This study reports on food selection by Nearctic migrants, especially thrushes (Catharus), wood-warblers (Oporornis, Seiurus, Wilsonia), and tyrant flycatchers (Empidonax) in a humid forest of central Panama. We attempt to determine how these migrants integrate into the resident bird community based on local food abundance and diet comparison between resident and migrant species. At our study site, migrants are most numerous during migration periods, especially in October when abundance of arthropods is low and fruits are plentiful. Migrants feed equally on fruits during both migrations, but overall depend more on an invertebrate diet. Actually, migrants feed extensively on a few invertebrate taxa that are mostly of two types: (1) small hard-bodied foliage-dwelling insects such as beetles and ants, which are of low nutritional value; and (2) invertebrates well known for producing distasteful or toxic chemicals, such as nonflying termites, millipedes, and centipedes. In contrast, the resident species feed more extensively on invertebrates of higher nutritional value (spiders, insect pupae, alate ants) and on large prey that are plentiful at tropical latitudes (orthopterans, lizards). Dietary relationships among species show that diet of migrants overlap little with that of resident species, even those with which they share a similar foraging substrate. However, unlike residents, migrants belonging to the same foraging guild have a highly similar diet. These results suggest that past competitive interactions between migrants and residents were more important than the ones among migratory species in determining food selection by migrants. Whether the resulting high potential for competition among migrants is related to the short stay of most species at our study site or is a characteristic of several migrant populations at tropical latitudes needs further investigation. Received 8 June 1995, accepted 6 September 1995.

Smithsonian Tropical Research Institute, P.O. Box 2072, Balboa, Ancon, Republic of Panama THE FEEDING ECOLOGY of Nearcftc migrants in the Neotropics has received considerable atten- tion over the last few decades (e.g. Keast and Morton 1980, Hagan and Johnston 1992, Martin and Finch 1995). Several lines of evidence sug- gest that migrants either complement resident species by filling unused niches (Hespenheide 1980, Rappole and Warner 1980) or depend on temporally superabundant resources that are underexploited by resident birds (Willis 1966, DesGranges and Grant 1980, Lefebvre et al. 1994). Considering the wide diversity of Neo- tropical habitats used by migrants and the dif- fering spatial and temporal use they make of these habitats, ecological roles of migrants within tropical communities are certainly di- verse, varying according to species, habitat, and geographic range, as well as over the course of the wintering period. Although food selection by a species is a ma- jor aspect of its ecological niche within the bird community, few studies (Morton 1980, Le- febvre et al. 1992a) have investigated the diet of migrants per se. Information on migrants' diet is mostly limited to observation of birds at flowers (DesGranges and Grant 1980), fruiting trees (Leck 1972, Hilty 1980, Greenberg 1981, Scott and Martin 1984), or army-ant swarms (Willis 1966, 1980). Most studies on migrants' feeding habits have focused on foraging meth- ods and microhabitats used. Although these fac- tors will influence the range of prey available to a bird species, it provides no information on how opportunistic or selective are birds in their prey choice. Direct examination of diet is a use- ful tool to evaluate feeding interactions among species because, in addition to providing in- direct information on the foraging substrate and method used, it gives further insight on the diversity of prey taken, as well as their nutri- tional value. There are no previous studies on bird-community ecology comparing diets of resident and migrant species exploiting the same Neotropical habitat. We report on food selection by migrants, es- pecially thrushes (Catharus), wood-warblers ( Oporornis, Seiurus, W ilsonia ), and tyrant flycatch- ers (Empidonax) that exploit a humid forest in central Panama over the wintering period, but mostly during migration. We attempt to deter- mine how migrants integrate into the resident community based on local food abundance and diet comparison between resident and migrant species. In particular, we answer the following questions: Is migrant abundance related to the exploitation of specific food types that are par- ticularly abundant during their presence at our study site? Do resident and migrant species dif- fer in the use they make of the different food taxa available? How similar are the diets of mi- grant and resident species using the same for- aging substrate? Is diet segregation among mi- grants from the same foraging guild compara- ble to that of their resident counterparts? STUDY AREA AND METHODS Our study was conducted in a second-growth hu- mid forest near Gainboa in central Panama (9ø10'N, 79ø7'W). The study site was in Soberanla National Park, covering an area of 3.2 ha along Pipeline Road between the Frijoles and Frijolito rivers. Mist netting was conducted for 7 h beginning at sunrise during three consecutive days twice monthly from Septem- ber 1993 through November 1994. Thirty-six nets (3 m x 10 m, 32-ram mesh) were placed every 25 m along eight transects 35 m distant. Each bird caught was banded and forced to regurgitate, following the method of Poulin and Lefebvre (1995). However, only the emetic samples collected between September and May (i.e. when migrants were present at our study site) have been used to determine the diet of both migrant and resident species in this paper. Food items from the regurgitations were preserved in 70% ethanol and identified to order (invertebrate) or species (fruit, pollen). Invertebrate prey within each order were grouped into 5-ram-interval size classes. Subadult forms of insects were classed as eggs, pupae, or larvae without taxonomic distinction. Over- all, 27 animal taxa, 65 fruit species, and 89 pollen species were identified in diet samples. To calculate diet similarity among bird species, animal prey taxa were clumped into seven ecological categories: non- flying air/foliage arthropods (Araneae); flying air/ foliage arthropods (Hymenoptera [wasps, alate ants], Diptera, Homoptera, Heteroptera, Orthoptera, Odo- nata, Lepidoptera, Neuroptera, Psocoptera); hard- bodied foliage arthropods (Coleoptera, Hymenoptera [ants]); soft-bodied foliage arthropods (Dermoptera, Isoptera [termites]); inactive foliage arthropods (in- sect egg, nymph, pupae, larvae); ground-dwelling in- vertebrates (Gastropoda [snails], Isopoda, Diplopoda [millipedes], Chilopoda [centipedes], Crustacea [crabs], Scorpionidae, Pseudoscorpionidae); and vertebrates (frogs, lizards). The sizes of animal prey were eval- uated through the numbers of items belonging to each of these three categories: (1) 0-5 ram; (2) >5-15 ram; and (3) >15 min. Similarly, the 65 fruit species found in the regurgitations were classified into five fruit types: small seed fruits; midsize seed fruits; large seed fruits; dry fruits; and large fruits. Number of fruits taken was extrapolated from the number of seeds counted in the regurgitation. Nectar intake was es- timated by the number of species of pollen found in each sample. Relative abundance of arthropods was estimated twice monthly from sweep-net samples. The first 2 m of vegetation were swept with a standard insect net during 20 rain in late morning. Arthropods were identified, sized, and counted as for the regurgitation samples. All taxa taken by the birds were sampled with sweep net. Furthermore, the proportion of ar- thropods from each taxa in sweep-net samples was positively correlated with that found in regurgita- tions (r, = 0.720, df = 23, P < 0.001), suggesting that sweep-net samples provide a reasonable estimate of food available to birds. Abundance of fruit was esti- mated twice monthly using 60 litter traps (0.5 m x 0.5 m) placed 0.25 m above ground every 10 m along the net transects. This method samples fruits (seeds) failing from trees and shrubs, as well as those defe- cated or regurgitated by animals. Only fruit species identified in regurgitation samples were considered to determine fruit abundance over time. To reduce bias associated with patchy distribution of fruits, their abundance was estimated by counting the number of traps in which a fruit species was found, instead of the total number of fruits sampled from that species. Diet preference based on animal prey taxa, animal prey size, and types of fruit taken was compared be- tween residents and migrants as a whole using G-tests. At the species level, diet similarity (overlap) was cal- culated using Pianka's formula (SIMI; Pianka 1973, Rotenberry 1980): SIMI = _ (x,y, x. 2 y , (I) where x, is the proportion of a specific food category in the diet of one bird species, y, is the proportion of a specific food category in the diet of a second bird species, i is the food category, and n is the number of food categories in the diet of the two bird species. This index was calculated based on the proportion of food items among 16 categories including the seven ecological categories of animal prey, the three prey- size classes, the five fruit types, and nectar intake. Only the bird species represented by five emetic sam- ples or more totalling at least 20 identified food items were considered in the analysis. With the exception of the Canada Warbler (Wilsonia canadensis, n = 5), all species were represented by I0 diet samples or more, which should provide an adequate sampling effort for diet estimation (Sherry 1984). The similarity ma- trix was submitted to a principal coordinates analysis (Legendre and Legendre 1983) to reveal diet rela- tionships among 31 resident and 7 migrant bird spe- cies. To facilitate interpretation of dietary relation- ships among birds, each species was assigned to a broad foraging guild based on the main food type and feeding substrate they use (Willis 1966, Ridgely and Gwynne 1989, Stiles and Skutch 1989). Foraging guilds were nectarivores, frugivores, and insecti- vores, the latter being divided into aerial foragers, foliage gleaners, bark foragers, and ground foragers. RESULTS Migrant abundance.--We caught 1,484 resi- dents and 143 migrants between September and May. Nearctic migrants were captured through- out the wintering period, but mostly during fall migration, when they represented up to 25% of the mist-netted birds. Thrushes (Catharus ustu- latus, C. minimus, C. fuscescens) were the most abundant migrants with a major peak in Oc- tober and a second one in April (Fig. 1). Two 50- 40- 20- 0 r'-] s Fig. I. :[J Thrushes [] Tyrant flycatchers ß Wood-warblers J A S 0 N --- dry season--, Temporal variation in migrant abundance. Wood Thrushes (Hylocichla mustelina) were also captured in both February and November. Wood-warblers (Oporornis formosus, Seiurus nov- eboracensis, S. aurocapillus, Wilsonia canadensis, Protonotaria citrea, Mniotilta varia, Dendroica stria- ta, and D. pensylvanica) were caught throughout the winter, but mostly during migration (Fig. 1). Finally, two flycatcher species were sam- pled--Contopus virens during fall migration, and Empidonax virescens throughout the wet season (Fig. 1). Food abundance.--The study site has an aver- age annual precipitation of 2,133 mm and is characterized by a severe dry season from Jan- uary through April. During the wet season, mean monthly rainfall is constant from May through September, peaks in October, and reaches a low in December. Arthropods sam- pled with sweep net showed a maximal abun- dance at the beginning of the wet season (May- June) and minimal numbers in October when rainfall was highest (Fig. 2). This pattern was not related to a few abundant taxa, since most invertebrate categories followed that same trend (Fig. 2). Fruit abundance pattern differed among the various fruit types (Fig. 2). Berrylike fruits with numerous tiny seeds, such as Miconia, showed a major peak in the late wet season (November) and a second one in the late dry season (April). Fruits bearing from one to five medium-sized seeds (e.g. Psychotria) peaked a month earlier. Fruits bearing large seeds (>8 mm long) were sampled mostly during the dry season (January through April). Dry fruits were abundant in June and September. Finally, large fruits from which only a portion could be taken by the birds were more common during the 600 500 .c_ -o 400 300 E 200 Z 100 60 ÷ m 40 20 Ground-dwelling [] invertebrate Inactive foliage [] arthropod Soft-bodied foliage [] arthropod Hard-bodied foliage [] arthropod Nonflying air/foliage [] arthropod Flying air/foliage ß arthropod ß [] Dry fruit [] Large fruit [] Large seed fruit [] Medium seed fruit ß Small seed fruit OND J FMAMJ JASON Fig. 2. Temporal variation (October 1993 through November 1994) in abundance of invertebrates sampled by sweep net and of fruits collected in fruitfall traps, with distinction of ecological prey category (see Methods for taxonomic affinity) and fruit type. second part of the wet season (August-Octo- ber). Bird diet.--Migrant and resident species dif- fered in the use they made of the different in- vertebrate taxa available (Table 1). Overall, mi- grants fed more frequently on nonflying ants, beetles, and insect larvae than did resident spe- cies, although these insect prey were common in the diet of the resident species as well. Gas- tropods (snails) represented a higher propor- tion of the diet of migrants compared to resi- dents, while Catharus, Seiurus, and Oporornis reg- ularly fed on them. Finally, migrants fed sig- nificantly more on nonflying termites, centi- TABLE 1. Diet comparison between migrant and res- ident species. Significant differences (G-tests) shown by asterisks (***, P < 0.001; **, P < 0.01; *, P < 0.05). No. items/100 samples TABLE 2. Fruit intake by migrant species. Prey taxa Per- cent of No. sam- No. No. fruit pies sam- fruits spe- with Sta- Bird species pies taken cies fruits tisti- Catharus ustulatus (fall) 25 54 17 80 Mi- cal Resi- C. ustulatus (spring) 11 27 6 73 grants dif- dents No. C. minimus 15 14 6 47 (n = fer- (n = items C. fuscescens 10 27 12 60 143) ence 1,484) taken Hylocichla mustelina 4 4 1 25 Other species 78 9 5 9 Hymenoptera (ants) 167.8 Coleoptera (beetles) 130.8 Isoptera (termites) 80.4 Diplopoda (millipedes) 44.1 Insect larvae 36.4 Gastropoda (snails) 15.4 Chilopoda (centipedes) 2.1 Insect eggs 33.6 Hymenoptera (wasps) 17.5 Heteroptera 9.8 Diptera 2.8 Isopoda 0.7 Odonata 0.7 Lepidoptera (adults) 0.0 Frogs 0.0 Hymenoptera (alate ants) 0.0 Insect pupae 0.0 Lizards 1.4 Homoptera 2.1 Orthoptera 7.7 Araneae (spiders) 19.6 Prey size 0-5 mm 423.1 >5-10 mm 109.1 >10-15 mm 23.1 >15-20 mm 4.9 >20-25 mm 3.5 >25 mm 3.5 Fruit type Small seed fruits 25.9 Medium seed fruits 65.0 Dry fruits 4.2 Large seed fruits 2.8 Large fruits 0.7 pedes, and millipedes than the residents. Mil- lipedes were not taken only by species that for- age typically on the ground such as thrushes, but also by tyrant flycatchers (Contopus, Empi- donax), and wood-warblers (Oporornis). These seven taxa accounted for 83% (682/822 items) of the invertebrate diet of all migrant species. In contrast, residents fed significantly more on spiders, orthopterans, homopterans, lizards, insect pupae and alate ants than migrants (Table *** 21.8 563 *** 46.2 873 *** 0.1 116 *** 2.6 101 1). These taxa accounted only for 38% (1,493/ * 25.7 433 3,888 items) of their animal diet. *** 0.4 28 Some differences also were observed in prey ** 0.1 5 35.0 568 size (Table 1). Migrants fed more extensively 11.4 194 on small invertebrates (74% of all items taken), 9.7 158 whereas prey longer than 15 mm consistently 3.6 58 were more important in the diet of the resident 0.9 15 species. 0.9 15 0.5 8 Finally, migrants showed a clear preference 1.3 19 for berrylike fruits with seeds of small to me- * 2.0 29 dium size, and seemed to avoid feeding on large ** 2.8 42 fruits. The thrushes clearly were more frugiv- * 5.1 77 * 6.1 94 orous than any other migrant species (Table 2), ** 16.3 253 but Ernpidonax virescens, Dendroica pensylvanica, *** 68.3 1,042 D. striata, and Seiurus aurocapillus also fed on fruits. Psychotria (61 fruits taken from nine spe- *** 160.5 2,987 cies), Miconia (25 fruits from two species), Pal- *** 52.2 930 icourea (19 fruits), and Conostegia (9 fruits) rep- 21.8 356 resented 84% of all fruits taken by migrants. 7.1 112 The occurrence of these fruit species in the 7.0 109 *** 13.1 199 emetic samples was significantly higher for mi- grants than residents (G = 68.8, df = 1, P < *** 12.3 219 0.001). Nevertheless, animal food was more im- *** 7.7 208 portant than fruits in the diet of any migrant 3.2 54 species, in terms of both number of items taken 2.3 38 and proportion of samples in which each food 1.9 29 type appeared. Dietary relationships.--Dietary relationships among the 38 bird species submitted to the principal coordinates analysis are summarized in Figure 3. The Euclidean representation of the first two axes of the analysis is adequate based on the criteria suggested by Cailliez and Pages (1976). The clumping of the various food taxa into a few categories allowed segregation of diets, since the bird species are almost uniform- ly distributed in the ordination diagram. Among resident species, only nectarivores and frugivores showed a relatively high diet simi- 0.2 0.1- 1145[] 17 & 18& rh2 16 [] m33 11 + ß 34 lO+ [] 3 [] 30 [] 6 8 9 ß 20 21 ß e28 22 ß e27 23 ß 26 36m ß ß 35 ß 25 29 ß 13+ ß 38ß 37 26A -0.2 -0.2 012 Axis 1 Ground-foraging ß insectivore Bark-foraging ß insectivore Foliage-gleaning ß insectivore Aerial-foraging + insectivore ß Frugivore r Nectarivore [] Migrant Fig. 3. Ordination of bird species based on diet similarity using principal coordinates analysis. Species corresponding to each number listed in Table 3. larity among them, each one of these guilds being confined to a small portion of the dia- gram. These birds feed primarily on nectar (hummingbirds) or fruits (manakins), and their arthropod diet is limited to a few prey taxa (Ta- ble 3). The high numbers of nectarivores and frugivores sampled, as well as their low recap- ture rates, are consistent with a high diet over- lap, since the birds' movements allow them to feed on locally abundant flowers and fruits (Loiselle and Blake 1992). There is no tendency for the resident insec- tivores belonging to a same foraging group to be clumped together. The low diet overlap among these residents is related to the variety of arthropods taken from different prey cate- gories. This was especially obvious for aerial foragers that are widely distributed in the or- dination diagram (Fig. 3): Mionectes oleagineus fed almost exclusively on spiders and small fruits; Attila spadiceus fed extensively on lizards in addition to flying insects; finally, Terenotric- cus erythrurus and Oncostoma olivaceum fed most- ly on the wing, but supplemented their diet with nonflying and subadult insects (Table 3). The other three groups of resident insectivores are intermixed and occupy the portion of the diagram on the right. There is little grouping of these species by their foraging guild and, frequently, the species most similar in diet do not use the same foraging substrate. Resident gleaning insectivores supplemented their diet with either fruits, seeds or flying insects, and also differed by the use they make of verte- brates, subadult insects and spiders (Table 3). Among bark foragers, Xiphorhynchus differed from the other two species by its high intake of insect larvae and low intake of spiders and ants (Table 3). Finally, resident ground foragers differed mostly in the use they made of spiders, beetles, and ants, as well as in the size of prey taken (Table 3). In contrast, the diets of migrant species be- longing to the same foraging guild were highly similar, as shown by the tight grouping of mi- grant foliage gleaners and migrant ground for- agers in the ordination diagram (Fig. 3). Dif- ferences in diet overlap between migrants and residents are obvious when comparing the mean diet similarity values between species from a same foraging group (Table 4). Even the resi- dent bark foragers--represented by only three species and two genera--overlapped less in their diet than migrants from any foraging guild (Ta- ble 4). Diet overlap among migrants was rather similar to the values obtained for nectarivorous and frugivorous resident species (Table 4). However, in contrast to nectarivores and fru- givores, Catharus thrushes fed on a large variety of food types (Table 3). Their high diet simi- TABLE 3. Diet of most-abundant resident and migrant species. Number of items taken indicated for each food category: (A) flying air/foliage arthropods (orthopterans, bugs, wasps, others); (B) nonflying air/foliage arthropods (spiders); (C) hard-bodied foliage arthropods (beetles, ants); (D) soft-bodied foliage arthropods (termites); (E) inactive foliage arthropods (insect larvae, eggs and pupae); (F) ground-dwelling arthropods (snails, millipedes, others); (G) vertebrates (frogs and lizards); and plant food (fruit, pollen). A B C D E Prey size Plant F G_<5 >5 food Bird species ortbugwasoth spi bee ant ter lar e/psnamilothver mm mm fru pol Nectarivores 1 Phaethornis longuemareus 9 9 42 2 P. superciliosus 3 1 99 1 2 2 100 8 2 242 3 Thalurania colombica 3 1 16 1 21 7 4 Amazilia amabilis 1 44 23 133 4 1 5 207 4 74 5 Damophila julie 2 61 27 137 4 1 7 237 2 69 Frugivores 6 Manacus vitellinus 1 1 36 7 Pipra coronata 3 2 3 4 19 26 1 6 2 55 11 124 8 P. mentalis 2 1 33 8 2 17 5 48 20 352 Aerial-forager insectivores 9 Mionectesoleagineus 1 221 2 2 1 194 33 111 3 10 Terenotriccus erythrurus 6 1 1 15 3 18 18 1 52 11 11 Oncostoma olivaceum 7 4 3 27 13 94 7 7 58 182 38 3 12 Empidonax virescens a 5 7 4 6 41 30 2 1 1 53 44 3 13 Attila spadiceus 1 1 3 2 2 10 5 14 3 Foliage-gleaner insectivores 14 Seiurus noveboracensis a,b 1 4 25 62 8 2 100 2 15 Oporornisformosus a,b 6 2 5 2 12 47 108 4 22 9 4 1 196 27 16 Wilsonia canadensis a 1 4 6 16 2 23 6 17 Xenops minutus 8 1 2 3 8 42 1 3 52 16 18 Hylophylax naevioides 4 3 3 9 7 104 9 6 29 4 1 14 124 69 19 Myrmotherula axillaris 4 1 15 12 2 2 23 13 20 Cyanocompsa cyanoides 4 11 1 8 8 21 Eucometis penicillata 5 2 2 1 5 10 5 2 1 1 14 20 22 Myrmotherulafulviventris 35 4 1 2 73 43 11 2 53 2 4 98 132 23 Tachyphonus luctuosus 7 1 4 2 9 3 1 10 17 24 Thamnophilus punctatus 38 48 17 19 49 9 45 109 2 9 96 250 25 Myrmeciza exsul 5 2 1 4 6 2 8 8 2 6 9 34 26 Schiffornis turdinus 1 1 44 1 45 36 13 19 29 Bark-forager insectivores 27 Dendrocincla homochroa 5 1 2 12 10 1 2 2 12 22 28 D. fuliginosa 18 15 11 9 35 42 13 1 74 1 4 16 109 130 29 Xiphorhynchusguttatus 7 10 2 7 7 25 27 2 4 83 Ground-forager insectivores 30 Catharusfuscescens a 1 1 8 2 15 1 2 11 2 28 12 27 31 C. ustulatus a 2 15 15 29 20 1 3 15 2 62 38 81 32 C. minimus a 1 23 9 71 18 423 1 1 94 49 14 33 Sclerurus guatemalensis 4 15 54 1 6 3 8 4 4 58 38 34 Formicariusanalis 8 3 1 4 38 50 9 11 11 5 8 60 88 1 35 Cyphorhinus phaeocephalus 3 1 12 3 1 3 2 1 10 16 36 Gymnopithys leucaspis 28 5 5 45 22 8 9 9 1 3 8 43 99 1 37 Phaenostictus mcleannani 5 1 5 1 1 2 1 14 38 Baryphthengus martii 2 2 5 3 1 1 3 1 1 17 6 Total 215 116 158 169 972 769 396 115 259 447 23 89 29 90 2,400 1,430 860 438 Migrant species. Also ground foragers. TABLE 4. Comparison of diet-similarity values be- tween each foraging guild with distinction of mi- grant and resident species. Diet-similarity value Foraging guild Mean Minimum Resident species Nectarivores 0.9175 0.8366 Frugivores 0.9525 0.9457 Aerial-forager insectivores 0.6013 0.2831 Foliage-gleaner insectivores 0.6620 0.0253 Bark-forager insectivores 0.7870 0.7107 Ground-forager insectivores 0.6239 0.2851 Migrant species Foliage-gleaner insectivores 0.9518 0.9172 Ground-forager insectivores 0.9263 0.9044 larity is related to the fact that they fed on the same arthropod taxa and fruit types in similar proportion. Thrushes also have a ratio of small versus large prey taken of 2:1, which is unusual among the bird community (Table 3). The high diet overlap among wood-warblers is mostly related to their similar intake of hard-bodied and inactive foliage arthropods, and flying and nonflying air/foliage arthropods, as well as to their avoidance of fruits and preference for small animal prey (Table 3). Diets of migrants, however, showed little overlap with that of resident species, even those with which they share a similar foraging mode (Fig. 3). Migrant thrushes differed from the res- ident ground foragers by their high intake of termites, millipedes, and fruits, as well as by their low intake of spiders and vertebrates (Ta- ble 3). Wood-warblers, especially Oporornis and Seiurus, differed from most resident gleaners by their intake of ants and gastropods, whereas Wilsonia fed more importantly on flying insects (Table 3). Finally, Ernpidonax, the only migrant aerial forager for which enough data were col- lected, differed from its resident counterparts by the extensive use it made of both beetles and ants (Table 3). DISCUSSION Because fall migration coincides with the tropical wet season, it is often assumed that mi- grants arrive at their wintering ground during a period of high arthropod availability (Russell 1980, Schwartz 1980). However, several studies show that arthropods peak primarily at the on- set of the rainy season and decrease afterward (e.g. Buskirk and Buskirk 1976, Willis 1976, Smythe 1982, Levings and Windsor 1985, Pou- lin et al. 1992). At our study site, foliage-dwell- ing arthropods reach their lowest numbers in October when migrant abundance is highest. Although there are some reports of fruit in- take by migrants during fall migration (Leck 1972), frugivory generally is considered to be more important during the dry season and spring migration (Morton 1971, Greenberg 1981, Scott and Martin 1984). Our results show that migrants feed equally on fruits during both mi- grations, although fruits as a whole are most abundant during the fall. Most migrant species, however, depend mainly on an invertebrate diet at our study site. They do not feed opportunistically on supera- bundant arthropods, but rather specialize on a few invertebrate taxa which are basically of two types. The first type are nonflying termites, mil- lipedes, and centipedes, invertebrates rarely taken by the resident birds and well known for producing distasteful or toxic chemicals (Cloud- sley-Thompson 1968, Moore 1969, Hopkin and Read 1992). The second group are small hard- bodied insects, such as beetles and nonflying ants, which are commonly taken by the resident species as well, although to a lesser extent. Be- cause these insects are found on every substrate and are relatively easy to prey upon, they rep- resent an important part of the arthropod diet of several species in tropical bird communities (Lefebvre et al. 1992b, Poulin et al. 1994, this paper). However, their low nutritive value (low fat content, high proportion of cuticle not di- gested by most birds; Bell 1990) and the fact that they are taken frequently but always in small amounts and, more importantly, when overall abundance of arthropods is low (Poulin unpubl. data), suggest that these prey are just a "better-than-nothing" food to the birds. The number of beetles and ants in sweep-net sam- pies was especially low in September-October and, consequently, their high intake by mi- grants cannot be related to local abundance. The reason why migrants forage on these taxa even more frequently than the resident birds could be related to: (1) the migrants' lower foraging success on other prey types; or (2) the birds' recent arrival after migration involving high physiological demands. Temporal intake of beetles and ants by migrants suggests that the second hypothesis is most likely correct. Ants and beetles were taken most frequently in Sep- tember and March, and least frequently in win- ter; they were avoided in April-May when the birds are expected to build up fat reserves. Alternatively, resident birds feed more ex- tensively than migrants on two prey types. The first one involves spiders, insect pupae, and alate ants, which are characterized by a low chitin and/or high fat content (Dalingwater 1987, Bell 1990). The other type refers to orthopterans and lizards, which are large prey particularly abun- dant and diverse at tropical latitudes. Overall, the diet of migrants shows little overlap with that of the resident species, even with those using a similar foraging substrate. Moreover, diet relationships among insecti- vores belonging to the same foraging group were totally different when comparing mi- grants and residents. While the resident species differed in several aspects of their diet, migrant species from a same foraging group were highly similar in their food choice. A lower diet over- lap among resident species is probably related to the fact that these birds coexist year-round, whereas most migrants spend only a few days or weeks at our study sites. Resident species will either feed on a wide variety of animal prey in differing proportions, or specialize on different taxa, which in both cases decrease diet overlap within the bird community. However, migrants of the same foraging guild tend to feed consis- tently on the same food taxa in similar propor- tion. Catharus thrushes were the most abundant migrants at our study site. They commonly feed on fruits during both migrations and this re- source is considered as the key factor to their presence in that habitat (Martin 1985). Thrushes were present at our study site when fruits were particularly abundant, suggesting that they are tracking that resource. In several habitats in Costa Rica (Blake and Loiselle 1992) and in Co- lombia (Hilty 1980), the same species of thrush- es are reported to feed on fruits throughout winter. Nonetheless, migrant thrushes feed more extensively on invertebrates than on fruits at our study site. These birds are unique in the sense that they are the almost exclusive con- sumers of termites, millipedes, and centipedes. As many as 56 millipedes and 115 termites were found in 65 emetic samples from thrushes. Be- cause millipedes occurred in more than 60% of the samples, and termites represented as much as 35% of all invertebrates taken, the distribu- tion of these invertebrates could also be im- portant in habitat selection by migrant thrush- es. At our study site, the intake of millipedes by the resident species peaks in June and Oc- tober, being especially low from December to March. Assuming that the actual number of mil- lipedes taken by the resident bird community is a good estimate of their local availability, the presence of thrushes at our study site would coincide with the period of highest abundance of millipedes. Nonflying termites are presum- ably available year-round, but are ignored as food source by the resident species. Winter habitat selection by migrants is prob- ably a compromise between the availability of appropriate food types and the minimization of competitive interactions with resident species. During their short stay in a humid forest of central Panama, migrants feed mostly on small fruits and low-quality invertebrates easy to prey upon. Furthermore, in contrast to resident spe- cies, migrants feed little on arthropods that are plentiful at tropical latitudes and/or character- ized by a low-chitin or high-fat content. The low quality and diversity of animal prey taken by migrants, as well as the highly overlapping diet among migrant species, suggest that mi- grants are not well integrated into the tropical bird community. On the other hand, migrants differ from residents in several aspects of their diet, and feed on different food taxa that each represents only a fraction of the diet of several resident species. As a result, there is little diet overlap between migrants and residents using a similar foraging substrate which, alternative- ly, suggests that each migrant foraging group is well integrated into the Neotropical bird community. Dietary relationships suggest that past competitive interactions between migrants and residents were more important than the ones among migratory species in determining food selection by migrants. Whether the re- suiting high potential for competition among migrants is related to the short stay of most species at our study site or is a characteristic of several migrant populations in the Neotropics needs further investigation. ACKNOWLEDGMENTS This study was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC). We are grateful to the Smithsonian Tropical Research Institute for providing logistical support and to INRENARE for issuing the research permit to work in Soberanla National Park. We are indebted to var- ious colleagues at the Smithsonian Tropical Research Institute, especially Osvaldo Calder6n and Joseph Wright, for their invaluable assistance in identifying fruits. Neal Smith, Peter Wrege, Allen Keast and John Blake provided helpful comments on the manuscript. LITERATURE CITED BELL, G. P. 1990. Birds and mammals on an insect diet: A primer on diet composition analysis in relation to ecological energetics. Pages 416-422 in Avian foraging: Theory, methodology, and ap- plications (M. L. Morrison, C. J. Ralph, J. Verner, and J. R. Jehl, Eds.). Stud. Avian Biol. 13. BLAKE, J. G., AND B. A. LOISELLE. 1992. Fruits in the diets of Neotropical migrant birds in Costa Rica. Biotropica 24:200-210. BUSCIRC, R. E., aD W. H. BUSIPaC. 1976. Changes in arthropod abundance in a highland Costa Ri- can forest. Am. Midi. Nat. 95:288-298. CaLLIEZ, F., AND J.-P. PAG,S. 1976. Introduction & l'analyse des donnes. Socit de Mathmatiques appliques et de Sciences humaines, Paris. CLOUDSLY-THoM?soN, J. L. 1968. Spiders, scorpi- ons, centipedes and mites. Pergamon Press, Ox- ford. DALINGwATER, J. E. 1987. Chelicerate cuticle struc- ture. Pages 3-15 in Ecophysiology of spiders (W. Nentwig, Ed.). Springer-Verlag, New York. DESGRANGES, J.-L., AND P. R. GRANT. 1980. Migrant hummingbirds' accommodation into tropical communities. Pages 395-409 in Migrant birds in the Neotropics: Ecology, behavior, distribution and conservation (A. Keast and E. S. Morton, Eds.). Smithsonian Institution Press, Washington, D.C. GREENBERG, R. 1981. Frugivory in some migrant tropical forest wood warblers. Biotropica 13:215- 233. HAGAN, J. M, III, AND D. W. JOHNSTON. 1992. Ecology and conservation of Neotropical migrant land- birds. Smithsonian Institution Press, Washing- ton, D.C. HESPENHEIDE, H.A. 1980. Bird community structure in two Panama forests: Residents, migrants, and seasonality during the nonbreeding season. Pages 227-237 in Migrant birds in the Neotropics: Ecol- ogy, behavior, distribution and conservation (A. Keast and E. S. Morton, Eds.). Smithsonian In- stitution Press, Washington, D.C. HILT3/4, S.L. 1980. Relative abundance of North Tem- perate Zone breeding migrants in western Co- lombia and their impact at fruiting trees. Pages 265-271 in Migrant birds in the Neotropics: Ecol- ogy, behavior, distribution and conservation (A. Keast and E. S. Morton, Eds.). Smithsonian In- stitution Press, Washington, D.C. HOPKIN, S. P., AND H. J. RD. 1992. The biology of millipedes. Oxford Univ. Press, Oxford. KEAST, A., AND E. S. MORTON. 1980. Migrant birds in the Neotropics: Ecology, behavior, distribu- tion and conservation. Smithsonian Institution Press, Washington, D.C. LECK, C.F. 1972. The impact of some North Amer- ican migrants at fruiting trees in Panama. Auk 89:842-850. LEFEBVRE, G., B. POULIN, AND R. MCNEIL. 1992a. Abundance, feeding behavior, and body condi- tion of Nearctic warblers wintering in Venezu- elan mangroves. Wilson Bull. 104:400-412. LEFESVRE, G., B. POULIN, AND R. McNEIL. 1992b. Set- tlement period and function of long-term terri- tory in tropical mangrove passerines. Condor 94: 83-92. LEFEBVRE, G., B. POULIN, AND R. MCNEIL. 1994. Tem- poral dynamics of mangrove bird communities in Venezuela with special reference to migrant warblers. Auk 111:405-415. LEGENDRE, L., AND P. LEGENDRE. 1983. Numerical ecology. Elsevier Scientific Publishing Company, Amsterdam. LEVINGS, S.C., AND D. M. WINDSOR. 1985. Litter ar- thropod populations in a tropical deciduous for- est: Relationships between years and arthropod groups. J. Anim. Ecol. 54:61-69. LOISELLE, B. A., AD J. G. BLAKE. 1992. Population variation in a tropical bird community. Bio- Science 42:838-845. MARTIN, T. E. 1985. Selection of second-growth woodlands by frugivorous migrating birds in Panama: An effect of fruit size and plant density? J. Trop. Ecol. 1:157-170. MTN, T. E., AND D. M. FINCH. 1995. Ecology and management of Neotropical migratory birds: A synthesis and review of the critical issues. Oxford Univ. Press, New York. MOORE, B.P. 1969. Biochemical studies in termites. Pages 407-432 in Biology of termite, vol. 1. (K. Krishna and F. M. Weesner, Eds.). Academic Press, New York. MORTON, E. S. 1971. Food and migration habits of the Eastern Kingbird in Panama. Auk 88:925-926. MORTON, E.S. 1980. Adaptations to seasonal changes by migrant land birds in the Panama Canal Zone. Pages 437-453 in Migrant birds in the Neotropics: Ecology, behavior, distribution and conservation (A. Keast and E. S. Morton, Eds.). Smithsonian Institution Press, Washington, D.C. PIANKA, E. R. 1973. The structure of lizard com- munities. Annu. Rev. Ecol. Syst. 4:53-74. POULIN, B., D G. LEFESVRE. 1995. Additional in- formation on the use of emetic in determining the diet of tropical birds. Condor 97:897-902. POULIN, B., G. LEFESVRE, AND R. McNEIL. 1992. Trop- ical avian phenology in relation to abundance and exploitation of food resources. Ecology 73: 2295-2309. POULIN, B., G. LEFEBVPE, ^Nr R. McNL. 1994. Diets of land birds from northeastern Venezuela. Con- dor 96:354-367. RAPPOI, J. I-I, AND D. W. WARNER. 1980. Ecological aspects of migrant bird behavior in Veracruz, Mexico. Pages 353-393 in Migrant birds in the Neotropics.' Ecology, behavior, distribution and conservation (A. Keast and E. S. Morton, Eds.). Smithsonian Institution Press, Washington, D.C. RIrCEE3/4, R. S., ^Nr J. A. GWYNNE. 1989. A guide to the birds of Panama with Costa Rica, Nicaragua, and Honduras. Princeton Univ. Press, Princeton, New Jersey. ROTENBERRY, J.T. 1980. Dietary relationships among shrubsteppe passefine birds: Competition or op- portunism in a variable environment. Ecol. Mon- ogr. 50:93-110. RVSSELE, S. M. 1980. Distribution and abundance of North American migrants in lowlands of north- ern Colombia. Pages 249-252 in Migrant birds in the Neotropics: Ecology, behavior, distribution and conservation (A. Keast and E. S. Morton, Eds.). Smithsonian Institution Press, Washington, D.C. SCW^RTZ, P. 1980. Some considerations on migra- tory birds. Pages 31-34 in Migrant birds in the Neotropics: Ecology, behavior, distribution and conservation (A. Keast and E. S. Morton, Eds.). Smithsonian Institution Press, Washington, D.C. SCOTT, P. E., AND R. F. MtRTN. 1984. Avian con- sumers of Bursera, Ficus and Ehretia fruits in Yu- catan. Biotropica 16:319-323. SI-IERRY, T. W. 1984. Comparative dietary ecology of sympatric, insectivorous Neotropical flycatchers (Tyrannidae). Ecol. Monogr. 54:313-338. SMYThe, N. 1982. The seasonal abundance of night- flying insects in a Neotropical forest. Pages 309- 318 in The ecology of a tropical forest: Seasonal rhythms and long-term changes (E.G. Leigh, Jr., A. S. Rand, and D. M. Windsor, Eds.). Smithson- Jan Institution Press, Washington, D.C. STES, G., ANr A. F. SKUTC. 1989. A guide to the birds of Costa Rica. Cornell Univ. Press, Ithaca, New York. Wts,E.O. 1966. Theroleofmigrantbirdsatswarms of army ants. Living Bird 5:187-231. WLLIS, E.O. 1976. Seasonal changes in the inver- tebrate litter fauna on Barro Colorado Island, Panama. Rev. Brasil. Biol. 36:643-657. WI,S, E. O. 1980. Ecological roles of migratory and resident birds on Barro Colorado Island, Panama. Pages 205-225 in Migrant birds in the Neotropics: Ecology, behavior, distribution and conservation (A. Keast and E. S. Morton, Eds.). Smithsonian Institution Press, Washington, D.C.