrants, the small mammal of the wood turtle Clemmysinsculpta (Le Conte) assemblageand the remains of at least two individ- from a Pleistocenedeposit in Middle Tennessee. uals of A. funereusand P. canadensissuggest a locally Amer. Midi. Natur. 105: 413-416. establishedpopulation of these boreal birds during RIGGINS,J. N. 1977. Fourth Nashville area record of the Wisconsinan. Saw-whet Owl. Migrant 48: 12-13. We wish to thank W. M. Wright for the photo- WRtGHT, H. E., JR. 1981. Vegetation east of the graphsand B. W. Creechfor typing the manuscript. Rocky Mountains 18,000 years ago. Quaternary We gratefullyacknowledge the TennesseeValley Au- Research 15: 113-125. thority under whose auspices(Contracts No. TVA TV-49244A and TVA TV-53013A) the excavation of Received27 May 1981, accepted20 October1981.

Flightlessnessin Flying Steamer-Ducks

PHILIP S. HUMPHREY AND BRADLEy C. LIVEZEY Museumof NaturalHistory and Department of Systematicsand Ecology, University of Kansas, Lawrence, Kansas 66045 USA Steamer-duckscomprise a single, widely distrib- that Flying Steamer-Ducks"are capable of getting uted, flying species,Tachyeres patachonicus (Falk- well into the air within a spaceof 2 meters, even in lands and Fuego-Patagonia),and three flightless a dead calm," and "they rarelydived at the approach species, T. brachypterus(Falklands), T. pteneres of... [Beck's]craft, but preferred to take wing .... (Tierra del Fuego and the southerncoast of Chile), some of them flew heavily, not rising far above the and the recentlydiscovered (Humphrey and Thomp- water, while others mounted high into the air and son 1981) T. leucocephalus(coast of Chubut, Argen- passedquite out of sight." Reynolds(in Lowe 1934: tina). This paper presentsdata on wing loading and 474) wrote that "Tachyerespatachonicus is capableof useof flight for escapeby FlyingSteamer-Ducks and sustainedand rapid if somewhatheavy flight.... evidencethat someindividuals are permanentlyin- At times T. patachonicusmay well be taken for a capableof flight. flightlessbird, and some of the confusiondistin- Flying Steamer-Ducksoccur in freshwater and guishingbetween the two species[patachonicus and marine coastalenvironments in Patagonia,Tierra del pteneres]is probably due to the unaccountablere- Fuego, and the Falkland Islands (Weller 1976, Hum- luctance to rise so often demonstrated by this phrey and Thompson1981). They feed on sessileor species." slow-moving molluscs,crustaceans, and insect lar- During field seasonsin 1979 (Humphrey and Max vae, which they often obtain by diving, using hind C. Thompson)and 1980-1981(Humphrey and Liv- feet and partly spreadwings for propulsion(Murphy ezey), 57 specimensof Flying Steamer-Duckswere 1936:960; Humphrey and Livezey unpubl. data). collectednear Ushuaia, Tierra del Fuego (5-17 No- Weller (1976: 47) noted the "hesitancy of ... vember 1979, 24-27 December 1980, 4-6 January Flying SteamerDucks to fly .... "Murphy (1936: 1981), Puerto Deseado, Province of Santa Cruz (10-16 970), writing of Beck's collecting activities, stated October 1979, 20-26 January1981), and Puerto Melo, April 1982] ShortCommunications 369

TABLE1. Weights, wing areas, and wing loadings of Flying Steamer-Ducksby sex and locality.

Weight Wing area Wing loading (g) (cm2) (g/cm2) Locality and latitude Sex n Mean Range n Mean Range n Mean Range Puerto Melo, Males 1 2,350 -- 1 1,160 -- 1 2.03 -- 44001 ' Females 0 -- -- 0 -- -- 0 -- -- Total 1 2,350 -- 1 1,160 -- 1 2.03 -- Puerto Deseado, Males 16 2,994 2,350-3,350 11 1,307 1,076-1,422 11 2.29 2.03-2.53 47o46' Females 18 2,416 1,950-2,900 14 1,207 1,090-1,290 14 1.95 1.65-2.35 Total 34 2,688 1,950-3,350 25 1,251 1,076-1,422 25 2.10 1.65-2.53 Ushuaia, Males 10 3,155 2,800-3,480 10 1,272 1,156-1,358 10 2.49 2.21-2.83 54048' Females 10 2,440 2,150-2,650 8 1,243 1,124-1,369 8 1.98 1.75-2.13 Total 20 2,798 2,150-3,480 18 1,259 1,124-1,369 18 2.26 1.75-2.83 All localities Males 27 3,029 2,350-3,480 22 1,284 1,076-1,422 22 2.37 2.03-2.83 Females 28 2,425 1,950-2,900 22 1,220 1,090-1,369 22 1.96 1.65-2.35 Total 55 2,722 1,950-3,480 44 1,252 1,076-1,422 44 2.17 1.65-2.83

Province of Chubut (29 September1979), Argentina. information on wind conditions at the time of col- We recorded details of escapebehavior while each lection for 21. Each specimen was weighed within bird was being pursued during 1980-1981;limited 2 h following collection, and additional data were data of the same kind were recorded during 1979. recorded, including: outline of fully extended wing We collected44 Flying Steamer-Ducksfor which (Raikow 1973:297), weight of contentsof esophagus wing-loadingdata couldbe recorded;for 30 of these, and gizzard, sex, gonad measurements,weight of we recordedwhether or not they flew, and we have pectoral musculature,and molt. Wing area for each

TABLE2. Mean weights, wing areas,and wing loadingsof 22 speciesof diving-ducks.Sources: (1) Hartman 1961, (2) Magnan 1912, (3) Magnan 1913, (4) Magnan 1922, (5) Meunier 1951, (6) Mtillenhoff 1885, (7) Poole 1938, (8) Raikow 1973.

Wing Weight Wing area loading Species n (g) (cm2) (g/cm2) Source Aythya affinis 20 615 474 1.30 1, 7, 8 A. collaris 3 757 460 1.65 7 A. ferina 1 842 615 1.37 6 A. fuligula 1 741 474 1.56 6 A. marila 1 675 621 1.09 6 A. nyroca 2 510 577 0.88 3, 6 A. valisineria 1 910 701 1.30 8 Bucephalaalbeola 6 393 341 1.16 7, 8 B. clangula 5 748 539 1.40 4, 6, 8 Clangulahyemalis 2 980 551 1.78 6, 7 Melanitta fusca 4 1,320 866 1.52 2, 5, 8 M. nigra 1 870 679 1.28 4 M. perspicillata 1 827 649 1.27 8 Lophodytescucullatus 1 630 401 1.57 Presentstudy Mergus albellus 1 495 431 1.15 4 M. merganser 1 1,470 853 1.72 4 M. serrator 1 818 589 1.39 4 Oxyura dominica 7 363 352 1.03 1 O. jamaicensis 13 530 331 1.60 7, 8 Somateriafischeri 1 1,457 719 2.03 8 S. mollissima 2 2,056 1,210 1.79 5, 8 S. spectabilis 1 1,700 824 2.06 8 370 ShortCommunications [Auk, Vol. 99

.o-39- •ID =patachomcus WingLoading = 2.5 g/cm2•/// / • j 38- •2,=leucocephalus ((•O/// • 37- •3= pteneres ß O O /• A=21.07 +1.03W ß•1•ß O•^ • • 3635-- •=0.35 • -•-•. / •u --• •

• 31- / • / • 3o- / D / / m 29- /

Weighty• (g) :W •. •. Re•ess•o• o• cube•oot o• body weight •d s•ua•e •oot o• w• a•ea •o• • •]y• Steame•-Duc•s. Ope• symbols•ep•ese•t ma]es•b]ac• symbols•ema]es. Data •o• •ht]ess 7. •o•ph• (•5) a•d 7. p•n•r• (•5) were •ot used J• ca]cu]atin• the •e•ess•o• e•uadom Note that• except•o• •o• •em•e/•o•p• • points •o• the •o specieso• •htless steame•-duc•s •a• to the •ht o• the dashed ]•e •ep•esem• the th•eshoM o• •ht]ess•ess a•d have w• a•eas below those p•ed•ctedby the •e•ess•o• e•uado• calc•ated

specimen was measured using a planimeter and for Flying Steamer-Ducks(Fig. 1). In addition, wing doubled to give total wing area. Wing loading in loadings of flightlesssteamer-ducks are for the most g/cm2 was calculatedby dividing totalweight by total part much higher; wing-loading values of heavy wing area. male T. patachonicusoverlapped those of female T. Wing loadings of 44 Flying Steamer-Ducksaver- leucocephaluswith low body weights (Fig. 1). aged 2.17 g/cm2 (Table 1). Basedon major reviews of Weights and wing loadings of Flying Steamer- wing loading (MQllenhoff 1885, Banks 1933, Poole Ducks differed between sexesand among the three 1938, Savile 1957, Hartman 1961, Greenewalt 1962, localities of collection (Table 1). Males averaged Raikow 1973),Flying Steamer-Ducks have amongthe heavier than females at Puerto Deseado (t = 6.19, highest wing loadingsrecorded for flying birds. P < 0.001) and Ushuaia (t = 7.94, P < 0.001). Ex- The relationship between means of wing area and treme weights of males and females overlapped at body weight in 23 speciesof diving-ducks,including Puerto Deseado but not at Ushuaia (Table 1). Weller T. patachonicus(Tables 1 and 2), is expressedin the (1976:47) found no overlap in weight between sexes power equation: of three speciesof steamer-ducks.Males also exceed- ed females in wing area at Puerto Deseado (t = 2.88, Wing Area = 7.36 Weight 0.6•(R 2 = 0.85, n = 23) P < 0.005)but not at Ushuaia(t = 0.87, P < 0120). and thus approximates the expected, wherein the Also, male Flying Steamer-Ducks were heavier (t = allometric coefficient is 0.67, given that wing area 1.75, P < 0.05) at higher-latitude Ushuaia than at increasesas the squareand body weight as the cube. PuertoDeseado, in agreementwith Bergmann'sRule The relationship between wing area and body (James 1970, McNab 1971); females, however, were weight for Flying Steamer-Ducksis negativelyallo- similar in weight at the two localities (t = 0.24, metric: P > 0.40). Wing Area = 147.4Weight 0.27(R s = 0.36, n = 44). Of 22 malespecimens (not in wing molt) collected Wing areas of two speciesof flightlesssteamer- in 1979 and 1980-1981, six had wing loadings of 2.5 ducks (T. pteneres,T. leucocephalus)are below those g/cm• or greater, the limit beyond which Meunier expectedfrom the linear regressionof the squareroot (1951: 430) estimated aerial flight to be impossible. of wing area on the cube root of weight calculated The highest wing loadings of birds known to have April 1982] Short Communications 371

flown were 2.25 and 2.43 g/cm2;both birds were Depending on the availability of wind, Flying males and were collectedon windy days. We col- Steamer-Ducksmay be reluctant to fly with wing lected sevenbirds, all males, with wing loadingsof loadings as low as 2.3 g/cm'; they are unable to fly 2.25 g/cm2 and greaterthat swam, steamed,or dived with wing loadings of 2.5 g/cm• and above. Wing but did not fly when pursuedby a noisy motorboat. loadings and incidence of flightlessnessin Flying Two of the birds had slight amounts(34 g, 35 g) of Steamer-Ducksare higher in malesand in birds from food in esophagusand gizzard and the rest only higher latitudes. clean grit (average15 g, range 9-22 g). Of the latter, We are now investigating factors that affect two had wing loadings above 2.6 g/cm2;both had weight, wing area, and wing loading of steamer- little fat, flightworthyremiges, and pectoralmuscu- ducks and testing hypothesesconcerning the mor- lature at or near maximum weight (17% of body phology, adaptive significance, and evolutionary weight;pers. data). One was collectedon a calmday origins of flightlessnessin Tachyeres. and the other in a strongwind when other steamer- This study was supported by NSF grant DEB- duckswere flying. We believe these birds were in- 8012403to P.S. Humphrey and by the KansasUni- capable of flight. versity Endowment Association, Southwestern Col- Assumingthe validity of Meunier's (1951)flight- lege, Prof. Max C. Thompson,R. Hamilton, Dr. and lesshessthreshold of 2.5 g/cm2, 50% of the nonmolt- Mrs. W. Saul, Mr. and Mrs. L. A. Osborne, Dr. T. ing maleswe collectednear Ushuaiawere flightless, Mastin, Dr. R. T. Peterson,and the Humphrey fam- as were 8% of those collected near Puerto Deseado. ily. We thank B. Mayer, Mr. and Mrs. F. V. T. J. The higherincidence of flightlessmales in the south Fauring, Dr. O. Kuhnemann, P. Medina, and J. Sesti is related to greater body weights and resultant for warm hospitality and facilities for fieldwork; Dr. higher wing loadings.Nevertheless, at both locali- J. M. Gallardo, Dr. J. Navas, R. A. Bockel, and their ties most male Flying Steamer-Duckswith wing colleaguesin the Museo Argentino de Ciencias Nat- loadings below 2.5 g/cm• are closeto the threshold urales "Bernardino Rivadavia" for generous assis- of flightlessness(Fig. 1). tance; Dr. E. O. Gonzales Ruiz and his colleagues, We weighed the contentsof gizzard and esopha- Direcci6n Nacional de Fauna Sylvestre,Ing. Qca. L. gus of all specimenscollected in 1980-1981; maxi- O. Saigg de Chialva, Directora de Protecci6n Am- mum weight of gizzard and esophagealcontents of biental, Provincia de Chubut, and Dr. D. H. Soria, Flying Steamer-Duckswas 176 g, and mean weight Director General de Ganaderia, Provincia de Santa (n = 32) was32 g. Becauseour measureof weight of Cruz, for issuingappropriate permits; Dr. A. Tarak ingestedfood did not include intestinalcontents, the and his colleagues,Direcci6n Nacional de Parques maximal weight of food ingestedper feeding bout Nacionales, Dr. G. A. Giaroli, Subsecretario de Re- is probably twice (or more) the 176 g measuredfor cursosNaturales Renovables y Ecologiade la Naci6n, contentsof esophagusand gizzard, or 350+ g. and F. Lobbe and F. Villar, Ceremonial, Ministerio Sixteen male Flying Steamer-Duckshad wing de Economiafor assistancein making arrangements loadings of less than 2.5 g/cm•. We calculatedthe for fieldwork in Argentina; Mr. and Mrs. R. Gibson weight increase needed for each of these birds to and family, Dra. B. de Ferradas,Prof. L. Orquera, havea wing loadingof 2.5 g/cm2. The meanincrease Lic. E. Piana, Lic. G. Piacentinofor warm hospitality required was 304 g, and for 11 birds the increase and assistanceof many kinds; W. Conway, R. Stra- needed was lessthan 350 g. Therefore,we hypoth- neck, F. Erize, P. Canevari, M. A. E. Rumboll for their esize that approximately64% of male Flying Steam- interestin our researchand sharingtheir knowledge er-Ducks at Puerto Deseado and 80% at Ushuaia that and impressions about steamer-ducks; Dr. F. B. are not already functionallyflightless for other rea- Crossand staffof KUMNH for their help;J. Rawlings sons(wing molt, wing loadingof 2.5 g/cm• or greater) for typing and other help; and D. Bennett for draw- could become temporarily flightless because of ing the figure. weight gained from ingestedfood. A strategythat would avoid substantialtemporary increasesin weight from feeding would be to eat BAui•s,E. 1933. The relation of weight to wing area small amountsof food at any one time and to feed in the flight of animals.J. Malayan Brit. Asiatic often during the day. Weller (1976:49) stated that Soc. 8: 334-360. "flightlessforms [pteneres,brachypterus] take large GREEUEWALT, C. H. 1962. Dimensional relation- food items in brief but intense feeding periods, ships for flying animals. Smithsonian Misc. whereas Flying SteamerDucks seem to feed more Coil. 144: 1-46. continuously on smaller items." HARTMAU, F. A. 1961. Locomotor mechanisms of We concludethat the thresholdof flightlessnessin birds. Smithsonian Misc. Coll. 143: 1-91. FlyingSteamer-Ducks is at wing loadingsof approx- HUM•,H•¾, P.S., & MAx C. THo•'sou. 1981. A imately2.5 g/cm2. Roughly25% of maleT. patachon- new speciesof steamer-duck(Tachyeres) from icusexceed this thresholdand approximatethe wing Argentina. Univ. KansasMus. Nat. Hist. Occas. loadingsof the lightestof their flightlesscongeners. Pap. 95: 1-12. 372 Short Communications [Auk, Vol. 99

JAMES,F. C. 1970. Geographicsize variation in birds MEUtqIER,K. 1951. Korrelation und umkonstruktion and its relationship to climate. Ecology 51: in dengr6ssbenbeziehungen Zwischen vogeltiff- 365-390. gel und vogelk6rper.Biol. Gen. 19: 403-443. LowE, P. R. 1934. On the evidence for the existence M/3LLE•HOFF,K. 1885. Die grosseder flugfluchen. of two speciesof steamerduck (Tachyeres),and Pflueger'sArch. Ges. Physiol.35: 407-453. primary and secondaryflightlessness in birds. MURPHY, g. C. 1936. Oceanic birds of South Amer- Ibis 4: 467495. ica, vol. 2. New York, Amer. Mus. Nat. Hist. MAGtqAtq,M. A. 1912. Modifications organiques PooLE,E. L. 1938. Weights and wing areasin North cons•ctives chez les oiseaux • l'absence de vol. American birds. Auk 55: 511-517. Bull. Mus. Hist. Nat. Paris 18: 524-530. RAIKOW, g. J. 1973. Locomotor mechanisms in --. 1913. Rapport de la surfacealaire avec le North American ducks. Wilson Bull. 85: 295-307. poids du corpschez les oiseaux.Bull. Mus. Hist. SAWLE,D. B. O. 1957. Adaptive evolution in the Nat. Paris 19: 45-52. avian wing. Evolution 11: 212-224. --. 1922. Les caracteristiquesdes oiseaux sui- WELL,R, M. W. 1976. Ecology and behaviour of vant le mode de vol. Ann. Sci. Nat. (Ser. 10) 5: steamer ducks. Wildfowl 27: 45-53. 125-334. Received24 August1981, accepted 4 November1981. McNAI•, B. K. 1971. On the ecologicalsignificance of Bergmann'srule. Ecology52: 845-854.

Observationson migratoryTurkey Vultures and LesserYellow-headed Vultures in northern Colombia

FRIEDEMANN KOESTER1 Institutode Investigaciones Marinas "Punta de Betin," Santa Marta, Colombia,South America

TurkeyVultures (Cathartes aura) breed from south- somebirds of the flockhad conspicuousorange-yel- ern Canadato Tierra del Fuego,while LesserYellow- lowish heads and dirty-whitish wing patches that headed Vultures (Cathartes burrovianus) occur from identified them as Lesser Yellow-headed Vultures. M•xico through Central America to northern Argen- The resultspresented here are basedmainly on ob- tina and Uruguay (Meyer de Schauensee1966). It has servationsand line-transect counts made along the long been known that Turkey Vultures breeding in Caribbean coastbetween the ports of Barranquilla the northern parts of their rangeare migratory, and and SantaMarta in northern Colombia.They give a data are available on their arrival at or departure first accountof the magnitudeand ecologicalaspects from their breeding groundsin spring or fall (Bent of vulture migrations in South America. 1937). Likewise, parts of their migration routesand Countsof vultureswere madeby driving alongthe winter ranges in Central America are well docu- Troncal del Caribe highway between Puente de la mented (Chapman1933, Wetmore 1965, Smith 1980). Barra, near the town of Cienaga, and Los Cocos,near Migrating TurkeyVultures were observedin north- the port of Barranquillaon the MagdalenaRiver. This western Colombiaby Dugand (1947),Haffer (1959), area includes most of Isla de Salamanca National and Olivares (1959), but there are no data on their Park. A detailed descriptionof this region, including abundanceor ecology. Becauseseasonal changes in a portrayalof its flora and fauna, is found in Franky their abundance in Panam• have been observed, it and Rodriguez (1976). Visibility along most of the is believed that the northern Lesser Yellow-headed highway was not impaired by vegetation, so nearly Vultures are also migratory (Ridgley 1976). all roosting or soaring vultures could be counted A tight flockof at least150 soaring Turkey Vultures within a broadcorridor of up to morethan 500m on was observed on 14 October 1978, some 16 km to the each side of the highway. Twenty-eight line tran- east of the town of Barranquilla, Colombia. Such an sects,each coveringa distanceof 44 kin, were made unusuallylarge number suggestedmigrating birds. between 1000 and 1500 from 17 October 1978 to 17 A closer view confirmed this; none of the vultures January1980. One persondrove, holding the speed showedthe whitish or dull yellow bandsacross the at a constant70 km, while a secondperson, using a backs of their necks that characterize the resident hand-operatedcounter, counted vultures with the race, Cathartesaura ruficollis.Although similar in nakedeye. Due to similaritiesin their plumagecol- generalappearance and size to the Turkey Vulture, oration and silhouettes,Turkey Vultures and Lesser Yellow-headed Vultures, when soaring or roosting, were difficult to separate at a distance. Separate Present address: Charles Darwin Research Station, CasiIla 58• 39, counts for resident Turkey Vultures and the race of Guayaquil, Ecuador,South America. migratingTurkey Vultureswere even more difficult. April 1982] ShortCommunications 373

TABLE 1. Date, relative abundance, and average number of vultures (C. aura and C. burrovianus) observed per line-transectcount from 17 October 1978 to 17 January 1980.

Number Num- 2 of indi- TotalJ ber of indi- Date viduals month counts viduals

17 Oct 1978 132 20 Oct 1978 137 353 3 117.7 23 Oct 1979 84 3 Nov 1978 210 4 Nov 1978 162 741 4 185.3 15 Nov 1978 117 26 Nov 1978 252 7 Dec 1978 154 14 Dec 1978 165 458 3 152.7 21 Dec 1978 139 8 Jan 1979 174' 17 Jan 1980 92 266 2 133.0 16 Feb 1979 47 Fig. 1. Yearly distribution of relative abundances 22 Feb 1979 100 147 2 73.5 16 Mar 1979 39 39 1 39.0 of Cathartesvultures in the study area, by months (J = January), from 28 line-transectcounts 17 Octo- 24 Apr 1979 27 27 1 27.0 4 May 1979 19 ber 1978 to 17 January 1980). Black areas indicate 15 May 1979 16 35 2 17.5 migration period. 3 Jun 1979 19 19 1 19.0 12 Jul 1979 24 20 Jul 1979 21 45 2 22.5 gration period, compared to a smaller proportion 2 Aug 1979 8 during the restof the year. Therefore,it appearslike- 5 Aug 1979 19 66 4 16.5 ly that migratory populationsof C. burrovianusalso 14 Aug 1979 24 wintered in the area. 20 Aug 1979 15 Blake (1977and pers. comm.) identifiesmigrating 4 Sep 1979 83 Turkey Vultures in Colombia, including the Santa 18 Sep 1979 72 223 3 74.3 26 Sep 1979 68 Marta region, as belonging to the race C. aura meri- dionalis.Indeed, basedon the known rangesof the races that could be possiblemigrants to the study area and on the measurementsof a female specimen collectedon 7 January1979, the Turkey Vultures re- The characteristicyellowish nape of resident birds ported here as wintering birds may well have be- was visible only at closerange and when birds were longed to this race rather than to C. aura septentrio- sitting or flying low. Becauseof theseuncertainties, nalis, as was assumed by Meyer de Schauensee line-transectcounts of residentand migrant Turkey (1964). The specimen,collected at PozosColorados, Vultures and Lesser Yellow-headed Vultures made Santa Marta, was cataloguedunder Nr. IND-A-3186 throughout the study were lumped into the single at the collectionof the Laboratoriode Investigacion categoryof Cathartesvultures. The only othervulture de Fauna Silvestre, Div. de Fauna Terrestre, INDER- of this genus, the Greater Yellow-head (C. melam- ENA, Bogota, D.E., Colombia. It measures as fol- brotus), is an Amazonian bird whose occurrence in lows: wing length (flattened),535 mm; tail length, the study area is highly unlikely. 270 mm; tarsus, 63.5 mm; culmen (from cere), 24.5 The abundance of vultures varied seasonally, in mm; body weight, 1,420 g; ovary small, in nonre- accordancewith the general migration pattem for productive state. The nominate race, C. aura aura, so nearctic migratory bird species (Table 1, Fig. 1). far reportedas a winter migrant only as far south as Thus, it is almostcertain that flocksof Turkey Vul- Panamfi,and C. aurajota, essentiallya highland bird tures observedalong the ColombianCaribbean coast not reported north of central Colombia, can both during the northernfall, winter, and spring seasons probably be ruled out as being involved in these were migrating or wintering birds. In addition, migrations(Blake pers. comm.). With respectto the Lesser Yellow-headed Vultures were observed in the Lesser Yellow-headed Vultures that were observed study area in greaternumbers during the migration to winter togetherwith the Turkey Vultures, it seems period, often mixed in with the flocksof Turkey Vul- reasonableto assign them to the nominate race C. tures. Roughly, 1 LesserYellow-headed Vulture to burrovianusburrovianus, which rangesfrom southern every 5-10 Turkey Vultures was seen during the mi- M6xico to Colombia. Previous records in northern 374 ShortCommunications [Auk, Vol. 99

TA•L• 2. Densities (individuals/km) of Cathartes vul- of only 30-50 m. During the first observationsin tures in the study area; 28 line-transect counts October 1978, the flocks of migrants roused in this along 44 km of coastline each for a total of 1,232 way had a rather spectaculareffect upon the local km; 2,419 vultures counted. BlackVultures (Coragypsatratus). Soon after the first TurkeyVultures had begunto soarabove their roost- Num- ing place,Black Vultures came in from all sides.The ber Number of of indi- speedof their flight and the way they flew in on a Period counts km viduals Density straight line to wheel about in the cloud of roused migrants suggestedthat they came to find carrion. Migration 18 792 2,227 2.8 On suchan occasionon 17 October,several groups Nonmigration 10 440 192 0.4 of two and three King Vultures (Sarcoramphuspapa), Totals 28 1,232 2,419 (• = 1.6) almost never seen in this coastal area, were observed flying high above. Previousobservations of the be- havior of King, Black,and Turkey vulturesfeeding on carrion have shown (Koester and Koester-Stoew- esand 1978)that Turkey Vultures in northern Colom- SouthAmerica for the largerC. burrovianusurubutin- bia invariably are the first to appearat the carcass, ga include only April through October, suggesting attractingthrough their presenceBlack Vultures that migration from its southern range (Blake 1977). On in turn may lure King Vultures to the site. It seems 10 January1979 a specimensoaring with a flock of possible,therefore, that the latter two speciesmis- five wintering Turkey Vultureswas shotat the same interpreted the unusuallylarge flock of soaringmi- placewhere the Turkey Vulture describedabove had grantsas an indication of the presenceof food. previously been collected. It measured as follows: It seemsgenerally acceptedthat migrants in the wing length (flattened),495 mm; tail length, 255 mm; tropics harvest superabundantand/or sporadically tarsus,58.0 mm; culmen (from cere), 23.6 mm; body available resourcesand are usually subordinate to weight, 1,200 g; sex undetermined.Preservation of the residentspecies (Karr 1976). This may apply only this specimenwas not possible. in part at the subspecieslevel, when one considers If the data from the line transects are subdivided wintering versus resident Turkey Vultures. Al- into numbers of Cathartesvultures seen during the though no detailed data on this subjectwere gath- migration and nonmigration periods (according to ered, the amount of what appearedto be suitable the presenceor absenceof groupsof 10 or more Tur- food for the localTurkey Vultures did not seemto be key Vulturesthat soaredor roostedtogether without more abundant during the wintering period com- obvious presenceof food: black areasin Fig. 1), then pared to the rest of the year. Thus, if a relatively the relative abundance of vultures observedduring constantavailability of food permitted a relatively the migration period from Septemberto March (;?= stablepopulation of C. aura ruficollisto exist in the 123.7 individuals per transect count) was signifi- study area, the presenceof additional Turkey Vul- cantly higher than during the nonmigrationperiod tures, C. aura meridionalis,with presumablysimilar from April to August (• = 19.2 individuals per tran- feeding habits during the northern winter, should sect count):[ = 8.00• 2.86= t19; P • 0.01. The have resultedin a generalfood shortageand com- density of Cathartesvultures, expressedas the num- petition between these two races.Indeed, resident ber of individuals per kilometerof line transect,was Turkey Vultureswere never observedto sharea car- seven times higher (2.8 individuals/km) during the casswith their wintering conspecifics,which ap- migration period than during the nonmigrationpe- peared to be more powerful flyers and generally riod (0.4 individuals/km) (Table 2). strongerbirds. In a remarkablecontrast to the resi- Accordingto Smith (1980),migrating Turkey Vul- dents, migrants seemed to cope easily with the tures in Panam• tend to abstain from food. In this agressive feeding behavior of the Black Vultures. study, however,Turkey Vulturesthat had arrived in This was observedon 15 November1978, when eight great numbers were frequently observed feeding. migrants and two Black Vultures fed on the carcass Therefore,it may be inferred that they acceptedthis of a dog near the village of Tasajeras,Isla de Sala- area as their winter home and did not extend their manca. migration farther to the south. For severaldays after During the migrationperiod, residentTurkey Vul- their arrival, flocks of 100 and more individuals were tures were more numerous in certain locations in observed soaring and roosting jointly. Thereafter, which they had not been as frequentlyencountered these large flocks dissolvedinto smaller groups or during the rest of the year. At times, in Tayrona single individuals. Especially when roosting in NationalPark to the eastof SantaMarta, for example, groups, wintering vultures took flight readily at the the number of C. aura ruficollisnearly equaledthe approach of an observer even when he was still numbers of the common Black Vultures. It may be 100-200m away,while TurkeyVultures flushed dur- inferred from this that the wintering race displaced ing the nonmigrationperiod, i.e. residentbirds, usu- a certainpercentage of the residentrace to other hab- ally took flight when an observerwas at a distance itats. Bent (1937)wrote that Turkey Vultures are ef- April1982] ShortCommunications 375 fectivein cleaningstreets, garbage piles, and dump ticular interest. Becausevultures are large, conspic- heaps in cities and are often seen feeding on the uous birds whose feeding habits can be studied main highwaysin Floridawhere snakes, turtles, and without too muchdifficulty, studiesof their behavior small mammals have been killed by passing auto- and ecologycould contributesubstantially to our mobiles or where fish have been thrown away by presentknowledge of the ecologyof bird migration. fishermen.It is noteworthy that, in contrastto this I am indebted to my wife, Heide, for her constant statement,resident Turkey Vultures were virtually assistance with the transect counts. Dr. Emmet R. never seen feeding on garbageheaps and strictly Blake, Field Museum of Natural History, Chicago, avoided towns and cities where Black Vultures are and Dr. Michael K. Rylander, TexasTech. Univer- abundant.But, like Turkey Vulturesin Florida, C. sity, Lubbock, made valuable suggestionsand as- auraruficollis feeds on road kills and smallercarcasses sistedwith the Englishtext, respectively. alongthe highway in the study area. Comparedto residents,therefore, wintering Turkey Vultures ap- LITERATURE CITED pearedto exploita wider spectrumof food, as they Bœ•qT,A. C. 1937. Life histories of North American were observedfeeding on garbagenear Cienaga as well as on road kills. birds of prey. U.S. Natl. Mus. Bull. No. 167: 1-44. Seventimes as many Cathartesvultures existed in BLAKœ,E. R. 1977. Manual of neotropicalbirds. Chi- the studyarea during the migrationperiod as during cago,Univ. ChicagoPress. the restof the year (Table2). Perhapsthe exploitation CHAPMA•q,F. M. 1933. The migrationof Turkey Buz- of a wider rangeof food sources,comprising those zards as observed on Barro Colorado Island, Ca- fed upon by the local Turkey Vultures as well as nal Zone. Auk 50: 30-34. those used by the local BlackVultures, enableswin- DUGA•qD,A. 1947. Aves del Departamentodel At- teringTurkey Vultures to coexistwith both the res- lantico. Caldasia 4: 499•548. ident race and the Black Vultures successfully.No FRANKY,S. M., & P. I. RODRIGUEZ.1976. Parque such observationswere possiblein the case of the Nacional Isla de Salamanca. Bogota, INDER- LesserYellow-headed Vultures. Throughout the year ENA. they were seenmore frequently in the wetlandsand HAFFER,J. 1959. Notas sobre las aves de la region mangroveswamps south and north of the Troncal de Uraba Lozania 12: 1-49. del Caribehighway, where the majorityof the birds KARR,J. g. 1976. On the relative abundancesof mi- believed to be wintering were observed. grantsfrom the northtemperate zone in tropical The breedingseasons of the BlackVultures and habitats. Wilson Bull. 88: 433-458. Lesser Yellow-headed Vultures in the study area KOESTER,F., & H. KOESTER-STOEWESAIqD.1978. partly coincidewith the migration period of Turkey and Lesser Yellow-headed vultures from the north. Koenigsgeier--Beobachtungenim Tayrona-Na- While more detailed information on season and tionalparkim Norden Kolumbiens,Suedameri- ka. Zeitschr. Koelner Zoo, 21: 35-41. breeding biology of the BlackVulture in the area is --, & ---. In press.Zur Brutbiologiedes Ra- available (Koesterand Koester-Stoewesandin press), bengeiers,Coragyps atratus (Bechstein 1793), in our presentknowledge of the breedingseason of the Nordkolumbien, Suedamerika. Bonn. zool. LesserYellow-headed Vulture is limited to a single Beitr. observationof an adult accompaniedby two recently MEYERDE SCHAUENSEE,g. 1964. The birds of Co- fledged young on 11 May 1979 in the mangrove lombia. Narberth, Pennsylvania,Livingston. swampssouth of the CienagaGrande de SantaMar- --. 1966. The speciesof birds of SouthAmerica ta, known as the natural reserve area E1 Santuario. with their distribution. Acad. Nat. Sci. Phila- Nothing so far is known of the annualcycle of the delphia. residentTurkey Vulturesand how their cyclesmight OLiVAReS,A. 1959. Aves migratoriasen Colombia. comparewith those of the migrants. According to Rev. Acad. Col. Ciencias, vol. 10, No. 41: Smith (1980), however, Turkey Vultures in PanamPi 371-442. commenceegg laying in Februarythrough early RIDtrEY,R. S. 1976.A guideto the birdsof Panama. April. A similartiming of the breedingseason for C. Princeton,New Jersey,Princeton Univ. Press. auraruficollis in the studyarea is likely. Thus, during SMITH, N. G. 1980. Hawk and vulture migration in the northern winter, Turkey Vultures along the the neotropics,Pp. 51•55 in Migrant birds in the northerncoast of Colombiaare likely to be separated neotropics(A. Keast and E. S. Morton, Eds.). into breeding residentand nonbreedingwintering Washington,D.C., SmithsonianInst. Press. races. WœtMOR•,A. 1965. The birds of the Republic of A more extendedecological study of the relation Panama. Washington, D.C. SmithsonianMisc. of the migratoryvultures with the residentstreated Coil. No. 150. hereinand the resultinginter- and intraspecificcom- petition for food and its possibleeffects on the Received7 November1978, accepted7 October1981. breedingsuccess of the residentsshould be of par- 376 ShortCommunications [Auk, Vol. 99

The Noff_hernLimit of the HummingbirdGenus Oreotrochilus in South America

FERNANDO I. ORTIZ-CRESPO 1 AND ROBERT BLEIWEISS 2 •Institutode Ciencias,Universidad Catdlica, Quito, Ecuador,and 2Museumof ComparativeZoology, Harvard University,Cambridge, Massachusetts 02138 USA The hummingbird genus Oreotrochilusis repre- logical barriers, the range extensionwe record here sentedby populationsliving at high altitudesin the is of biological significance. South American Andes from north-central Ecuador These new recordspermit us to emphasizeseveral to south-centralChile and adjacentArgentina (Ber- factsabout the biogeographyof Oreotrochilusin gen- lioz and Jouanin 1942, Zimmer 1951, Peters 1945, eral and of the Ecuadorianpopulations in particular. Carpenter 1976,Olrog 1978). Ecuadorianpopulations First, membersof Oreotrochilusseem to occurin dry constitute a morphologicallydistinct, allopatric tax- puna (cf. Vuilleumier and Simberloff1980) or puna- on, which is related to the northern-most Peruvian like grasslandhabitat through much of their range. and Bolivian species,O. estella,but is characterized Our new recorddocuments a parallelingbetween the by largely blue-headedadult malesand is often treat- distribution of Oreotrochilus and these habitats, be- ed as a separate species, Oreotrochiluschimborazo causethe Cotacachi-YanaUrco region is coveredby (Vuilleumier and Simberloff 1980, but see Zimmer dry, puna-like grassland.It should be pointed out, 1951: 40). The northern-most of the three Ecuadorian however, that Oreotrochilusapparently does occur in O. chimborazotaxa is assignedto O. c. jamesoni(Jar- a variety of habitats, from dry, brushy slopesto wet- dine 1849), previously recordedin the western Cor- ter pararno, elsewherein Ecuador (Vuilleumier 1967), dillera as far north as Mt. Pichincha and in the east- so grasslandhabitat may not be a necessarycondi- ern Cordillera as far north as the Guamani Pass,just tion for Oreotrochilus at its northern limits. Further- north of Mt. Antisana (Orces 1944, Peters 1945, Cor- more, the grasslandsof Cotacachi-YanaUrco are fre- ley-Smith 1969, Vuilleumier 1976). We here report a quently burned and heavily grazed by cattle northward range extensionof birds presumed to be throughout the year, and, therefore, it is difficult to of the jamesoniform, a presumptionbased upon two say whether the presenceof Oreotrochilusrepresents independent setsof visual observationsin the head- a recent invasion due to the creation of favorable waters of the Rio Pantavi, Imbabura Province, in the conditions by man or is part of the historic pattern Mt. Cotacachi-Mt. Yana Urco de Piti fin portion of the of differentiation of O. chimborazo in Ecuador. The western range, about 70 km north of Mt. Pichincha presenceof O. c. jamesoniat the Guamani Pass, the (see Fig. 1). northern-most record for Oreotrochilus in the eastern Ortiz-Crespo visited the area on 22 December1975 Cordillera of Ecuador, has been suggestedto result and saw a femaleOreotrochilus feeding at the flowers from the creation of favorable habitat due to human of an isolatedthicket of Chuquiragainsignis HB (Com- disturbance(Corley-Smith 1969). The Guamani Pass positae) amid clumps of pararno grassesat an ele- is an exceedinglywet site (Vuilleumier 1976), which vation of 3,600m. A secondvisit to the area by Blei- might be unfavorablehabitat for Oreotrochilusunder weiss on 23 June 1980 documented the presence of natural conditions. Similarly, in the high-altitude a population, which was then tentatively assigned regions immediately to the north of the presently to jamesoni;at that time two recordsof adult males, revisedwestern portion of the range, rainfall appears both showing the all-blue head characteristicof the to be much greater than within the range (Ortiz- taxonjamesoni, as well as two of females,were made Crespo 1974). This additional correlation of Oreotro- at elevationsof 3,550-3,650 m in open or protected chilus distribution with climatic factors in western stream ravines where Chuquiragawas common. Ecuador,in light of the generalecology of the genus, In Ecuador, Oreotrochiluspopulations are found suggeststhat Oreotrochilusis not a recentinvader to only above an elevation of 3,500 m up through the the Cotacachi-YanaUrco region. highest vegetatedzones to the snow line. North of Also of interest is the co-occurrenceof the plant Mt. Pichincha, the western Andes are insular with Chuquiragaand Oreotrochilus.Ortiz-Crespo has nev- respect to appropriate Oreotrochilushabitats. The er observedthe birds feedingfrom plantsother than deep, arid gorgecut by the Rio Guayllabambaeffec- Chuquiragainsignis in Ecuador, and the availability tively isolates Pichincha from the Cotacachi-Yana of this plant may be requisite for the presenceof Urco massif to the north, which in turn is separated Oreotrochilusin the Ecuadorianpart of its range. A from the high mountains farther north by the course high relianceof Oreotrochiluson Chuquiragais cor- of the upper Rio Mira running roughlyparallel to the roborated by several other investigators. Corley- Rio Guayllabamba and from high elevations to the Smith (1969)also notes encounteringO. chimborazo east by an extensionof the inter-Andean valley run- only in the presenceof Chuquiraga,and Carpenter ning north-south in northern Ecuador (see Fig. 1). (1976) found, for some Peruvian populationsof O. Thus, in terms of both horizontal distance and eco- estella,an exclusivedependence on Chuquiragaspi- April 1982] ShortCommunications 377

80 ø 79 ø 78 ø 77 ø

COLOMBIA

ECUADOR

0oI Mt Pich[n C

Fig. 1. Map of north-centralEcuador showing localities mentioned in the text. noza as a native nectar source during the winter, this species(Ortiz-Crespo 1974). Both occur in drier nonbreeding season.Curiously, there are no Oreo- habitats,however, and may be similarlylimited fur- trochilusrecords from Mt. Cayambe,the next major ther northward by the increasedrainfall in northern- peak north of Mt. Antisana in the easternCordillera, most Ecuador. although Chuquiragaoccurs there (L. Holm-Nielson There is no reason to believe that the currently pets. comm. 1980). However, Oreotrochilusis also revised northern limit of Oreotrochilus is definitive. adapted in roosting, breeding, and other feeding The rangeextension we recordhere, however, high- habits to the vegetation and physiographicfeatures lightssome of the majorfeatures in the geographical of drier, high-altitude habitats (Carpenter1976), and ecologyof this Andeanhummingbird. these factors may also influence distribution. The We would like to thank Dr. Francois Vuilleumier apparent absenceof Oreotrochilusfrom the relatively for a criticalreading of a draft of the manuscript.R. undisturbed wet paramos of Mt. Cayambe may be Bleiweisswould like to acknowledgethe generous due to these factors. support of the World Wildlife Fund U.S. for his field- Our record of Oreotrochilus in the Cotacachi-Yana work in Ecuadorand the Departamentode Parques Urco area also points out a close parallel in the ap- Nacionalesy Vida Silvestre, Direccion General de parent northern terminus of this widespread genus DesarrolloForestal of Ecuadorfor logisticalsupport with that of the Giant Hummingbird (Patagonagi- in the field. Dr. Vicki Funk of the New York Botanical gas), anotherwidespread species characteristic of the Gardenskindly identifiedthe Chuquiraga. southernAndes. Patagonahas previouslybeen re- corded on the Cotacachi-Yana Urco massif from the LITERATURE CITED Lake Cuicocha area near Cotacachi (Ortiz-Crespo 1974), only 20 km south of our presentrecord of Or- BERLIOZ,J., & C. JOUANIN. 1942. Revision critique eotrochilus.During Bleiweiss'svisit, a singleP. gigas desTrochilides du genereOreotrochilus. L'Oiseau was observedclose to the Oreotrochiluspopulations, et al. R.F.O.: 1-13. at 3,150 m on the west slope of Yana Urco, above CARPENTER,F. L. 1976. Ecologyand evolution of an Pitian, and the specieswas commoneast of this point Andean hummingbird (Oreotrochilusestella). on the west side of the inter-Andean valley. P. gigas Univ. Calif. Publ. Zool. 106: 1-74. differs ecologicallyfrom Oreotrochilus,occurring at CORLE¾-SMITH,G. T. 1969. A high altitude hum- lower elevations and in more xeric habitats, and the mingbird on the volcano Cotopaxi. Ibis 111: Guayllabambacanyon is apparentlynot a barrier for 17-22. 378 ShortCommunications [Auk, Vol. 99

JARDINE,E. 1849. Ornithology of Quito. Jardine's birds. Unpublished Ph.D. dissertation. Cam- Contribution to Ornithology 1849:67. bridge, Massachusetts,Harvard Univ. OrRoo, C. C. 1978. Nueva lista de la avifauna Ar- ß 1976. On the occurrenceand identity of Or- gentina. Opera Lilloana 27. eotrochiluschimborazo at the Guamani PassßEc- ORCES,V. G. 1944. Notas sobrela distribucion geo- uador. Ibis 118: 425. grafica de algunas aves neotropicas.Flora 4: ß & D. SIMBERrOFF.1980. Ecologyversus his- 103-123. tory as determinantsof patchy and insular dis- ORTIZ-CREsPO,F. I. 1974. The giant hummingbird tribution in high Andean Birds. Evol. Biol. 12: Patagonagigas in Ecuador. Ibis 116: 347-359. 335-379. PETERSßJ. L. 1945. Check-list of birds of the world, ZIMMER, J. T. 1951. Studies of Peruvian birds No. vol. 5. Cambridge, Massachusetts, Harvard 60. Amer. Mus. Novitates No. 1513. Univ. Press. Received3 June1981ß accepted I October1981. VUILLEUMIER,F. 1967. Speciationin high Andean

Differential Occurrenceof Yearlingand Adult Male Gadwallsin Pair Bonds

ROBERTJ. BLOHM• Departmentof WildlifeEcology, University of Wisconsin, Madison, Wisconsin 53706 USA Most age-specificinformation on breedingwater- most infrequentcombination was the yearling male/ fowl has been concernedwith the reproductivechar- adult femalecategory (3 of 39, 8%). acteristics of femalesß and few data are available on The importance of the observedfrequencies can- the effectsof age on male breeding biology. Weller not be evaluated without a consideration of the rel- (1964)and Lack (1968)have extensivelyreviewed the ative availability of young and adult males during literature on breeding age in the Anatidae. Lackßin pairing. In southern Manitoba, I found that most particular,noted that malesof the Anatini, especially Gadwalls arrived already pairedßand Paulus (1980), of the genusAnas, probablybreed as yearlings.In working on the ecology of Gadwalls on their win- the Athyini, a group in which a largeexcess of males tering grounds, statedthat most birds were already competefor smallernumbers of females,drakes may paired upon departure from Louisiana during late not breed until the secondbreeding season.In the March-early April. Although the age structureof the Mergini, most malestake part in the breeding activ- Gadwall on wintering areas is unknown and not ities at the age of 3 (Lack 1968). As part of my re- readily determined, a conservative approach is to searchon the breedingecology of the Gadwall (Anas determine the proportion of yearling males present strepera)in southern Manitoba, ! obtained informa- during the breeding season,using generalpopula- tion on the age of paired malesin order to evaluate tion projection models (see Martin et al. 1979: 214). the relationship of age to breeding acitivity. For exampleßI used the following values [ e.g. see During 1972-1975at Delta and Marshy Point, Man- Anderson 1975ßMallard (Anas platyrhynchos);Blan- itoba, I gathered data on 79 Gadwall pairs from ob- din in prep.ß Black Duck (Ariasrubripes)] as reason- servationsfollowing banding or directly from birds able mean parameterestimates for a dabbling duck collected.All duckswere aged as yearlings (1 yr old) population: or adults (2 yr old or older) at the time of captureor S^M = 0.60 = average annual survival rate for collection (Blohm 1977). adult males; In 39 casesßthe ages of both pair members were S^r = 0.55 = average annual survival rate for known. Adult males were involved in 29 (74%) pair adult females; situations,while yearling maleswere involved in 10 SyM = 0.45 = average annual survival rate for (26%). In the40 casesin whichonly the agesof males young males; were availableßadult drakeswere presentin 31 (78%) SyF = 0.45 = average annual survival rate for pair bonds, and yearling maleswere involved in the young females; remaining9 (22%). Adult male/adultfemale pairs (20 R = 0.50 = sex ratio of young birds expressed of 39, 51%) predominatedover all possiblesex-age as proportion male; combinations.Next in order of frequencywere adult P= 2.0 = average annual recruitment rate or male/yearling female (9 of 39, 23%) and yearling preseasonage ratio (young/adultfe- male/yearlingfemale (7 of 39, 18%) pair bonds. The male in the fall population). Accordingly, this modelling effort predicted a val- t Presentaddress: Office of MigratoryBird Management,U.S. Fish ue of 0.40 as the asymptoticproportion of yearling and Wildlife Service,Laurel, Maryland20708 USA. males in a breeding-seasonpopulation. Thenß as- April 1982] Short Communications 379

suming this value to be representativeof Gadwalls stairs,R. Gatti, P. Herzog, C. Madson,R. Rodgers,L. in southern Manitoba, I evaluated the observed fre- Sopuck,R. Wishart, and D. Young for their valuable quenciesof malesin pairing situations.In both sam- advice and assistance in the field. plesthe observednumber of yearlingmales partici- pating in pair bonds was significantlyless than LITERATURE CITED expectedin a situationin which40% of the breeding ANDERSON,D. R. 1975. Population ecologyof the populationwas comprisedof first-yeardrakes (one- Mallard: V. Temporaland geographicestimates tailed Binomial test: 10 of 39 pairs, P < 0.05; 9 of 40 of survival, recovery, and harvest rates. U.S. pairs, P < 0.02, Snedecor and Cochran 1967). Fish Wildl. Serv. Res. Publ. 125. The participationof a yearlingor adult male in a BLOHM,R. J. 1977. A capture and age determination pair bond is dependentupon severalfactors that in- method for the Gadwall (Anasstrepera). Unpub- fluencecourtship activity. Recent studies of the An- lished M.S. thesis. Madison, Wisconsin, Univ. atini haveindicated that first-yearmales may be less Wisconsin. competitive during pair bonding due to plumage DANE, B., • W. G. VAN DER KLOOT. 1964. An anal- and behavioralimmaturity. Stottsand Davis (1960) obtainedinformation on 20 pairs of BlackDucks in ysis of the displayof the Goldeneye(Bucephala Marylandand foundno juvenilemales in pair bonds clangula).Behaviour 22: 282-327. KALT•NHAOSER,D. 1971. Ober evolutions vorgange during early pairing periods.This lack of participa- in der Schwinnentenbalz.Z. Tierpsychol.29: tion in pairingwas attributedto sexualimmaturity 481-540. in the first-yearcohort (p. 132). McKinney (1965) KORSCHGEN, C. E., & L. H. FREDRICICSON. 1976. found that plumagedevelopment and pairing oc- Comparativedisplays of yearlingand adult male curred later in young captive Northern Shovelers Wood Ducks. Auk 93: 793•807. (Anas clypeata) than among adults. In addition, LACiC,D. 1968. Ecologicaladaptations for breeding youngbirds tendedto be lesscompetitive in mating in birds. London, Methuen. situations.Definite differences between the displays MARTIN, F. W., R. S. POSPAHALA,& J. D. NICHOLS. of young and adult males were also documentedin 1979. Assessmentand population management the CommonGoldeneye (Bucephala clangula, Dane of North Americanmigratory birds. Pp. 187-239 and Van der Kloot 1964)and, morerecently, in the Gadwall (Kaltenhauser 1971) and the Wood Duck in Environmental biomonitoring, assessment, (Aix sponsa,Korschgen and Fredrickson1976). Oring prediction,and management-•certaincase stud- ies and related quantitative issues(J. Cairns, G. (1969) describeddifferences in wing plumagebe- P. Patil, and W. E. Waters, Eds.). Statistical tween yearling and adult male Gadwalls.In addition, he suggestedthat yearlingmales were lesslikely to EcologySeries, vol. 2. Burtonsville,Maryland, breed than adults and that some drakes did not at- Intern. Co-op. Publ. House. McKINN•¾, F. 1965. Spacing and chasingin breed- tempt to mate. ing ducks. Wildfowl Trust, Ann. Rept. 16: In southernManitoba, the observedfrequencies 92-106. suggest that age-related differences in neuroendo- ORING,L. W. 1969. Summerbiology of the Gadwall crine development,plumage growth, or courtship at Delta, Manitoba. Wilson Bull. 81: 44-54. behavior may bestow a competitiveadvantage to older, more experiencedmales during early court- PAULUS,S. L. 1980. The winter ecologyof the Gad- ship, therebyexplaining the preponderanceof adult wall in Louisiana. Unpublished M.S. thesis. Grand Forks, North Dakota, Univ. North Da- drakesselected by femalesin pair situations. kota. The field work was supportedby the Delta Water- SNEDECOR,G. W., & W. G. COCHRAN. 1967. Statis- fowl ResearchStation and the Collegeof Agriculture tical methods. Ames, Iowa, Iowa State Univ. and Life Sciences,University of Wisconsin.I thank Press. R. A. McCabe, W. W. Blandin, T. J. Dwyer, R. S. STOTTS,V. D., & D. E. DAVIS. 1960. The Black Duck Pospahala, R. I. Smith, and V. D. Stotts for their in the ChesapeakeBay of Maryland: breeding reviews of this manuscript.J. D. Nichols provided valuable technicalsupport and statisticalassistance. behaviorand biology. ChesapeakeSci. 1: 127- 154. I am especiallyindebted to the ownersof East Mead- ows Ranch, the Peter Curry and Arthur Vincent fam- WELL,R, M. W. 1964. The reproductivecycle. Pp. 35-79 in The waterfowl of the world, vol. 4. (J. ilies, and to LawrenceKing, manager,for their gen- erosityduring my stay at MarshyPoint, Manitoba. Delacour, Ed.). London, Country Life Ltd. Finally, specialthanks must go to A. Afton, R. Car- Received12 March 1981, accepted2 October1981. 380 ShortCommunications [Auk, Vol. 99

Prey Handling in Yellow-crownedNight Herons

MARK F. RIEGNER Departmentof Ecologyand Evolution, State University of New Yorkat StonyBrook, StonyBrook, New York11794 USA Although an extensiveliterature existson the for- Prey sizeswere estimatedby comparingcaptured agingecology of wadingbirds (reviewedby Kushlan large prey with the length of the heron'sbill (mean 1978),detailed studies of the Yellow-crownedNight bill length + 1 SD = 72.7 mm + 3.3, from measure- Heron (Nyctanassaviolacea) are lacking.The yellow- ments of 15 male and 15 female museum specimens, crown is unique among the ardeids in that it spe- unpubl. data; dimensionsof malesand femaleswere cializes on crustaceanprey (Palmer 1962, Hancock pooled,because sexes were indistinguishablein the and Elliott 1978) and exhibits behavioral and mor- field). Published data (Mulstay 1975) and measure- phologicaladaptations suited to this diet. Analyses ments of prey remains from nest sites were used to of prey handling are presentedhere. estimate sizes of small prey. Prey-handling times Feedingdata were collectedat Mount Sinai Harbor were recorded with a stopwatch. (I defined prey- and John F. KennedyWildlife Sanctuary,Long Is- handling time as the time elapsedbetween initial land, New York and StoneHarbor, New Jerseybe- graspingof prey and swallowing.) tween May and August 1979and 1980.Adult yellow- Handling time varied directlywith prey size with- crowns were observed foraging in tide channels, in Brachyura(Fig. 1). A Model I regressionof loga- tide-pool depressions,Spartina grass, and on mud rithmically transformedhandling times on rank or- flats.Prey includedfiddler crabs(Uca pugnax), marsh der of crabs by increasingsize (1 through 4) was crabs (Sesarmareticulatum), rock crabs (Cancer irro- highly significant (0.001 < P < 0.005; regression ratus), and American eels (Anguillarostrata). Data equation:y = 0.611x+ 0.165;Sokal and Rohlf 1969), were also collectedat a commercialshrimp farm sit- justifying the above conclusion.All observations uated on a cleared mangrove swamp in Chomes, were pooledwithin a given prey categoryregardless PuntarenasProvince, Costa Rica during Octoberand of geographiclocality. Crabs exhibited no marked December 1980. Yellow-crowns foraged in man- behavioraldifferences among speciesafter capture; made basins containingpenaeid shrimps, portunid therefore,size was consideredthe major determinant crabs, and various fishes. All observations were of handling time. There was a direct associationbe- made during daylight hours, althoughfeeding also tweenhandling time and prey length within Anguilla occurredat night. rostrata (Fig. 2, Kendall Tau coefficient= 0.566; 0.008 < P < 0.014; Gibbons 1976). Typically, all eel size classeswrapped their body around the yellow- crown's bill during manipulation, contributing to 400-

300:

800-

700-

200- 600-

r• 500- Z

T 400- I00-

300-

200-

20 4o 6o 8'o 1004 CARAPACE WIDTH (ram) ,25 ,75 200 2;0 Fig. 1. Handling time of Yellow-crownedNight EEL LENGTH (mm) Herons in relation to prey size within Brachyura. Horizontal lines represent means, vertical lines de- Fig. 2. Handling time of Yellow-crownedNight pict ranges, rectanglesenclose mean + 1 SD, and Herons in relationto prey length within Anguillaro- points indicate individual observations. strata. Points represent individual observations. April 1982] Short Communications 381

TAI•LE1. Frequenciesof prey droppingby Yellow-crownedNight Herons.a

Mean Percentage carapace of prey dropped Prey width (mm)b Prey (n) Drops/prey 51 time Uca pugnax 13.2 240 0.10 7.5 Sesarma reticulatum 22.7 29 0.90 31.0 Portunidae 75.0 17 4.24 82.4 Cancer irroratus 130.0 2 9.00 100.0

• Data are from field observationsin New York and New Jersey,except for Portunidae(not identified to genus),which were observedin Chomes, Puntarenas Province, Costa Rica. b Meansizes of Portunidaeand Cancerwere estimated in the field(see text). The mean size of Ucawas determined from 17 carapacesfound in prey remains at nest sites. The mean size of Sesarmawas determined from measurements of 248 individuals from Mount Sinai Harbor, New York (Mulstay 1975).

handling time. Eels also demonstratethis behavior I thank A. Galli and D. Ward for directing me to when handled by other ardeids (Recherand Recher favorablestudy sites in New Jerseyand R. A. Arm- 1968). Amphipods and small fishes (•20 mm) con- strong, R. D. Bayer, M. A. Bell, H. A. Raffaele, R. E. stituted the category of smallest prey captured by Smolker, and T. J. Van't Hof for constructive criti- yellow-crownsduring my field observations.Con- cismson an earlier draft of the manuscript.Financial sumption of these prey occurred almost instanta- support was provided by a Frank M. Chapman Fund neouslyafter grasping. grant from the AmericanMuseum of Natural History The frequencyof prey (crab)dropping during ma- and a Sigma Xi grant-in-aid of research.This con- nipulation is reported in Table 1 (an item that fell tributionis number405 from the Programin Ecology from the bill while the heron's head was raised was and Evolution at the State University of New York considereda "drop"). Larger crabs were dropped at Stony Brook. more frequently than smaller ones (RxC test of in- dependence; G = 60.34, df = 3, P • 0.005; Sokal LITERATURE CITED and Rohlf 1969). Yellow-crownstransported greater GIBBONS,J. D. 1976. Nonparametric methods for than 95% of largerprey, includingrock crabs,por- quantitative analysis.New York, Holt, Rinehart tunid crabs,large marshcrabs, and eels, away from and Winston. submerged areas to exposed terrestrial patches. HANCOCK,J., & H. ELLIOTT. 1978. The herons of the Smaller prey (fiddler crabs and small fishes) were world. New York, Harper and Row. rarely carried to other sites before ingestion.The KUSHLAN,J. g. 1978. Feeding ecologyof wading transport of prey to exposed surfacesis common birds. Pp. 249-297 in Wading birds (A. Sprunt, amongwading birds (Kushlan1978) and is obviously IV, J. C. Ogden, and S. Winkler, Eds.). Natl. a measure associatedwith reducing escapes(Recher Aud. Soc. Res. Rept. 7. and Recher1968). Prey transportedby yellow-crowns --. 1979. Feedingecology and prey selectionin moved awkwardly on exposed surfacesand failed to the White Ibis. Condor 81: 376-389. escapewhen dropped. The herons usually impaled MULSTAY,R. E. 1975. Aspectsof the life history and these prey repeatedlywith their bills by jabbing socialbehavior of the purple marsh crab, Sesar- their heads downward. Herons removed the crabs' ma reticulatumSay. Unpublished Ph.D. disser- chelaeand legs by gripping either the appendages tation. StonyBrook, New York, StateUniv. New or carapaces in their bills and then vigorously York. shaking their heads. The carapaceswere crushed PALMER, R. S. 1962. Handbook of North American between the mandibles and swallowed with char- birds. Vol. 1. New Haven, Connecticut, Yale acteristicupward head tilting; the appendageswere Univ. Press. eaten subsequently.After the entire prey was con- QUINNEY, T. E., & P. C. SMITH. 1980. Comparative sumed, the heron returned to the water and drank foragingbehavior and efficiencyof adult and ju- to facilitate swallowing. venile Great Blue Herons. Can. J. Zool. 58: In summary, the duration of prey handling by Yel- 1168-1173. low-crownedNight Heronsis directlyrelated to prey RECHER,H. F., & J. g. gEeHER. 1968. Comments on size. This relationshiphas been found in other wad- the escapeof prey from avianpredators. Ecology ing birds (Recher and Recher 1968, Kushlan 1979, 49: 560-562. Quinney and Smith 1980).Prey transportby yellow- SOKAL,R. R., & F. J. ROHLF.1969. Biometry.San crownsaway from submergedareas to exposedsur- Francisco, California, W. H. Freeman and Co. facesis a behaviorused to reduceprey escapesdur- ing manipulation. Received23 June1981, accepted4 November1981. 382 ShortCommunications [Auk, Vol. 99

SynchronousFluctuations in ChristmasBird Countsof CommonRedpolls and PitionJays

CARL E. BOCXC Departmentof Environmental,Population, and Organismic Biology, University of Colorado, Boulder, Colorado 80309 USA Bockand Lepthien (1976)described synchronous cella1980). If PitionJays "erupted" in the high years southwarderuptions of eightNorth Americanboreal 1963-1964, 1965-1966, 1968-1969, 1969-1970, and seed-eatingbird species,based upon an analysisof 1971-1972(Table 1), then they shouldhave been de- ChristmasBird Count (CBC) data. These findings tectedon more CBC'sin those5 yr than in the other were confirmedby an analysisof banding data for 5. I testedthis by determiningthe number of counts two of the species--Common Redpoll (Carduelis in each year that reported at least one Pition Jay, fiammea)and Purple Finch (Carpodacuspurpureus; adjustingthese data to 1962levels to compensatefor Kennard 1976, 1977). Between 1962 and 1971, syn- the overall increasein CBC's each successiveyear. chronouseruptions of the eight speciesoccurred in Resultsshowed that Pition Jayswere detectedon sig- the winters of 1963-1964, 1965-1966, 1968-1969, nificantly more countsin eruption years (x = 17.6, 1969-1970,and 1971-1972.Available evidencesug- SD = 2.06) than in noneruption years (x = 13.0, geststhat theseeruptions were triggeredby wide- SD = 3.86; t = 2.34, P < 0.05). spread seed-cropreductions among high-latitude One interpretation of these results is that some tree species--especially spruce (Picea) and birch common environmental factor affected movements (Betula).Inspection of our 1962-1971CBC data bank and/or population sizes of the two speciesbetween indicatedthat only the Pition Jay (Gymnorhinuscy- 1962 and 1971. Becauseboth speciesare dependent anocephalus)among 616 additionalspecies fluctuated upon irregularly availabletree seeds,it is tempting in a manner stronglyresembling that of the boreal to speculatethat pition pine conecrops fluctuated in birds. synchrony with seed crops of more boreal tree In Table 1, mean numbers of Pition Jayscounted speciesduring this period. per party-hourover the 10-yr periodfor which I have Resultsof this study also could be due to some computerizedCBC data are comparedto the equiv- artifact of CBC data, but Pition Jaysand Common alent numbers of Common Redpolls--the species Redpollsconcentrate in winter in very differentparts that showed the boreal-bird eruption pattern most of North America, and it doesnot seempossible that strongly (Bock and Lepthien 1976). The Pearson CBC observerscould be influencingone anotherbe- product-momentcorrelation between jay and redpoll tween these regions. Certain nationwide weather numberswas 0.908 (P • 0.0003);the nonparametric patternsmight be influencingboth CBC observers Spearmanrank correlationfor the two data setswas and the birds, but this shouldhave resultedin syn- 0.720 (P • 0.01). chrony between redpoll numbersand countsof a Pitionpines (Pinusedulis and P. monophylla)pro- great many species.No such pattern is apparent in duce sporadicallyabundant cone crops, and Pition CBC data. Jaysare known to move opportunisticallyin search The synchronyin fluctuationsof redpolland Pition of them (Baldaand Bateman1972, Ligon 1978,For- Jay numbers could be due to chance,but it is very difficult to calculatesuch a probability exactlybe- cause of the small sample size involved (n = 10). Spearmanrank correlationalmost certainly is an un- TABLE1. Mean number of CommonRedpolls and realisticallyconservative estimate in the presentcase Pition Jayscounted per party-hour per year on (P • 0.01), becauseit gives much weight to rank mainland U.S. and Canadian Christmas Bird differencesbetween pairs of valuesthat actuallydif- Counts between 1962-1963 and 1971-1972 (n = fer very little in their absolutepositions in the two 8,129). data sets (Table 1). Neither variable differs signifi- cantlyfrom a normaldistribution, so use of the para- Common metric r is justified (P • 0.0003). Small sample size, Year (fall) Redpolls PitionJays however, makes tests for normality comparatively 1962 0.82 0.09 1963 2.48 0.29 There are 616 speciesin our CBC data bank not 1964 0.40 0.15 consideredto be boreal tree-seed-eatingbirds. It 1965 3.03 0.35 seemsprobable that one or two would show sucha 1966 0.28 0.09 strongcorrelation with redpollnumbers by chance 1967 0.23 0.13 alone.The probabilitythat it would occurby chance 1968 3.42 0.33 between redpolls and another notoriously vagrant 1969 3.57 0.28 tree-seedpredator seems very slight,although such 1970 0.15 0.15 1971 2.55 0.22 a possibility cannot be be discounted.If the rela- tionship between jays and redpolls is real, and if it April 1982] ShortCommunications 383

is due to fluctuationsof their food supplies, the re- FORCELLA,F. 1980. Cone predation by pition cone lationshipsuggests that geographicand interspecific beetle (Conopthorusedulis; Scolytidae): depen- synchronyamong tree-seed crops may be muchmore denceon frequencyand magnitude of cone pro- widespread than previously suspected. duction. Amer. Natur. 116: 594-598. This study was supported in part by NSF grants KENNARD,J. H. 1976. A biennial rhythm in the win- GB 36860 and DEB 79-10784. I thank Russell Balda, ter distribution of the Common Redpoll. Bird- RichardBradley, Michael Grant, Terry Root, Stephen Banding 47: 231-237. Vander Wall, and John Wiens for their comments --. 1977. Biennial rhythm in Purple Finch mi- and assistance. gration. Bird-Banding 48: 155-157. LICON,J. D. 1978. Reproductiveinterdependence of LITERATURE CITED Pition Jays and pition pines. Ecol. Monogr. 48: 111-126. BALDA,R. P., & G. C. BATEMAN.1972. The breeding biology of the Pition Jay. Living Bird 11: 5-42. Received8 May 1981, accepted11 November1981. BocI(, C. E., & L. W. LEPTHIEN.1976. Synchronous eruptions of boreal seed-eating birds. Amer. Natur. 110: 559-571.

Mortality of DucklingAttributed to Separationfrom Mother and Subsequent ProtractedExposure to Low Ambient Temperature

NORMAN R. SEYMOUR1 Facultyof Agriculture,Macdonald College of McGill University,Sainte Anne de Bellevue, QuEbecH9X 1C0, Canada

On 17 May 1981I was observingthe breeding be- but then swam to about 30 m offshore. It remained havior of dabbling ducks on the St. Lawrence River there in water 2 m deep, swimming in circles with near Lancaster, Ontario. A female Mallard (Anas its head continually dipping down. Within 60 min platyrhynchos)and her nine 1-2-day-old ducklings of separationfrom the female, it was dead. were feeding as they slowly moved alongthe marshy Temperatures on this day were the lowest since edge of St. Francis Island. At 1130, when they 1944, and that night they were 0øCor below. Night- reached the point closestto Hamiltons Island (ca. 0.3 time temperatures had dipped to 2øC for the two km away), the female called and swam into open nights before this incident, and they were to 0øCfor water, followedby eight of the ducklings.The duck- the two nights thereafter; wind velocity reached a lings clusteredtogether and stayedwithin 1 m of the steady35-40 km/h the day after. When I first saw the female during the entire crossing;the female called brood, air temperature under bright sun was ap- frequentlyand kept her head closeto her body. The proximately 4øCand water temperaturewas approx- shoreline where the female reached Hamiltons Island imately3øC. This ducklingappeared healthy and was was rocky; to the south the shore remained rocky, dry when retrieved. It swam strongly, despite the but a cattail marsh was approximately300 m to the wind from the northwest and a 10-12 km/h current north. The female (followed closelyby the bmod) flowing north between the two islands; it reached swam to within 5 m of shore and then followed the the island at exactlythe same location as the female. shore to the marsh. She took the brood to shore, Presumably the duckling was cold and could have where I lost sight of them. conservedor even gained body heat had it gone to The ninth duckling left St. FrancisIsland at the shore and sat on the stones in the sun. Periodic sameplace about 5 min after the othersand remained brooding by the female could potentiallyhave pro- about 100 m behind them. It presumablysaw the vided an additional sourceof heat, which may have others depart, becauseit followed the same trajec- been why she took the young to shoreas soon as she tory. The wind was blowing at 12-15 km/h from the reached the marsh. The relatively short (about 5 northwest and 0.5-m-high waves of short interval min), but critically timed, separationfrom the female buffetted the duckling. It reachedHamiltons Island was apparentlyenough for the duckling to become about 8-10 min after the others; by this time, how- lost and subsequentlyto succumbto exposure. ever, the brood presumablywas out of sight of the Dabbling ducksin this habitat nest on islands and lone duckling. It stayed there approximately45 min move their newly hatched broods to marshy areas associated with islands. They frequently move broodsbetween islands, presumably to new feeding • Presentaddress: Department of Biology,Saint FrancisXavier Uni- areas.Movements of up to 2 km are commonin Mal- versity, Antigonish, Nova ScotiaB2G 1C0, Canada. lards, Black Ducks (A. rubripes),American Wigeon 384 Short Communications [Auk, Vol. 99

(A. americana),and Gadwall (A. strepera).There is April, during an unusuallymild early springbut cool considerablepotential for ducklingsto become sep- and wet late spring. This incident is significantbe- arated from their mothers. Risk of exposureto ad- causeit emphasizesthe hazardsassociated with the verseclimatic conditions is great,especially early in timing of nesting for females and the consequences the yearwhen coldwater and coldair oftencombine of even a brief lapseof attentivenessby young and/ with rain and high wind. In addition, femalesthat or females. nest early in the seasonalways risk havingbroods hatch when food is scarce or unavailable. This female Received3 June1981, accepted18 November1981. musthave begun egg laying during the firstweek of

RockhopperPenquin (Eudyptes chrysocome) Record at Palmer Station, Antarctica

KATHRYN MATTHEW Academyof NaturalSciences, 19th and the Parkway, Philadelphia, Pennsylvania 19103 USA An adult male Rockhopper Penquin (Eudyptes mighthave recently been relieved by his mate(males chrysocome)with a broken wing was collectedon usuallyincubate first, for approximately25 days)and Cormorant Island near Palmer Station, off the south gone out to sea to feed, which would explain his coast of Anvers Island, Antarctic Penninsula (64ø8'S, occurrence far from the Falkland Islands. A broken 63ø58'W)on 29 December 1980. The specimenis de- wing might well have hinderedhis ability to return positedat the Academyof NaturalSciences of Phila- to the Falklands. This is the first record for this delphia (uncatalogued).This localityis approximate- speciesat Palmer Station (confirmedby D. Parme- ly 1,100km from the nearestbreeding area of the lee). species,the islands off Cape Horn. The bird was I am gratefulto P. Tirrell, who first spottedthe standing at the edge of a colony of nesting Adelie Rockhopper.I also wish to thank (in alphabetical Penguinsand appearedalert and in good condition order) R. Fraser, D. Murrish, D. Parmelee, P. Peitz, exceptfor its injury. Dissectionand weighing pro- and D. Wiggin. R. E. Ricklefsand G. E. Watsonpro- ducedthe followingwet weights:total, 2,532 g; pec- vided critical comments on the manuscript. This toral muscles,440.44 g; thigh muscles,219.62 g; liver, specimenwas collectedduring researchsupported 76.86g; heart,28.14 g; brain, 10.91g; stomach,18.45 by the U.S. AntarcticResearch Program under grant g; the stomachwas emptyand therewere no visible number DPP-79-08570 awarded to R. E. Ricklefs, subcutaneousfat deposits.Both testeswere present; Universityof Pennsylvania. they were 180mm by 80 mm in lengthand width and, together, weighed 1.23 g. Color notes were LITERATURE CITED madefrom life. The eyeswere dull red, the feetwhit- CARINS,M. 1974. Facial characteristicsof Rockhop- ish pink with black soles, and the bill reddish per Penquins. Emu 74: 55-57. brown. PETTtNCtLL,O. S., JR. 1960. Crache behavior and Male Rockhopperbody weightson MacquarieIs- individual recognitionin a colonyof Rockhop- land average2.7 kg (range,2.1-3.2) (Warham1963); per Penguins.Wilson Bull. 72: 213-221. the individual collected near Palmer Station was PREVOST,J., & J. L. MOUCHLtN. 1970. Guide des therefore well within the expected weight range. oiseaux et mammif•res des terres Australes et Unless this individual was taken onboard ship, Antarctiques Franqaises,Guides Natur. Neu- transported,and then released,it most likely came chatel, Switzerland, Delachaux et Niestl• Edi- from the Falkland Islands (G. E. Watson pers. teurs. comm.),where Rockhoppers breed abundantly (Pet- STRANGE,J. L. 1965. Beauchine Island. Polar Record tingill 1960,Strange 1965). The facial characteristics 12: 725-730. and head plumes were consistentwith the dark- WARHAM, J. 1963. The Rockhopper Penguin, Eu- faced form (chrysocome)expected for a bird from dypteschrysocome, at MacquarieIsland. Auk 80: South America or the Falkland Islands(Carins 1974), 229-256. althoughthe underwingpattern was more like that 1972. Breeding seasonsand sexual dimor- of the subtropical form moseleyi (Prevost and phism in RockhopperPenguins. Auk 89: 86-105. Moughlin 1970). The peak egg-layingdate for Rockhopperson the Received13 July 1981, accepted4 November1981. Falkland Islands is estimated to be 20 November (Warham 1972).If this individual were breeding, he April 1982] ShortCommunications 385

Distribution of the Quetzal in Honduras

DONALD A. HANSON Departmentof Fisheriesand Wildlife,Michigan State University, EastLansing, Michigan 48824 USA I studied the Quetzal (Pharomachrusmocinno) on tafia de Celaque; Altantida: (17) Montafia Pico Bo- behalf of the Department of Wildlife and Environ- nito. The following Quetzal locationsalso were re- mental Resources, Honduras from October 1977 to ported, but no literature or map referenceto them November 1980, in part to determine its nationwide could be found either in Honduras or in the United distribution. Quetzals were reported from the fol- States. Santa Barbara: Montafia Paso Viejo; E1 Pa- lowing mountainsby AmericanPeace Corps volun- raiso:Montafia Jogon;Francisco Morazan: Montafia de teers who respondedto a questionnaireregarding Palo Alto. This could reflect local names for the the distribution of the species (numbers are those mountains. used in Fig. 1). The questionnairerequested rural- The above are additional Quetzal locations to those living volunteersto interview residentsof the local- reportedby Monroe (1968).He listed the following. ity regardingthe presenceof Quetzals.These resi- Santa Barbara: (18) Montafia Santa Barbara; Ocote- dentsare very knowledgeableconcerning the species peque:(19) Montafia E1 Chorro; FranciscoMorazan: of wildlife present. (20) Montafia Monsettat; Lempira:(21) Montafia de The locations of the mountains are indicated in Puca;La Paz: (22) Montafia La Cruz; Comayagua:(23) Fig. 1 by number. Ocotepeque:(1) Montafia Monte- mountainssouth and eastof the plain of Comayagua. cristo; Yoro: (2) Montafia de Pijol, (3) Montafia de Appreciation is expressedto former PeaceCorps Yoro; Intibuca: (4) Montafia Pacayas, (5) Montafia volunteers in Honduras for assistancein supplying Parnina; Cortes: (6) Montafia Cusuco, (7) Montafia distributional information. San Ildefanso; E1 Paraiso: (8) Montafia Yuscaran; FranciscoMorazan: (9) Montafia La Tigra, (10) Mon- LZTEV,•TURE CZTED tafia E1 Chile, (11) Montafia de la Flor, (12) Montafia MONROE,B. L. A distributional survey of the birds de Uyuca;Copan: (13) Montafia San Jose and/or Mon- of Honduras. Ornithol. Monogr. No. 7. tafia San Antonia; Olancho: (14) Sierra de Algalta, (15) Montafia de Almendares; Lempira: (16) Mon- Received25 June1981, accepted11 November1981.

89ø 88ø 87ø 86ø 85ø 84ø

15ø

5 9 I0 14ø 16 23 15/ • 2212•• KM 80 13ø • HONDURAS Fig. 1. Map of Hondurasshowing the distributionof the Quetzal.Map locationsare codedin the text. COMMENTARY

Macaroni Penguins:Comment on Mistaken King George Island BreedingRecord and SoutherlyRange Extension

NICHOLASJ. VOLKMAN,1 WAYNE Z. TRIVELPIECE?NEIL P. BERNSTEIN? AND PAUL C. TIRRELL 3

On the basis of a personal communication from were verifiable, Bernstein and Tirrell's contention that Polish researchers at Arctowski Station, Point Thom- it representsa southward extension of their range as, Admiralty Bay, King George Island (62ø10'S, was erroneous.Breeding Macaroni Penguins have 58ø30'W), Bernstein and Tirrell (1981) stated that for been reported on Nelson Island (Andersson 1905, the first time a few Macaroni Penguins (Eudyptes Watson et. al. 1971), Gibbs Island (Watson et al. chrysolophus)bred on King GeorgeIsland during the 1971),Deception Island (62ø57'S;not 58øSas in Bern- 1979-1980 austral summer. Bernstein and Tirrell fur- stein and Tirrel11981) (Gain 1914, Watson et al. 1971), ther suggestedthat Macaroni Penguin populations and Livingston Island (J. Croxall, pers. comm.), all of might be increasingin the area of the Antarctic Pen- which are south of King George Island. In addition, insula. Volkman and Trivelpiece have communicated Watson collecteda Macaroni Penguin 10 km north severalinaccuracies in Bernsteinand Tirrell's paper, of Palmer Station on Dream Island (G. Watson, pers. and we wish jointly to correctthese errors. comm.). Volkman and Trivelpiece did research on pygo- We wish to thank J.P. Croxall for his communi- scelid penguins at Pt. Thomas during the austral cationand G. E. Watsonfor his suggestionsand crit- summers of 1976-1977, 1977-1978, and 1980-1981 icisms. and saw no evidence of breeding Macaroni Pen- guins. Furthermore, B. Jablonski(pers. comm.), a LITERATURE CITED Polish ornithologist who studied penguins at Pt. ANI)ERSSON, K. A. 1905. Das h6rere Tierleben im Thomas during the austral summers of 1978-1979, antarktischen Gebiete. Wiss. Ergebn. Schwed- 1979-1980, and 1980-1981, did not observe any ischen S/idpolarexpedition, 1901-1903, Bd. 5 breeding Macaroni Penguins. On severaloccasions, Zool. I, Lief. 2: 1-57. however, Volkman and Trivelpiece have seen molt- BERNSTEIN, N. P., & P. C. TIRRELL. 1981. New ing MacaroniPenguins late in the season.For what- southerlyrecord for the Macaroni Penguin (Eu- ever reason, the personal communication to Bern- dypteschrysolophus) on the AntarcticPeninsula. stein and Tirrell of breeding Macaroni Penguins at Auk 98: 398-399. Pt. Thomas appears to be inaccurate. GAIN, L. 1914. Oiseaux antarctiques. Pp. 1-200. In Even if this record of breeding Macaroni Penguins Deuxi•me expedition antarctique frangaise (1908-10). Paris. • Point ReyesBird Observatory,4990 ShorelineHigh- WATSON, G. E., J.p. ANGLE, P. C. HARPER, M. A. way, StinsonBeach, California 94970 USA. BRIDGE, R. P. SCHLATTER,W. L. N. TICKELL, J. • FieldBiology Program, Bell Museum of Natural His- C. BoYi), & M. M. BoYv. 1971. Birds of the Ant- tory, Universityof Minnesota,Minneapolis, Minnesota arcticand sub-Antarctic.Antarc. Map Folio Set., 55455 USA. Folio 11. New York, Amer. Geog. Soc. 3 Departmentof Biology,State Universityof New York, Binghamton,New York 13901 USA. Received10 August1981, accepted5 January1982.

Fellows and Elective Members are reminded that tions for Fellow of the A.O.U. alsomust be received nominationsfor ElectiveMembers may be submit- by that date. Nominationsfor President-Elect,Vice ted to the Secretaryon the prescribedform up until Presidentand three Elective Councilors may be made five months prior to the opening of the next Stated in writing to the Secretaryat any time prior to the Meeting. The deadlinefor 1982is 10 May. Nomina- Annual Meeting.

386 242 GP•ANTET AL. [Auk, Vol. 99

--, A. OLSZOWKA, & A. AR. 1974. The avian WHITTOW.1976. Regulationof incubation water egg: surfaceareaß volume and density. Condor lossin sevenspecies of terns. Physiol.Zool. 49: 76: 319-325. 245-259. PETTIT, T. N., G. S. GRANT, G. C. WHITTOW, H. RICKr•FS, R. A., & H. RAHN. 1979. The incubation RAHN,• C. V. PAGANELLI.In press.Respiratory temperature of Leach's Storm-Petrel. Auk 96: gas exchangeand growth of Bonin Petrel em- 625-627. bryos. Physiol.Zool. ROB•I•TS,B. 1940. The life cycle of Wilson's Petrel RAHN,H., • A. AR. 1974. The avian egg:incubation Oceanitesoceanicus (Kuhl). Scient. Rep. Br. Gra- time and water loss. Condor 76: 147-152. ham Ld Exped. 1934-37, 1: 141-194. , & --. 1980. Gas exchangeof the avian ROUDYBUSH,T., L. HOFFMAN, & H. RAHN. 1980. egg: time, structure, and function. Amer. Zool. Conductance,pore geometry and water loss of 20: 477-484. eggsof Cassin'sAuklet. Condor82' 105-106. , & W. R. DAWSON. 1979. Incubation water TUrL•TT, S. G., & R. G. BoAar>. 1977. Determinants loss in eggsof Heermann'sand Western Gulls. of avian eggshellporosity. J. Zool., London 183: Physiol. Zool. 52: 451-460. 203-211. ß C. V. PAGANErrS, & A. AR. 1975. Relation VL•CK, C. M., & G. J. KœNAG¾.1980. Embryonic of avian egg weight to body weight. Auk 92: metabolism of the Fork-tailed Storm-Petrel: 750-765. physiologicalpatterns during prolongedand in- , R. A. ACKERMAN,& C. V. PAGANELLI. 1977. terruptedincubation. Physiol. Zool. 53: 32-42. Humidity in the avian nest and egg water loss WHITTOW,G. C. 1980. Physiologicaland ecological during incubation. Physiol. Zool. 50: 269-283. correlatesof prolongedincubation in seabirds. , C. V. PAGANELLI, I. C. T. NISBET, & G. C. Amer. Zool. 20: 427-436.

The sixthannual meeting of the ColonialWaterbird Group will be held 4-7 November1982 in Washington, D.C. A symposiumon the feedingbiology of waterbirdsis planned.Papers given at the meetingare eligible, after refereeing,for publicationin ColonialWaterbirds. Anyone wishing to contributeto either the sym- posium (deadline 1 September)or general session(deadline 15 September)should contactDr. Michael Erwin, U.S. Fish and Wildlife Service,Patuxent Wildlife ResearchCenter, Laurel, Maryland 20708.In- formationconcerning registration can alsobe obtained from Dr. Erwin. April 1982] Migrationand Winter Mortality in Juncos 259

dominance and the advantageof subordinance STUEBE,M. M., & E. D. KETTERSON.1982. A study in a badge signalingsystem. Evolution 35: 441- of fasting in Tree Sparrows(Spizella arborea) and 454. Dark-eyedJuncos (Junco h. hyemalis):ecological ROSEBERRY,J. L. 1962. Avian mortality in southern implications. Auk 99: 229-308. Illinois resulting from severe weather condi- U.S. WEATHER BUREAU BULLETIN W. 1932. Clima- tions. Ecology43: 739-740. tography of the United States.Sections 1-106. SEARCY, W. A., & K. YASUKAWA. 1981. Sexual size VERNER,J., & M. F. WILLSON.1969. Mating systems, dimorphism and survival of male and female sexual dimorphism, and the role of male North blackbirds (Icteridae). Auk 98: 457-465. American passerinebirds in the nesting cycle. SELANDER,R. K. 1965. On mating systemsand sex- Ornithol. Monogr. 9: 1-76. ual selection. Amer. Natur. 99: 129-141. WILEY, R. H. 1974. Evolution of socialorganization SPAANS,A. L. 1977. Are Starlings faithful to their and life-history patterns among grouse. Quart. winter quarters?Ardea 64: 83--87. Rev. Biol. 49: 201-227.

The AOU Student Awards Committee issues the following reminders to all students interested in com- peting for the AOU Student Paper Award and the two Nellie JohnsonBaroody Student Paper Awards at the 1982 AOU meetings:(1) To qualify for the competition,each student must submit an expandedab- stract of the paper directly to the chairpersonof the AOU Student Awards Committee (Douglas Mock, Departmentof Zoology, University of Oklahoma, Norman, Oklahoma 73069)by the "Call for Papers" deadline in the meeting announcement.(2) Appropriate applicationswith expanded abstractsmust be submitted separatelyfor the student paper awards and for the Marcia Brady Tucker travel awards. (3) A studentis ineligiblefor studentpaper awardsif she/heshares authorship of a paper, i.e. the studentmust be the exclusive author.

The AmericanOrnithologists' Union will hold its annualmeeting in Chicago,Illinois from 10-14 October 1982. A formal call for papers is enclosedwith this issue of The Auk, with a deadline for submission of abstractsfor papers and poster sessionsof 20 May 1982. For further information on the scientific program, contactJames R. Karr, Program Chairman, Department of Ecology, Ethology, and Evolution, Universityof Illinois, 606 E. Healey, Champaign,Illinois 61820USA. April1982] Hybridizationin Flycatchers 291

R•s•tac, J. D. 1969. A multivariate assessmentof in- SVEtaSSON,L. 1975. Identification guide to European terbreedingbetween the chickadees,Parus atri- passerines. Stockholm, Naturhistoriska Riks- capillus and P. carolinensis. Syst. Zool. 17: museet. 160-169. VEPS•L•IIqEN, K., • O. J•RVINEN. 1977. Ein Bastard- ROSVALL,S. 1979. HalsbandsflugsnapparensFice- M•nnchen Halsband-Trauerschn•pper(Ficedula dula albicollisankomst till Gotland och (•land albicollisx F. hypoleuca)briitet in Finnland. J. v•ren 1979. Bl•cku 5: 73-78. Ornithol. 118: 436-437. SHORT,L. L. 1969. Taxonomic aspectsof avian hy- , K. KEYIq•S, & P. SALMINEIq. 1975. bridization. Auk 86:84-105. S;pelsiepotFicedula albicollis Hankoniemell• --. 1972. Hybridization, taxonomy and avian 1971-74. Lintumies (Helsinki) 10: 118-122. evolution. Ann. Missouri Bot. Garden 59: WEst, D. A. 1962. Hybridization in grosbeaks 447-453. (Pheucticus)of the Great Plains. Auk 79: 339-424.

REVIEWERS FOR THE AUK, 1980-1981 TheAuk is fortunateto benefitfrom the servicesof many individuals who act as reviewersof manuscripts that are submitted. Their efforts are considerable,and their care and constructivenesscontinue to impress me as an Editor and as a sometimeauthor. The individuals listed below helped make publication of a journal with high scientificand scholarlystandards possible; my thanks to all of them. Individuals who have contributed reviews of two or more manuscriptsare indicated by an asterisk. Ian Abbott, David G. Ainley, JohnAlcock*, Thom- Fitch, JohnW. Fitzpatrick*, Brian K. Follett, Eric D. as Alerstam, Dean Amadon, Daniel W. Anderson, Forsman*,W. J. Francis,Leigh Fredrickson,Herbert Ted R. Anderson, C. Davison Ankney, Einar Arna- Friedmann, Carl Gans, C. Lee Gass*, A. J. Gaston, son, John C. Avise, Robert O. Bailey, Allan Baker, Sidney A. Gauthreaux, W. George, Frank B. Gill*, Myron C. Baker, Robert J. Baker, George S. Bakken, Robert E. Gill, P. D. Gingerich, BradleyM. Gottfried, Russell P. Balda, Thomas G. Balgooyen*, David F. Gilbert Gottlieb, Walter D. Graul, RussellGreenberg, Balph, Martha Balph*, Richard C. Banks*, Luis F. Paul J. Greenwood, T. C. Grubb, Jr., JosephA. Grzy- Baptista*,George F. Barrowdough*,George A. Bar- bowski, Svein Haftorn, Caldwell D. Hahn, F. Reed tholomew, Jr., ThomasS. Baskett,Range Bayer*, Jeff Hainsworth, William J. Hamilton, III, M.P. Harris, Baylis, Michael Beecher*, Colin Beer*, Marc Bekoff, Richard J. Harris, JeremyJ. Hatch, Helen Hays, Rob- Frank C. Bellrose, Jr.*, Louis B. Best*, T. R. Birkhead, ert E. Hegner, Dennis Heinemann*, Carl W. Helms*, CharlesR. Blem, Peter Boag,Walter Bock*, William Charles J. Henny*, Wayne Hoffman*, A. J. Hogan- R. P. Bourne,John H. Brackenbury,Richard A. Brad- Warburg, Richard T. Holmes, Dominique G. Hom- ley*, Eliot Brenowitz, I. Lehr Brisbin*, Pierce Brod- berger,John L. Hoogland*,Henry F. Howe*, Donald korb*, Richard G. B. Brown*, Charles J. Burke, M. F. Hoyt*, JohnP. Hubbard, George L. Hunt, Jr.*, R. Byrd, William A. Calder, Jr.*, Thomas Caraco, Cyn- Hutton, Jerome A. Jackson*, Frances C. James*, Dan- thia Carey*, Lynn F. Carpenter, Susan M. Case, iel H. Janzen*, Olli Jarvinen*, JosephR. Jehl, Jr.*, Graeme Caughley, Mary H. Clench, Charles T. Col- Paul A. Johnsgard*,Douglas H. Johnson*,David W. lins*, Michael Collopy, Edward F. Connor, Peter G. Johnston, Richard F. Johnston*, James R. Karr*, Janet Connors, Michael R. Conover*, Fred Cooke, John Kear, J. Allen Keast, Bretton W. Kent, Paul Kerlinger, Cooper, Kendall W. Corbin, Richard Cowie, Joel Lloyd F. Kiff, Lawrence Kilham, James R. King*, Cracraft*, Robert B. Craig, Richard D. Crawford, Warren B. King, Erwin E. Klaas*, Richard W. Knap- John Croxall, E. Curio, Thomas W. Custer, John Da- ton, Carl E. Korschgen, Gary Krapu*, John R. vis, David DeSante, Diane De Steven, Robert W. Krebs,DonaldE. Kroodsma,James A. Kushlan, Mary Dickerman, James J. Dinsmore, David S. Dobkin, K. LeCroy, Louis Lefebvre,Ross Lein*, Robert E. Richard A. Dolbeer, Hilary Dow, R. D. Drobney, Lemon*, K. Lessells,J. David Ligon*, C. D. Little- William H. Drury, Alfred M. Dufty, Jr., Arthur E. field*, Clare S. Lloyd, William J. Mader, Chris Marsh, Dunham, Erica H. Dunn*, Thomas J. Dwyer*, Eu- Carl D. Marti*, Dennis J. Martin, Elden W. Martin, gene Eisenmann,Phillip F. Elliott, John T. Emlen, L. D. Martin*, Steven Martindale, Barbara W. Mas- Robert Epting, Anthony J. Erskine, R. Michael Er- sey, Harold F. Mayfield*, Donald A. McCrimmon, win*, P. G. H. Evans,Roger M. Evans*,David Ewert, Jr., ChristopherMcGowan, Judith W. Mcintyre, Ann C. J. Feare*, J. Alan Feduccia, Peter Feinsinger, Mil- McNabb, A. H. Meier, Charles Meslow, L. Richard licent Ficken*, JamesS. Findley, Hans Fisher, Walter Mewaldt, David Miller*, Don E. Miller*, DouglasW.

(continuedon p. 315) April 1982] CowbirdMovements and Activities 327

plumages of the Brown-headedCowbird and headed Cowbird on Empidonaxflycatchers in Red-winged Blackbird. Condor 62: 202-214. Michigan. Auk 78: 266-268. SIEGEL,S. 1956. Nonparametricstatistics for the be- WEST, M. J., A. P. KING, &: D. H. EASTZER.1981. havioral sciences. New York, McGraw-Hill. The cowbird: reflectionson developmentfrom TOMPA, F. S. 1962. Territorial behavior: the main an unlikely source.Amer. Sci. 69: 56-66. controllingfactor of a local Song Sparrow pop- WIENS,J. A. 1963. Aspectsof cowbirdparasitism in ulation. Auk 79: 687-697. southern Oklahoma. Wilson Bull. 94: 130-139. VAN V•LZ•N, W. T. 1972. Distribution and abun- --. 1973. Pattern and processin grasslandbird dance of the Brown-headedCowbird. Jack-Pine communities. Ecol. Monogr. 75: 237-270. Warbler 50: 110-113. WILLSON,M. F. 1966. Breeding ecologyof the Yel- V•RNV.R, J. 1964.Evolution of polygamyin the Long- low-headed Blackbird. Ecol. Monogr. 36: 51-77. billed Marsh Wren. Evolution 18: 252-261. ZIMMV.mWAN, J. L. 1971. The territory and its density WALKINSHAW,L. H. 1949. Twenty-five eggs appar- dependent effect in Spiza americana.Auk 88: ently laid by a cowbird.Wilson Bull. 61: 82-85. 591-612. ß 1961. The effect of parasitismby the Brown-

An International Symposiumon the Tamaulipan Biotic Provincewill be held at La Quinta Royale Motor Inn, CorpusChristi, Texas,on 28-30 October1982. The symposiumwill emphasizeall ecologicalaspects of the biome: vegetation,invertebrates, vertebrates, ecological structure and function,biological resources (use and effects),and management.Abstracts of contributedpapers are due 1 August 1982;abstracts for invited papers must be receivedby 1 September1982. For information on the program, contactGene W. Blacklock, Curator, Welder Wildlife Foundation, P.O. Drawer 1400, Sinton, Texas 78387 or David Riskind, Head, ResourceManagement Section, Texas Parks and Wildlife Department, 4200 Smith School Road, Austin, Texas 78744.For informationon registration,contact Jimmie R. Picquet, Director, John E. Conner Museum, Texas A&I University, P.O. Box 2172, Kingsville, Texas 78363.

The joint 8th International Conference on Bird Census Work and the 6th Meeting of the European Ornithological Atlas Committee will be held at Newland Park College, Chalfont St. Giles, Buckingham- shire, United Kingdom, on 5-9 September1983. The conferencewill include papers on validation and testingof censusmethods, application of censustechniques to bird communitystudies within and between habitats,description of vegetationand habitats for use in ornithologicalstudies, atlas studies,and data processingmethods. For further information, contact R. J. Fuller, British Trust for Ornithology, Beech Grove, Tring, Hertfordshire, HP23 5NR, United Kingdom. COMMENTARY

Macaroni Penguins:Comment on Mistaken King George Island BreedingRecord and SoutherlyRange Extension

NICHOLASJ. VOLKMAN,1 WAYNE Z. TRIVELPIECE?NEIL P. BERNSTEIN? AND PAUL C. TIRRELL 3

On the basis of a personal communication from were verifiable, Bernstein and Tirrell's contention that Polish researchers at Arctowski Station, Point Thom- it representsa southward extension of their range as, Admiralty Bay, King George Island (62ø10'S, was erroneous.Breeding Macaroni Penguins have 58ø30'W), Bernstein and Tirrell (1981) stated that for been reported on Nelson Island (Andersson 1905, the first time a few Macaroni Penguins (Eudyptes Watson et. al. 1971), Gibbs Island (Watson et al. chrysolophus)bred on King GeorgeIsland during the 1971),Deception Island (62ø57'S;not 58øSas in Bern- 1979-1980 austral summer. Bernstein and Tirrell fur- stein and Tirrel11981) (Gain 1914, Watson et al. 1971), ther suggestedthat Macaroni Penguin populations and Livingston Island (J. Croxall, pers. comm.), all of might be increasingin the area of the Antarctic Pen- which are south of King George Island. In addition, insula. Volkman and Trivelpiece have communicated Watson collecteda Macaroni Penguin 10 km north severalinaccuracies in Bernsteinand Tirrell's paper, of Palmer Station on Dream Island (G. Watson, pers. and we wish jointly to correctthese errors. comm.). Volkman and Trivelpiece did research on pygo- We wish to thank J.P. Croxall for his communi- scelid penguins at Pt. Thomas during the austral cationand G. E. Watsonfor his suggestionsand crit- summers of 1976-1977, 1977-1978, and 1980-1981 icisms. and saw no evidence of breeding Macaroni Pen- guins. Furthermore, B. Jablonski(pers. comm.), a LITERATURE CITED Polish ornithologist who studied penguins at Pt. ANI)ERSSON, K. A. 1905. Das h6rere Tierleben im Thomas during the austral summers of 1978-1979, antarktischen Gebiete. Wiss. Ergebn. Schwed- 1979-1980, and 1980-1981, did not observe any ischen S/idpolarexpedition, 1901-1903, Bd. 5 breeding Macaroni Penguins. On severaloccasions, Zool. I, Lief. 2: 1-57. however, Volkman and Trivelpiece have seen molt- BERNSTEIN, N. P., & P. C. TIRRELL. 1981. New ing MacaroniPenguins late in the season.For what- southerlyrecord for the Macaroni Penguin (Eu- ever reason, the personal communication to Bern- dypteschrysolophus) on the AntarcticPeninsula. stein and Tirrell of breeding Macaroni Penguins at Auk 98: 398-399. Pt. Thomas appears to be inaccurate. GAIN, L. 1914. Oiseaux antarctiques. Pp. 1-200. In Even if this record of breeding Macaroni Penguins Deuxi•me expedition antarctique frangaise (1908-10). Paris. • Point ReyesBird Observatory,4990 ShorelineHigh- WATSON, G. E., J.p. ANGLE, P. C. HARPER, M. A. way, StinsonBeach, California 94970 USA. BRIDGE, R. P. SCHLATTER,W. L. N. TICKELL, J. • FieldBiology Program, Bell Museum of Natural His- C. BoYi), & M. M. BoYv. 1971. Birds of the Ant- tory, Universityof Minnesota,Minneapolis, Minnesota arcticand sub-Antarctic.Antarc. Map Folio Set., 55455 USA. Folio 11. New York, Amer. Geog. Soc. 3 Departmentof Biology,State Universityof New York, Binghamton,New York 13901 USA. Received10 August1981, accepted5 January1982.

Fellows and Elective Members are reminded that tions for Fellow of the A.O.U. alsomust be received nominationsfor ElectiveMembers may be submit- by that date. Nominationsfor President-Elect,Vice ted to the Secretaryon the prescribedform up until Presidentand three Elective Councilors may be made five months prior to the opening of the next Stated in writing to the Secretaryat any time prior to the Meeting. The deadlinefor 1982is 10 May. Nomina- Annual Meeting.

386