The Condor89~779-787 0 The Cooper OrnithologicalSociety 1987

ECOLOGICAL FITTING: USE OF FLORAL NECTAR IN STILESII DANIELS BY THREE OF

FRANK B. GILL The Academyof Natural Sciences,Philadelphia, PA 19103

Abstract. Three speciesof hermit hummingbirds-a specialist( aquikz), a gen- eralist (Phaethornissuperciliosus), and a thief (Threnetesruckerz]-visited the nectar-rich flowers of Heliconia stilesii Daniels at a lowland study site on the Osa Peninsula of Costa Rica. Unlike H. pogonanthaCufodontis, a related Caribbean lowland specieswith a less specialized flower, H. stilesii may not realize its full reproductive potential at this site, becauseit cannot retain the services of alternative pollinators such as .The flowers of H. stilesii appear adapted for pollination by Eutoxeres, but this rarely visited them at this site. Lek male Phaethornisvisited the flowers frequently in late May and early June, but then abandonedthis nectar sourcein favor of other flowers offering more accessiblenectar. The strong curvature of the perianth prevents accessby Phaethornis to the main nectar chamber; instead they obtain only small amounts of nectar that leaks anteriorly into the belly of the flower.

Key words: Hummingbird; pollination; ;foraging; Heliconia stilesii; nectar.

INTRODUCTION Ultimately affectedare the hummingbird’s choice Species that expand their distribution following of flowers and patterns of competition among speciationenter novel ecologicalassociations un- hummingbird species for nectar (Stiles 1975, related to previous evolutionary history and face 1978; Wolf et al. 1976; Feinsinger 1978; Gill the challenges of adjustment to new settings, 1978). Use of specificHeliconia flowers as sources called “ecological fitting” (Janzen 1985a). In the of nectar by particular species of hermit hum- case of mutualistic species, such as plants and mingbirds, however, varies seasonally and geo- their pollinators, new ecological settingsmay in- graphically (Stiles 1975). Comparative studies of clude new arrays of speciesvarying in ability to the foragingecology of hermit hummingbirds and function as partners. -pollinated plants in a the pollination biology of Heliconiaflowers could new setting, for example, will face new selection help us to understand the loosening and tight- pressureson the form of floral display, the ac- ening of mutualistic relationships in different cessibility of nectar, and the phenology of flow- ecological settings. ering, all of which affect ability to compete for In this paper I examine the use of nectar in the services of hummingbirds (Brown and Ko- flowers of Heliconia stilesii Daniels by three dric-Brown 1919, Kodric-Brown and Brown speciesof hermit hummingbirds at one locality 1979, Stiles 1980). in the Pacific lowlands of southern Costa Rica. Hermit hummingbirds (Trochilidae, Phae- Belonging to different genera, the three species thorninae) and Heliconiaflowers (Zingiberales: of hermit hummingbirds differ strikingly in bill Heliconiaceae) provide striking examples of spe- form: Phaethornissuperciliosus (Long-tailed cialized pollination mutualisms (Snow and Snow Hermit) has a long (38 to 39 mm) decurved bill; 1972, 1980; Stiles 1975, 1979; Feinsinger 1983; Threnetesruckeri (Band-tailed ) has a Dobkin 1984). Promoting the parallel evolution shorter (28 to 29 mm) nearly straight, sharp- of bills and flowers is the effect of the precise fit tipped bill; Eutoxeresaquilu (White-tipped Sick- between the two on the hummingbird’s rate of lebill) has a sharply bent, stout bill (photos in nectar extraction and the associatedprobability Stiles 1975). The differencesin bill form affected of pollen transfer (Wolf et al. 1972, Stiles 1980). their abilities to extract nectar from H. stilesii flowers, which were abundant next to a large lek ’ ’ Received24 October 1986. Final acceptance 13 of P. superciliosus,and thus an obvious potential May 1987. sourceof energy for their breeding efforts.Nectar

[7791 780 FRANK B. GILL

TABLE 1. Floral characteristicsof two speciesof Hel- iconia.

Character H. rmbricata ’ H. stilesii

Flower (perianth) A c Length (cm) 2.5-3.0 5.5 I I Curvature slight strong ,0”l?l Nectar Concentration f 1 SD (O/asucrose equivalents) 22.4 t 2.1 29.1 k 5.4 Energy content f 1 SD (J/pi) 3.6 f 0.3 4.9 * 0.9 Production’ (&hr) 19 18

I Ratesof nectarproduction declined during the day (seeStiles 1975; Gill, m press).The valuespresented here are averageworking estimates for early to midmorning. FIGURE 1. Mid-longitudinal section of Heliconia imbricata(upper) and H. stilesii(lower) flowers. A = into Panama (Daniels and Stiles 1979). It was main nectar chamber; B = belly of flower where nectar accumulatesafter overflowing from the main chamber; common at several forest edge localities at Si- C = site of bill insertion by Phaethornissuperciliosus rena. It was scarceelsewhere in Corcovado Na- and Eutoxeresaquila. tional Park. H. stilesiiflowers throughout the year at Sirena with a general peak of bloom during from the quite similar H. pognonantha Cufo- the rainy seasonstarting in late May. Each flower dontis fuels reproduction by P. superciliosusat lasts only half a day, wilting conspicuously by La Selva in the Caribbean lowlands on the op- early afternoon. The flowers are long and sharply posite side of Costa Rica (Stiles and Wolf 1979). bent, making accessto the main nectar chamber Contrary to expectation, however, the flowers of extremely difficult. A tight passagewayat the an- H. stilesii served only as a temporary resource terior end of the chamber compounds the chal- for P. superciliosusin our study area. lenge of nonlinear accessto the distant nectar chamber. MATERIALS AND METHODS H. imbricata is an abundant, widespread I conducted this study of color-marked hum- speciesin both the Caribbean and Pacific low- mingbirds (see Stiles and Wolf 1973 for proce- lands of Central America, and was one of the dures) in 1980 and 198 1 in Corcovado National dominant plants in the wet second growth hab- Park on the Osa Peninsula of Costa Rica. Allen itat at Sirena. The dark red, compact, vertical (1956)andHartshorne(l983: 132-136)describe inflorescence produces short, slightly curved the forests of this region. Our study site was lo- flowers, which allow direct accessto the nectar cated on the edge of rain forest near the park chamber by a straight bill or capillary tube. The headquartersat Sirena at the base of a small ridge anterior opening to the nectar chamber of H. next to the park headquarters and adjacent to a imbricata flowers allows easy passage. Both lek of P. superciliosus.Both Phaethornis and straight-billed hummingbirds, such as Thalura- Threneteswere common at this locality, but Eu- nia fircata and Amazilia decora, and hermit toxereswas rare, as reflectedin the relative abun- hummingbirds can reach the floral nectar cham- dance of captures during our study, namely 155 ber. Phaethornis, 76 Threnetes,and 7 Eutoxeres. The ease of measuring floral nectar contents Table 1 summarizes the floral and nectar char- facilitates study of energetic rewards available to acteristics of H. stilesii, the principal botanical hummingbirds. Nectar concentrations in H. sti- subjectof this study, and of H. imbricata (Kuntze) lesii flowers were measured in terms of percent Baker, the main alternative sourceof nectar used sucroseequivalents with a temperature-compen- by Phaethornis in our study area in May to July. sated hand refractometer and converted to H. stilesii is found on the Pacific side of Central J/flower based on grams of solute per 100 ml America up to 1,000 m elevation from Par&a, (Bolten et al. 1979). The nectar contained fiuc- Costa Rica south through the Golfo Dulce region tose, sucrose, glucose, and unidentified amino THREE HERMITS 781

0

0

. .

07,. I,, I I I. I I 20 24 29 1 5 9 13 17 21 25 29 3 7 11 15 19 23 27 31 4 9 MAY JUNE JULY AUG. FIGURE 2. Standingcrops of nectar energy in Heliconiastilesii (squares) and in H. imbricata(closed circles). Black squares= mean total nectar per H. stilesiiflower at 06:OO;white squares= mean total nectar per H. stilesii flower at 10:OO;black circles = mean total nectar per H. imbricataflower * 1 SD indicated for values after mid- July (deviations in May and June were similar but omitted to simplify figure).Nectar energyavailable as overflow in the bellv of H. stilesiiflowers is shown at the bottom left of the figure:horizontal bars = mean; vertical black bars = i 1 SD. acids (Gill, unpubl. data). Nectar volumes in studies.One lek male, color-marked Pink-White- plucked flowers were measured using 100 ~1 cap- Red (PWR), accounted for 83% of the visits to illary tubes, first from the belly of the flower “I” in 1981. anterior to the staminode and then separately To determine the patterns of nectar removal from the main chamber (Fig. 1). The presence from H. stilesiiflowers by Phaethornis,we bagged of insect larvae or ants in the nectar chamber bracts with new flowers at dawn and then re- was noted. Most flowers also contained floral moved the netting in midmorning to await hum- mites (see Colwell 1973, Dobkin 1984). In this mingbird visits. Baggedflowers remaining on an paper the term “standing crop” refers to nectar inflorescence served as controls. Flowers with present in flowers open to visitors of all kinds. beetle or fly larvae were excluded from the anal- Nectar production was estimated as the accu- ysis of nectar removal; suchflowers typically were mulation in flowers bagged with mesh cloth be- rejected by Phaethornis. Brief or aborted flower fore dawn. visits (less than 5% of total) also were discarded To establish the temporal patterns of flower to insure that the flower contained substantial visitation, we undertook continuous vigils at nectar and that the hummingbird fed without stands (= presumed clones) of H. stilesii. We interference from larvae inside the flower or from monitored all visits by hummingbirds to these Trigona bees. flowers from 07:OO to 12:00, and into the late afternoon on some days. On certain days in 198 1, RESULTS we numbered 80 to 90% of the bracts with fresh flowers, which enabled us to record the specific NECTAR AVAILABILITY flowers that a hummingbird visited. The cohort Large amounts of energy were present in H. sti- sizes of the monitored flowers were 13 1 on 26 lesii flowers. Standing crops averaged 200 to 300 May, 140 on 27 May, 13 1 on 28 May, 97 on 11 J/flower in the early morning and 300 to 500 June, 88 on 12 June. Data in this paper are based J/flower by midmorning (Fig. 2). Some flowers on 167 monitor hours from June 6 to July 8, contained over 842 J and one (unvisited) flower 1980 and 100 monitor hours from May 22 to contained 1,374 J by 14:O0. Variations among July 9, 198 1. One large stand (designated “I”) flowers in accumulated nectar reflected con- with >40 inflorescences was the focus of our sumption by insectsand as well as variable 782 FRANK B. GILL

a 0 20 40 60 60 100 120 140 160 160

TOTAL NECTAR VOLUME (&I) FIGURE 3. The amount of nectar that overflows from the main nectar chamber anteriorly to the belly of Heliconiastilesii flowers increaseswith total production. Sixty flowers with 22 to 142 ~1 of nectar in the rear chamber had no nectar anterior to the chamber (baseline stippling) in the belly of the flower. Open circles = unbaggedflowers; closed circles = baggedflowers. nectar production per se. Nectar volumes of less HUMMINGBIRD VISITS than 100 ~1 were restricted to the main nectar Visits to H. stilesii flowers increased during the chamber in most flowers. As production contin- morning. Phaethornisrarely visited theseflowers ued, however, nectar flowed forward into the bel- before 07:OOand only occasionally before 08:00, ly of some flowers, causing their anterior nectar but visited over 70 flowers per hour from 10:00 volumes to increase with total accumulation (Fig. to 12:OO(Table 2). Such increased feeding activ- 3). Flowers with more than 115 ~1 total had 10 ity in late morning correspondedto the increas- to 60 ~1available anteriorly. The average amount ing amount of nectar in the belly of the flower. of nectar available anteriorly at 10:00 most Despite continued availability of nectar in H. mornings in 1981 was less than 6 ~1 per flower. stilesii flowers, use of the nectar by Phaethornis H. stilesii flowers offered more nectar (total) declined dramatically in late June. Phaethornis than did H. imbricata flowers, which (in late May) visited H. stilesii standsup to four times an hour contained only 42 to 80 J/flower at 10:00 and in late May and early June in 1980 and up to six accumulated up to 4 17 J/flower by midday. As times an hour in 198 1. After mid-June Phae- H. imbricata Aower abundance increased, mid- thornis visited the stands less than three times morning nectar volumes also increased to stable per hour, usually less than once per hour (Fig. levels of 150 to 200 J/flower for the rest of the 4). Visits to stand “I” virtually ceasedafter mid- summer. Although much less than the total nec- June 198 1, even though though there was no tar energy content of H. stilesiiflowers, the nectar obvious decline through July in the number of energy content of H. imbricata flowers was about new flowers available each day. Decreased use eight times that available anteriorly in the belly in terms ofthe number of flowersvisited per hour in H. stilesii flowers. after mid-June was especially pronounced (Fig. 5). By late June, we observedPhaethornis feeding primarily on increasingly abundant H. imbricata TABLE 2. Flower visitation by a lek male (PWR) flowers. Phaethornissuperciliosus during the morning and as The proportion of available stilesiiflowers that nectar volumes increasedin the accessiblebelly of Hel- iconiastilesii flowers. Values are mean number of flow- were visited once by 12:00 declined with the ers visited per hour (+ 1 SD) for three successivedays changein Phaethornis ’ feeding preferences,from in late May 1981. 80% in late May to 12% in mid-June 198 1 (Fig. 6). Fifty-five to 60% of the flowers received two Early morning MidmornIng Late morning (07:0&08:00) (08:0&10:00) (lo:Oc-12:OO) or more visits by midday in mid-May and 20% received four or more visits. No flowers were 16.6 ?z 6.1 56.2 +- 13.7 74.2 f 9.6 visited twice by midday on 12 June. THREE HERMITS 783

3.0 . 0 0 0 22 26 30 3 7 1, 15 19 23 27 1 5 9

. l o . MAY JUNE JULY 1 0 2.0. 0 FIGURE 5. The number of marked Heliconia stilesii o CQ flowers hummingbirds visited each hour declined coo o 0 sharply in June 1981. Closed circles = visits by a lek 0 0 0 male (color-marked PWR) Phaethornissuperciliosus; 1.0. 0 . open circles = visits by other P. superciliosus;X = %ib 0 0 visits by Threnetesruckeri. 0 P e Go0 0 0 l 0 .~.,.~I...,~.~~...~,~11~~ flowers, probe deeply into the flower to extract 22 26 30 3 7 11 15 19 23 27 1 5 9 nectar, and, while doing so, brush their foreheads MAY JUNE JULY against the anthers and stigma, conspicuously FIGURE 4. Phaethornissuperciiiosus visited stands of Heliconia stilesiiless frequently during the morning coating their foreheads with whitish H. stilesii (before 12:OO)as the rainy seasonprogressed and other pollen. These hummingbirds typically remove sourcesof nectarbecame available. Open circles= 1980, most of the nectar from the belly of the flower, r = ~0.40; Closed circles = 1981, r = -0.47. but little (8%) from the main nectar chamber. The averageresidual nectar volume in the bellies Threnetes regularly visited H. stilesii flowers, of visited flowers was only 1.1 ~1 compared to especially soft wilted flowers in the afternoon. 12.7 ~1in controls (P < .Ol; Wilcoxon’s Matched They accounted for 18 to 20% of all flower visits Pairs Signed-Rank Test, n = 12). The average before 12:00 on May 26 to 28, 1981, and for 54 to 65% of all flower visits in the afternoon on residual volume (73.9 ~1 + 17.9 SD) in the main chamber of visited flowers was only slightly less these same days. Visit rates by Threnetes were than that (80.2 ~1 ? 18.7 SD) in unvisited con- below those of Phaethornis (Fig. 5). trols (0.01 < P < 0.025; Wilcoxon’s Matched Eutoxeres rarely visited the abundant H. sti- Pairs Signed-Rank Test, one-tailed, n = 23). Pos- lesii flowersin May to July. We recordedno visits during our regular monitoring, but did observe sibly some of the nectar in the main chamber is drawn forward as the hummingbird removes the two visits at other times. We noted this hum- mingbird at H. stilesii flowers primarily during overflow. the dry season in February to March, when we Unlike Phaethornis,both Eutoxeres and Thre- netesextract nectar from the main nectar cham- recorded a total of 43 site visits and 119 flower visits during 25 hr of morning hour monitors bers of the H. stilesii flowers they visit, but in (06:OO to 11:OO)at two different stands of H. different ways. Eutoxeres probes the flower di- Phaethornis, but does so while stilesii. Rates of site visitation varied from 0.8 rectly as does clinging to the bract instead of hovering in front to 3.2 times per hour and rates of flower visita- tion varied from 2.2 to 7.6 flowers per hour. One of the flower. Prolonged contact with the anthers of the Eutoxeres responsible for these data was causespollen to be deposited in the groove on territorial at a large stand of H. stilesii. their foreheads.The nectar chambers of the four flowers we checked immediately after sicklebill NECTAR REMOVAL feeding visits were completely empty. Threnetes Feeding Phaethornis insert their bill through the doesnot usually probe the flowersdirectly through anterior opening in the perianth of H. stilesii the anterior opening in the perianth, but instead 184 FRANK B. GILL

FORAGING EFFICIENCY Differences among the three hermits in average times per H. stilesii flower visit and in their for- aging costs determine each species’ net energy gain (Table 3). Extraction time is the total time that a hummingbird takes to insert its bill into a flower, to extract nectar, and then to remove its bill from the flower (Wolf 1975). Phaethornis fed more quickly at H. stilesii flowers than did Eutoxeres or Threnetes. The rapid flower visits by Phaethornis are consistent with the uptake of small leakage nectar volumes only compared to FIGURE 6. Cumulative probabilityof a Heliconiu the entire contents of the nectar chamber and, stilesiiflower being visited once by Phaethornissuper- for Threnetes only, the time required to pierce ciliosus duringthe morning in lateMay and June198 1. A switchby thesehummingbirds to other flowersin the flower tissue. mid-Junecaused the majorityof the flowersto remain To determine if Phaethornisachieved the same unvisited. net gain as the other hermits by rapidly visiting more flowers each with less nectar, we recorded average times per flower visit achieved during piercesthe lower side of the perianth and drinks 72 foraging bouts of variable lengths. The times directly from the main nectar chamber without included extraction time at a flower plus time making contact with a flower’s reproductive spent moving between flowers. In these foraging structures. No or little nectar remained in the bouts, which were restricted to flowers within rear chamber of 11 of 13 visited flowers. Sub- one large stand of H. stilesii, Phaethornis aver- stantial nectar volumes (69 and 83 ~1) were left aged 4.38 -t 1.53 (SD) set per flower visited. in two flowers, but these were less than in the Thus, Phaethornis potentially visits four flowers paired control (109 and 89 ~1). In contrast to six in the same amount of time (17.3 set) Eutoxeres control flowers, only one visited flower contained takes to visit one flower, obtaining a net energy some nectar anterior to the main nectar chamber. gain of 72 J compared to 345 J for Eutoxeres. I conclude that pirating Threnetesextract much of the nectar available in H. stilesii flowers. In DISCUSSION summary, accessto the main nectar chamber of The roles of the three hermit hummingbirds that H. stilesii flowers potentially yields much more visit the flowers of H. stilesii are not unique to nectar to Eutoxeres and Threnetes than Phae- this assemblageof species.Combinations of spe- thornis can get from the belly of the flower. cialists, generalists,and thieves are characteristic

TABLE 3. Foragingefficiencies of threehermit hummingbirdsat Heliconiastilesii.

Long-tailed Band-tailed White-tipped Hermit Barbthroat Sicklebill (Phaethornis superciliosis) (Threwem mckrz) (Eutoxeres aquila) Weight(g) 6.0 6.5 10.0 Metabolism(J/sec)l 1.O (hover) 1.1 (hover) 0.3 (sit) Foragingcosts2 Extractiontime 2.0 f 1.0 6.5 f 4.2 17.3 + 11.2 (n) (85) (33) (I I) Energy(J/flower) 2.1 7.2 5.2 Net gain* Anterior at 20 J/flower 17 13 15 Total at 350 J/flower - 3432 345 ’ ’ Assumeshovering costs = 0.175 J/g.secand sitting at 304: = 0.03 J/g.sec(DeBenedictis et al. 1978, Gill 1985). 2 Assumesremoval of all nectar;does not accountfor residualvolumes. THREE HERMITS 185 of several well-studied hummingbird assem- Phaethornis (Stiles and Wolf 1979; Gill, pers. blages(Colwell et al. 1974, Feinsinger 1976, Wolf observ.). et al. 1976, Feinsinger and Colwell 1978). The Diffuse mutualisms characterize pollination three montane speciesof sunbirds that often feed systems(Janzen 1980, 1983, 1985a; Feinsinger together at Leonotis mint flowers in East Africa 1983; Schemske 1983a). Most pollinators visit (Gill and Wolf 1978) provide a striking analogue, a variety of flowerson a nonexclusive basis; most in which differences in the fit between bill and flowers attract a variety of visitors that vary flower affect rates of nectar uptake and patterns greatly in their effectiveness as pollinators of flower use. In this case, the largest species, (Schemske 1983b, Jennersten 1984, Schemske Nectarinia reichenowi has a strongly decurved and Horvitz 1984). Becausethe distributions and bill and pollen-carrying groove similar to those abundances of partners independently reflect of Eutoxeres.Like Eutoxeresat certain Heliconia varied historical and ecological factors, the par- flowers, N. reichenowiis a specialized mutualist ticipants in pollination mutualisms also vary of the Leonotis flowers. Resembling Phaethornis radically in time and space. New associations at H. stilesii flowers, the sunbird N. farnosa has form and old bonds break, preventing the de- a long straight bill that fails to negotiate the flow- velopment of stable equilibria (Janzen 1985a, er curvature and to probe accurately into the 1985b). basal nectar chamber. Like Threnetes, the sun- The pollination biology of H. stilesii Daniels bird N. venustais a thief that piercesfloral tissue at Sirena in 198 1 doubtfully resembles the con- directly into the nectar chamber and performs ditions under which this plant originally evolved. little or no pollination service. The use of Leono- My study area was in abandoned pasture at the tis flowers by N. famosa and N. venustadepends edge of rain forest. Much of the surrounding re- on whether N. reichenowiis the primary visitor gion had been clearcut for livestock or banana of the flowers and on the local availability of crops, favoring an abundance of second-growth other flowers yielding higher foraging efficiency. plants, such as H. imbricata, and an abundance Hummingbirds are sensitive to subtle varia- of opportunistic hummingbirds and insects, in- tions in net energetic rewards, and quickly shift cluding Trigona bees. The attraction of such op- to the best available sourcesof nectar (Wolf et portunists and robbers to unharvested nectar can al. 1972, Feinsinger 1978, Wolf and Hainsworth thwart visits by pollinators and depressseed set 1983, Montgomerie 1984, Montgomerie et al. (Lyon and Chadek 197 1, Carpenter 1979, 1984). Even though H. stilesii produces flowers McDade and Kinsman 1980, Willmer and Cor- rich in nectar throughout the year and primarily bet 1981, Gill et al. 1982, Roubik 1982, Snow during the 6-month rainy season,P. superciliosus 1982, Inouye 1983). Studies of seed set by H. only uses them for a several week period and stilesii at Sirena in relation to pollinator visita- then shifts to other, higher-yield nectar sources. tion at different times of the year would be of This abbreviated seasonalinterest reflectsPhae- interest. Unless autogamy or self pollination by thornis ’ inability to remove the majority of nec- floral mites (Dobkin 1984) prevails, we expect tar available; the length and severe curvature of that the ratio of fruits/flowers declines markedly the perianth of H. stilesii flowers prevented ac- during the rainy seasonat Sirena. Even if autog- cessto the main nectar chamber. The abandon- amous fruits are formed, the effectsof inbreeding ment of H. stilesii flowers contrasts with this depression on seed quality may be profound hummingbird’s sustained 6 to 7 month relation- (Schemske 1983b). Also, the probability of pol- ship with H. pogonantha in the lowlands of lination and seed set will depend on the exact northeastern Costa Rica (Stiles 1975,1979; Stiles times of hummingbird visits during the morning and Wolf 1979). The difference lies in the fact in relation to temporal patterns of stigma recep- that Phaethornis extracts the full nectar reward tivity and anthesis,details of which are not known from the basal chamber ofH. pogonantha flowers for this species. and thus is not restricted to anterior overflow in Many speciesof Heliconia have evolved floral these flowers (Stiles, pers. comm.). The accessi- attributes that exclude casual foragersand attract ble nectar also attracts a large, territorial hum- regular visits by traplining hermit hummingbirds mingbird, Chalybura urochrysia,which defends (Snow and Snow 1972, Stiles 1979, Feinsinger some clumps of H. pogonanthaagainst traplining et al. 1982). The speciesof Heliconia with pen- 786 FRANK B. GILL dant inflorescencesin the H. pogonantha assem- straightening of the corolla to allow easier access blage, including H. stilesii, have coevolved with to the nectar chamber i.e., convergence towards sicklebills to varying degrees (Stiles 1979, also floral morphology of H. pogonantha. Kress 1982 for taxonomic issues).H. vellerigera Poeppig and H. ramonensis Daniels and Stiles, ACKNoWLEDGMENTS which have long (45 to 55 mm) perianths bent A. Mack, P. Marcotullio, R. Ray, C. Thompson, and 90” in the middle, are the most specialized for J. Wunderle assistedin thesestudies. I am grateful also Eutoxeres visitation. The sharply bent flowers of to J. H. Brown. R. K. Colwell, D. H. Janzen, D. W. Schemske,F. G. Stiles and L. L. Wolf for their com- H. stilesii Daniels and H. curtispatha Petersen ments on this manuscript and for making their un- flowers are intermediate in length, but still seem published manuscriptsavailable to me. The National geared primarily for Eutoxeres. H. pogonantha, Park Service of Costa Rica and especially the staff of on the other hand, has a short (40 cm), moder- CorcovadoNational Park made our field work pleasant and productive.This studywas supported by the Na- ately curved, lessspecialized perianth. The vary- tional Science Foundation (DEB 7906034, DEB ing degreesof specialization correspond to the 8019731). distribution of Eutoxeres in Central America. Eutoxeres is most common at moderate eleva- LITERATURE CITED tiOnS in the wet foothills, but ranges SeaSOnallY ALLEN,P. H. 1956. The rainforests of Golfo Dulce. into the lowlands (Slud 1964, Ridgely 1976). The Univ. Florida Press, Gainesville, FL. Reissued severely curved H. vellerigera,which is restricted 1977. StanfordUniv. Press,Stanford, CA. BOLTEN,A. B., P. FEINSINGER,H. G. BAKER,AND I. to a narrow band of 800 to 1,000 m on the Ca- BAKER. 1979. On the calculation of sugar con- ribbean slopesof the Cordillera Central, occupies centrationin flowernectar. Oecologia (Berl.) 41: prime sicklebill habitat. Likewise, H. ramonensis 301-304. &so occupiesprime sicklebill haditat up to 1,200 BROWN,J. H., AND A. KODRIC-BROWN.1979. Con- m on the southwesternslopes of Costa Rica. The vergence,competition, and mimicry in a temper- ate communityof hummingbird-pollinatedflow- less-curved H. pogonantha is restricted to the ers. 60:1022-1035. Caribbean lowlands below 350 m and thus below CARPENTER.F. L. 1979. Competition between hum- the primary distribution of Eutoxeres, and it also mingbirds and insects for nectar. Am. Zool. 19: ranges north to Honduras, where sicklebills do 1105-1114. COLWELL,R. K. 1973. Competition and coexistence not occur. The slightly curved flowers of H. po- in a simple tropical . Am. Nat. 107: gonantha attract generalized hermit humming- 737-760. birds, such as P. superciliosus,throughout the COLWELL,R. K., B. J. BETTS,P. BUNNELL,F. L. CAR- flowering season.Eutoxeres visits the Caribbean PENTER,AND P. FEINSINGER.1974. Competition lowlands at Finca La Selva only irregularly, ap- for the nectar of Centropogonvalerii by the hum- mingbird Colibri thalassinusand the flower-pierc- parently descending there to feed at H. pogon- er Diglossaplumbea, and its evolutionary impli- antha and other flowers during periods of food cations. Condor 76~447-452. shortage at higher elevations on the Caribbean DANIELS,G. S., AND F. G. STILES. 1979. The Heli- slopes. conia taxa of Costa Rica. Keys and descriptions. Brenesia 15 (Suppl.):l-l 50. E. aquila, one of the pollinators with which DEBENEDICTIS,P., F. B. GILL, F. R. HAINSWORTH,G. H. stilesii probably coevolved, rarely visits Si- H. PYKE,AND L. L. WOLF. 1978. Optimal meal rena, a coastal locality. Stiles and Wolf (unpubl. size in hummingbirds. Am. Nat. 112:30l-3 16. data) observed that when the primary mutualist DOBKIN,D. S. 1984. Flowering patternsof long-lived does not visit the flower and harvest the nectar, Heliconia inflorescences:implications for visiting and residentnectarivores. Oecologia (Berl.) 64:245- overflow of nectar from the chambers of the spe- 254. cialized species of Heliconia with long flowers FEINSINGER,P. 1976. Organization of a tropical guild provides a “fail-safe” mechanism for the attrac- ofnectarivorousbirds. Ecol. Monogr. 461257-291. tion of generalized pollinators, such as P. super- FEINSINGER,P. 1978. Ecologicalinteractions between plants and hummingbir& in a successionaltrop- ciliosus.A survey of geographicvariation in H. ical communitv. Ecol. Monoar. 48:269-287. stilesiifloral morphology in relation to Eutoxeres FEINSINGER,P. 1683. CoevolutTonand pollination, pollination service would be most interesting. H. p. 282-3 10. In D. J. Futuyma and M. Slatkin [eds.], stilesii at Sirena is at the fringes of its range and . Sinauer Assoc., Sunderland, MA. partnership with Eutoxeres. Here, perhaps, it FEINSINGER,P., AND R. K. COLWELL. 1978. Com- munity organization among neotropical nectar- faces new selection pressuresto bring its mutu- feeding bir&. Am. Zool. 18!779-795. alistic relationships into better accord with the FEINSINGER,P., J. A. WOLFE,AND L. A. SWARM. 1982. Long-tailed Hermit, including shortening and Island ecology:reduced hummingbird diversity and THREE HERMITS 181

the pollination biology of plants, Trinidad and SCHEMSKE,D. W. 1983a. Limits to specializationand Tobo. West Indies. Ecoloav 63:494-506. coevolution in plant- mutualisms, p. 67- GILL, F. B. 1978. Proximate-costsof competition in 109. In M. H. Nitecki [ed.], Coevolution. Univ. nectar-feedingbirds. Am. Zool. 18:753-763. of Chicago Press, Chicago. GILL, F. B. 1985. Hummingbird flight speeds.Auk SCHEMSKE,D. W. 1983b. Breeding system and hab- 102:97-101. itat effects on components in three neo- GILL, F. B. In press. Effect of nectar removal on rates tropical (Zingiberaceae). Evolution 37:523- of nectar accumulation in Heliconia imbricata 540. flowers. Biotropica. SCHEMSKE,D. W., AND C. C. HORVITZ. 1984. Vari- GILL, F. B., AND L. L. WOLF. 1978. Comparative ation among floral visitors in pollination ability: foraging efficiency of montane sunbirds. Condor a precondition for mutualism specialization. Sci- 80:391-400. ence 225:519-521. GILL, F. B., A. L. MACK, ANDR. T. RAY. 1982. Com- SLUD,P. 1964. The birds of Costa Rica: distribution petition between hermit hummingbirds Phae- andecology.Bull. Am. Mus. Nat. Hist. 128:l-430. thominae and insectsfor nectar in a Costa Rican SNOW,A. 1982. Pollination intensity and potential rain forest. Ibis 124~44-49. seedset in Pass&&a vitifofiu.Oecologia (Berl.) 55: HARTSHORNE,G. S. 1983. Plants. Introduction, p. 231-237. 118-157. In D. H. Janzen [ed.], Costa Rican nat- SNOW,B. K., ANDD. W. SNOW. 1972. Feeding niches ural history. Univ. of Chicago Press, Chicago. of hummingbirds in a Trinidad valley. J. Anim. Ecol. 41:471-485. INOWE, D. W. 1983. The ecologyof nectar robbing, SNOW,D. W., AND B. K. SNOW. 1980. Relationships p. 153-173. In B. Bentley and T. Elias [eds.], Bi- between hummingbirds and flowers in the Andes ology of nectaries. Columbia Univ. Press, New ofColombia. Bull. Br. Mus. (Nat. Hist.) Zool. 38(2): York. ~ I 105-139. JANZEN,D. H. 1980. When is it coevolution? Evo- STILES,F. G. 1975. Ecology,flowering phenology, and lution 34:6 1l-6 12. hummingbird aollination of some Costa Rican JANZEN,D. H. 1983. Seed and pollen dispersal by Heliconia species.Ecology 56:285-30 1. : convergencein the ecology of contami- STILES,F. G. 1978. Temporal organization of flow- nation and sloppy harvest. Biol. J. Linn. Sot. 20: ering among the hummingbird foodplants of a 103-l 13 tropical wet forest. Biotropica 10:194-2 10. JANZEN,D. H. 1985a. On ecologicalfitting. Oikos 45: STILES,F. G. 1979. Notes on the natural history of 308-310. Heliconia (Musaceae)in Costa Rica. Brenesia 15 JANZEN,D. H. 1985b. The natural history of mu- (Suppl.):15 l-180. tualisms, p. 40-99. In D. H. Boucher [kd.], The STILES,F. G. 1980. Ecological and evolutionary as- bioloav of mutualism. Croom Helm. London. pectsofbird-flower coadaptations.Proc. XVII Int. JENNERST&,0. 1984. Flower visitation and polli- Congr. Omithol. (1978):1173-l 178. nation efficiency of some North European butter- STILES,F. G., ANDL. L. WOLF. 1973. Techniquesfor flies. Oecologia 63:80-89. color-marking hummingbirds. Condor 75:244- KODRIC-BROWN,A., AND J. H. BROWN. 1979. Com- 245. petition between distantly related taxa in the co- STILES,F. G., AND L. L. WOLF. 1979. Ecology and evolution of plants and pollinators. Am. Zool. 19: evolution of lek mating behavior in the Lona-tailed 1115-1127. Hermit Hummingbird. Ornithol. Monogr. No. 27, K~ESS,W. J. 1982. New Central American taxa of American OrnithologistsUnion,’ Washington,DC. Heliconia. Evolution 37~735-744. WILLMER,P. G., AND S. A. CORBET.1981. Temporal LYON,D. G., AND C. CHADEK. 1971. Exploitation of and microclimatic partitioning of the floral re- nectar resourcesby hummingbirds, bees (Bom- sourcesof Justicia aurea amongst a concourseof bus),and Diglossabaritula and its role in the evo- pollen vectorsand nectarrobbers. Oecologia (Berl.) lution of Penstemonkunthii. Condor 73:246-248. 5 1:67-78. WOLF, L. L. 1975. Energy intake and expenditures MCDADE,L. A., AND S. KINSMAN. 1980. The impact in a nectar feeding sunbird. Ecology 56:92-104. of floral parasitism in two neotropical humming- WOLF, L. L., AND F. R. HAINSWORTH. 1983. Eco- bird-pollinated plant species.Evolution 34:944- nomics of foragingstrategies in sunbirdsand hum- 958. mingbirds, p. 223-264. In E. P. Aspey and S. L. MONTGOMERIE,R. D. 1984. Nectar extraction by Lustick [eds.], Behavioural energetics:the cost of hummingbirds: response to different floral char- survival in vertebrates. Ohio State Univ. Press, acters. Oecologia (Berl.) 63:229-236. Columbus. MONTGOMERIE,R. D., J. M. EADIE,AND L. D. HARDER. WOLF, L. L., F. R. HAINSWORTH,AND F. G. STILES. 1984. What do foraginghummingbirds maximize? 1972. Energeticsof foraging: rate and efficiency Oecologia (Berl.) 63:357-363. of nectarextraction by hummingbirds.Science 176: RIDGELY,R. S. 1976. A guide to the birds of Panama. 1351-1352. Princeton Univ. Press, Princeton, NJ. WOLF, L. L., F. G. STILES,AND F. R. HAINSWORTH. ROUBIK,D. W. 1982. The ecologicalimpact ofnectar- 1976. Ecologicalorganization of a tropical high- robbing bees and pollinating hummingbirds on a land hummingbird community. J. Anim. Ecol. 45: tropical shrub. Ecology 63:354-360. 349-379.