The Condor 93:904-915 0 The CooperOrnithological Society I99 I

DIETS OF FOUR SYMPATRIC AMAZONIAN (DENDROCOLAPTIDAE)’

ANGELACHAPMAN~ AND KENNETH V. ROSENBERG Museum of Natural Scienceand Department of Zoology and Physiology,Louisiana State University, Baton Rouge, LA 70803-3216

Abstract. Contents of 78 stomachsfrom four widespread and sympatric speciesof Am- azonian woodcreepers(Dendrocolaptidae) were examined to assessthe role of diet in resource partitioning. Orthopterans (2%35%) and beetles (lO-32%) dominated the diets of all four species,despite large differencesin foraging behavior. A higher proportion of spiders in the diet of Dendrocinclamerula was associatedwith specialized ground-foragingat army ant swarms; however, specialization on dead leaves by Xiphorhynchusguttatus resulted in no significant dietary differences from the substrategeneralist X spixii. Overlap in diet was much greater than overlap in behavior for all speciespairs, and the degree of diet special- ization was unrelated to behavioral specialization. Taxonomic representation of prey in stomach contents differed significantly from field sampling of available prey in the three speciestested, with orthopterans apparently selectedby all speciesand beetles selectedby D. fuliginosa. We suggestthat behavioral differences may have evolved to reduce overt aggression(interference competition) but may not serve to reduce diffuse competition for food among syntopic species.Segregation in substrateuse, however, may allow high diet overlap, suggestinga case of niche complementarity among these species. Key words: Dendrocolaptidae;diet;foraging ecology; resource partitioning;specialization; woodcreepers;tropical forest .

INTRODUCTION Munn 1985). Others follow swarms ofarmy ants How large numbers of closely related speciesco- (E&on spp.) to feed on escaping arthropods exist in diverse tropical communities has in- (Willis and Oniki 1978). In addition, many spe- trigued ecologistsfor decades.Although studies cies do not feed directly on trunk or branch sur- of resource partitioning among potential avian faces,but investigateclusters of dead leaves,vine- competitors are numerous, studies that comple- tangles, or palm fronds, or sally after flying prey ment data on habitat selection and foraging be- (e.g., Pierpont 1986). The extent to which these havior with data on diet are few (e.g., Rotenberry behavioral differences may result in differences 1980, Robinson and Holmes 1982, Rosenberg in diet is unknown. et al. 1982, Dahlsten et al. 1985). This is es- Becauseup to 17 speciesof woodcreepersmay pecially true for species-richtropical forestguilds; coexist in parts of the Amazon Basin (e.g., Ter- only Sherry (1984) has investigated the dietary borgh et al. 1984) their foraging relationships relationships among sympatric neotropical in- are a significant component of the overall com- sectivores,using 16 flycatcher speciesfrom Costa munity dynamics. In this paper, we describe the Rica. diets of four widespread speciesat two sites in The woodcreepers (Dendrocolaptidae) are a southwesternAmazonia. We choseDendrocincla large and characteristic family of neotropical fuliginosa (Plain-brown ), D. mer- forest birds. Although superficially similar in ex- ula (White-chinned Woodcreeper), Xiphorhyn- ternal morphology and in their tendency to climb thus guttatus (Buff-throated Woodcreeper), and vertical trunks and branches, woodcreeper spe- X. spixii (Spix’s Woodcreeper)because these were cies exhibit a wide range of foraging behaviors. the most common large (> 30 g) woodcreepers Many species typically join mixed-species for- at each of these two well-sampled sites. Both aging flocks in the understory or canopy (e.g., species of Dendrocincla are “professional” fol- lowers of army-ant swarms throughout much of Amazonia (Willis 1972, 1979), although D. fu- ’ ’ Received 12 February 1991. Final acceptance 7 liginosa Xi- June 1991. also forages away from ants. Both z Presentaddress: Department of Biology, Universiti phorhynchus spp. rarely follow army ants but Brunei Darussalam, Bandar Seri Begawan, 3 186 Bru- routinely join mixed-species flocks in the un- nei, Darussalam. derstory and subcanopy (Munn 1985, Pierpont

[9041 DIETS OF AMAZONIAN WOODCREEPERS 905

1986; Rosenberg, unpubl. data). To address such as beetles. We believe, therefore, that with whether diet was important in niche partitioning knowledge of the particular fragments repre- by these species, we asked the following ques- senting each food-type, we could detect hard- tions: (1) do differencesin foraging behavior cor- and soft-bodied prey equally well. Further ex- respond to differencesin diet? (2) do thesespecies perimentation would be needed to verify our specialize on particular prey? and (3) how do impression. species’ diets relate to food availability? The proportion of each prey category was de- termined separatelyfor each stomach; diets were METHODS then determined as the average of the propor- Diets were determined by direct observation of tions in the individual stomachs of each species stomach contents from 78 individual birds: 32 (i.e., samples were not pooled). Diet overlap be- X. guttatus,18 X. spixii, 11 D. jiuliginosa,and tween pairs of woodcreeperswas calculated as: 17 D. merula(Table 1). Specimenswere collected 0, = ~(P,,P,,)/\/(~Pz,.)(~Pz,~), where Pi, and P,. from June through August, mainly with mist nets, are the proportions of prey category “a” in the near Cobija, Dpto. Pando, extreme northwestern diets of species“i” and “j” respectively (Pianka Bolivia, in 1986 (see Parker and Remsen 1987) 1974, May 1975). Niche breadth, or prey-type and near Abujao, Dpto. Ucayali, eastern Peru, diversity, was calculated for every stomach using in 1987 by field parties from Louisiana State B = (Zp,Z))‘, where p, is the proportion of taxon University Museum of Natural Science “i” in the stomach (Levins 1968). An average (LSUMNS). The two sites are in continuous, niche breadth was then calculated for each spe- lowland rainforest, separatedby roughly 200 km. cies. It is unclear whether pooled frequency dis- Additional foraging and insect availability data tributions of prey items from multiple stomach were gathered at the Tambopata Reserve, Dpto. samples can be compared statistically, because Madre de Dios, southeastern Peru, from 1987- of potential pseudoreplication (Hurlbert 1984). 1989. Although in most casesindividual arthropod prey Stomachs were preserved in 70% ethanol and captured by woodcreepersprobably representin- housed permanently in the LSUMNS collection, dependent events (except for rare attacks on ant along with referenceskins or skeletons.Contents nests),we have conservatively chosen not to ap- of each stomach were examined using a stereo- ply goodness-of-fit statistics to compare diets microscope(6-25 x) with a micrometer. Because among species. We believe, however, that the only a small number of items could be identified patterns in dietary overlap are clear enough to to family, most prey were identified to order or indicate biologically significant trends, and our suborder, using Borror and White (1970), Borror overall conclusions would not be altered. et al. (1981), and Ralph et al. (1985). Prey items For comparisons with the foraging behavior were sorted, measured, and counted, and sketch- of these species,we drew upon data from several es of identified material were made to facilitate sources.First, woodcreeperswere observed for- later identification. Prey sizewas determined from aging as part of a general community study at measurable fragments using regressionequations the Bolivia study site and at the Tambopata Re- in Calver and Wooller (1982) or computed from serve in southeastern Peru (J. V. Remsen and arthropods collected in the study areas (Rosen- Rosenberg, unpubl. data). These observations berg, unpubl. data). Reference series of identi- consisted of opportunistic sightings along forest fied, mounted fragments, drawings, and photo- trails, for each of which we recorded foraging graphs (Fig. 1) were prepared for future com- height (estimated to the nearest 1 m), foraging parisons. substrate, perch site, prey-capture method, and The question of differential digestion of hard- an estimate of foliage density around the , and soft-bodied prey is pertinent to any analysis as well as associatedspecies (e.g., mixed-species of stomach contents (Rosenberg and Cooper flocks) and general habitat features. Data nota- 1990, and referencestherein). Hard-bodied items tion followed Remsen and Robinson (1990). take longer to digest and may persist longer in Usually, fewer than five consecutive observa- the stomach. However, it seemed that the hard tions were made on each bird, although repeated fragments of soft-bodied prey, such as orthop- sightings of (probably) the same individuals in teran mandibles and spider fangs, were as prev- the same mixed-speciesflocks were included. Be- alent in our samplesas thoseof hard-bodied prey, cause these efforts focused on mixed-species- 1. FIGURE 1. Fragmented prey items found in the stomachsof four woodcreeperspecies. A. Homoptera(head); B. Lepidopteralarva (mandible);C. Coleoptera(elytron); D. lizard (jaw); E. Curculionidae (head); F. ant (head with mandibles); G. roach (head with mandibles); H. spider (fang); I. Coleoptera (head with mandibles); .I. Orthoptera (mandible part); K. spider (chelicerae with fangs); L. Orthoptera (mandible). Black bar indicates I mm. flocking species,our samples are adequate only then computed using the same procedure as for for the two Xiphorhynchusspecies. For the two diets, based on the proportional use of seven species of Dendrocincla, we rely primarily on categoriescommon to all studies (see Fig. 7). published accounts of their behavior in Willis In these comparisons, we assume that geo- (1972, 1979). Willis (1972) studied D. jiiliginosa graphic variation in each species’ behavior in lowland rainforest in Panama from 1960-l 97 1 throughout the lowland rainforests is consider- by observing behavior both at and away from ably less than variation among the four species army ants. Observations consisted of quantifi- at our study sites. Although quantitative data on cations of behavioral characteristics, including geographic variation are lacking, our observa- height, diameter, and angle of perches,and sub- tions suggestthat these data sets represent the stratesfrom which prey were taken. Willis (1979) foraging roles of these four speciesin Bolivia and used identical procedures to study D. me&a in southern Peru. For example, D. merula is known northern and eastern Brazil (Manaus, Belem) in to be an obligate army-ant follower throughout 1973 and 1974. To compare these data with our its range, and its behavior was found to vary little observations of Xiphorhynchus spp., we calcu- among 19 Amazonian study sites (Willis 1979); lated frequency distributions of foraging height our few observations of this specieswere within and substrate use from Willis’ published tables. 1 m of the ground at ant swarms. Our small Overlaps in substrate use among species were sample of observations of D. fuliginosa, both at DIETS OF AMAZONIAN WOODCREEPERS 907

TABLE 1. Characteristicsof four Amazonianwoodcreepers. Weights and measurementsare averagesof five male and five female specimensof each speciesat LSUMNS. Bill lengthis the exposedculmen; bill width measuredat nares(both in mm).

Number of stomachs Soecies Bill lenath Bill width PWU Bolivia

Dendrocinclamerula 44.6 25.9 5.9 4 13 Dendrocinclafulieinosa 30.4 27.2 6.4 6 5 Xiphorhynchls s&ii 36.4 32.3 5.6 13 5 Xiphorhynchusguttatus 57.8 38.3 6.1 19 13

and away from army ants, also is qualitatively speciesconsisted of Coleoptera and Orthoptera similar to thoseofwillis in Panama (seeResults). (Fig. 2). D. merula had the highest percentageof We compared diets with data on prey avail- spiders (21%) and the lowest percentage of Co- ability for three of the four woodcreeper species. leoptera (10%). Spidersin the other three species For the two ant-following Dendrocincla spp., we ranged from 10-l 1% of total items in the diet. used data on arthropods flushed by army ants in Ants were eaten consistently by all speciesexcept Costa Rica (Otis et al. 1986). This study con- D. jiiliginosa. Weevils (Curculionidae) were ab- sistedof paired samples of leaf-litter plots before sent from D. merula but were particularly well and after the passing of an ant swarm and also representedin the two Xiphorhynchus species(33- direct observationsof arthropodsfleeing the ants. 50% of beetles). Minor prey items (l-10%) in all We estimated prey availability to birds by com- four species included roaches, lizards or frogs, bining counts of fleeing arthropods (from their bugs (Heteroptera), and insect larvae. Table 1) with the number of insects and spiders The four speciesoverlapped considerably with reduced on the paired plots. Other arthropods respectto diet composition (Table 2). The great- (e.g., isopods, ticks) found in the leaf-litter sam- est overlap was between X. guttatus and X. spixii, ples were not included because they were not and the least was between D. jiiliginosa and D. observed fleeing from ants nor were they found me&a. Dendrocincla merula ate more soft-bod- in the diets of the woodcreepers.The diet of X. ied spiders, roaches and orthopterans, whereas guttatus was compared with arthropods found in its congener preyed more on beetles and verte- suspendeddead leaves at the Bolivian study site brates. Overall dietary diversity was similar in (Rosenberg1990). Samplesconsisted of 275 dead these species,ranging from 4.8 in D. merula to leaves,individually placed in zip-lock plastic bags 6.2 in X. guttatus, out of a possible 8.0. and sprayed with insecticide; arthropods were To assessthe adequacy of our samples, we then collected and later identified to lowest tax- examined the increasein the number of prey taxa onomic level possible. No data were available represented in cumulative samples of randomly on arthropods in the microhabitats used by X. selectedstomachs of each species(Fig. 3). In all spixii. These comparisons are necessarily crude, species,the important arthropod orders and sub- but they give a first approximation of the selec- orders considered in this study were represented tivity of prey items by these species. after samples of only one to five stomachs. The additional prey taxa that accumulated in each RESULTS species’ diet included rare items such as earwigs, flies, scorpions, and centipedes, which together DIET COMPOSITION never exceeded 5% of the diet. To test further Diet composition within a speciesdid not differ our sample sizes, we determined the number of significantly between the two sites (G-test, all P stomachs necessary to estimate the proportion > 0.06), except for D. me&a, in which the Bo- of each prey category within + or -0.05 of the livian samples (n = 13) contained a greater pro- total diet for each speciesat each site. In most portion of spiders (24% vs. 10%) and fewer ants samples, the cumulative estimates of diet com- (9% vs. 3 1%) than those from Peru (n = 4). For position was within 0.05 for all prey categories further comparisonsamong species,data for both after five stomachs(Fig. 4). The exceptions were sites were pooled. for ants in X. guttatus (up to eight stomachsneed- The majority (54-60%) of the diet of all four ed) and in the small sample of D. me&a from 908 ANGELA CHAPMAN ANDKENNETH V. ROSENBERG

n Beetles 0 Roaches q Heteroptera q Larvae El Other Q Oflhopteraq Spiders q Ants II0 Vertebrates

D. menila D. fuliginosa X. guttafus x. spixii Amy ants Dead leaves (171236) (11 1107) (32 1430) (18 1357) (157) (275)

Species FIGURE 2. Diet compositionof four woodcreeperspecies compared with prey flushedby army ants (data from Otis et al. 1986) and prey availablein deadleaves. For bird species,numbers in parenthesesare number of stomachs/numberof prey items;for availability samples,number of arthropods.

Peru, in which total diet may not have been well liginosa foraged as high as 10 m (Fig. 6) and represented. These results indicate low levels of sallied to trunks, vines, foliage, and air, as well individual variation in these speciesand suggest as to the ground (Fig. 7; data from Willis 1972). that our pooled samples of 10 or more stomachs At our study sites, D. filiginosa was seen mostly for each speciesadequately represent their diets. from 5 to 8 m above ground (n = 34 observa- tions), and it foraged from a variety of substrates PREY SIZE including trunks (3 lo/o), air (3 lo/o), live foliage (23%) and epiphytes (8%). Therefore, the char- All four woodcreeper speciesate a similar size acterization of this species as an arboreal gen- range of the most important prey (Fig. 5) in spite eralist, compared to the more restricted ground- of considerable differences among the wood- foraging of D. merula, appears representative of creepersin bill size and body weight (Table 1). Dendrocincla merula took, on average, slightly their foraging at our study sites as well as those of Willis. In Peru and Bolivia, X. guttatus stayed larger beetles and spiders than the other species, mostly above 6 m (Fig. 6), and about two-thirds but none of the differences were significant (t- of its foraging was at suspended dead leaves, tests; all P > 0.10). The sizes of caterpillars, liz- especially large Cecropia leaves and palm fronds ards, and other uncommon items could not be (Fig. 7). Xiphorhynchus spixii foraged generally determined from digested fragments; however, from 2 to 10 m above ground (Fig. 6), usually they probably contributed little, if any, to dif- ferencesin prey size among thesespecies. In spite of substantial differencesin bill length there was TABLE 2. Dietary and behavioraloverlaps among no relationship to prey size. four woodcreeperspecies. Values above diagonalare for diet composition;values below diagonalare for substrateuse. Equation for overlapgiven in text. COMPARISON WITH FORAGING BEHAVIOR X. X. D. fili- D. In Brazil, D. merula perched usually within 1 m gun**us spixri ginosa merula

of the ground (Fig. 6) and about 90% of its for- X. guttatus - 0.999 0.986 0.906 aging consistedof sallies to the ground in pursuit X. spixii 0.419 0.957 0.830 of arthropods fleeing the ants (Fig. 7; data from D. jiiliginosa 0.196 0.649 0.780 D. met-da 0.017 0.034 0.444 - Willis 1979). At ant swarms in Panama, D. fu- DIETS OF AMAZONIAN WOODCREEPERS 909

A. Bolivia

16 1 I”-;

4-

0 I 0 4 8 12 16 20

B. Peru

v D. fuliginosa + X. guttatus

0; -. 1. ..‘ 1.‘ 1..‘ ’ 0 4 8 12 16 20

Number of pooled stomachs FIGURE 3. Number of prey taxa in cumulativesamples of stomachsfrom four woodcreeperspecies at two

pecking at trunks or probing under vine stems terms of prey orders, from D. merula and D. or epiphytic moss and lichens (Fig. 7). fuliginosa but was nearly identical to X. spixii. The overlap in diets among these four species Diversity of feeding behaviors was not con- in all caseswas much greater than their overlap sistently related to diet diversity. Dendrocincla in foraging substrate use (Table 2). In the two merula, with the most restricted mode of for- ant-following species,almost total segregationin aging, also had the least diverse diet, whereasthe foraging height and substrate use may have re- somewhat specialized X. guttatus had a slightly sulted in a shift in the proportions of main prey more diverse diet than X. spixii. categories,with more spiders and roachesin the diet associatedwith the ground-foraging D. rner- PREY AVAILABILITY ulu and more beetles associatedwith foraging on Basedon the behavioral summary above, we were trunks and vines in D. jiiliginosa. Xiphorhynchus able to compare the diets of three specieswith guttutus,which foragesoften on suspendeddead data on prey availability. Arthropod prey dis- leaves, had a noticeably different diet, at least in placed by army ants in Costa Rica (Otis et al. 910 ANGELA CHAPMAN ANDKENNETH V. ROSENBERG

PERU BOLIVIA 0.6 0. fuliginosa D. fuliginosa 0.5 3

0.6 0.6 D. mew/a

0.5

0.4

0.3

0.2

0.1

0.0 0 5 10 15 20 0 5 10 15 20 0.6 X. gottatus 0.6 X. guttatus 0.5 0.5 . 0.4

0.6 0.6 x. spixii

0.5

0.4 - Orthoptera

0.2 - Ant

0.1

0.0 0 5 10 15 20

NUMBER OF POOLED STOMACHS FIGURE 4. Proportionsof prey taxa in cumulativesamples of stomachsfor four woodcreeperspecies at two sites.

1986) consisted mainly of spiders, roaches, and more beetlesand orthopterans than expectedand ants, whereas prey available in suspendeddead fewer roaches, spiders, and ants. In comparison leaves consistedof beetles, roaches, spiders, and with prey availability in dead leaves, X. guttatus orthopterans (Fig. 2). In all comparisons, the ate slightly more orthopterans and fewer roaches; woodcreepers’ diets were different from’ the dis- all other prey were eaten in proportion to their tributions of available prey. The ant-following availability. D. merula appeared to select orthopterans and avoided small roaches, but took other prey ap- DISCUSSION proximately in proportion (+ or - 10%) to that In this study we have documented dietary dif- flushed by army ants (Fig. 8). D. fuliginosaate ferencesin four woodcreeperswith different for- DIETS OF AMAZONIAN WOODCREEPERS 911

A. Beetles

XG

B. Orthoptera 35 30 25 20 15 10 5 xs XG 0 25 30 35

C. Spiders 2.5 -

2.0 -

1.5- , 1 .o -

0.5 DM DF XG

25 30 35 40 Bill length (mm) FIGURE 5. Comparison of bill length with prey size in four woodcreeper species.Bars indicate one standard deviation; vertical lines indicate range. DM = Dendrocinclame&a, DF = D. fuliginosa, XS = Xiphorhynchus spixii, XG = X. guttatus.

aging habits. Although some differences in diet Dendrocincla merula and D. fuliginosa com- were detected, they nonetheless represent only Pete for food at ant swarms (Willis 1979). D. subtle shifts in the proportions of major prey merula is usually dominant, occupying the center types. Dietary overlap was high for most species- of the swarm, close to the ground, where prey is pairs, and all specieswere more similar in diet most abundant. D. merula also is less adept at than in foraging behavior. arboreal foraging,with lessstiffened rectrices than 912 ANGELA CHAPMAN AND KENNETH V. ROSENBERG

--(c LX menda (n= 1603)

- D. fU/i@KISa (n = 7966) - x. gUft&flS (11~247) * X. spixii (n= 105)

” . 0.0 0.2 0.4 0.6 0.6 1 .o

Proportion of observations

FIGURE 6. Foraging heights of four woodcreeperspecies. Data for D. merulufrom Willis (1979); data for D. fuliginosa from Willis (1972); data for X g&tutus and X spixii from this study. other woodcreepers and more “clumsy” sallies Dendrocincla spp. to be members of a sallying to substrates above the ground (Willis 1979). guild of woodcreepers at Manu National Park, Thus, the behavioral differencesand consequent Peru. She found D. merula to forage more fre- dietary shifts between the two Dendrocincla are quently above the ground at palms and dead actively maintained by a dominance hierarchy leaves, where it takes flushed prey normally hid- at their sharedforaging sites. The diet of D. mer- den inside these substrates.Foraging of D. fuli- ula was more similar to our estimate of prey ginosa was more restricted to trunks and vine availability at ant swarms than was that of D. surfaces,and less often at ants. fuliginosa. This difference,and especiallythe large Willis (1972, 1979) rarely observed Dendro- number of beetles eaten by D. fuliginosa, may cincla spp. preying on beetles, although 30-42% relate to increasedforaging away from army ants of the observed prey in those studies (especially by this species.Pierpont (1986) considered both small prey) were not identifiable. The small num-

n Ground q Vine stem 0 Air El Trunk Live foliage PI Dead leaf Epiphyte 1 .oo z .o 0.80 5 z 3 0.60 8 ‘ii 0.40 5 \\\\\\\\\ \\\\\\\\. .- \\\\\\\\\I I I , I I I , \\\\\\\\\, , , , I , , , r I I I , I I , , \\\\\\\\\I I , , I , I I \ , \ , \ I \ , \ I \ , \ , \ I \ E. 0.20 I I , , , , I I \ ,I,,\ \ \ \ 1,,,\ \ \ \ 2 \.\\\\\\\ a 0.00 D. menda D. Miginosa x. guttatus x. s&li_X/i

(1795) (1622) (217) (69) Species FIGURE 7. Foraging substrateuse by four woodcreeper species.Samples sizes in parenthesesare number of observations. Data sourcessame as Figure 6. DIETS OF AMAZONIAN WOODCREEPERS 913

A. Dendrocincla spp. versus army ants

33 1 w D. fnelu/a 0 D. fuliginosa 20

10

0

-10

-20

-30 ! 4 Beetle Orthopt Roach Spider Bugs Ant Larva

B. X. gulfatus versus dead leaves

-20 ! , Beetle Orthopt Roach Spider Bugs Ant Larva

Prey categories FIGURE 8. Comparison of diet and prey availability for three woodcreeper species. Horizontal mid-line indicatesuse equal to availability; bars above and below horizontal representprey categoriesselected or avoided, respectively. A. D. merulaand D. fuliginosacompared with arthropods flushed by army ants (Otis et al. 1986); B. X. guttatuscompared with arthropods in dead leaves in Pando, Bolivia. ber of beetles seen at an army-ant swarm by Otis (x2 = 0.30, ns). Although these two speciesdif- et al. (1986) also may not reflect their true avail- fered in substratessearched for food, overlap in ability to woodcreepers.Data from leaf-litter traps their diets was nearly complete. Xiphorhynchus in the same area showed that beetles made up guttatus might be expectedto compete more di- about 12% of the arthropod fauna (Otis et al. rectly with other large-bodied flock members that 1986), and Pearson and Derr (1986) found that searchdead leaves, such as foliage-gleaners(Au- beetles made up 13% of the litter arthropods in tomolusand Philydor spp.) and barbets (Eubucco terra jirme forest in Peru. spp.). These various species will occasionally In the mixed-species flocks in which Xipho- search the same clusters of dead leaves without rhynchuswoodcreepers forage, overt aggression exhibiting overt aggression(Rosenberg 1990) al- between X. guttatusand X. spixii is rare (Pierpont though Pierpont (1986) observed several dis- 1986), and competition for food, if present, is placements of foliage-gleaners by X. guttatus. diffuse. Xiphorhynchusguttatus and X. spixii do However, diet composition of X. guttatus over- not avoid flocks in which the other species is lapped more with the other arboreal woodcreep- present. Of 9 1 mixed-species flocks censusedby ers (X. spixii and D. jiiliginosa) than with any Rosenberg in Peru, 36 had X. guttatus, 14 had dead-leaf foraging species (Rosenberg 1990). X. spixii, and 7 had both species;i.e., these spe- Furthermore, X. guttatus ate smaller prey, es- cies were distributed independently among flocks pecially smaller orthopterans, than most of the 914 ANGELA CHAPMAN AND KENNETH V. ROSENBERG dead-leaf foraging species,overlapping most in letting the specimensused in this study. Fieldwork in prey size with the much smaller Myrmotherula Bolivia and Peru was supported by grants from the National GeographicSociety and contributionsby John antwrens (Rosenberg 1990). S. McIlhenny, and H. Irving and L. Schweppe. Per- Compared with birds in temperate commu- mission to work in Bolivia was granted by the Museo nities, tropical birds are generally thought to be National de Historia Natural (La Paz), through Ing. more specialized, in association with greater spe- Magin Zubieta V. and Dr. Luis Felipe Hermann, and cies-packing and more finely divisible resources by the Centro de DesarolloForestal, through Jose Tuesta H. and Ing. For. Luiz Cabrera Vaca. Rosenberg’s re- (e.g., Orians 1969, Terborgh 1980). Several lev- searchin Peru was supportedby NSF Dissertation Im- els of specialization may occur, as seen in the provement Grant BSR 8800905 and a Charles M. Fug- four woodcreepersstudied. In none of thesecases, ler fellowship from LSUMNS. We thank C. L. Canaday however, did specialized behaviors result in a and M. Israel for assistingwith insect identifications. J. M. Bates, K. M. Brown, C. P. Kofron, C. A. Munn, specialized diet. Sherry (1990) has argued that J. V. Remsen, S. K. Robinson, and T. W. Sherry offered foraging specialization may result in stereotyped, helpful suggestionsduring the study or improved drafts but not necessarilynarrow, diets. The low intra- of this manuscript. This study was originally a project population variability in diets seen in these spe- by A. Chapman for J. V. Remsen’s ornithology class. cies suggeststhat the woodcreepers are evolu- LITERATURE CITED tionary (strategic) specialists, yet may be ecological (tactical) generalists (sensu Sherry BORROR,D. J., D. M. DE LONG, ANDC. A. TRIPLEHORN. 1990). 1981. An introduction to the study of insects. SaundersCollege Publ., New York. Why then did these various behaviors evolve BORROR,D. J., AND R. E. WHITE. 1970. A field guide if these speciescontinue to exploit a broadly sim- to the insects of America north of Mexico. The ilar resourcebase? First, it is possiblethat similar Peterson Field Guide series, Houghton Mifflin, morphology and reliance on vertical perches Boston. CALVER,M. C., AND R. D. WOOLLER. 1982. A tech- constrains the diets of these woodcreepersto be nique for assessingthe taxa, length, dry weight, similar. Second, it is possible that behavioral and energycontent of the arthropod prey of birds. modifications evolved largely as a means of Aust. Wildl. Res. 9:293-301. avoiding physical interactions with other species DAHLSTEN,D. L., M. L. MORRISON,D. L. RO\KNEY, while foraging, thus reducing interference com- M. WILSON,ANDY. COHEN. 1985. Bird diets and prey availability in the Sierra Nevada, California. petition (Pierpont 1986) but not diffuse com- Cahf. Fish and-Game 71:172-178. petition for food. Finally, it is likely that these HURLBERT.S. H. 1984. Pseudoreolication and the speciesare taking different prey, but our level of design of ecological field experiments. Ecol. prey identification could not detect these differ- Monogr. 54: 187-2 11. LEVINS, R. 1968. Evolution in changing environ- ences. For example, beetles in the family Cur- ments: some theoretical explanations. Princeton culionidae were prevalent in diets of Xiphorhyn- University Press, Princeton, NJ. thus speciesbut not in Dendrocincla species. MAY, R. M. 1975. Some notes on estimating the Becausethese woodcreepersdiffer in their use competition matrix. Ecology 56:737-741. MUNN, C. A. 1985. Permanent canopy and under- of substratesand foraging space,high overlap in story flocksin Amazonia: speciescomposition and diet is perhaps not unexpected. This may rep- population density, p. 683-712. In P. A. Buckley, resent a case of complementarity of niches, with M. S. Foster, E. S. Morton, R. S. Ridgely, and F. segregation in use of substrates being sufficient G. Buckley [eds.], Neotropical ornithology. Or- to allow similar diets. However, other dead-leaf nithol. Monogr. No. 36. ORIANS,G. H. 1969. The number of bird speciesin foraging species,which overlap greatly in sub- some tropical forests. Ecology 50:783-801. strate use, also overlap broadly in composition OTIS, G. W., E. SANTANAC,D. L. CRAWFORD,AND M. of their diets (Rosenberg 1990). Until we un- L. HIGGINS. 1986. The effect of foraging army derstand more precisely the dietary relationships ants on leaf-litter arthropods. Biotropica 18:56- 61. of other species in these diverse communities, as PARKER,T. A. III, ANDJ. V. REMSEN,JR. 1987. Fifty- well as the distribution of prey among foraging two bird speciesnew to Bolivia. Bull. Brit. Omith. substrates, the role of foraging specialization in Club 107:94-107. maintaining high species richness will remain PEARSON, D. L., AND J. A. DERR. 1986. Seasonal vague. patterns of lowland forest floor arthropod abun- dance in southeastern Peru. Biotropica 18:244- ACKNOWLEDGMENTS 246. PIANKA,E. R. 1974. Niche overlap and diffuse com- We thank members of expeditions from the Louisiana petition. Proc. Natl. Acad. Sci. USA 71:2141-2145. State University Museum of Natural Science for col- PIERPONT,N. 1986. Interspecific aggressionand the DIETS OF AMAZONIAN WOODCREEPERS 915

ecology of woodcreepers (Aves: Dendrocolapti- shrubsteppe birds: competition or dae). Ph.D.diss. Princeton Univ., Princeton, NJ. opportunism in a variable environment? Ecol. RALPH, C. P., S.E. NAGATA,ANDC. J. RALPH. 1985. Monogr. 50:93-l 10. Analysis of droppings to describe diet of small SHERRY,T. W. 1984. Comparative dietary ecology birds. J. Field Ornithol. 56:165-174. of sympatric, insectivorousneotropical flycatchers REMSEN,J. V., JR., AND S. K. ROBINSON. 1990. A (Tyrannidae). Ecol. Monogr. 54:313-338. classificationscheme for foragingbehavior of birds SHERRY,T. W. 1990. When are birds dietarily spe- in terrestrial habitats, p. 144-160. In M. L. Mor- cialized? Distinguishing ecological from evolu- rison, C. J. Ralph, J. Vemer, and J. R. Jehl, Jr. tionary approaches,p. 337-352. In M. L. Moni- teds.], Avian foraging: theory, methodology, and son, C. J. Ralph, J. Vemer, and J. R. Jehl, Jr. [eds.], applications. Studies in Avian Biology No. 13. Avian foraging: theory, methodology, and appli- ROBINSON,S. K., ANDR. T. HOLMES. 1982. Foraging cations. Studies in Avian Biology No. 13. behavior of forest birds: the relationship among TERBORGH,J. 1980. Causes of tropical species di- searchtactics, diet. and habitat structure.Ecoloav __ versitv. Proc. XVII Inter. Omithol. Conar. 955- 63:1918-1931. 961. - ROSENBERG,K. V. 1990. Dead-leaf foraging special- TERBORGH,J., J. W. FITZPATRICK,AND L. EMMONS. ization in tronical forest birds. Ph.D.diss. Loui- 1984. Annotated checklist of bird and mammal siana State Univ., Baton Rouge, LA. speciesof Cocha Cashu Biological Station, Mat-m ROSENBERG,K. V., AND R. J. COOPER. 1990. Ap- National Park, Peru. Fieldiana (Zoology) No. 2 1. proaches to avian diet analysis, p. 80-90. In M. WILLIS. E. 0. 1972. The behavior of Plain-brown L. Morrison, C. J. Ralph, J. Vemer, and J. R. Jehl, Woodcreepers, Dendrocinclafuliginosa. Wilson Jr. [eds.], Avian foraging: theory, methodology, Bull. 81:377-420. and applications. Studies in Avian Biology No. WILLIS, E. 0. 1979. Behavior and ecology of two 13. forms of White-chinned Woodcreepers (Dendro- ROSENBERG,K. V., R. D. OHMART, AND B. W. cincla merula, Dendrocolaptidae) in Amazonia. ANDERSON. 1982. Community organization of Pap. Avulsos Zool. 33~27-66. riparian birds:response to an annualresource peak. WILLIS, E. O., AND Y. ONIKI. 1978. Birds and army Auk 99:260-274. ants. Ann. Rev. Ecol. Syst. 9:243-263. ROTENBERRY,J. 1980. Dietary relationships among