184 ShortCommunications [Auk, Vol. 102

and communication (P. Marlet and J. Vanden- 1977. Local variations in time of breeding bergh, Eds.).New York, Plenum Press. of female Blue Grouse. Condor 79: 185-191. 1981. social behavior. Boston, Dux- --, & J. F. BENDELL. 1972. Blue Grouse, habitat, bury Press. and populations. Proc. 15th Intern. Ornithol. ZWICKEL,F.C. 1972. Someeffects of grazingon Blue Congr.: 150-169. Grouseduring summer. J. Wildl. Mgmt. 36: 631- 634. Received12 December1983, accepted 8 July 1984. 1973. Dispersion of female Blue Grouse during the brood season.Condor 75: 114-119.

SelectiveAlarm Calling by Downy Woodpeckersin Mixed-speciesFlocks

KIMBERLY SULLIVAN 1 Instituteof AnimalBehavior, Rutgers University, 101 WarrenStreet, Newark, New ]erse¾07102 USA

Many speciesof and mammals give alarm (Morse 1970).All 5 speciesare vulnerable to preda- signalswhen disturbedby predators.Although these tion by raptors and respond to each others' alarm alarm signals may appear altruistic, the alarm calls calls (Bent 1937, 1938; Gaddis 1980). Yet within these could benefit the caller. Individual fitnessmight be flocks2 of the ,chickadees and titmice, give promotedby decreasingthe probabilityof an attack, almost all of the alarm calls (Gaddis 1980). if the alarm call servesas a pursuit deterrent signal I examined the conditionseliciting alarm calls for or discouragesthe predatorfrom hunting in the im- 3 of these species(Black-capped Chickadees, Tufted mediate area (Smythe 1970, Trivets 1971, Woodland Titmice, and Downy )in winter flocks et al. 1980);by misdirectingthe predatorand increas- during encounterswith naturally occurring preda- ing the probability that another individual is at- tors (raptors)and predatormodels. Results indicate tacked (Perrins 1968, Charnov and Krebs 1977, Ow- interspecificdifferences that can be interpretedin ens and Goss-Custard1976); and by warning kin or the contextof benefit/costlogic. a mate (Maynard Smith 1965, Williams 1966, Sher- Observationswere made on alarm calling in mixed- man 1977). speciesflocks during 3 winters from November1979 The balance between the risks and benefits asso- to March 1982at the Great Swamp National Wildlife ciated with alarm calls presumablyinfluences the Refuge in New Jersey.I recordedthe flock compo- conditions under which alarm calls are given. For sition and alarm calls during encounterswith natu- example, the net benefit of alarm calling and the rally occurringpredators [Sharp-shinned Hawks (Ac- probabilityof giving an alarm call vary with age, sex, cipiterstriatus), Cooper's Hawks (Accipitercooperii), and and reproductive status (Sherman 1977). The eco- American Kestrels (Falco sparverius)]and predator nomics of alarm calling also may differ between models. speciesin mixed-speciesgroups, so that one species In the first predator-modeltests, a stuffedRed-tailed is more likely to call than another in a given situa- Hawk (Buteojamaicensis) was mounted on a pole and tion. Mixed-specieswinter flocks provide an oppor- presentedfrom a blind for 10 s to woodpeckersfor- tunity to examinethe situationsin which severaldif- aging 3-15 m away. I collecteddata only on wood- ferent speciesgive alarm calls. During the winter, peckerswith this model.I later replacedthis model mixed-species flocks composed of Black-capped with a stuffed Sharp-shinnedHawk mounted on a Chickadees(Parus atricapillus), Tufted Titmice (Parus pulley that ran down a wire betweentwo trees.The bicolor),and several follower species[Downy Wood- model was released from a blind and immediately peckers (Picoidespubescens), Hairy Woodpeckers (Pi- pulled back into the branchesof the tree. In tests coidesvillosus), and White-breasted Nuthatches (Sitta with this model, I collected data on woodpeckers, carolinensis)]are common in eastern North America titmice, and chickadeesforaging 2-15 m from the blind. I presentedpredator models from blinds to 20 in- •Present address:Department of BiologicalSci- dividually marked Downy Woodpeckersthat were ences,State University of New York, Albany, New foraging with a mixed-speciesflock, with a conspe- York 12222 USA. cific, or alone. Two days generally were allowed be- January1985] ShortCommunications 185 tween testson an individual ,although 2 birds. The female was not banded, and I never saw occasionallytwo woodpeckersforaging together were her either before or after that day. I observedall oth- exposedto the predatormodel twice in the sameday. er pairs foraging togetherat varioustimes through- Suet-filledtraps and feederswere used to capturethe out the winter. woodpeckersfor bandingand to attractwoodpeckers Chickadeesand titmice gave alarm calls signifi- to the blinds. Thesefeeders were set out sporadically cantly more often in responseto potential predators throughout the winter and removed from a 's than did Downy Woodpeckersboth in the absence territory at leasttwo daysprior to testingthe bird of conspecifics(X 2 = 71.57, n = 80, P < 0.01) and in with a predator model. Becausesquirrels were very the presenceof conspecifics(Fisher ExactProbability successfulat emptying the feeders, suet was never test, n = 36, P = 0.01). Black-cappedChickadees and available to woodpeckersfor more than two consec- Tufted Titmice gave alarm calls in 9 of 11 instances utive days. (82%; 5/5 for chickadees,4/6 for titmice) when for- The Black-capped Chickadees' high-frequency aging without conspecificsand in 18 of 21 instances whistle ("high zees")and the Tufted Titmice's"seet" (86%; 17/19 for chickadees,1/2 for titmice) when for- call were scoredas alarm callsfor thesespecies (Morse aging with at least one conspecific.For the pooled 1970, Ficken and Witken 1977, Gaddis 1980). The loud chickadeeand titmousedata the probability of alarm "tichrr" call or rattle call was scored as an alarm call callingdid not differ signfiicantlybetween when they for the Downy Woodpeckers(Lawrence 1967, Wink- were foraging with conspecificsand when they were let and Short 1978). foraging without conspecifics(Fisher ExactProbabil- Downy Woodpeckersresponded to the predator ity test, n = 32, P = 0.36). models, the naturally occurring predators, and the Both Downy Woodpeckers and Black-capped alarm calls of flock members in a similar manner Chickadees(Witkin and Ficken1979) appear to alarm (Sullivan in press).They froze for about 10 s, then call in the winter to protect their mates. Downy exhibited a high level of vigilance (measuredby Woodpeckersgive alarm calls less frequently and headcocks/min)and ceasedforaging for severalmin- more selectivelythan Black-cappedChickadees or utes. Chickadeesand titmice responded to the pred- Tufted Titmice.What differencesbetween parids and ator models,naturally occurringpredators, and alarm picids in the species-specificrisks and benefits of callsby diving into bushesand remaining quiet for alarm calls could account for this? several seconds to several minutes. It is doubtful that kin selectioncan be usedto fully Downy Woodpeckersgave alarm callsin only 3 of explain the chickadees'and titmice's propensity to the 11 encounters(27%) with live raptors, in only 1 alarm call. The Black-cappedChickadees in these of the 19 tests(5%) with the Red-tailed Hawk model, mixed-speciesflocks tend to be pairs from the pre- and in only 3 of the 50 tests (6%) with the Sharp- vious seasonand unrelated young (Wallace 1941, shinned Hawk model. In contrast, chickadees and Fickenet al. 1981),and pairs move from flock to flock titmice frequentlygave alarm calls.Chickadees gave during the winter (Ficken et al. 1981). alarm calls in all 7 encounters (100%) with live rap- Chickadees and titmice fill different roles than tors and in 14 of the 16 tests (88%) with the Sharp- woodpeckersin these flocks,which may affect the shinned Hawk model. Titmice gave alarm calls in 2 probability that they benefit from giving an alarm encounters (100%) with live raptors and in 3 of the call. Chickadees and titmice constitute the nucleus of 6 tests(50%) with the Sharp-shinnedHawk. these mixed-speciesflocks and usually are found in Downy Woodpeckersnever gavealarm callswhen the company of conspecifics(Morse 1970). Wood- foraging alone (0/46 instances),when with a flock peckers,on the other hand, are peripheral flock that did not include another woodpecker(0/23 in- membersand often drop out of the flock as the flock stances),or when with a Downy Woodpeckerof the leavesthe woodpecker'sterritory (Morse 1970,Sul- samesex (0/6 instances).This contrastswith the high livan 1984). Ficken et al. (1981) found strong associ- frequencyof alarm calls from woodpeckersforaging ationsbetween mated pairs of Black-cappedChicka- with a conspecificof the oppositesex (7/9 instances). deesin winter flocks.I observedDowny Woodpeckers Of the 7 alarm calls, 3 were given by femalesand 4 foragingwith their presumedmates in only 82 of 250 by males.On 6 of these7 occasionsthe woodpeckers (33%) sightings during 1979-1981. During the winter were foraging as a pair, and on the seventh occasion Black-cappedChickadees (and possiblyTufted Tit- the pair was foraging with 2 chickadeesand 2 nut- mice) may spendmore time foragingwith their mates hatches.The differencein propensityto give alarm than woodpeckersspend with their respectivemates, calls when foraging with a member of the same or and therefore the parids' mates may be more likely opposite sex was highly significant (Fisher Exact to hear and benefit from the warning calls. Probability test, n = 15 opportunities to alarm call, Differencesin how these speciesevade predators P = 0.006). The 2 instances when neither member of may affect the costof alarm calling. Chickadeesand a heterosexualpair gave an alarm call occurredon titmice immediatelydove into low busheswhen alarm the sameday (18 March 1982)and involved the same calls were given or the predator model was pre- 186 ShortCommunications [Auk, Vol. 102

sentedßIn contrast,Downy Woodpeckersfroze on 1938. Life history studiesof North Ameri- the side of a tree trunk or large branch, where they canbirds of prey, vol. 2. New York, Doverß may be more vulnerableto predatorsthan the chick- CHARNOV,E. L., & J. R. KREBS.1977. The evolution adees and titmice. The predator models were pre- of alarm calls: altruism or manipulation? Amer. sented at short distancesfrom the birds. This may Natur. 109: 107-112. have affected the relative rates of alarm calls if wood- FICatEN,M. S., & S. R. WIT•CIN. 1977. Responsesof peckershad a longer flight distancethan the parids Black-cappedChickadee flocks to predators.Auk and panicked during the tests. 94: 156-157. There are also differences in the acoustic charac- --, --, & C. M. WEISE. 1981. Associations teristicsof the alarm callsamong these species. High- among membersof a Black-cappedChickadee frequencyalarm callsmay be difficult for predators flock. Behav. Ecol. Sociobiol. 8: 245-249. to locate (Marlet 1955, 1957; Perrins 1968). I found GADDIS,P. 1980. Mixed flocks,accipiters and anti- the chickadees'"high zee" call and the titmice's"seet" predatorbehaviorß Condor 82: 348-349ß call more difficult to locate than the woodpeckers' GREIG-SMITH,P. W. 1981. The role of alarm re- lower-pitched "tichrrr" call, although I have no evi- sponsesin the formationof mixed-speciesflocks dence that hawks also found these calls more difficult of heathland birdsß Behav. Ecol. Sociobiol. 8: 7- to locate. 10. While the relative costs and benefits of alarm call- LAWRENCE,L. 1967. A comparativelife historystudy ing for paridsand picidsare speculative,chickadees of four speciesof woodpeckersßOrnithol. Mono- and titmice may experience greater benefits (the gr. No. 5. probability that a mate hearsthe warning call) and MARLER,P. 1955. Characteristicsof someanimal calls. smaller costs(the probability of a predatorlocating Nature 176: 6-8. and catchingthe alarm caller) when they give an 1957. Specificdistinctiveness in the com- alarm call than do Downy Woodpeckers.The chick- municationsignals of birdsßBehaviour 11: 13- adees'and titmice'spropensity to give alarmcalls can 37. then be exploitedby peripheralflock members, such MAYNARDSMITH, J. 1965. The evolution of alarm asthe DownyWoodpeckers. Greig-Smith (1981) found calls. Amer. Natur. 99: 59-63. that flocks apparently formed around Stonechats MORSE,D.H. 1970. Ecologicalaspects of somemixed- (Saxicolatorquata) because Stonechats gave alarm calls speciesforaging flocksof birds. Ecol. Monogr. at longer flight distances than the other species. 40: 119-168. Downy Woodpeckers benefit from foraging with OWENS,N. W., & J. D. GOss-CuSTARDß1976. The chickadee-titmouseflocks by decreasingthe time they adaptive significanceof alarm calls given by spendon vigilance,increasing their feedingrate, and shorebirdson their wintering groundsßEvolu- respondingto the other species'alarm calls(Sullivan tion 30: 397-398. 1984, in press).The conditionsunder which these PERRINS,C. 1968. The purposeof the high-intensity mixed-speciesflock members give alarm callssuggest alarm call in small passerines.Ibis 110:200-201. that the woodpeckersare exploiting chickadeesand SHERMAN,P. 1977. Nepotism and the evolution of titmice as sentinelsrather than participatingin a co- alarm calls. Science 197: 1246-1253. operative situation. SMYTHE,N. 1970. On the existenceof "pursuit in- This work was performedin partial fulfillment of vitation" signalsin mammals.Amer. Natur. 104: the Ph.D. requirements at Rutgers University. The 491-494. author is grateful to C. G. Beer, T. Caraco,M. Chai- SULLIVAN,K. 1984. The advantagesof social forag- ken, K. Egid, M. Ficken,L. Heisler, and two anony- ing in Downy WoodpeckersßAnim. Behav.32: mous reviewers for their helpful commentson this 16-22ß manuscript.I wish to thank the managerand staffof ß In pressßInformation exploitation by Downy the Great SwampNational Wildlife Refugefor allow- Woodpeckersin mixed-speciesflocksß Behav- ing me accessto the studysite and for their assistance iour. in the field. This researchwas supportedby a grant TRIVERS,R. L. 1971. The evolution of reciprocalal- from the Eastern Bird Banding Associationand the truismßQuartß Rev. Biol. 46: 35-57. Charles Tobach and Peter Nikolic Awards from the WALLACE,G.J. 1941. Winter studiesof color-banded T. C. Schneirla Research Fund. This is contribution chickadeesßBird-Banding 12: 49-64ß number 388 of the Institute of Animal Behavior, Rut- WILLIAMS,G.C. 1966. Adaptationand naturalselec- gers University. tion. Princeton, New Jersey,Princeton Univ. Pressß

LITERATURE CITED WINKLER,H., •t L. L. SHORT. 1978. A comparative analysisof acousticalsignals in piedwoodpeck- BENT,A. C. 1937. Life history studies of North ers (Aves, ).Bull. AmerßMus. Nat. Hist. Americanbirds of prey,vol. 1. New York,Dover. 160: 1-109. January1985] ShortCommunications 187

WIT•CIN, S. R., & M. S. FZC•CEN.1979. Chickadee alarm The "pursuit deterrent"function of alarm sig- calls:does male investmentpay dividends?Anim. nals. Amer. Natur. 115: 748-753. Behav. 27: 1275-1276. WOODLAND,D. J., Z. TAAFAR,& g. L. KNIGHT. 1980. Received20 January1984, accepted 30 July 1984.

Daily EnergyExpenditure by FemaleSavannah Sparrows Feeding Nestlings

JOSEPHB. WILLIAMS• AND KENNETHA. NAGY2 •NaturalScience Division, Pepperdine University, Malibu, California 90265 USA, and 2Laboratoryof Biomedical and Environmental Sciences, University of California, 900 VeteranAvenue, Los Angeles, California 90024 USA

The adaptivesignificance of many life-historytraits of femaleSavannah Sparrows while they were feed- of birds presumably can be understoodwithin the ing nestlings.Specifically, we wanted to comparethe context of patterns of energy allocation (Cody 1966, DEE of femaleswhile feeding a normal brood of 3 King 1974).Yet preciousfew dataexist for the energy youngwith thosefeeding 2 young,late in the nest- expenditureof free-living birds during their repro- ling period when energy demandspresumably are ductive period. In particular, little is known about greatest. the energy requirementsof females while they are Thestudy area and birds.--Our study area lay in the feeding their young. middle and upper littoral zone of a large saltmarsh Most extant information on the energeticcost of locatedon the Point Mugu Naval Air Station,Pt. feeding young has been gathered indirectly by mea- Mugu, California (34ø07'N,119ø07'W). The vegeta- suring the weight lossof parentsas they feed differ- tion consistedof, in decreasingorder of importance, ent numbersof young (Hussell 1972) or from studies Salicorniavirginica, Frankenia grandifolia, Batis maritima, of cagedparents (Brisbin 1969). With the advent of and Monanthochloelittoralis. Large, barren salt pans the doubly labeled water (DLW) technique(Lifson et and shallow tidal channelsintermixed with patches al. 1955,Nagy 1975),reasonable estimates of the dai- of vegetationto form a mosaicof sitesused by Sa- ly energy expenditure(DEE) of free-ranging birds vannah Sparrowsfor foragingand nesting. now can be obtained (Nagy 1980). In essence,the The breedingbiology of the SavannahSparrow at techniqueinvolves isotopiclabeling of an animal's Pt. Mugu has been studiedextensively by J. B. Wil- body water with oxygen-18and tritium or deute- liams and will be reported in detail elsewhere. In rium. From the difference between the turnover rates brief, thesesmall, ground-nestingpasserines reside of the two isotopes,the rate of CO2 production can there year-roundand commencebreeding in early be measured.When we concurrently comparedCO2 April. Males defend territories (ca. 0.1 ha) and aid in production in Savannah Sparrows (Passerculussand- provisioningthe young, but only femalesincubate wichensis)using DLW and standard laboratory tech- eggs.Of 83 nestsfound in 1978 and 1979,78% con- niques,we found a mean differenceof +6.5% (range tained 3 eggs,18% held 4 eggs,and only about4% -0.2 to +11.0, n = 7), suggestingthat acceptablees- held 2 eggs.Mean clutchsize for these83 nestswas timates of CO2 can be obtained while these birds are 3.16 + 0.46SD. In this studywe removeda nestling functioning in their normal environment (Williams from broodsof 3 to makebroods of 2 youngseveral and Nagy 1984a).Furthermore, Nagy and Costa(1980) days prior to DLW measurement. have shown that water flux rates estimated with tri- Field and laboratorymethods.--We first determined tiated water vary within +10% of actual flux rates in the impactthat capturingand handlingfemales had most situations. on their subsequentbehavior and the time required Using the DLW method, Utter and LeFebvre(1973) after disturbancefor femalesto resumenormal pa- calculatedthat female Purple Martins (Prognesubis) rental activities.To this end, we gave the birds sham with nestlingsmetabolized an averageof 183.6 kJ/ injections of distilled H20 and restrained them in a day (n = 2; mean wt = 47.7 g), which is somewhat cloth bag for 1 h. We then releasedthe female bird higher than that of malesduring the sametime pe- and watched from a blind for her return. In most riod (142.9 kJ/day; n = 2). Males apparentlydid not cases, females handled 3-4 h before sunset resumed feed nestlings as much as did females. Unfortu- caring for young within 2-3 h after their release; nately,the numberand ageof the nestlingswere not however, about 20% of the females that we netted reported. Hails and Bryant (1979) found that female abandonedtheir young.All birds for which we pres- CommonHouse-Martins (Delichon urbica, 20 g) feed- ent data in this studyresumed feeding their young ing young metabolized 75.3 kJ/day, but the DEE of within 3 h after their release. Each female was color- female birds was not correlated with brood mass. bandedto facilitaterecognition. In this study, we used DLW to measure the DEE We capturedbirds by placingmist netsaround their