436 ShortCommunications [Auk, Vol. 103

Frequencyof Arthropods in Stomachsof Tropical

J. v. gEMSEN,JR., • F. GARY STILES,2 AND PETERE. SCOTT1 tMuseumof Zoologyand Department of Zoologyand Physiology, Louisiana State University, BatonRouge, Louisiana 70803 USA, and 2Escuelade Biologfa,Universidad de CostaRica, Ciudad Universitaria "Rodrigo Facio," Costa Rica

Although flower nectar is the most conspicuous per we do not record in detail the kinds of arthro- and energeticallyefficient food sourceof humming- pods consumed,except insofar as this may affect the ,it is notablydeficient in amino acidsand other frequency of detectablearthropod remains. A de- essential nutrients (Baker and Baker 1975, Hains- tailed study of the arthropod diets and foraging tac- worth and Wolf 1976). Therefore, hummingbirds re- ticsof hummingbirdsis in preparation(Hespenheide quire an additional source of proteins, lipids, and and Stiles unpubl. data). other nutrients. In most or all ,these nutrients The specimensreported here were collected (1) are obtainedby consumingsmall arthropods.Yet ar- from 1980 to 1985 by personnel of the Museum of thropod-foragingby hummingbirdsremains very lit- Zoology,Louisiana State University (LSUMZ) or (2) tle studied comparedwith nectar-foraging(Gass and from 1971 to 1985 by Stiles and his students.Ap- Montgomerie1981, Hespenheide and Stilesunpubl. proximately 70% of all specimenswere collectedin data). The few available time-budget studies of for- Bolivia or Peru, 25% in Costa Rica, and the remainder aging hummingbirds(reviewed by Gassand Mont- in northwestern Ecuador, Venezuela, or Darign, Pan- gomerie 1981) indicate that arthropod-hunting con- ama.Twenty recentspecimens of 15 speciesfrom Ec- stitutesonly 2-12%of total foragingtime exceptwhen uador and Peru depositedin the Academyof Natural nectar is scarceor unavailable.An incubating female Sciences,Phildelphia, also were included. Although Broad-tailed (Selasphorusplatycercus) specimenswere obtained from throughout the year, consumedonly arthropodsfor about 10 days when the samplefrom Peru and Bolivia is mostly from May nectar was unavailable (Montgomerie and Redsell throughAugust, the "dry" season.Most (>95%) birds 1980). Thus, under some field conditions a hum- from group ! above were captured in mist nets, mingbird can satisfy energeticas well as nutritional which were checked from dawn to dusk at 3-4-h needsby feeding exclusivelyon arthropods. intervals;the remainder were shot. Birds from group The contentsof stomachsand crops of collected 2 were collected from mist nets checked at least hour- birds representan obvioussource of dietary infor- ly, or were shot,in roughly equalproportions. Stom- mation. Becausethese data are recordedroutinely on achsand cropsof group ! birds were scoredon spec- specimenscollected by our institutions,the speci- imen labels as containing "insects"(including other menscan provide an indicationof the frequencyand arthropods),as containingunidentifiable material, or generalityof arthropod-foragingin hummingbirds. as being "empty." Group 2 birds were scored in Accordingly,we recordedthe presenceor absenceof greater detail, but the data were convertedto one of arthropodremains in the stomachsand cropsof !,629 the three categoriesused for group ! birds. Although specimensof 140 speciesof hummingbirds(41% of not usually recorded on the labels, stomachsthat the approximately340 speciesin the family Trochil- containedarthropods were almostalways packed with idae). material rather than containing just a trace of parts. Presenceof arthropodsin the digestive tract of a Becausesoft insectsand spiderscan be digestedcom- specimenindicates only that the had recently pletely within 3-4 h (Stilesunpubl. data), our sam- fed on arthropods;it tells us nothing aboutthe rel- pling procedure for mist-netted birds is conserva- ative frequencyof arthropodsvs. nectarin the diet. tively biasedagainst detection of arthropods.Also, It is well known that different kinds of foods have many of the stomachcontents of group ! specimens different digestibilities,rates of gut passage,and de- scoredas "unidentifiable" were presumably the re- tectabilities; nectar and arthropods exemplify this mains of soft insectsand spiders. problemin the extreme.Nectar is virtually 100%as- Most (1,279 of 1,629, 79%; see Appendix) hum- similable (Hainsworth 1974, Hainsworth and Wolf mingbird stomachsand crops containedarthropod 1972)and is rarely detectedin the stomachbecause remains. Individuals of 133 (95%) of the 140 species it usuallypasses directly from esophagusto intestine examined had been feeding on arthropods.Of the 7 (the openingsof theseare closelyapposed in hum- speciesfor which arthropodswere not recorded,the mingbird stomachs).Arthropods are extensivelydi- largestsample size was only 5 individuals. gestedin the stomach,following only brief storage Few significantregional differences were detected in the crop.Soft-bodied arthropods, such as most small between samplesof the samespecies; therefore, data spiders,may be digestedrelatively quickly, except from all areaswere pooledin the Appendix.The main for easilyoverlooked poison claws. Many chitinous exceptionswere several speciesof hermits (Phae- insectparts (e.g. wings, head capsules) are easilyde- thorninae: genera Glaucisthrough ).In tectablefor considerablylonger periods.In this pa- stomachsof these speciesarthropods occurred with April 1986] ShortCommunications 437

T^I•LE1. Percentageof individual nonhermit hum- TABLE2. Chi-squarevalues for pairwise compari- mingbirds feeding on arthropods in six humidity- sons of arthropod-feeding frequency between elevation categories.Species found in more than nonhermit hummingbirds in different humidity- one category(see Appendix) were not included. elevation categories.a HM = humid montane, HF = humid foothills, HL = humid lowlands, SL = Percent semihumid lowlands, AH = arid highlands, AL = Humidity-eleva- arthro- No. of No. of arid lowlands. tion category pods birds species HM HF HL SL AH AL Humid montane 86.3 401 26 Humid foothills 79.9 313 32 HM -- 5.24* 5.61' 1.81 1.28 1.89 Humid lowlands 78.4 176 17 HF -- 0.15 3.69 4.21' 4.95* Semihumid lowlands 95.8 24 8 HL -- 4.08* 4.75* 5.50* Arid highlands 92.0 50 11 SL -- 0.38 0.16 Arid lowlands 93.5 46 9 AH -- 0.08 AL --

a* =p < 0.05. significantlylower frequency in SouthAmerican than in CostaRican birds [pooled samplesof all hermits, 60% for South America (n = 231) vs. 78% for Costa ing arthropods.Species were assignedto one or more Rica (n = 138); X2 = 12.26, df = 1, P < 0.001]. We sus- of six humidity-elevation categories.The data on pect that the discrepancybeween South American nonhermitsare summarizedby humidity-elevation and Costa Rican samplesis due to a difference in categoryin Table 1. The null hypothesisthat the specimen-collectingtechniques, which is relevant in probabilityof a stomachcontaining arthropods would the caseof hermitsbecause they take a much higher be the same for birds in the six different habitats was proportion of soft-bodiedspiders than do most non- rejected(X 2 = 17.33,df = 5, P < 0.01).Results of pair- hermits (Hespenheideand Stilesunpubl. data). Non- wise comparisonsbetween birds in different cate- hermits take more hard-bodiedinsects that persist goriesare shown in Table 2. In general, the cooler longer in the gut. The time that netted humming- or drier an area, the more likely a bird was to have birds remainedentrapped prior to collectionwas on arthropodsin its stomach.Arthropods were mostfre- average 3-4 times longer for the South American quent in the stomachsof hummingbirdsof arid or birds.The time of day of collectionmay be another semihumid areas. Within humid areas, montane contributingfactor. Most hermitsare lek speciesthat hummingbirdsmore often had consumedarthropods maycommence lek activitybefore feeding, following than had lowland species(Table 1). These relation- a dawn flight between roost and lek (cf. Stiles and shipsobtained in CostaRican and in South American Wolf 1979).By opening nets at dawn, South Ameri- samples(except that no arid regions exist in Costa can workers might well have captured a number of Rica). Becausethe two samplesare large and were hermits that had not yet fed. By contrast,virtually gatheredindependently, we think the relationships all netting to collect hummingbirds in Costa Rica reflect differencesin hummingbird feeding patterns commenced at least I h after dawn, and it was more between environments. likely that the birdshad fed beforebeing collected. The differencesmay reflect variation between en- Because hermits and nonhermits take different vironmentsin availability and quality of nectarand kinds of arthropods,a comparisonof frequencyof arthropods.Dry tropicalforests support relatively few arthropod feeding is problematical.The bias result- species of hummingbird-pollinated flowers com- ing from morerapid digestionof arthropodsin her- paredwith wetter areas(Stiles 1981), and humming- mits' stomachs is minimized in the Costa Rican sam- birds in dry areasmay make relatively greater use of pie. Arthropodsoccurred less frequently in hermits' arthropodsto meet energeticneeds. Nevertheless, the than nonhermits' stomachs(78% for hermits; 85% for energetic importance of good nectar supplies is in- nonhermits,n = 312),but the differencewas signifi- dicated by the facts that dry-forest hummingbird cant only at the 0.10 level (X2 = 3.66, df = 1). The dif- communitiescontain relatively few speciesand show ferencewas highly significantin the SouthAmerican low morphologicaland behavioral diversity com- sample (60% for hermits; 83% for nonhermits, n = pared with wet-forestcommunities (Stiles 1981). It is 930; X2 = 57.01, df = 1, P < 0.0001), but, as discussed lessclear why the frequencyof arthropodconsump- above,we believe this to be largely an artifact of the tion should increase with elevation. There is evi- longer time SouthAmerican birds spent in nets be- dence that nectar from highland areas tends to be fore being collected. more dilute than that of lowland flowers (Hains- Perhapsthe mostinteresting pattern to emergefrom worth and Wolf 1972, Wolf et al. 1976). Arthropod the data on nonhermits(see Appendix) is a relation- consumptionmay help highland hummingbirdssur- shipbetween the elevationand humidityof the col- vive cold nights by providing a longer-lastingsource lectingareas and the proportionof stomachscontain- of energy than nectar. Data on time budgets of for- 438 ShortCommunications [Auk, Vol. 103 aging birds and seasonaland daily fluctuations in LITERATURE CITED resourceavailability in different areaswould be re- quired to testhypotheses suggested by differencesin BAKERßH. G., & I. BAKER. 1975. Studies of nectar- frequency of arthropod consumption. constitutionand pollinator-plantcoevolution. Pp. Although significant habitat-relatedvariation in 100-140 in Coevolution of and plants (L. feedingpatterns apparently exists, we emphasizeour E. Gilbert and P. H. Ravenß Eds.). Austin, Univ. general finding that almost all tropical humming- Texas Press. birds routinelyfeed on arthropods.The percentageof GASS, C. L., & R. D. MONTGOMERIE. 1981. Hum- stomachsand cropscontaining arthropodswas high mingbird foraging behavior: decision-making in all habitatsßranging from 78% to 96% among non- and energy regulation. Pp. 159-194 in Foraging hermits. Among hermits, which specializeon soft- behavior:ecological, ethological, and psycholog- bodied spiders that are digested rapidly, the value ical approaches(A. C. Kamil and T. D. Sargentß obtainedfrom the CostaRican sample(78%) repre- Eds.). New York, Garland STPM Press. sents our best estimate. The data indicate that most HAINSWORTH,F.R. 1974. Foodquality and foraging hummingbirds,at any given momentduring the day, efficiency:the efficiencyof sugarassimilation by are digesting arthropods.From this it may be in- hummingbirds.J. Comp.Physiol. 88: 425-431. ferred that most hummingbirds feed on arthropods , & L. L. WOLF. 1972. Crop volume, nectar on a daily basis, and probably at regular intervals concentration and hummingbird energetics. throughoutthe day. Comp. Biochem.Physiol. 42A: 359-366. LSUMZ fieldwork in the Neotropicshas been gen- ß & --. 1976. Nectar characteristics and erouslysupported by JohnMcIlhenny, BabetteOdom, food selectionby hummingbirds.Oecologia 25: H. Irving and Laura Schweppe,and E. Mudge. Field- 101-113. work in CostaRica hasbeen supportedby the Amer- MONTGOMERIE,g. D., & C. A. REDSELL. 1980. A nest- ican Museumof Natural History;the WesternFoun- ing hummingbirdfeeding solelyon arthropods. dation of Vertebrate Zoology; the Vicerrectorla de Condor 82: 463-464. Investigaci0n,Universidad de CostaRica; and CON- STXLES,F. G. 1981. Geographical aspectsof bird- ICIT. For permissionto work in Bolivia, we thank flower coevolution,with particularreference to the Direcci0nde Cienciay Tecnologla(through Gas- Central America. Ann. Missouri Bot. Gard. 69: ton Bejarano)and the Academia Nacional de Cien- 323-351. cias,La Paz;for permissionto work in Peru, we thank ß& L. L. WOLF. 1979. Ecologyand evolution the Direcci6nGeneral Forestaly de Fauna. The Or- of lek mating behavior in the Long-tailed Her- ganization for Tropical Studies has facilitated work mit hummingbird. Ornithol. Monogr. No. 27. in CostaRica on many occasions.We are grateful to WOLF, L. L., F. G. STILES,& F. R. HAINSWORTH. 1976. all those fieldworkerswho carefully examined the Ecologicalorganization of a tropical, highland hummingbirds.Mark B. Robbinsgraciously provided hummingbirdcommunity. J. Anim. Ecol.45: 349- data on stomachcontents of specimensat the Acad- 379. emy of Natural Sciences,Philadelphia. We thank T. Received14 May 1985, accepted18 November1985. J. Davis, T. Marr, T. A. ParkerßK. V. Rosenberg,A. Saxton, and L. L. Wolf for comments on the manu- script.

APPENDIX.Stomach and crop contentsof hummingbirds. The number of individuals in each of three cate- gories(AR = arthropodspresent; E = empty; 0 = contentsunidentifiable) is given, and for those species with n > 9, the proportionof individualswith arthropodsis given in parentheses.H-E refersto humidity- elevationcategories: HL = humid lowlands,under 1,000m elevation;HF = humid foothills, 1,000-2,500 m; HM = humid montane, 2,500-3,600 m; SL = semihumid, subtropicallowlands, under 1,500 m; AL = arid lowlands, under 2ß000m; AH = arid and semiarid highlandsßabove 2ß000m; X = does not fit other categories.For speciescollected in more than one H-E categoryßthe first categorylisted is the one in which most specimenswere taken.

Species AR E 0 H-E Glaucis aenea 25 (0.76) 6 2 HL G. hirsuta 15 (0.68) 7 0 HL Threnetes leucurus 12 (0.50) 12 0 HL T. ruckeri 17 (0.74) 5 1 HL Phaethornisyaruqui 0 2 2 HL P. guy 10 (0.48) 11 0 HF P. syrmatophorus 0 3 0 HF P. superciliosus 90 (0.75) 27 3 HL, HF April 1986] Short Communications 439

APPENDIX. Continued.

Species AR E 0 H-E P. hispidus 8 (0.53) 7 0 HL P. bourcieri 3 2 0 HL P. philippii 3 5 1 HL P. koepckeae 8 (0.40) 12 0 HF P. pretrei 0 0 3 SL P. subochraceus 4 1 1 SL P. ruber I 1 0 HL P. griseogularis 0 1 0 X P. longuemareus 12 (0.67) 6 0 HL Eutoxeresaquila 14 (0.61) 9 0 HF E. condamini 23 (0.77) 7 0 HF Androdonaequatorialis 1 0 0 I-IF Doryferajohannae 11 (0.73) 4 0 HF D. ludoviciae 16 (0.89) 2 1 HF Phaeochroa cuvierii 3 0 0 AL, HL Campylopteruslargipennis 21 (0.81) 5 0 HF C. hemileucurus 6 1 0 I-IF C. villaviscensio 16 (0.76) 5 0 HF Eupetomenamacroura 2 0 0 SL Florisugamellivora 20 (0.83) 4 0 HL Col•l•ridelphinae 8 0 0 I-IF C. thalassinus 6 0 0 HM C. coruscans 14 (0.82) 2 I AH, HM, HF Anthracothoraxprevostii 5 0 0 AL A. nigricollis 4 0 0 HL Chrysolampismosquitus 1 0 0 SL Klaisguimeti 1 I 0 HL Lophornisadorabilis 3 0 0 HL Popelairiapopelairii I 0 0 HL Discosura conversii 2 0 0 I-IF Chlorostilbon aureoventris 9 0 0 SL C. canivetii 5 I 0 AL C. stenura 2 0 0 AH Thaluraniafurcata 65 (0.76) 19 I HF, HL T. colombica 27 (0.73) 10 0 HL Panterpeinsignis 25 (0.93) 2 0 HM Hylochariseliciae 5 1 0 HL H. chrysura 2 1 0 SL Chrysuroniaoenone 13 (0.72) 5 0 HF Goethalsia bella 0 5 0 I-IF Polytmusguainumbi 5 0 0 SL Leucippusbaeri 3 0 0 AL L. chlorocercus 4 0 0 HL Taphrospilushypostictus 2 0 0 HF Amaziliachionogaster 13 (0.93) 1 0 AH A. viridicauda 3 3 0 AH, HF A. fimbriata 3 0 0 HL A. amabilis 5 3 0 HL A. rosenbergi 2 0 0 HL A. boucardi 4 1 0 X A. decora 10 (0.83) 2 0 HL A. saucerrottei 12 (0.86) 2 0 AL, AH A. tobaci I 0 0 SL A. edward 3 0 0 HF, HL A. rutila 4 1 0 AL A. tzacatl 14 (0.82) 3 0 HL A. amazilia 4 0 0 AL Eupherusaeximia I 2 0 I-IF E. nigriventris 4 3 0 I-IF Elvira chionura 4 1 0 I-IF E. cupreiceps 8 1 0 HF 44O Short Communications [Auk, Vol. 103

APPENDIX. Continued.

Species AR E 0 H-E Microchera albocoronata 5 0 0 HF, HL Chalyburabuffonii 1 3 0 X C. urochrysia 30 (0.81) 7 0 HL Lampornishemileucus 5 1 0 HF L. calolaema 14 (0.93) 1 0 HF L. castaneoventris 3 0 0 HM Adelomyiamelanogenys 17 (0.81) 4 0 HF Urostictebenjamini 1 0 0 HF Phlogophilusharterti 1 0 0 HF Polyplanctaaurescens 8 (0.53) 6 1 HL Heliodoxa rubinoides 6 0 0 HF H. leadbeateri 36 (0.86) 6 0 HF H. jacula 8 (0.73) 3 0 HF H. schreibersii 7 (0.70) 3 0 HL, HF H. gularis 3 I 0 HF H. branickii 2 1 0 HF Eugenesfulgens 6 0 0 HM Oreotrochilusmelanogaster 2 0 0 AH O. estella 7 0 0 AH O. adela 2 0 0 AH Patagonagigas 3 0 0 AH Aglaeactiscupripennis 2 1 0 AH A. aliciae 1 0 0 AH A. castelnaudii 17 (0.77) 5 0 AH, HM A. pamela 0 I 0 HM Lafresnayalafresnayi 6 2 0 HM Pterophanescyanopterus 19(0.90) 2 0 HM Coeligenacoeligena 18(0.82) 4 0 HF C. torquata 30 (0.88) 4 0 HM, HF C. bonapartei 1 0 0 HM C. lutetiae 6 0 0 HM C. violifer 86 (0.88) 12 1 HM Ensiferaensifera 9 (0.90) 1 0 HM Boissonneauamatthewsii 6 2 0 HM, HF Hellangelusspencei 2 I 0 HM H. amethysticollis 35 (0.90) 4 0 HM H. strophianus 2 0 0 HM H. regalis I I 0 HF Eriocnemis luciani 3 1 0 FIM E. mosquera 1 0 0 HM E. alinae 9 (0.75) 3 0 HF Haplophaediaaureliae 1 I 0 HF Ocreatus underwoodii 5 1 0 HF Lesbia nuna 1 0 0 AH Sapphosparganura 10 (0.83) 2 0 AH Ramphomicronmicrorhynchum 7 0 0 HM Metalluraphoebe 3 0 0 AH M. odomae 3 0 0 HM M. theresiae I 0 0 HM M. aeneocauda 21 (0.84) 4 0 HM M. eupogon 51 (0.77) 15 0 HM M. tyrianthina 40 (0.83) 8 0 HM Chalcostigmaruficeps 4 2 0 HM C. olivaceum 1 0 0 HM C. stanleyi 2 0 0 HM Aglaiocercuskingi 3 2 0 HF Schistesgeoffroyi 7 0 0 HF Heliothryxbarroti 6 0 0 HL, HF H. aurita I 0 0 HL Hellomasterconstantii 10 (0.91) I 0 AL H. longirostris 2 0 0 SL,HL April 1986] Short Communications 441

APPENDIX. Continued.

Species AR E 0 H-E H. furcifer 2 0 0 SL Thaumastura cora 4 0 0 AL Calliphloxmitchellii 0 1 0 HL C. bryantae 1 0 0 HF C. amesthystina 1 0 0 SL Myrtis fanny 4 0 0 AL Myrmia micrura 4 0 0 AL Acestrura mulsant 5 0 0 HM, AH, HF Selasphorusscintilla 4 0 0 HF, HM S. fiammula 6 0 0 HM