Effects Of Plant Species And Foliage Structure On The Foraging Behavior .

674ROBINSONANDHOLMES[Auk,Vol. 101(seeSiccamaet al. 1970);theseare rarely usedas foraging substratesby the bird speciesconsideredinthis paper (Holmes et aL 1979).al. 1979 and Robinson and Holmes 1982, for moreobservedsquare tests.detailed definitions): (1) Glean--a perched bird attempts to capture prey from a nearby substrate;noflight is involved for either bird or prey. (2) Hover-a bird flies out to catch prey from a more distantMETHODSsubstrate;the bird maypauseand hovermomentarilyBird foraging behavior was observed during theat the substrateor may snatchthe prey from the subbreedingseasons,approximatelylate May to mid-July, strate as it passes.(3) Hang--a modified glean inin the years 1974-1979;mostintensivesamplingef- which the bird hangs,usuallyupsidedown,and oftenforts were in 1975, 1976, and 1978. Foraging rates tearsapart or in someway manipulatesthe substratewere quantifiedonly during the nestling and early (e.g. a curled leaf). (4) Hawk--a bird flies out from afledgling stagesof the nesting cycles(mid-Juneto perch to catchflying prey. (5) Flush-chaseor tummid-July),when birds were feeding intensively.Inble--a bird chaseselusive or evasive prey flushedeachyear,we systematicallywalkedthroughthe study from foliage, often in a long downward flight. Staplot and observedasmanydifferentbirdsaspossible, ny of which were individually color-marked.We the prey-attackcategorieswere determined by Chiat least 20-25individualadults of the morecommonspecies(Red-eyedVireo, redstartand BlackthroatedBlue Warbler)eachyear and 4-6 of the lessabundantPhiladelphiaVireo. Wheneveran activelyforaging bird was encountered,we followed it for aslong as possible(which, becauseof the densefoliage,was usually 60 s) and recordedits locomotorandforaging patterns on a portable tape recorder. Wespecificallynotedthe numberof perchchanges(eitherhopsor flights)madeby the bird while searchingforprey,the frequencyof attacksit directedtowardprey,the plant species,and height above ground. This informationwas transcribedfrom the tapesby meansof a stopwatch.Becauseindividual birds seldomremainedlong in the samestratumor tree species,manyof the sequencesobtained were very short ( 30 s).For this reason,rather than calculatinga separateratefor each sequence,as we did in a previous paper(Robinson and Holmes 1982), here we have addedtogether the sequencesoccurringwithin each foliagecompartment to obtain a single compositesequence,for which we then calculateoverall flight, hop, andprey attackrates.In this procedurewe follow Moermond (1979a, b) and Eckhardt (1979).For statisticalcomparisonsof locomotor patterns,Lengthsof hover flights were used as an index ofthe bird's maximum effective search radius (cf. Robinson and Holmes 1982); becauseof their skewed dis-tributions,hover lengthswere comparedstatisticallywith a median test. For testsinvolving many multipie pairwise comparisons(H/F ratios, hover-flightlengths), we adjustedalpha levels, accepting 0.001as our minimum level of significance.This was necessary,becauserepeatedpairwise useof the samestatistical test invalidates the assumptionof independenceand raisesthe likelihood of Type ! error. Forother tests, statistical significancewas acceptedatP 0.05.RESULTSRed-eyedVireo.--At Hubbard Brook, the Redeyed Vireo breeds commonly throughout theforest.Its dietconsistsprimarilyof Lepidopteralarvae and beetles,which it capturesby hovering or sometimesby gleaning from leaf sur-faces(Robinsonand Holmes 1982).It foragesmostlyin the subcanopyand lower portionsofthe canopy (Robinson 1981) and shows littlewe summedthe number of originally observedforpreferencefor foragingin any particulartreeaging sequencesof 10 s or more in which the hop/flight (H/F) ratio was greater than 1 or lessthan or species (Holmes and Robinson 1981). Whenequal to 1. Using Chi-squaretests,we then compared foraging,it searchesfoliage within abouta 0.5frequenciesof each bird speciesamong the foliage 1-m radiusand then flies at an averagerate ofcompartments.With this method, we avoided the about 9/rain to a new perch. These searchproblem of serial dependenceof flights and hops flightsaverage1.1m in length.Betweenflights,within a sequence(Williamson 1971;Smith 1974a,b).We used the attack rate as an index to how oftensuitable prey were encounteredin a particular foliage compartment.Becausethe goal of food-searching behavior is to find prey, the attackrate reflects"searchingsuccess."We recognizethat this is not thesame as "foraging success,"which would requirepreciseinformation on the percentageof successfulattacks and the size, kinds, and nutritional value ofprey taken.Attackswere categorizedas follows (seeHolmes etthis vireo hops along branchesand twigs at arateof 7/rain while it visuallysearchesnearbyfoliage (Robinson and Holmes 1982).Theseforagingpatternsof the Red-eyedVireo, however, vary in specificways among thedifferent foliage compartmentsof this forest(Table 1). In white ash and the shrub layer, itfliesfrequentlybut seldomhops;in the canopyof the three dominanttree species,it hopsandflies at intermediate rates, whereas in the sub-

El. Foragingratesand searchradii of Red-eyedVireosin eachof the majortree speciesand strataofthe Hubbard Brook Forest.The flight rate is for searchflights only and does not include flights used inattackingprey. The searchradiusis indicatedby the mean length of prey-attackinghover flights.Foragingrate (number/min)SearchStratumCanopySubcanopyShrubTree species(t)aWhite ashPreyattacksflights(F)Search radiusHops(H)(in cm) ? SD (n)b(764)2.114.53.458 30 (22)Yellow birchAmerican beech(2,330)(721)2.42.28.57.46.45.772 36 (59)68 36 (45)Sugarmaple(1,523)2.66.26.173 66 (104)Yellow birch(2,934)2.25.212.374 71 (90)American beech(2,631)2.45.911.373 35 (158)Sugar maple(2,013)2.56.012.5(523)2.413.72.8All74 40 (154)68 33 (47)cumulativenumber of secondsthat actively foraging birds were observed.number of hover flights observed.canopy of thosesametree species,it hops moreoften and flies lessfrequently (Table 1). Statis-tical comparisonsof the hop/flight (H/F) ratiosin each tree speciesand stratum (Table 2) substantiate these trends. The patterns in ash andin the shrub layer are similar and often statistically distinctfrom thosein other foliage compartments. Although there are no significantdifferencesin the H/F ratios among the threedominant tree specieswithin layers, 7 of the 9possiblecomparisonsbetween layersare statistically different (Table 2). Thus, within a stratum, the searchingpatternsof this vireo appearto be unaffectedby differencesin foliage structure among trees (except for ash). Betweenstrata,however, their behavior often changes,even within a particular tree species.The maneuvers used by the Red-eyed Vireoto captureprey from foliagediffer only slightlyamongtree speciesor strata(Fig. 1). The onlystatisticallysignificantdifference is that theyhover more often in canopythan in subcanopyyellow birch (P 0.001,Fig. la). Mean lengthsof hovering attacksrange from 58 cm in ash to74 cm in subcanopybirch and maple (Table 1),but there are no statisticallysignificantdiffer-TABLE2. Resultsof pairwisestatistical(X2)comparisonsof hop/flight (H/F) ratiosof Red-eyedVireosamongtree speciesand strata[* P 0.001 chNSNS--.NSNS--*NSNSBeechShrubH/F - hrubAllPercentageof foraging sequencesin which searchflights were more numerousthan hops.n number of observedforaging sequences.

676ROBINSONANDHOLMES[Auk,Vol.101RED-EYED VIRED.CANOPYBeechIII8)S or maple(212)Yellowbirch(180 AshSUBCANOPY.SHRU BBeech(326) Sugarmaple(398) Yellowbirch(240) Allspecies(14I)Ha GiHab)HaGIHaHaGI HaHaGI ellowbirch[210) Ash(114)Beech(76)Sugarmaple(125)Yeliowbirch( )Ha GI HoHaGI HaBOSUBCANOPY8OC)SHR.UBAllspecies[109)HaGI HeHaGI ellowbirch[1 8)Sugarmaple[330)Yetlowbirch[ 39)Allspecies(B7)Ha GI FI Hk SUBCANOPYBeech021) GI FI HkAsh (34)SHRUBGI FI HkGI FI HkBLACK-THROATED BLUE WARBLERd) SUBCANOPYBOBeech(193)S[ m mOple(229)Yetlawbirch(315)Str maple(42)SHRUB80Beech(82)S[ . ngpreyin eachofthemajortreespeciesandstrataof theHubbardBrookForest.Numbersof observedpreyattacksin eachfoliagecompartmentaregivenin parentheses.Seetextfordefinitionsof strataandprey-attackmaneuvers.Ho hover,G1 glean,Ha hang, F1 flush-chase,Hk hawk.encesin thesehover lengthsamongfoliage to differentstrataandto whiteashby changingcompartments.We concludethat the Red-eyedVireo is relatively consistent in its use of attack maneu-its hop and flight rates.Perhapsas a resultoftheseadjustments,the vireo'soverallprey attackrateremainsnearlyconstantamongthesevers,hasa constantsearchradius,but responds foliage compartments(Table 1).

E3. Foragingratesand searchradii of PhiladelphiaVireosin eachof the majortree speciesand strataof the Hubbard BrookForest.The flight rate is for searchflights only and doesnot include flights used inattackingprey. The searchradius is indicatedby the mean length of prey-attackinghover flights.Foragingrate (number/min)SearchStratumCanopyTree speciesWhite ashHops(H)(in cm)œ SD (n)b(721)3.613.52.256 24 (34)3.61.612.49.33.13.963 29 (89)78 16 (7)Sugar maple(825)2.311.62.467 22 (22)Yellow birch(688)3.89.23.765 33 (47)(1,276)2.810.55.268 29 (59)Sugarmaple(701)4.69.95.168 34 (71)All(742)3.212.32.871 38 (44)American beechShrubsflights(F)(1,874)(441)Yellow birchAmerican beechSubcanopy(t)aSearch radiusPreyattackscumulativenumber of secondsthat actively foraging birds were observed.number of hover flights observed.PhiladelphiaVireo.--At Hubbard Brook, thePhiladelphiaVireo is patchily distributedin theforest,occurringin areaswhere white ash andyellow birch are well represented in the canopy (Robinson 1981). In addition, it exhibitsstrong preferences for foraging and nestinghigh in the canopy,particularly in ash and yellow birch (Holmes and Robinson 1981, Robin-son 1981). Its foraging behavior and diet areotherwise very similar to those of its congener,exceptthat it tendsto fly more often while foraging (Robinsonand Holmes 1982).Hopping and flight ratesof the PhiladelphiaVireo while searchingfor prey are similar inthe variousfoliage compartments(Table 3), especially when they are contrastedwith thoseof the Red-eyedVireo (seeTable 1). In eachtreespeciesand stratum, the Philadelphia flies 2-5times more frequently than it hops. Like thered-eye, the Philadelphia Vireo tends to hopmore frequently in the subcanopythan in thecanopy,although the differencesare slight (Table 3), and the H/F ratiosbetweenfoliage compartments are not significantly different.Unlike its congener, however, the Philadelphia's attack rates vary among foliage compartments (Table 3). Within the canopy, it attacksprey almostthree times more often in ashand yellow birch than in beech. In fact, in 11of the 18 timesthat we observeda PhiladelphiaVireo feeding in a beechcanopy,it left withoutmaking a single attack. This contrastswith 12of 71 visits in canopyyellow birch (P 0.001)and 8 of 31 in ash(P 0.01).In the subcanopylayer, the Philadelphia'sattackrate is higher insugar maple and yellow birch than in beech.Likewise, attack ratesare higher in the subcanopiesof sugarmaple and beechcomparedwiththosein the canopiesof thesesametree species(Table 3). Thus, in contrastto the Red-eyed Vireo, the Philadelphia appearsto find prey moreoften in some foliage compartmentsthan inothers.The Philadelphia Vireo at Hubbard Brookprimarily hovers for prey, gleans never accountingfor more than 22%of the leaf-directedattacks(Fig. lb). There are no statisticallysignificant differences(P 0.10) in the frequenciesof maneuversusedamongthe foliagecompartments. Likewise, hover distances of thePhiladelphia Vireo do not vary significantlyamong foliage compartments(Table 3), indicating that this species,like its congener,has arelatively constantsearchradius.Thus,the Philadelphia Vireo is relatively stereotypedin its foraging patterns;it searchesforand attacksprey in the samemanner regardlessof tree speciesand strata. Its rate of prey encounter varies among foliage compartments,however, being highestin canopyash and yellow birch and in subcanopyyellow birch andsugar maple.AmericanRedstart.--Theredstartforagesat allheights, from on or near the ground to the topof the canopy (Holmes et al. 1978, 1979). Itsearchesthe different tree speciesapproximately in proportion to their occurrence (Holmesand Robinson 1981). While foraging, it hops

678ROBINSONANDHOLMES[Auk,Vol. 101TABLE4. Foragingratesand searchradii of AmericanRedstartsin eachof the majortree speciesand strataof the HubbardBrookForest.The flight rateis for searchflightsonly and doesnot includeflightsusedinattackingprey. The searchradiusis indicatedby the meanlengthof prey-attackinghoverflights.Foraging rate (number/min)SearchSearch radiusTree species(t)aPreyattacksCanopyWhite ashYellow birchAmerican 9.711.513.510.29.416.916.218.050 47 46 51 SubcanopyYellow birchAmerican beechSugar 915.413.948 42 (71)56 34 (75)43 24 (158)ShrubAll(800)4.118.29.647 26 (144)Stratumflights(F)Hops(H)(in cm) SD (n) 23 (10)26 (50)29 (26)30 (19)cumulativenumberof secondsthat activelyforagingbirds were observed.numberof hover flightsobserved.and fliesrapidly betweenperches,makesshorthovering maneuversto attack prey on leaves,and often chasesprey it flushes(RobinsonandHolmes 1982).It frequently fans and pumpsitscontrastinglypatterned tail from side to sideand flicksits wings while hopping throughfoliage,a behaviorthat appearsto flushprey (Rootple (P : 0.001). In the canopy,the patternsinyellow birch differ significantlyfrom thoseinboth beech(P 0.001)and sugarmaple (P 0.001). These differencesoccurlargely because1967, Robinsonand Holmes 1982). It also hawksnificant differences (P 0.001) in manueuverinsectsfrom the air more often than any othercommon bird species at Hubbard Brookfrequenciesbetween canopy and subcanopywithin both birch and sugar maple (Fig. lc).Although the lengthsof redstarthoversvaryfrom 43 to 56 cm amongthe foliage compart-(Holmes et aL 1979).redstarts use the flush-chasemaneuvermoreoften in yellow birch than in the other treespecies(Fig. lc). There are alsostatisticallysig-When searchingfor prey, the redstart hopsand flies at similar ratesin all foliage compart- ments (Table 4), the differences are not statisments,exceptin white ashand the shrub layer, tically significant.Hence, again, as with thewhere the flight ratesare elevated and greater vireos, the redstart's searching radius seemsthan the hop rates (Table 4). Its H/F ratios are relatively fixed.We alsoquantified the occurrenceof tail-fanstatisticallyindistinguishablein canopy andsubcanopyfoliageof the dominanttreespecies, ning behavior by redstartsin eachfoliage combut, in the comparisonsinvolving ash and the partment. The data indicate that redstarts fanshrub-layer data, there are significant differ- their tails more often in yellow birch canopyencesin 3 of 7 cases(more flights in both ash and subcanopythan in any other foliage unitand shrubscomparedwith canopy birch and (Table 5). Yellow birch alsohostssignificantlysubcanopybirch and beech). Thus, as in the more insect prey than any other tree speciesRed-eyed Vireo, the redstart changeslocomo- (Holmesand SchultzMS), especiallyleafhoptor patternswhen searchingfor prey in white pers (Homoptera: Cicadellidae). These oftenash and the understory.hop from their substrateswhen disturbedandThe rate at which redstartsattackprey is sim- appearto be the prey chasedmostoften by redilar in the different foliage compartments,ex- starts in the tumble-maneuver (Robinson andcept in yellow birch, in which prey were en- Holmes pers. obs.).Hence, the tree speciesincounteredslightly more often (Table 4). Its use which the redstart fans its tail most, yellowof specificattack maneuvers,however, often birch, is also that in which it encounterspreyvariesamongfoliagecompartments(Fig. lc). In at the highestrate (asindicatedby attackrates)the subcanopy,there is a significantdifference and in which it uses the flush-chase maneuverin maneuver use between birch and sugarma- most often.

679FoliageStructureand Bird ForagingOctober 1984]TABLE5. Frequency of tail-fanning by AmericanRedstartsin the Hubbard BrookForestin the majortree speciesand strata. Data are from the postfledgingperiod (late Juneto mid-July).often than it flies and searchesnearby substrateswhile rapidly moving from perch toperch. Its primary prey-attackingmaneuver isthe hover (Robinson and Holmes 1982).HopswithtailfannedStratumCanopyTree speciesWhite ashSubcanopyShrubna(%)6555.4Yellow ellow birch45574.4American38158.0Sugarmaple88861.3All plant species29846.6beechHop and flight rates of the Black-throatedBlue Warbler vary among plant species,fewerflightsoccurringin subcanopysugarandstripedmaple and more hops in shrub-layerhobblebush (Table 6). None of the differences in H/Fratiosamong foliage compartments,however,is significant.The locomotorypatternsusedbyforaging Black-throatedBlue Warblers, therefore, seemlargely unaffectedby differencesinfoliage structureamongunderstoryor subcanopy plant species.Attacknumber of hopsobserved.rates of the black-throatedbluearesimilar among plant species but generallyhigher in the shrub layer than the subcanopy(Table 6). Its foraging maneuversare not sig-nificantly different among any of the foliageIn summary, the foraging behavior of theredstartis more variablethan that of eithervir-eo species.Its locomotionpatterns,attackrates,use of foraging maneuvers,and searchingtactics often differ among tree speciesand strata.Some of these changesare related to foliagestructure,others to differencesin prey throated Blue Warbler foragesalmost entirelyin the shrub and lower subcanopylayers(Black1975).While searchingfor prey, it hops morecompartments;it consistentlyhovers about 65%and gleans about 35% of the time in each treespeciesand stratum (Fig. ld). Mean lengths ofhoversmadeby black-throatedbluesvary littleamong foliage compartments(Table 6). Theonly significantdifferencein hover lengths isbetween subcanopybeech and hobblebush.In summary, the foraging rates, locomotorpatterns,searchradius,and prey-attackingmaneuvers exhibited by the Black-throatedBlueWarbler are relatively similar among the different plant speciesand layers.TABLE6. Foragingratesand searchradii of Black-throatedBlue Warblersin eachof the majortree speciesand strataof the Hubbard BrookForest.The flight rate is for searchflights only and doesnot includeflights usedin attackingprey. The searchradiusis indicatedby the mean length of prey-attackinghoverflights.Foragingrate (number/min)SearchStratumSubcanopyShrubSearch radiusTree species(t)aPreyattacksflights(F)Hops(H)American beech(4,587)1.710.16.262 45 (121)SugarmapleYellow birchStriped 47 30 (146)59 47 (23)45 28 (24)American beechSugar mapleHobblebushStriped mapleAll mber of secondsthat activelyforagingbirds were observed. number of hover flights observed.(in cm) SD (n)b5057403745 35 (45) 40 (26) 25 (72) 24 (22) 32 (165)

680ROBINSONANDHOLMES[Auk, Vol. 101TABLE7. Summaryof changesin foragingtacticsby tree speciesand stratafor four passerinebird speciesat lueWarblerLocomotionpatterns Fliesmorein shrubs,hops Nonemore in subcanopyFlies more andNoneSearch radiusNoneNoneNoneAttackNoneNoneFewer eyedVireohops lessinshrubsmaneuverschases inshrubsNoneinNoneNoneNoneLocomotionpatterns Flies more in white ashNoneFlies more inwhite ashNoneSearch neAttackTreeNoneratesNonespeciesmaneuversmore in ctureas a determinantof avianforagingbehavior.--Holmes and Robinson(1981)suggestedthat leaf morphology and arrangement (e.g. size, shape, petiole length, distribution along branches, and elevation abovetwig) strong

species, which forage mainly in the canopy, the shrub stratum is treated here as a single foliage type. The structure and arrangement of foliage differ among tree species. Sugar maple and beech are shade- tolerant "climax" species that have monolayered fo- liage distributions (cf. Horn 1971). Beech leaves are