The Composition and Foraging Ecology of Mixed-Species Flocks in Pine Forests of Hispaniola’

The Condor98595-607 0 The CooperOrnithological Society 1996

THE COMPOSITION AND FORAGING ECOLOGY OF MIXED-SPECIES FLOCKS IN PINE FORESTS OF HISPANIOLA’

STEVEN C. LATTA*,~ AND JOSEPH M. WUNDEIUE, JR. International Institute of Tropical Forestry, U.S.D.A. Forest Service,P.O. Box 490 Palmer, Puerto Rico 00721

Abstract. We determined the flocking propensity of 48 species of birds occurring in native pine forest in the Cordillera Central of the Dominican Republic, and the species composition of 180 mixed-speciesflocks. Flocks were unusuallyubiquitous, with 46 species occurring in at least one flock, 11 speciesregularly present, and all insectivorous species and all migrant speciesparticipating. Most birds encounteredwere permanent residents,but winter residents (Nearctic migrants) were an important component of the flocks and, as a group,had the highestflocking propensity. Flockswere cohesiveand the residentinsectivore, the Black-crowned Palm Tanager (Phaenicophiluspalmarum) often served as the nuclear species.Censuses suggest species richness within flocks reflects the speciespresent in the habitat, but agonisticinteractions indicate that intraspecificaggression may limit the number of individuals of a speciesin these flocks. Speciesco-occurrence data indicate that species do not occur independently of one another in flocks. Positive associationswere far more common than negative co-occurrences,suggesting mutual habitat dependenciesor species interactions within flocks. A non-random associationof nearest neighbors also indicated that speciesmay be gaining feeding benefits from flocking by associatingas close neighbors with an individual of another species,but we were not able to rule out the possibility that predation is an important selective agent. Intraspecific comparisons of foraging behavior between flocking and solitary birds pro- vides some evidence that individuals modify foraging locations and foraging tactics upon joining mixed-species flocks, and that their foraging behavior tends to converge with the feeding behavior of the nuclear species.An increasein the feeding rate was recorded for one species.These data suggestthat at least some speciesmay accruefeeding advantagesas flock participants. Key words: mixed-speciesflocks; pine forest; Hispaniola; foraging behavior; Nearctic migratory birds;nuclear species.

Sinopsis. Determinamos la tendencia para hater bandadaspara 48 especiesde aves en 10sbosques de pinos nativos de la Cordillera Central de la Republica Dominicana, y la composici6n de 180 bandadas de especiesmixtas. Las bandadas son bien com6n, con 46 especiesque se encontrabanen por lo menos una bandada, 11 especiesque se encontraban regularmenteen bandadas,y todas las especiesinsectivoras y las especiesmigratorias par- ticipandose. Las mayorlas de las aves son residentespermanentes, pero residentesde1 in- vierno (aves migratorias)son una parte importante de las bandadas,y coma un grupo tienen una alta propensi6n para hater bandadas.Ias bandadasson cohesiva, y el residente insec- tivoro, Phaenicophiluspalmarum, es la especie nuclear. Datos de1 censo indican que la abundanciade las especiesdentro de las bandadasrefleja las especiesque estanen el habitat, pero datos sobre interacionesindican que agresi6n intra-especie limita el ntimero de individuos de una especie en las bandadas. Datos sobre el co-ocurriencia de especiesindican que las especiesno ocurren independiente en las bandadas. Asociaci6nes positivas estan mh comtm que co-ocurrienciasnegativas, indicando dependenciasmutual de1 habitat o interacci6nesentre especiesen las bandadas.Una distribuci6n no al azar de “vecinos cerca” tambi6n indica que las especies estan ganando mh comida cuando estan asociada con “vecinos cerca” de otras especies,pero no pudimos excluir la posibilidad que predation es un agente selectiva importante. Comparacionesintra-especies de1 comportamiento alimenticio entre aves en bandadasy aves solitarias provienen evidencia que individuos modifican localizaci6nesde alimenta- ciones y metodos alimenticios cuando estan en bandadas, y que su comportamiento ali-

I Received 4 January 1996. Accepted 3 May 1996. z Current Address: University of Missouri, Division of Biological Sciences, 110 Tucker Hall, Columbia, Missouri, 65211. 3Corresponding author.

15951 596 STEVEN C. LATTA AND JOSEPH M. WUNDERLE, JR.

menticio hate convergercon el comportamientoalimenticio de1especie nuclear. La razone de alimentacion subi para una especieen bandadas. La informaci6n indican que algonas especiesganan ventajas de alimentacionescoma miembras de bandadas.

INTRODUCTION sity of permanent-resident and winter-resident Many species of birds commonly associate in bird species,and flock organization in order to mixed-species flocks during the non-breeding determine whether there are non-random pat- season(Rand 1954, Powell 1985). Hypothesized terns of association within flocks. In order to selective advantages of flocking have been sum- understand some of the costs and benefits as- marized by Powell (1985), but the favored hy- sociated with participation in mixed-species pothesesare decreasedpredation and increased flocks, we also tested whether a speciesmodifies foraging efficiency. The predation hypothesis foraging locations, foraging tactics, or foraging suggeststhat the cost of vigilance is reduced for rates in the presenceof a mixed-species flock. each individual participating in flocks because more individuals are available to detect ap- STUDY SITE AND METHODS proaching predators(Elgar 1989). The protection STUDY SITE afforded flock participants may also allow the Mixed-species flocks were observed in pine for- birds to forage in more exposed sites and use est at six sites on slopes near Manabao (905- riskier behavior-foraging faster with less vigi- 1,050 m) and six sites near Jarabacoa(643-779 lance (Buskirk et al. 1972, Pulliam 1973). A bird m) in La Vega province in the Cordillera Central that enjoys the protection of a flock may also of the Dominican Republic from Oct. to Apr., suffer a cost through increasedinterspecific com- 1993-95. Native pine forests (Pinus occidental- petition or interference in food acquisition, or is), many of which have been selectively logged, because of the need to adjust movement rates remain on the steepermountain slopesand ridge- and other behaviors in order to stay with the tops, with variable amounts of broadleaf under- flock (Hutto 1988, Petit and Bildstein 1987). The story which may be affected by fire, cutting, or foragingfacilitation hypothesissuggests that birds grazing. A foliage height profile was determined benefit from group foraging through the flushing for representative pine habitat in an area where of insects by the activity of flock members (Belt feeding observations were collected and it shows 1874 cited in Powell 1985); locating or allowing a fairly open canopy, a denser mixed-broadleaf the use of new types of food, foraging locations, understory, and very little in the way of an in- or foraging tactics (Valburg 1992); or facilitating termediate layer (Wunderle and Latta, in press). a reduction of niche overlap (Morse 1967). Sup- Pine foliage predominated in the canopy with port for the foraging facilitation hypothesis does greatest cover in the 15-20 m height category, not exclude the possibility that a speciesbenefits although scatteredbroadleaf trees also extended from predator avoidance as well. into the canopy. Broadleaf specieswere not sam- Powell (1985) and Hutto (1988, 1994) have pled but included Ribes sp., Inga Vera, mango suggested that before distinguishing between (Mangijka indica), citrus (Citrus sp.), Cecropia competing hypotheses to explain the evolution sp., Syzygium jambos. Clusia sp., Miconia sp., of mixed-speciesflocking behavior, we first need as well as numerous unidentified shrubs. Al- more descriptive studies, particularly from the though the exact area of the pine forest is un- tropics, to determine whether foraging efficiency known, the remnant patcheswere all large enough increasesor decreasesas a result of flocking, and that trails could be followed for five or more to assesswhether flock members accrueany feed- hours without leaving the pine or returning to ing advantages at all. Information on group previously censusedhabitat. membership, patterns of movement and association, and foraging behavior of individuals in many kinds of mixed-species flocks can help us POINT COUNTS to better understand the costs and benefits as- Point count data were used to estimate abun- sociated with participation in such flocks. dances of all species.A total of 44 fixed-radius Here we presentdata describing mixed-species point counts (Hutto et al. 1986) were conducted foraging flocks occurring in native pine forests from 11 Jan.-9 Feb. 1994, in which an observer of the Dominican Republic. Our goals are to recorded all birds seen and heard during a ten- describe flock composition, the flocking propen- minute period at each point. Each point was at MIXED-SPECIES FLOCKS ON HISPANIOLA 591 least 100 m from all others. Counts were initiated ent in sequence data (Wagner 1981). This was at sunriseand terminated before 1 1:00, with most facilitated by the frequent presence of only one counts completed before 10:30. All counts were individual of each sex of each speciesin flocks. made by the same observer who estimated the Terminology for foraging maneuvers and for- minimum distance to each bird detected during aging site classificationswas derived from Rem- a count within a 25 m radius. We calculated the sen and Robinson (1990). Foraging maneuvers mean number of detections for each speciesper included glean, hang, reach-up, reach-out, reach- 25 m radius plot. To verify the independence of under, jump-up (leap), sally-strike, sally-stall, samplesfrom each point we compared mean de- sally-hover, sally-pounce, and probe. tections per 25 m radius plot at L 200 m intervals The location of a food item when attacked by with those at 100 m intervals and found no sig- the bird was identified as the foraging site and nificant differences.These data are presented in the following were noted: (1) estimated height of Wunderle and Latta (in press). the bird above the ground; (2) estimated height of the tallest tree within 15 m (canopy height); FLOCKING BEHAVIOR (3) horizontal position of the bird (inner l/3 of Flocking observations were made from 07:00- tree, middle l/3 of tree, outer l/3 of tree); (4) 18:00 hrs in pine forests where feeding activity the amount of light passingthrough an imaginary occurred all day. To reduce the possibility of 2.0 m diameter sphere surrounding the foraging encountering a given individual or flock more site (0 = 100% of light passesthrough, 1 = 95% than once, habitat was never censusedmore fre- 99%, 2 = 75%94%, 3 = 25%74%, 4 = 5% quently than once a week. Observations were 24%, and 5 = O-4%; (5) the location of the prey made by walking slowly through the habitat until item, or the “primary substrate type” (for ex- a bird was encountered. In all casesthe bird was ample a tree speciesor air); (6) the “secondary identified and categorized as either solitary or a substrate” or the location of the food item within member of a mixed-species flock. A flock was or on the primary substrate (for example leaf, defined as comprising at least three conspecifics twig, trunk, fruit, flower); and (7) substrate con- or two heterospecificswithin 25 m of one another dition (live or dead). Height estimates were and moving together for at least five min. Flocks checked daily with a rangefinder. were followed for as long as possiblebut the max- Becausecanopy heights varied somewhat be- imum time was 60 min. Generally, flocks were tween sites,bird height relative to canopy height not followed for more than 150 m, which reduced was calculated and used in further analyses. In the possibility of accumulating speciesthrough all statistical analyses, sample sizes within a cell turnover as the flock moved through adjoining were increasedwith the following condensations: territories. For each mixed-species flock we re- (1) horizontal positions 1 and 2 were combined; cordedthe number of individuals of each species, (2) foliage density scores of O-2 and 4-5 were the nearest neighbor (< 10 m) of focal individ- combined, (3) foraging methods were combined uals, agonistic interactions between birds, and into three categories:surface moves (glean, hang, the linear distance moved by the flock. Single- probe), near-surfacemoves (reach,jump-up), and speciesflocks occurred but were small and un- aerial moves (all sallies); and (4) primary sub- usual. These flockswere not included in our anal- strates and secondary substrateswere re-classi- yses. Only the Cattle Egret occurred exclusively fied into five categories:woody material (trunk, in single speciesflocks and so was excluded from branch, twig); leaves (including pine needles, ep- our analyses. We also excluded all raptors and iphytes, lichen, and mistletoe); fruit, seeds,and hummingbirds. pine cones; flowers; and air, ground, or herba- ceousground cover. For most speciescells were FORAGING BEHAVIOR further condensed by summing sparse cells (< Foraging observationswere made as often as pos- 5) in an “other” categoryprior to statistical anal- sible for both solitary birds and flock partici- ysis. pants. The first foraging event five secondsafter Feeding rates were determined by following an individual was initially detectedwas recorded insectivores and recording the time of each feed- to avoid a bias toward the more conspicuous ing attempt with a cassetterecorder. Birds were feeding techniques. We sought to record only a followed for up to 10 min or as long as consec- single foragingevent from an individual eachday utive feeding attempts could be observed. Rates to reduce the problem of autocorrelation inher- were calculatedas the averagetime between feed- 598 STEVEN C. LATTA AND JOSEPH M. WUNDERLE, JR. ing attempts, but no rates were determined un- We detected 180 mixed-species flocks, which lessthere were at least four consecutiveattempts. contained 77% of all individuals observed.Flocks averaged 7.1 + 0.2 speciesand 11.3 f 0.5 in- STATISTICAL ANALYSES dividuals. Of the 48 speciesobserved, 46 species The software package SYSTAT Version 5.2 were present in at least one flock, and eleven (Wilkinson 1992) was used to perform various species were recorded as occurring regularly statistical tests described in Sokal and Rohlf (present in > 25% of the flocks; Table 3). The (198 1). A probability of Type I error of 0.05 or speciesoccurring in the greatestnumber of flocks lesswas acceptedas significant, but greatervalues (Table 3) were Black-throated Blue Warbler (78% are shown for descriptive purposes. Data pre- of flocks), Black-crowned Palm Tanager (60%), sented are means -+ SE. When data were not Greater Antillean Pewee(60%) Black-and-white normally distributed, nonparametric statistics Warbler (53%), Pine Warbler (52%) and Bana- were used. naquit (490/o).Of the eleven regularly occurring Independent samples t-test (with pooled var- species, the Stripe-headed Tanager, Black- iances) or a Mann-Whitney U-test was used to crowned Palm Tanager, and the Bananaquit were test for intraspecific differences in the means of the most intraspecifically gregarious, averaging foragingheights and feeding rates.Regression was the highest number of individuals per flock (Ta- used to relate the occurrence of a speciesin a ble 1). flock to log transformed relative abundance from Flocking propensity, or the proportion of en- point counts. A 2 x 2 Test of Independence with counters with a species during which the bird a $-statistic or a Row x Column Test of In- was associatedwith a flock, was calculated for dependence with a G-statistic was used to test each species(Table 1) and for each guild (Table for the independence of speciesco-occurrences 2). Of the 23 specieswith more than 20 obser- within flocks, and to compare various feeding vations, flocking propensity ranged from 0.59 to behaviors and feeding site characteristicsof sol- 1.OO. Flocking propensity was highest for Antil- itary individuals and flocking individuals within lean Siskin (1 .OO), Prairie Warbler (1 .OO), Yel- a species.A Bonferroni correction for multiple low-throated Warbler (0.97) Black-and-white comparisons was not used in these comparisons Warbler (0.96), American Redstart (0.95) Yel- but the number of significant results expectedby low-rumped Warbler (0.92), and Greater Antil- chance alone is noted. In tests where cells were lean Pewee (0.91). Insectivores and granivores sparse (< 5) we used a x*-test with Yates’ cor- had similarly high flocking propensities (0.81). rection for continuity, or a G-test with Williams’ Omnivores had a moderately high flocking pro- correction. pensity (0.75), but nectarivores had a low propensity (0.6 1). As a group, though, migrants had the highest propensity for joining flocks (0.87) RESULTS and this differed significantly from that of per- SPECIESABUNDANCE AND manent residents (x2 = 5 1.8, df = 1, P -c 0.001). FLOCK COMPOSITION When analyzed by diet type, the proportionate We detected 2,667 birds representing 48 species abundance of individual insectivores and gran- in our surveys(Table 1, includes scientificnames). ivores participating in flocks was greater than The most numerous speciesoverall were Bana- proportionate abundances derived from point naquit, Black-crowned Palm Tanager, Black- count data (x2 = 23.8, df = 1, P < 0.001 and x2 throated Blue Warbler, Pine Warbler, Stripe- = 7.8, df = 1, P = 0.005, respectively). In con- headed Tanager, Greater Antillean Elaenia, and trast, the proportionate abundancesof individual Greater Antillean Pewee. Most birds encoun- omnivores and nectarivores in flocks was sig- tered were permanent residents (33 speciesand nificantly less than proportionate abundances 1,953 individuals; Table 2), but migratory winter derived from point count data (x2 = 19.9, df = residents were common (15 speciesand 7 14 in- 1, P < 0.001 and x2 = 14.3, df = 1, P -c 0.001, dividuals). Insectivores (25 speciesand 1643 in- respectively). The proportionate abundance of dividuals) and omnivores (16 species and 527 migratory individuals participating in mixed- individuals) predominated over granivores (5 speciesflocks was significantly higher than pro- speciesand 205 individuals) and nectarivores (2 portionate abundances derived from point count speciesand 292 individuals; Table 2). data (x2 = 11.8, df = 1, P < 0.001). The prob- MIXED-SPECIES FLOCKS ON HISPANIOLA 599

TABLE 1. The diet, resident status, abundance, flocking propensity, intraspecific gregariousness,number of agonisticinteractions, and number of times each of 48 speciesof landbirds were observed in pine forest in the Cordillera Central of the Dominican Republic.

Species Diet1 Res’ Abd3 %’ ?. n’ A@ n’

Common Bobwhite Colinus virginianus 0 PR 0.0 1.00 6.0 o/o 6 Scaley-napedPigeon Columba squamosa 0 PR 2.2 0.60 1.5 o/o 5 Mourning Dove Zenaida macroura 0 PR 11.1 0.35 1.4 o/o 20 Zenaida Dove Zenaida aurita 0 PR 0.0 0.00 0.0 o/o 9 White-winged Dove Zenaida asiatica 0 PR 0.0 0.00 0.0 o/o 1 Common Ground Dove Columbina passerina 0 PR 0.0 0.50 1.0 o/o 2 Hispaniolan Parrot Amazona ventralis 0 PR 48.9 1.00 2.0 o/o 2 Hispaniolan Lizard Cuckoo Saurothera longirostris I PR 6.7 0.30 1.0 o/o 10 Smooth-billed Ani Crotophaga ani I PR 2.2 1.00 2.0 o/o 2 Hispaniolan Trogon Temnotrogon roseigaster 0 PR 0.0 1.00 1.0 o/o 2 Narrow-billed Tody Todus angustirostris I PR 13.3 0.71 1.3 o/3 142 Broad-billed Tody Todus subulatus I PR 15.6 0.61 1.3 o/o 28 Hispaniolan Woodpecker Melanerpes striatus I PR 42.2 0.76 1.4 o/o 94 Grey Kingbird Tyrannus dominicensis I PR 0.0 0.50 1.0 o/o 2 LoggerheadKingbird Tyrannus caudtfasciatus I PR 2.2 0.56 1.0 o/o 9 Stolid Flycatcher Myiarchus stolidus I PR 2.2 0.80 1.0 o/o 5 Greater Antillean Pewee Contopus caribaeus I PR 22.2 0.91 1.3 l/4 151 Greater Antillean Elaenia Elaenia fallux 0 PR 37.8 0.85 1.7 11/7 161 Red-leggedThrush Mimocichla plumbea 0 PR 2.2 0.59 1.5 l/O 34 Rufous-throatedSolitaire Myodestes genibarbis 0 PR 42.2 0.40 1.0 o/o 15 Palmchat Dulus dominicus 0 PR 4.4 0.73 9.5 o/o 52 Black-whiskeredVireo Vireo altiloquus I M 0.0 0.67 1.3 o/o 6 Blue-winged Warbler Vermivora pinus I M 0.0 1.00 1.0 o/o 1 Northern Parula Par&a americana I M 0.0 1.00 1.0 o/o 1 Cape May Warbler Dendroica tigrina NM 8.9 0.72 1.5 o/2 25 Black-throatedBlue Warbler Dendroica caerulescens I M 46.7 0.79 1.3 3/2 233 Yellow-rumped Warbler Dendroica coronata I M 37.8 0.92 3.5 l/O 100 Black-throatedGreen Warbler Dendroica virens I M 0.0 1.00 1.0 o/o 1 Yellow-throated Warbler Dendroica dominica I M 2.2 0.97 1.6 o/2 63 Pine Warbler Dendroica pinus I PR 44.4 0.74 1.5 3/l 195 Prairie Warbler Dendroica discolor I M 2.2 1.00 1.1 o/o 21 Palm Warbler Dendroica palmarum I M 0.0 0.79 2.1 o/o 42 Black-and-white Warbler Mniotilta varia I M 20.0 0.96 1.1 o/o 110 American Redstart Setophaga ruticilla I M 11.1 0.95 1.1 o/o 76 Ovenbird Se&us auricapula I M 0.0 0.73 1.1 o/o 11 Common Yellowthroat Geothlypis trichas I M 4.4 0.56 1.0 o/o 16 Ground Warbler Microligia palustris I PR 13.3 0.67 1.8 3/l 81 Bananaquit Coereba flaveola N PR 95.6 0.60 1.8 6/O 267 Blue-hooded Euphonia Euphonia musica 0 PR 2.2 0.50 1.0 o/o 6 Stripe-headedTanager Spindalis zena 0 PR 46.7 0.80 2.9 3/l 192 Black-crownedPalm Tanager Phaenicophilus palmarum 8/O 243 Black-cowledOriole Icterus dominicensis :, PR 31.10.0 0.791.00 1.81.0 o/o 1 Antillean Siskin Carduelis dominicensis G PR 0.0 1.00 2.4 o/o 26 Greater Antillean Bullfinch Loxigilla violacea 0 PR 2.2 0.74 1.3 o/o 19 Yellow-faced Grassquit Tiaris olivacea G PR 2.2 0.82 2.5 o/o 71 Black-facedGrassquit Tiaris bicolor G PR 0.0 0.50 1.0 l/O 2 Indigo Bunting Passerina cyanea GM 0.0 0.75 3.0 o/o 8 Rufous-collaredSparrow Zonotrichia capensis G PR 22.2 0.76 1.7 o/o 98 I Diet basedon foragingobservations (this study) and Wunderle and Latta (in press):G = granivore, I = insectivore,N = nectarivore,0 = omnivore. 2Resident status: PR = permanentresident, M = migratorywinter resident. ’ Mean numberof birds per point count(x 100) from Wunderleand Latta (in press). ’ Proportionof(n) encounterswith a speciesdurin which the individual wasobserved foraging with a mixed-speciesflock. 5 Mean numberof individuals of this speciesin a / ock, gwen that at leastone wasdetected. 6 Number of intraspecificand interspecificagonistic interactions observed in mixed-speciesflocks. ’ Number of individuals encounteredeither alone 01 in a mixed-speciesflock. 600 STEVEN C. LATTA ANDJOSEPH M. WUNDERLE, JR. TABLE 2. The abundance,flocking propensity, and number of times birds of variousdiet typesor residency statuseswere encountered in pine forestsof the CordilleraCentral of the Dominican Republic.

All species 48 2,661 2,061 0.77 292 - Insectivores 25 1,643 1,325 0.81 144

ability of occurrenceof a speciesin a flock could, neighbors overall was found to be independent however, be predictedfrom log transformed point of focal species(G = 190.2, df = 49, P < O.OOl), count abundance data (r = 0.93, df = 1, P < and for each focal speciesthe identity of nearest 0.001). neighbors differed significantly from expected Of those flocks that were followed for > 10 proportions derived from point count abun- min, rate of movement was determined for 82 dances (Table 4). flocks. The average rate at which flocks moved We noted 64 agonistic interactions among 17 was 2.1 * 0.1 m.min-I. species(Table 1). We found a significant differ- Potential avian predators were occasionally ence in the number of intraspecific and inter- observed in the pine forest. These included Red- specificagonistic interactions (x2 = 52.5, df = 1, tailed Hawk (Buteojumaicensis), American Kes- P < 0.001; Table 1); intraspecific interactions trel (F&o sparverius),Sharp-shinned Hawk (Ac- were far more frequent, and interspecific inter- cipiter striatus), and Barn Owl (Tyto alba). We actions less frequent, than would be expected if once observedan accipiter make an unsuccessful birds were interacting at random with one an- pass on a mixed-species flock, and we suspect other. that this is the most effective predator of small FORAGING BEHAVIOR birds in the area. We analyzed six aspects of feeding behavior FLOCK ORGANIZATION AND for ten commonly-occurring species that were AGONISTIC BEHAVIOR present both as flock members and as solitary Species co-occurrenceswere calculated for the individuals to determine if changes in foraging 26 speciesthat occurred in more than 5 flocks behavior occurred in the presenceof mixed-spe- (Table 3). Of 325 possiblepairwise comparisons, cies flocks. These ten species(and sample sizes 48 were significantly different than the expected for foraging observations as flock members and number of co-occurrences. Forty-four of these as solitary birds) included: Bananaquit ( 17/ 19), significant comparisonswere positive, indicating Black-throated Blue Warbler ( 142/7 l), Gape May that thesepairs of speciesoccurred together more Warbler (12/ 18), Greater Antillean Pewee (68/ frequently than expected. Four pairwise combi- 33), Greater Antillean Elaenia (46/l 8) Hispan- nations were negative suggestingthat these spe- iola Woodpecker (22/ 17) Ground Warbler (17/ cies pairs may be avoiding one another. Three 3 l), Prairie Warbler (25/ 13), Black-crownedPalm of these negative pairwise combinations in- Tanager (33/l 3) and Stripe-headed Tanager (36/ volved the nectarivorous Bananaquit. Because 23). Adequate sample sizes of the sexually di- only 16 comparisons are expectedto be positive morphic Black-throated Blue Warbler allowed us based on chancealone at the P = 0.05 level, these to test the sexesseparately. Only three of these data indicate that not all birds are joining flocks comparisons are expected to be positive based independently of one another. on chance alone at the P = 0.05 level. Nearest neighbors were recorded for eight fre- We examined several variables to determine quently encountered species in mixed-species whether a species shifted foraging locations in flocks (Table 4). The distribution of nearest the presence of mixed-species flocks. Intraspe-

602 STEVEN C. LATTA AND JOSEPH M. WUNDERLE, JR.

TABLE 4. The occurrenceof pairs of nearest neighbors in mixed-species flocks. Values indicate the number of times a particular specieswas observed as the near neighbor of the focal specieson the left. The identity of near neighborsdiffers significantlyamong species(G = 190.2, df = 49, P -c 0.001).

Neighboringspecies Focalspecies 1 2 3 4 5 6 7 8 P’

1 Black-throated Blue Warbler 4 10 10 10 8 0.049 2 Black-crowned Palm Tanager 10 10 5 6 0.026 3 Greater Antillean Pewee 10 10 0.003 4 Black-and-white Warbler 10 z

cific comparisons of relative foraging heightsand ers used (as opposedto all other substratetypes) horizontal positions of flocking and solitary in- decreasedin the presenceof mixed-speciesflocks dividuals were non-significant for all species. from 79% to 53% of observations. Both the Prai- Nearly significant differences were found, how- rie Warbler and the Black-crowned Palm Tan- ever, in the relative foraging height of the insec- ager showed a near-significant trend towards the tivorous Ground Warbler with a mean height of use of more leaf and pine needle substrates(as 0.21 + 0.02 m for solitary birds, but 0.32 f 0.05 opposedto all other substratetypes) as members for flocking birds (t = 1.8, df = 46, P = 0.064). of mixed-species flocks (x2 = 3.1, df = 1, P = The normally skulking Ground Warblers tended 0.078; x2 = 2.7, df = 1, P = 0.098). Prairie War- to forage higher in the foliage in the presenceof blers and Palm Tanagers increased leaf and nee- interspecific flocks. Opposite trends were found dle use from 46% to 80%, and 69% to 74%, re- for the two nectarivores-the Bananaquit and spectively. We were unable to reject the null hy- the Cape May Warbler. These two speciesfor- pothesisof no intraspecific differencesin the use aged higher in the canopy (relative foraging of live or dead substrates for any species, but heights of 0.47 f 0.04 and 0.5 1 + 0.05, respec- significant differenceswere found in foliage den- tively) as solitary birds, but in the company of sity scores surrounding foraging sites for male mixed-species flocks the Bananaquit foraged at Black-throated Blue Warblers (G = 13.2, df = 2, a mean relative height of 0.36 k 0.04, and the P = 0.001) Greater Antillean Elaenia (G = 7.5, Cape May Warbler at 0.36 + 0.06. These trends df = 2, P = 0.024) Ground Warbler (G = 6.8, towards lower relative foraging heights for the df = 2, P = 0.033) and Stripe-headed Tanager nectarivores in the presence of flocks were also (G = 7.1, df = 2, P = 0.029). The two insectiv- nearly significant (t = - 1.9, df = 34, P = 0.065 orous warbler specieswere observed significantly and t = - 1.8, df = 27, P = 0.071). more often on substrateswith a more open fo- Intraspecific differencesin foraging location of liage density as interspecific flock members (G = solitary and flocking birds of each specieswere 13.2, df = 2, P = 0.001 for Black-throated Blue further analyzed with respect to the type of for- Warbler; G = 6.8, df = 2, P = 0.033 for Ground aging substrate,the use of live or dead substrates, Warbler). For Black-throated Blue Warblers, use and the foliage density surrounding the foraging of the most open category of foliage density rose location. Flocking Hispaniolan Woodpeckersoc- from 37% for solitary birds to 62% for flock curred significantly (x2 = 13.5, df = 1, P < 0.00 1) members; corresponding proportions for the more often on tree trunks or branches than did Ground Warbler were 23% and 59%. The om- solitary birds (90% vs 35%), which were ob- nivorous species showed an opposite trend. served foraging on leaves, epiphytes, or fruit 65% Flocking Greater Antillean Elaenia and Stripe- of the time. Nearly-significant differencesin sub- headed Tanagers decreaseduse of substrateswith strate use were found for Bananaquits (x2 = 2.8, more exposedfoliage density scores(G = 7.5, df df = 1, P = 0.096), where the proportion of flow- = 2, P = 0.024; G = 7.1, df = 2, P = 0.029), with MIXED-SPECIES FLOCKS ON HISPANIOLA 603 percent usage dropping from 83% to 56% of all in press). Birds appear to be generally conspic- foraging observations and from 70% to 43%, re- uous and more exposed in pine compared to spectively. broadleaf habitats. Shifts in foraging methods were observed for FLOCK ORGANIZATION AND two species in the presence of mixed-species AGONISTIC BEHAVIOR flocks. Ground Warblers shifted from primarily The high number of regularly occurring species near-surface reachesand jumps (55% of obser- in mixed-speciesflocks is unusual (Powell 1985) vations) as solitary birds to gleaning behavior as and indicates that there are many commonly oc- flock members (65% of observations; G = 6.9, curring specieswith a high flocking propensity df = 2, P = 0.033). Black-crowned Palm Tana- in this habitat. Nearly all bird speciesoccurring gers also shifted from near-surface maneuvers in the pine forest participated to some degreein (77%) outside of flocks to gleaning (6 1%) within flocks. Even speciesthat foraged on the ground flocks (G = 8.1, df = 2, P = 0.019). A near- or in dense understory vegetation appeared to significant trend towards the use of fewer non- join flocks,at least when the flock passedthrough surface(aerial) maneuvers by the Greater Antil- their presumed territory, but more detailed ob- lean Elaenia was observed (G = 5.7, df = 2, P = servations on joining and following are needed 0.058); outside of flocks 78% of feeding attempts to assessthe participation of ground-foraging were non-surface maneuvers, while only 56% of birds in flocks with canopy birds. Species such maneuvers within flocks were aerial sallies. as Palm Warbler, Rufous-collared Sparrow, and The scarcity of individuals apart from mixed- Yellow-faced Grassquit were regularly seen in speciesflocks limited the analysis of feedingrates, flocks,while Ground Warblers and Narrow-billed but adequate sample sizes to compare within- Todies also participated and were two species speciesforaging rates were obtained for two in- that most clearly adjustedtheir foragingbehavior sectivores,the Greater Antillean Pewee (n = 37), in the presence of flocks. The only non-partici- and the Black-throated Blue Warbler (n = 63). pating species were the three species of hum- The pewee fed significantly faster in the presence mingbirds, Cattle Egret, and the birds of prey. of mixed-species flocks (t = -3.1, df = 35, P = Some species were often seen in monospecific 0.004); the mean time between feeding attempts flocks, but at times associated (perhaps coinci- was 38.5 -t 3.8 set outside of flocks, but de- dentally) with mixed-species flocks. These spe- creasedto 23.0 -t 3.2 set within flocks. Foraging cies included Common Bobwhite, several doves, rates did not differ between male and female Hispaniolan Parrot, Smooth-billed Ani, Palm- Black-throated Blue Warblers in flocks (U = chat, Stripe-headed Tanager, and Antillean Sis- 259.5, P = 0.83), or as solitary birds (U = 15.0, kin. P = 0.68), so the two sexeswere combined. For- The most numerous participants in mixed- aging rates of flocking Black-throated Blues (X = speciesflocks were permanent residents,but mi- 9.8 f 1.2 sec.) were not significantly different grants formed an important component of all from those of non-flocking birds (X = 7.9 f 1.4 flocks and, as a group, had the highest flocking sec.; U = 380.5, P = 0.19). propensity. Most studiesof mixed-species flocks report migrants as only a minor component of DISCUSSION the flocks (Powell 1985) but in western Mexico Mixed-species flocks have not been previously Hutto (1994) found half the participants and half described from Hispaniola; in fact, very little the species were migrants. Migrants and resi- work has been done on the behavior and ecology dents were also reported in mixed-species flocks of birds on the island (Dod 198 1). We found in Cuba (Eaton 1953) and Jamaica (Lack and mixed-species flocks, however, to be ubiquitous Lack 1972) and in both the U.S. Virgin Islands in the pine forestsof the Cordillera Central, even (Ewert and Askins 199 1) and Puerto Rico (Stai- though many of the same flock constituents were cer 1992) flocks were described that consisted territorial and did not form flocksin nearby shade almost entirely of migrants. Most speciespartic- coffeeplantations and riparian habitat (pers. ob- ipating in mixed-species flocks in the Cordillera serv.). Pine forest differs from shadecoffee in the Central were insectivores, and all of the insec- Cordillera Central in being relatively open hab- tivorous speciesoccurring in the pine habitat were itat and in containing significantly fewer inver- recorded in mixed-species flocks. tebrates in the understory (Wunderle and Latta, The mean flock size of 7.1 speciesand 11.3 604 STEVEN C. LATTA AND JOSEPH M. WUNDERLE. JR.

individuals in the Cordillera Central is compa- quency of occurrenceof a speciesin flocksreflects rable to other studies, falling within the range of its relative abundance in the habitat as indicated flock sizes reported by Powell (1985). The mean by point counts. rate of movement was similar to that recorded Flocks in the Cordillera Central appear to be from the U.S. Virgin Islands, where Ewert and cohesive, unlike mixed-species flocks or aggre- Askins (199 1) found mixed-species flocks mov- gations reported from Puerto Rico (Post 1978, ing at 1.6 m min-I, nearly the same rate as re- Staicer 1992) and Jamaica (Lack and Lack 1972). ported here. Results from studies of understory We were often able to follow flocks for thirty flocks in the Neotropics, however, show flocks minutes or more. Elsewhere in the West Indies, moving much faster, with an average movement Eaton (1953) found cohesive flocks in Cuba, as rate of 5.0 m min-I (Powell 1985). did Ewert and Askins (1991) in the Virgin Is- Although most individuals were not color- lands, but care must be taken in such compari- banded, our impression was that most flock par- sonsbecause of differencesin methodologiesand ticipants remained with the hock as it moved definitions. through the pine forest and that there was very Nuclear species(sensu Moynihan 1962) may little turnover in speciesor individuals. This is be important in maintaining the cohesion of the same pattern described by Hutto (1987) and mixed-species flocks (Hutto 1994). Hutto char- Munn and Terborgh (1979). A few speciesap- acterized nuclear species(after Moynihan 1962) peared to join mixed-speciesflocks for only short as: (1) occurring more often in the forefront of periods of time. This was particularly true of flocks, as they are joined and followed by other those speciesthat also tended to flock monospe- speciesmore often than they themselvesjoin and citically. In thesecases monospecific and mixed- follow; (2) intraspecifically gregariouswith a high speciesflocks may have had overlapping terri- flocking propensity and a relatively greater num- tories or ranges and periodically foraged togeth- ber of individuals per flock than other species; er. Other speciesthat participated in mixed-spe- (3) regular participants of mixed-species flocks; cies flocks may have maintained smaller, indi- and (4) generally conspicuous in their plumage, vidual territories, and joined the flock only when calls, and behavior. A single speciesis generally the flock passed through their territory. These viewed as the nuclear species(Powell 1985), but species generally appeared very active in the Hutto (1994) found two nuclear speciesin west- presenceof mixed-speciesflocks, but did not ap- em Mexico, and Powell reports as many as six pear to move long distances. We suspect that speciescontributing to the maintenance of group Broad-billed and Narrow-billed todies, Greater cohesion in the Amazon. Resident species are Antillean Pewees, Ground Warblers, Rufous- most often reported as the nuclear species(Pow- collared Sparrows,and some Black-throated Blue ell 1985; but see Greenberg 1984, Hutto 1994). Warblers may have maintained such territories. Our data suggestthat the permanent resident Agonistic interactions have not been frequent- Black-crowned Palm Tanager is often a nuclear ly recorded in mixed-species flocks in the West speciesin these flocks. The palm tanager was the Indies (Ewert and Askins 199 1, Staicer 1992), second most frequently encountered species in but aggressionhas been commonly reported in the pine forest, is unusually gregariousand vocal, flocks in other regions and may regulate the has a relatively high flocking propensity, and is abundance of individuals or specieswithin the conspicuouslycolored. Although the Bananaquit flock (Morse 1970, Munn and Terborgh 1979, was more common, it is not insectivorous and Powell 1985). Our data, showing a higher than had a much lower flocking propensity. Several expected incidence of intraspecific interactions, other species had a higher flocking propensity suggestthat aggressionmay be used to limit the than the palm tanager, but were less common number of individuals per speciesin theseflocks. and less intraspecifically gregarious, or did not However, the highly significant correlation we appear to be conspicuous.The Greater Antillean found between point count censusdata and flock Pewee may also serve as a weak nuclear species composition indicate that flock composition is a in some instances,as it often appearedas a vocal reflection of the pool of species present in the component at the forefront of flocks, but the spe- habitat. This suggeststhat speciesrichness within cies is neither visually conspicuousnor intraspe- the flock is limited primarily by the number of cifically gregarious. speciesoccurring in the habitat, and that the fre- Speciesco-occurrence data indicate that spe- MIXED-SPECIES FLOCKS ON HISPANIOLA 605 ties do not occur independently of one another to the sometimes widespread and dispersedHis- in flocks.Negative co-occurrences,suggesting that paniolan pine flocks. In these flocks, mutual speciesmay be avoiding one another, were re- predator protection is a likely possibility for some corded for four pairs of species. Although ex- speciespairs which share similar feeding behav- ploitative competition or differences in micro- iors and foraging rates, and feeding facilitation habitat requirements between species may ex- may still explain other positive species co-oc- plain these negative co-occurrences, the most currences. likely explanation is that they are a statistical artifact of multiple pairwise comparisons. FORAGING BEHAVIOR Positive associationsbetween pairs of species Most measures of the impact of interspecific which occurredin numbers higher than expected, flocking behavior on niche overlap indicate that were far more common. While theseassociations flocking results in a reduction in foraging niche may also reflect a statistical artifact, the high diversity, or the convergence of feeding behav- number of associationssuggest biological mech- iors (Powell 1985). Several studieshave reported anisms are required to explain at least some of changesin avian feeding behavior resulting from the results. Positive associationsmay reflect mu- association with mixed-species flocks (Morse tual microhabitat dependencies which, though 1970, Valburg 1992) including the convergence difficult to recognize, may explain the co-occur- of the foraging behavior of individuals within rence of species pairs such as Yellow-throated the flock with that of the nuclear species(Buskirk and Yellow-rumped warblers, both of which for- 1976, Valburg 1992) but only Buskirk (1972, age in the higher reaches of pine trees, and the cited in Powell 1985) has recorded convergence associationof Ground Warblers and Rufous-col- of foragingmethods by insectivores. Intraspecific lared Sparrows, which forage in low, broadleaf comparisonsof flocking and solitary birds in this thickets. study provide some evidence that individuals The presenceof so many positive associations modify foraginglocations or foragingtactics upon may, however, result from speciesinteractions joining a mixed-species flock, and that these (Hutto 1994). Evidence from our nearest neigh- changes tend to converge with the feeding be- bor analysis supportsthe speciesinteraction hy- havior of the nuclear species.Caution must be pothesis, suggestingthat a non-random spatial taken in interpreting these results as environ- organization exists within the flock. First, the mental conditions were uncontrolled and may close neighbors of each of the most common have differed independently of flocking behavior. specieswere significantly different than those ex- Changes in the relative height of feeding ac- pected on the basis of the relative abundance of tivity were the most frequently recorded modi- species,and the distribution of nearest neighbors fication among flocking birds, and these changes differed significantly among these species.In ad- supported the hypothesis of niche convergence. dition, some of the same species-pairsthat oc- While the two nectarivorous species,the Bana- curred together more often than expectedon the naquit and the Cape May Warbler, which nor- basis of censusdata (such as Black-throated Blue mally feed relatively high in the vegetation, Warbler and palm tanager; Black-and-white moved lower in the vegetation in the presence Warbler and Pine Warbler; Black-and-white of flocks, the Black-throated Blue Warbler and Warbler and Yellow-throated Warbler; Table 3) the Ground Warbler moved from low and dense are the same species-pairsfound in close asso- foliage to relatively higher foraging sites. This ciation within flocks (Table 4). was also the pattern exhibited by both the Broad- The benefit of associatingwith another species billed and Narrow-billed tody (Latta and Wun- may be related to food acquisition or to increased derle, in press). protection from predators. If the primary benefit For the nuclear species,as well as flock mem- of group protection (auditory warnings), how- bers as a whole, gleaning was the most common ever, is afforded a flock member regardless of foraging method, and leaves and pine needles position within the flock, as Hutto (1994) sug- were the most common substrate.The observed gests,then flock participants may instead accrue shifts in foraging method and substrate use by feeding benefits from speciesassociations within flocking speciessupports the reduction of niche the flock. While this explanation may apply to diversity towards the gleaning of leaves and nee- tightly cohesiveflocks, it is more difficult to apply dles. The Bananaquit, for example, appeared to 606 STEVEN C. LATTA AND JOSEPH M. WUNDERLE, JR. be acting more insectivorous by using fewer flow- able one to follow individual birds both within ers for foraging sites, while the Prairie Warbler and apart from flocksto accurately assessfeeding and the Black-crowned Palm Tanager both used benefits accruing to the individual. a higher percentage of leaves when foraging in flocks. Only the Hispaniolan Woodpecker ap- ACKNOWLEDGMENTS peared to feed in more specializedlocations, with We gratefully acknowledgethe fine assistanceprovided more foraging on trunks and limbs. The signif- by Teodoro Lara, Esteban Terranova, and Eduardo icant shift towards more gleaning by both the Vazquez. The manuscript benefitted from the con- structive comments of Robert Askins, John Faaborg, Ground Warbler and the palm tanager, and the Wendy Gram, Richard Hutto, and an anonymous rd- nearly significant use of lessaerial maneuvers by viewer. Fundina was vrovided bv the National Fish the Greater Antillean Pewee, are further indi- and Wildlife F&ndatibn and the-John T. and Cathe- cations of niche reduction in the mixed-species rine C. MacArthur Foundation. flocks. The shift to gleaning in the presenceof mixed- LITERATURE CITED speciesflocks may be assumed to be less risky BELT,T. W. 1874. The naturalist in Nicaragua. Mur- behavior, as birds remain in the vegetation while ray Press, London. foraging. But birds may also be using riskier be- BUSKIRK,W. H. 1972. Ecology of bird flocks in a havior by foraging in more exposed locations. tropicalforest. Ph.D. diss.,Univ. California, Davis. This was observed in Black-throated Blue War- BUSKIRK,W. H. 1976. Social systemsin a tropical forest avifauna. Am. Nat. 1 l&293-310. blers and Ground Warblers, as well as in Broad- Busmarc.W. H.. G. V. N. POWELL.J. R. W~I-~ENBER- billed and Narrow-billed todies (Latta and Wun- GER; R. E. BUSKIRK,AND T. U. POWELL. 1972. derle, in press)which both foraged in sites with Interspecific bird flocks in tropical highland Pan- a lower foliage density in the presenceof mixed- ama. Auk 89:612-624. DOD, A. S. 1981. Guia de campo para las aves de la speciesflocks. Only the Greater Antillean Elaen- Republica Dominicana. Editora Horizontes, San- ia and the Stripe-headed Tanager were found to to Domingo, Dominican Republic. forage in significantly denser vegetation when in EATON,S. W. 1953. Wood Warblers wintering in flocks. Feeding in more exposed locations may Cuba. Wilson Bull. 65: 169-l 74. ELGAR,M. A. 1989. Predator vigilance and group be the result of individuals benefitting from the size in mammals and birds: a critical review of the protection offered by the flock through mutual emvirical evidence. Biol. Rev. 64: 13-33. vigilance or confusion of predators(Powell 1985). EWERT,b. N., AND R. A. ASKINS. 1991. Flocking Under the predation hypothesis, the protection behavior of migratory warblers in winter in the afforded by the flock may allow birds to forage Virgin Islands. Condor 93:864-868. GREENBERG,R. 1984. The winter exploitation sys- faster with less vigilance (Sullivan 1984). The tems of Bay-breasted and Chestnut-sided War- increased feeding rates we have recorded for the blers in Panama. Univ. Calif. Publ. Zool. 116:l- Greater Antillean Pewee may indicate that the 107. foraging behavior of this speciesis constrained Hurro, R. L. 1987. A description of mixed-species by predators. insectivorousbird flocks in western Mexico. Con- dor 89~282-292. Feeding advantagesmay be expectedto accrue Hwrro, R. L. 1988. Foraging behavior patterns sug- individuals in flocks under the food facilitation gest a possible cost associatedwith participation hypothesisas well. In addition, the high number in mixed-sveciesbird flocks. Oikos 5 1:79-83. of positive associationsof species-pairs,and the Hurro, R. L. -1994. The composition and social organization of mixed-species flocks in a tropical convergenceof some spatial parameters and for- deciduous forest in western Mexico. Condor 96: aging behaviors by flocking species,suggest that 105-l 18. some birds are finding enhanced foraging op- Hurro, R. L., S. M. PLETSCHET,AND P. HENDRICKS. portunities amidst flocks. Support for the food 1986. A fixed-radiusvoint count method for non- facilitation hypothesis does not exclude the pos- breeding and breeding seasonuse. Auk 103:593- 602. sibility that birds benefit from predator avoid- LACK,D., ANDP. LACK. 1972. Wintering warblers in ance as well (Powell 1985). In order to complete- Jamaica. Living Bird 11:129-153. ly understand the evolutionary significance of LATTA,S. C., AND J. M. WUNDERLE,JR. 1996. Eco- flocking behavior, we need more complete de- logical relationships of two todies in Hispaniola: effectsof habitat and flocking. Condor: in press. scriptionsof the structureand dynamics of mixed- MORSE,D. H. 1967. Foragingrelationships of Brown- speciesflocks from many habitats and geograph- headed Nuthatches and Pine Warblers. Ecology ic areas. Studies of color-banded birds would en- 48:94-103. MIXED-SPECIES FLOCKS ON HISPANIOLA 607

MORSE,D. H. 1970. Ecologicalaspects of somemixed- SOKAL,R. R., ANLIF. J. ROHLF. 1981. Biometry. W. speciesforaging flocks of birds. Ecol. Monogr. 40: H. Freeman, San Francisco. 119-168. STAICER,C. A. 1992. Socialbehavior of the Northern MOYNIHAN,M. 1962. The organizationand probable Pamla, Cape May Warbler and Prairie Warbler evolution of some mixed speciesflocks of Neo- wintering in second-growthforest in southwestern tronical birds. Smithsonian Misc. Collect. 143:l- Puerto Rico. In J. M. Hagan and D. W. Johnston 140. [eds], Ecology and conservation of Neotropical MUNN, C. A., AND J. W. TERBORGH.1979. Multi- migrant landbirds. Smithsonian Institution Press, speciesterritoriality in Neotropical foragingflocks. Washington, DC. Condor 81:338-347. SULLIVAN,K. A. 1984. The advantagesof social for- PETIT, D. R., AND K. L. BILDSTEIN. 1987. Effect of aging in Downy Woodpeckers.Anim. Behav. 32: group size and location within the group on the 16-22. foragingbehavior ofWhite Ibises.Condor 89:602- VALBLJRG, L. K. 1992. Flocking and frugivory: the 609. effect of social groupings on-resource use in the POST,W. 1978. Social and foragingbehavior of war- Common Bush-Tanaaer.Condor 94:358-363. blers wintering in Puerto Rican coastalscrub. Wil- WAGNER,J. L. 1981. Visibility and bias in avian son Bull. 90: 197-2 14. foraging data. Condor 83:263-264. POWELL,G. V. N. 1985. Sociobiologyand the adap- WILKINSON,L. 1992. SYSTAT: The system for sta- tive significanceof interspecific foraging flocks in tistics. SYSTAT Inc., Evanston, IL. the Neotropics. Ornithol. Monogr. 36:7 13-732. WUNDERLE,J. M., JR., AND S. C. LATTA. In press. PULL-, H. R. 1973. On the advantagesof flocking. Avian abundance in sun and shade coffee plan- J. Theor. Biol. 38:419-422. tations and remnant pine forest in the Cordillera RAND, A. L. 1954. Social feeding behavior of birds. Central, Dominican Republic. Omitologia Neo- Fieldiana: Zool. 36: 1-7 1. tropical. REMSEN,J. V., AND S. K. ROBINSON.1990. A clas- sification schemefor foragingbehavior of birds in terrestrial habitats. Stud. Avian Biol. 13:144-l 60.