April 1999] ShortCommunications 557

The Auk 116(2):557-559, 1999

FunctionalRoles in Mixed-SpeciesForaging Flocks: A Field Manipulation

ANDREW S. DOLBY• AND THOMAS C. GRUBB,JR. BehavioralEcology Group, Department of Evolution,Ecology, and Organismal Biology, Ohio State University, Columbus,Ohio 43210, USA

Mixed-speciesforaging flocks occur in a varietyof woodlots,intraspecific aggression of White-breasted habitats(Winterbottom 1949, Moynihan 1962, Mc- Nuthatchesand Downy Woodpeckersprohibited us Clure 1967, Morse 1970, Greig-Smith1978, Powell from adding extra satelliteindividuals to woodlots 1985,Eguchi et al. 1993),and participantsin such from which we had removedparids. flocksare thoughtto acquireforaging and predator- Methods.--Weconducted the experimentduring avoidanceadvantages (McClure 1967, Morse 1977, the winters of 1995-1996 and 1996-1997 in woodlots Barnardand Thompson1985). Descriptions of mixed- within the agriculturallandscape of Unionand Del- speciesflocks often include attemptsto categorize awarecounties, Ohio. These topographically flat de- participatingspecies according to their presumed ciduousforest fragments consisted primarily of oaks functionalroles in the flocks.Some species, referred (Quercusspp.), ashes (Fraxinus spp.), shagbark hick- to as nuclearor corespecies, appear to facilitateflock ory (Caryaovata), sugar maple (Acersaccharum), and formationand to initiateflock movements. Other spe- American beech (Fagusgrandifolia). Each woodlot cies,known as satelliteor attendantspecies, seem to was completely surrounded by cultivated fields, behaveas flockfollowers (Winterbottom 1943, Moy- lackedany connectionto otherwoodlands, and was nihan 1962, McClure 1967, Morse 1970, Austin and so small that it containedonly one mixed-species Smith1972, Buskirk 1976, Powell 1985, Hutto 1994). flock.During the study period, such flocks were nev- Suchfunctional roles have been applied to mem- er observedto crossopen ground into neighboring bers of mixed-speciesflocks that form during the woodlots. In addition to the speciesmentioned winterwithin the assemblageof bark-foragingbirds above,the studywoodlots usually contained one or in easternNorth America (Morse 1970). Developing two Red-bellied Woodpeckers(Melanerpes carolinus) during early autumn,these flocks typically consist of and occasionallya Hairy Woodpecker(Picoides vil- two parids,Tufted Titmouse(Baeolophus bicolor) and losus)or Brown Creeper (Certhiaamericana). eitherCarolina (Poecile carolinensis) or Black-capped Sixteenwoodlots were used in the experiment.Be- (P.atricapillus) chickadee, and severaladditional spe- fore eachfield season,four of the eightwoodlots to cies,including Downy Woodpecker(Picoides pubes- be usedin that yearwere randomly assigned to the cens)and White-breastedNuthatch (Sitta carolinen- controlgroup and four to the parid-removalgroup. sis).Within theseflocks, the paridshave been clas- Controlwoodlots averaged 5.3 _+SD of 2.2ha in size, sified as nuclearspecies, whereas the woodpeckers and removalwoodlots averaged 5.5 -+ 2.1 ha. Each and nuthatcheshave been categorizedas satellite woodlot was used only once. species(Morse 1970, Sullivan 1984a). Frommid-December to mid-Januaryeach year, we Althoughobservational evidence suggests that nu- trapped and mist netted birds attracted to feeders clear speciesfacilitate flock cohesion(Moynihan and wire-meshsuet cages. All DownyWoodpeckers 1962,Morse 1970, Powell 1985), this hypothesishas and White-breasted Nuthatches were fitted with receivedlittle experimentalsupport (but seeMonk- USFWS bands and uniquely colored plastic leg konenet al. 1996).We predictedthat if paridsmain- streamers for individual identification. Chickadees tain flockcohesion, then woodpeckersand nuthatch- and titmice in control woodlots were similarly es(i.e. satellitespecies) would occur together less of- marked, but those in treatment woodlots were re- ten within isolated woodlots from which titmice and moved and releasedapproximately 50 km away in chickadees had been removed than in control wood- suburban Columbus, Ohio. lots in whichthese nuclear species were present. By Observationsof flockingbehavior took place during removing parids from woodlots, we manipulated two-hour visits to woodlots conducted between 0800 flock size and flock composition.Therefore, our re- and 1600.To balancesampling across time of day and suitscould be interpretedas being attributable to ei- season,we alternated visits to treatment and control ther variable.Although we would havepreferred to sites. Each woodlot was visited 10 times, or until new hold group size constantin treatmentand control paridsbegan arriving in the treatmentwoodlots in late Februaryduring the juvenile dispersal period (T. C. Grubbpers. obs.). During eachvisit, we assigned • Presentaddress: Department of Biology,Univer- thelocation of individuallymarked birds to oneof the sity of SouthFlorida, Tampa, Florida 33620, USA. E- 25 x 25-mblocks that formed a grid thatcovered the mail: [email protected] entire area of the woodlot. 558 ShortCommunications [Auk, Vol. 116

TABLE1. Number of individual nuclear(i.e. titmiceand chickadees)and satellitespecies present per wood- lot initially (i.e.before removal) and on averageper visit overthe courseof the observationperiod in eight control(i.e. parids not removed)and eight treatment(i.e. parids removed) woodlots. Values are œ_+ SE.

Species Observationperiod Control Treatment Nuclear species Parids Initial 6.9 _+ 1.1 6.5 _+ 0.5 Average per visit 6.1 _+1.1 0.4 _+0.1 Satellite species Downy Woodpecker Initial 3.4 _+0.4 2.6 _+0.3 Average per visit 3.4 _+0.4 2.6 _+0.3 White-breasted Nuthatch Initial 1.9 _+ 0.1 1.6 _+ 0.5 Average per visit 1.9 _+0.1 1.5 ñ 0.5

Upon enteringa woodlot, we usually locatedbirds The proportionof woodlot visits during which at by sound.To minimize chancesthat our presencein least one White-breastedNuthatch and one Downy woodlots would cause the birds to move before their Woodpeckerwere observedin associationwas sig- positionscould be recorded,we identified the loca- nificantlyhigher in controlthan in treatmentwood- tion of eachbird as quicklyas possible instead of us- lots (control, • = 0.81 +_0.18; treatment, • = 0.44 _+ ing predeterminedtransects. The sameentry point 0.16; F = 6.27, df = 1 and 13, P = 0.026). During a was used for each visit, and all color-marked birds mean of 82.4 _+SE of 6.5% of the visits, at least one consistentlywere locatedwithin two hours. woodpeckerand one nuthatch were in the same Pravosudovand Grubb (1999) found that the ben- blockas the majorityof the parid group,or in blocks efits of flockingbegin accruingwith a groupsize of adjacentto the parid group. two. Therefore, we defined a satellite bird association Discussion.--DownyWoodpeckers and White- as an instancein which at leastone Downy Wood- breasted Nuthatches associated with each other more pecker and at least one White-breastedNuthatch often in control than in the treatment woodlots from were recordedwithin the sameor adjacent25 x 25- whichparids were removed. In addition,both species m block. Becausewe conductedmultiple surveysin were found in closeproximity to parids during most eachwoodlot, we basedour analyseson thepropor- of our visitsto controlwoodlots. These results suggest tion of visits to each treatment or each control wood- that the presenceof parids enhancesassociation be- lot during whichwe observeda satelliteassociation. tweenthe satellitespecies, and they supportthe hy- Speciescomprising mixed-species flocks are nonran- pothesisthat paridsfacilitate flock cohesion. domly distributed in space(Monkkonen et al. 1996). An alternativeexplanation for our data, however, Furthermore,bark-foraging birds may be nonran- is that reducedflock size aloneproduced the reduc- domly distributedin woodlotsowing to suchfactors tion in association between nuthatches and wood- as differentialexposure to wind and solarradiation peckers.It is possiblethat a critical group size needs (Grubb 1975). Consequently,we confinedour anal- to be reachedbefore flocking behavior is stimulated yses to differencesin satellite bird associationsbe- betweenthese species. Because the benefitsof flock- tween control and treatment groups. ing begin to accumulatewith just two individuals We usedanalysis of covariance(ANCOVA) to testfor (Pravosudovand Grubb 1999),satellite species seek- a differencebetween the mean proportion of visits(arc- ing benefitsspecific to group size without regard to sine-transformed)to treatment and control woodlots groupcomposition should have been just aslikely to during whichat leastone satellite association was ob- form groupswith eachother in treatmentwoodlots served. To control for different numbers of satellite in- as in controlwoodlots. The enhancedstimulus pro- dividualsamong woodlots (Table 1), we includedthe vided by increasedgroup size alone,however, may numberof satelhteindividuals per ha in eachwoodlot promote flock participation by woodpeckersand as a covariate in our ANCOVA model. The woodlot nuthatches. servedas our unit of statisticalindependence. The chickadees and titmice of North American Results.--Exceptfor two titmice during the first mixed-speciesflocks have several attributes in common winter and two titmiceand onechickadee during the with other nuclearspecies (Moynihan 1962, Hutto secondwinter, all parids were removed from the 1994).These parids (1) leadheterospecifics more often eight experimentalwoodlots during the initial cap- thanthey follow (Morse 1970), (2) consistentlyassociate ture period (Table1). The averagenumber of parids with conspecifics(Hogstad 1989) as well as with het- removedper treatmentwoodlot was 5.6. During the erospecifics(Morse 1970), and (3) give more frequent observationperiod, 73 and 74 visits were made to and conspicuousalarm callsthan do woodpeckersor treatment (2 = 9.1 ñ 0.9 visits) and control (• = 9.3 nuthatches(Sullivan 1985). Parids with comparable _+0.9 visits) woodlots, respectively. propertiesare thoughtto behaveas nuclearspecies in April 1999] ShortCommunications 559

othermixed-species flocks (Greig-Smith 1978, Szekely GREIG-SMITH,P. W. 1978. The formation, structure et al. 1989,Monkkonen et al. 1996). and function of mixed-speciesinsectivorous Monkkonenet al. (1996)found that heterospecifics bird flocks in West African savanna woodland. were attracteddifferentially to playbacksof Willow Ibis 120:284-295. Tit (Parus montanus)calls relative to control calls. GRUBB,T. C., JR.1975. Weather-dependent foraging Similarly,Downy Woodpeckers and White-breasted behaviorof somebirds wintering in a deciduous Nuthatchesmay be attractedto TuftedTitmouse and woodland. Condor 77:175-182. CarolinaChickadee vocalizations. This possibility HOGSTAD,O. 1989. Social organization and domi- mayexplain how paridsfacilitate flock cohesion. nancebehavior in someParus species. Wilson Wedo not claim, however, that parids actively recruit Bulletin 101:254-262. woodpeckersand nuthatchesto flocks.It is possible HUTTO,R. L. 1994.The compositionand socialor- thatbeing followed by otherspecies is a selectivelyneu- ganizationof mixed-speciesflocks in a tropical tralby-product of thecommunication system of parids deciduous forest in western Mexico. Condor 96: thatis exploitedby woodpeckersand nuthatches. In a 105-118. playbackexperiment, Sullivan (1984b) found that MCCLURE,H. E. 1967.The composition of mixedspe- woodpeckersresponded to playbacksof parid alarm cies flocks in lowland and sub-montane forests callsby freezingand thenincreasing their rate of head of Malaya. Wilson Bulletin 79:130-154. cocking.After an alarm reaction,they resumed their MONKKONEN, M., J. T. FORSMAN,AND P. HELLE. 1996. foragingactivity more quicklywhen the alarm-call Mixed-speciesforaging aggregations and het- playbackwas followedby playbackof parid contact erospecificattraction in boreal bird communi- ties. Oikos 77:127-136. calls (Sullivan 1984b).Sullivan's (1984b) experiment showedthat Downy Woodpeckers reduced their vigi- MORSE,D. H. 1970. Ecologicalaspects of some lancerates in responseto theinformation contained in mixed-speciesforaging flocks of birds.Ecologi- paridvocalizations. This behavior also may be exhib- calMonographs 40:119-168. MORSE,D. H. 1977.Feeding behavior and predator itedby White-breasted Nuthatches, and exploitation of avoidancein heterospecificgroups. BioScience vocalcommunication may be a primarybenefit gained 27:332-339. by bothsatellite species when they flock with titmice and chickadees. MOYNIHAN,M. 1962.The organizationand probable evolutionof somemixed-species flocks of Neo- Acknowledgments.--Wethank Elena Pravosudova tropicbirds. Smithsonian Miscellaneous Collec- andMichelle Crow-Dolby for theirassistance in the tions 143:1-140. field,and the Lowe, Michaels, Parrot, Spurgeon, Gey- POWELL,G. V. N. 1985. Sociobiologyand adaptive er,Thomas, Schmitter, John, Goddard, Styre, MayerE, significanceof interspecificforaging flocks in Gray,Raye, Miller, Mackan,and Nogglefamilies for theNeotropics. Pages 713-732 in Neotropicalor- allowing accessto their woodlots.The commentsof nithology(P. A. Buckley,M. S. Foster,E. S. Mor- T. A. Waite, W. M. Masters, J. B. Williams, G. G. Bernt- ton, R. S. Ridgely,and E G. Buckley,Eds.). Or- son,and W. H. Buskirk,and two anonymousreview- nithologicalMonographs, no. 36. ersimproved this manuscript.Funding was provid- PRAVOSUDOV,V. V., AND t. C. GRUBB,JR. 1999. Effects ed by theWilson Ornithological Society and the Ohio of dominance on vigilance in avian social State University Alumni Association.During the groups.Auk 116:241-246. courseof thisproject, TCG was funded by NSFgrant SULLIVAN,K. A. 1984a.The advantagesof socialfor- IBN-9522064. agingin DownyWoodpeckers. Animal Behav- iour 32:16-22. LITERATURE CITED SULLIVAN,K. A. 1984b.Information exploitation by Downy Woodpeckersin mixed-speciesflocks. AUSTIN,G. T., ANDE. L. SMITH.1972. Winter foraging Behaviour 91:294-311. ecologyof mixed insectivorousbird flocks in SULLIVAN,K. A. 1985. Selective alarm calling by oak woodland in southern Arizona. Condor 74: Downy Woodpeckersin mixed-speciesflocks. 17-24. Auk 102:184-187. BARNARD,C. J., AND D. B. A. THOMPSON.1985. Gulls SZEKELY,t., T. SzEP,AND T. JUHASZ.1989. Mixed spe- and plovers:The ecologyand behaviourof ciesflocking of tits (Parusspp.): A field experi- mixed-speciesfeeding groups. Croom Helm, ment. Oecologia 78:490-495. London. WINTERBOTTOM,J. M. 1943.On woodlandbird par- BUSKIRK,W. H. 1976.Social systems in a tropicalfor- ties in northern Rhodesia. Ibis 85:437-442. est avifauna. American Naturalist 110:293-310. WINTERBOTTOM,J.M. 1949.Mixed bird partiesin the EGUCHI, K., S. YAMAGISHI, AND V. RANDRIANASOLO. tropics,with specialreference to NorthernRho- 1993.The compositionand foragingbehaviour desia. Auk 66:258-263. of mixed-speciesflocks of forest-livingbirds in Received17 February1998, accepted 30 October1998. Madagascar.Ibis 135:91-96. AssociateEditor: L. J. Petit