Condor 85:36&364 0 The Cooper Omlthological Society 1983 LONG-TAILED HERMIT HUMMINGBIRD VISITS TO INFLORESCENCE COLOR MORPHS OF HELICONIA IRRASA LUCINDA A. McDADE ABSTRACT. -Visits by Long-tailed Hermit hummingbirds (Phaethornis super- ciliosus)to flowers of different inflorescencecolor morphs in a natural population of Heliconia irrasa were monitored indirectly using fluorescent powder. The two color morphs (red and yellow) were indistinguishablewith respect to amount and rate of reward (nectar) offered to pollinators. The birds did not appear to prefer flowers of either bract color morph. Fluorescent powder was dispersedto flowers of the two color morphs in the frequencies predicted by the relative abundance of the morphs in the study area. This result adds support from field data to earlier exuerimental work that has challenged the notion that hummingbirds innately prefer red flowers. Despite the observed associationbetween red- visits by Long-tailed Hermit hummingbirds colored flowers and hummingbird pollination, (Phaethornis superciliosus)to flowers of these carefully controlled experiments have shown two morphs. that hummingbirds have no innate color pref- erences(Bene 1945, Collias and Collias 1968, DESCRIPTION OF THE PLANT Stiles 1976, Goldsmith and Goldsmith 1979). Heliconia irrasa (Heliconiaceae) is a broad- Hummingbirds are, however, able to distin- leaved monocot herb found in the understory guish colors and they readily learn to favor of wet forests in Costa Rica and Panama. The feeders of the color containing the most desir- plants are rhizomatous, but individuals at my able artificial nectar (Stiles 1976, Goldsmith study site in Panama rarely have more than and Goldsmith 1979). These experimental re- two stems and do not form large or dense sults have clear implications for the foraging clumps. During the early wet season(May to behavior of hummingbirds in nature. Because June), individual shootsproduce solitary erect differencesin flower color among plant species inflorescenceswith three to nine large cincin- are likely to be correlated with floral morpho- nal bracts. Two distinct colors of bracts and logical differences affecting the efficiency of rachises are found among plants at this site nectar extraction by hummingbirds (e.g., Stiles (and elsewhere, cf. Daniels and Stiles 1979): 1975) pollinator preference for flowers of one pure red (“red-bracted”), and orange-yellow speciesover others cannot be causally related with only the revolute margin of the bract red to flower color differences alone. The floral (“yellow-bracted”). Although yellow bracts color polymorphisms that are known in many tend to become dull and darken with age, the species(reviewed by Kay 1978) provide ideal morphs were alwaysreadily distinguishableand material for investigation of the role of color I found no intermediates. Inflorescencespro- in pollinator visitation. In some plant species, duced from a common rhizome are uniform the color of flowers or inflorescenceschanges in color. Although I have no conclusive data, over time. George (1980) and Schemske(1980) bract color thus appears to be genetically de- showed that hummingbirds use morphologi- termined. Stiles’s observations (pers. comm.) cal and color changesto discriminate between of transplanted rhizomes of this speciessup- first-day, nectar-producing flowers and older, port this notion. non-rewarding flowers. Interindividual, pre- Each bract contains 8 to 2 1 flower buds that sumably genetic, floral color polymorphisms open one at a time over a period of up to three are also known in some hummingbird-polli- months. Floral buds are completely enveloped nated plant species.Results from the experi- by the cincinnal bracts and even at anthesis mental work cited above predict that hum- only the distal 1 cm of the approximately 6- mingbirds will not discriminate between flower cm long flower is exposed (Fig. 1). As in most color morphs unless the morphs differ in the speciesof this tropical genus,the colorful bracts amount or rate of nectar production. To test are thus the most conspicuousportion of the this prediction, I studied flowering phenology flowering plant. Perianths are solid yellow re- and nectar production in flowers of two color gardlessof bract color. Anthesis occurs at or morphs of Heliconia irrasa, and quantified shcrtly after dawn and flowers last a singleday, 1360, HUMMINGBIRD COLOR PREFERENCE 361 becoming discolored by late afternoon (17:OO). Owing to the subtendingbract that largely en- velops the flowers, perianths wither and de- compose in place without falling from the plants. My observations indicated that Heli- conia flowers at this site were visited by nectar- seeking Long-tailed Hermit hummingbirds. Other long-billed hermit species(Glaucis hir- suta, Threnetesruckeri) may occasionally visit FIGURE 1. Single cincinnal bract (a), with open flower these flowers, but are rare in this area (Ridgely (b), from an inflorescenceof Heliconia zrrasu. 1976) and I did not see them. METHODS marked flower as frequently as other open This study was conducted in Parque National flowers in the vicinity. I removed all open Soberania, approximately 10 km NNW of flowers on all Heliconia stems at the site be- Gamboa, Colon Province, Panama. tween 13:00 and 15:00 on each of the exper- All flowering stems of H. irrasa were num- imental days. The one or more flowers from bered, marked and mapped along a 200-m for- each stem were placed in separate, appropri- ested stretch of “Pipeline Road,” including an ately numbered bags. Each flower was then abandoned spur (Fig. 2). In addition to deter- inspected in the laboratory under ultraviolet mining the proportion of flowering stems of light with a 10X lens and scored as being with each color at the site, I made morphological or without fluorescentpowder. No attempt was and phenological measurementsto test for dif- made to quantify either the amount of powder ferencesbetween the morphs. Size of plants (to placed on the source flower or the amount of base of inflorescence), number of cincinnal powder on recipient flowers. Becauselittle is bracts, and number of flowers produced daily known about the dynamics of powder pick-up, were counted for all marked shoots. The total carry-over or deposition, it is not clear that number of flowers per bract was counted for any monotonic relationship between powder the basal and distal bracts of a subsample of quantity and number of visits can be expected. the study plants of each morph. Cumulative I thus interpreted the presence of powder on nectar quantity and sugarconcentration (sugar a flower as conservatively as possible: any weight/total weight) in flowers baggedprior to flower that had powder must have received at anthesiswere measured every 2 h from 07:OO least one hummingbird visit during the ex- to 17:O0. At least 10 flowers from different perimental period. individuals of each color morph were used for At the end of the study, the limits of the area each sample hour, and I sampled over a two- to be usedin data analysiswere setat the plants week interval. Those aspectsof data collection farthest from the sourcesthat received powder that might affect hummingbird foraging (i.e., during the study. Thus, although stemsbeyond baggingplants, destruction of bracts to count 75, 47 and 5 1 at all extremes of the “transect” flowers) were conducted after the experimental were originally included in the study, flowers portion (see below) of the study was complet- from thesestems were excluded from the anal- ed. ysis (Fig. 2). This adjustment should allow Three closely spaced red-bracted plants lo- hummingbird foraging patterns and the dy- cated near the mid-point of the site were cho- namics of transport and deposition of fluores- sen for use as the source for marking with flu- cent powder to set the relevant patch size for orescentpowder. Between 06:30 and 07:OOon this experiment. Within this area, if hum- eight days spread through the peak flowering mingbirds did not discriminate between season (July and August) in 198 1, an open morphs, then flowers could be expected to re- flower on one of the three plants was marked. ceive powder in proportion to the relative The marked flower was on plant 26 except abundance of the color morphs. Flowers be- when this plant produced no flowers on a study yond this area did not receive powder but I day. In this case, a flower on plant 55 or 56 have no evidence that powder can ever be was marked (see Fig. 2). The upper sepal of transported so far from the sourceplant in this the selected flower was opened gently and or- system and thus cannot relate the observed ange fluorescent powder (color code A- 14-N, absenceof fluorescent powder to lack of pol- Day-G10 Color Corp.) was dustedliberally over linator visitation. the anthers using a pipet. Foraging humming- Expected frequencies of flowers of each birds did not appear to discriminate between morph receiving powder were calculated from marked and unmarked flowers. My observa- the overall proportion of flowers of each morph tions indicated that hummingbirds visited the collected from within the area of observed 362 LUCINDA A. McDADE 5m A Yellow 0 Red 0 Source Red 1N FIGURE 2. Map of study site. Locationsof all flowering stems of each bract color morph of Heliconia irrasa indicated by symbols. Numbered open circlesindicate locationsof three stems used as sourcesof fluorescentpowder. Numbered symbols at extremes of site indicate maximal dispersalof fluorescentpowder by hummingbirds. powder flow on each day. Chi-square analysis of any spatial association of color morphs. At was used to test for differences between ob- selectedradii (10, 25, 50 and 75 m) from the served and expected frequencies of flowers re- source plants, the proportion of red- and yel- ceiving powder. low-bracted stemsdid not differ from the over- all frequencies. The two morphs did not differ RESULTS AND DISCUSSION significantly with respect to the features of flo- Of the 56 flowering stems in the study area, ral presentation and phenology measured (Ta- 21 (38%) bore red-bracted inflorescences.