Report

Adaptation in a Plant-Hummingbird Association

See allHide authors and affiliations

Science  25 Apr 2003:
Vol. 300, Issue 5619, pp. 630-633
DOI: 10.1126/science.1080003

Abstract

Sexual dimorphism in bill morphology and body size of the Caribbean purple-throated carib hummingbird is associated with a reversal in floral dimorphism of its Heliconia food plants. This hummingbird is the sole pollinator of H. caribaea and H. bihai, with flowers of the former corresponding to the short, straight bills of males, the larger sex, and flowers of the latter corresponding to the long, curved bills of females. On St. Lucia, H. bihai compensates for the rarity of H. caribaea by evolving a second color morph with flowers that match the bills of males, whereas on Dominica, H. caribaea evolves a second color morph with flowers that match the bills of females. The nectar rewards of all Heliconia morphs are consistent with each sex's choice of the morph that corresponds to its bill morphology and energy requirements, supporting the hypothesis that feeding preferences have driven their coadaptation.

Understanding natural adaptations of organisms to their environment has been the focus of evolutionary investigations since the time of Darwin (1) and Wallace (2). Plant-animal interactions, especially those between flowering plants and their animal pollinators, provide classic examples of hypothesized coadaptations that confer advantages to both mutualists (3, 4). The most convincing investigations of adaptation require (i) the presence of discrete character polymorphisms in natural populations, (ii) evidence for the genetic basis of this variation, (iii) fitness measures, and (iv) comparative studies among populations (5). Here we present comparative data from two Lesser Antillean islands on contemporary character polymorphisms in sympatric species of the tropical plant genus Heliconia and their common pollinator, the purple-throated carib hummingbird, which support the principle of coadaptation between these mutualists.

The purple-throated carib hummingbird, Eulampis jugularis, is an example of sexual dimorphism that results from ecological causation: Although the wings and body masses of males average 8.6 and 25%, respectively, larger than those of females, the bills of females average 30% longer than those of males and are 100% more strongly curved (6, 7). Temeles et al. (7) showed that E. jugularis is the sole pollinator of two species of Heliconia on St. Lucia, a green-bracted H. bihai whose long, curved flowers match the bills of females and a red-bracted H. caribaea whose short, straight flowers match the bills of males (Fig. 1, A to E and G). Each sex prefers and feeds most efficiently from the Heliconia species whose flowers correspond to its bill size and shape, which is consistent with the predictions of ecological causation and feeding adaptation (7). At forest reserves on St. Lucia where the red-bracted H. caribaea is absent or rare, a red-green–bracted morph of H. bihai takes its place and has shorter, straighter flowers with a morphology intermediate between those that correspond to the bill morphologies of males and females (Fig. 1F). Male E. jugularis at these sites are associated primarily with the red-green–bracted morph of H. bihai, whereas females visit both the red-green– and green-bracted morphs.

Fig. 1.

Polymorphisms in bills of E. jugularis (A and B) and in flowers (C and D) and inflorescences (E to J) of Heliconia species on St. Lucia (E to G) and Dominica (C to D and H to J), West Indies. (A) E. jugularis, female bill. (B) E. jugularis, male bill. (C) H. bihai, flower. (D) H. caribaea, flower. (E) H. bihai, green inflorescence morph, St. Lucia. (F) H. bihai, red-green inflorescence morph, St. Lucia. (G) H. caribaea, inflorescence, St. Lucia. (H) H. bihai, red-and-yellow–striped inflorescence, Dominica. (I) H. caribaea, red inflorescence morph, Dominica. (J) H. caribaea, yellow inflorescence morph, Dominica.

E. jugularis on St. Lucia provides compelling evidence for a tight coevolutionary association between sexual dimorphism in its bill morphology and the floral morphology of its Heliconia food plants. We tested the generalities of this plant-pollinator system with comparative field studies on a second Antillean island, Dominica.

Our visits to Dominica in April of 2001 and 2002 coincided with peak flowering of the Heliconia species and the breeding season of the hummingbirds. We conducted roadside censuses of Heliconia species along six transects (8). As on St. Lucia, the principal understory food plants of E. jugularis on Dominica are H. bihai and H. caribaea. In contrast to their sympatric distribution on St. Lucia, however, on Dominica the two species occur allopatrically along an altitudinal gradient, with H. caribaea occupying lower elevations (100 to 600 m) and H. bihai occupying higher elevations (600 to 900 m). In addition, the pattern of floral dimorphism on Dominica is completely reversed from the pattern on St. Lucia: H. bihai, which has two color morphs on St. Lucia (green or red-green), has only one color morph on Dominica (red with a yellow stripe) (Fig. 1H). In contrast, H. caribaea, which has only one color morph on St. Lucia (red), has two color morphs on Dominica (red or yellow) (Fig. 1, I and J).

To determine whether the sexes of E. jugularis differed in their use of these two Heliconia species, we watched birds at patches of both species of Heliconia (8). Males were associated exclusively with H. caribaea, which they defended against conspecifics, and were not observed to visit H. bihai, although they fed from the flowers of other plants at sites where H. bihai was common. In contrast, females fed at H. caribaea and were the sole visitors to H. bihai (20 of 49 females, but 0 of 21 males, were observed to visit H. bihai; P < 0.001; χ2 = 12.0; df = 1).

To examine the relationship between flower use and bill morphology, we measured flower lengths and curvatures of H. bihai and the two H. caribaea morphs along four transects having all three kinds of Heliconia (Table 1). For all four transects, the flowers of H. bihai were significantly longer and more strongly curved than the flowers of either morph of H. caribaea, consistent with our observations of feeding preferences [P < 0.05; one-way analyses of variance (ANOVAs) with Tukey multiple comparisons].

Table 1.

Flower lengths and curvatures [mean ± SE (n, where n is the number of flowers)] of H. bihai and the red and yellow morphs of H. caribaea along six transects in Dominica, West Indies.

Transect H. bihai Red H. caribaea Yellow H. caribaea
Flower length (mm)
Salisbury Loop 47.6 ± 0.3 (16) 39.7 ± 0.4 (23) 36.2 ± 0.3 (15)
Mt. Diablotin 47.2 ± 1.2 (8) 40.0 ± 0.4 (20) 37.2 ± 0.2 (15)
Freshwater Lake 48.5 ± 0.4 (25) 38.3 ± 0.3 (14) 36.6 ± 0.4 (17)
Morne Trois Pitons 48.9 ± 0.7 (13) 38.1 ± 0.2 (16) 35.4 ± 0.3 (9)
Layou River Absent 36.4 ± 0.2 (19) 35.8 ± 0.2 (15)
Central Forest Absent 36.2 ± 0.4 (16) 36.4 ± 0.4 (12)
Flower curvature (degrees)
Salisbury Loop 28.8 ± 0.5 (16) 24.0 ± 0.7 (12) 20.2 ± 0.4 (13)
Mt. Diablotin 30.0 ± 0.4 (8) 22.6 ± 0.6 (15) 19.4 ± 0.7 (12)
Freshwater Lake 29.7 ± 0.5 (12) 24.9 ± 0.6 (12) 18.4 ± 0.6 (10)
Morne Trois Pitons 31.3 ± 0.9 (13) 23.9 ± 0.6 (16) 20.7 ± 0.5 (9)
Layou River Absent 20.0 ± 0.3 (19) 19.2 ± 0.3 (15)
Central Forest Absent 20.8 ± 0.4 (7) 20.4 ± 0.5 (7)

Further support for the hypothesis of floral specialization by males and females came from an analysis of the floral morphology and visitation frequencies to the two color morphs of H. caribaea. At four of our six study sites (Salisbury Loop, Mt. Diablotin, Freshwater Lake, and Morne Trois Pitons) (Table 1), flowers of the red morph were slightly but significantly longer and more strongly curved than flowers of the yellow morph (P < 0.05; one-way ANOVAs with Tukey multiple comparisons). On the basis of Temeles et al.'s (7) finding that preferences by male and female E. jugularis maintain a floral dimorphism in H. bihai on St. Lucia, we hypothesized that a similar difference in preferences between the sexes maintained the floral dimorphism of H. caribaea on Dominica. The four sites at which flowers of the red morph were longer and more curved than flowers of the yellow morph included contact zones between H. caribaea and H. bihai. Because female E. jugularis were the sole pollinator of the red-and-yellow–striped H. bihai on Dominica, we hypothesized that the longer, more curved flowers of the red morph of H. caribaea might result from greater use by females of the red morph at sites where the two Heliconia species came into contact. In support of this hypothesis, 15 of 19 females were associated with red clumps of H. caribaea at contact zones, whereas only 2 of 13 males were associated with red clumps of H. caribaea at contact zones (P < 0.0005, χ2 = 12.5; df = 1). At low elevation sites lacking a contact zone between H. bihai and H. caribaea, equal proportions of males and females visited the red and yellow morphs (n = 8 males and 8 females).

The sexual differences in bill morphology of E. jugularis may result from resource partitioning of Heliconia species on the basis of patch rewards (7): Larger males, with correspondingly higher energy requirements (8, 9), dominate females and establish priority of access to the more effusive H. caribaea. Comparisons of bract numbers—a measure of flower numbers per inflorescence (8)—and energy rewards among and between Heliconia species and morphs on Dominica and St. Lucia supported this hypothesis. On both islands, H. caribaea had significantly more bracts per inflorescence than H. bihai and at seven of eight sites offered a higher energy reward per inflorescence, consistent with male preferences for flowers of this species (P < 0.05; one-way ANOVAs with Tukey multiple comparisons) (Tables 2 and 3). On Dominica, the red-bracted morph of H. caribaea had significantly fewer bracts per inflorescence than the yellow-bracted morph, but only at sites where it came into contact with H. bihai (P < 0.05; one-way ANOVAs with Tukey multiple comparisons) (Table 2). The reduction in bract numbers of the red-bracted morph at these sites is consistent with greater use of the red-bracted morph by smaller female E. jugularis with their correspondingly lower energy requirements. In contrast, on St. Lucia, the red-green morph of H. bihai had significantly more bracts per inflorescence than the green-bracted morph, but only at sites where it geographically replaced H. caribaea and was used by larger male E. jugularis, with their correspondingly higher energy requirements (P < 0.05; one-way ANOVAs with Tukey multiple comparisons) (Table 3).

Table 2.

Number of bracts [mean ± SE (n, where n is the number of plants)] and joules per inflorescence of H. bihai and the red and yellow morphs of H. caribaea along six transects with and without contact zones between H. bihai and H. caribaea on Dominica, West Indies. Calculations of energy rewards in joules were based on measured average sucrose concentrations per flower of 22% and average 24-hour volumes per flower of 80, 90, and 125 μl for the yellow, red, and H. bihai morphs, respectively.

Transect H. bihai Red H. caribaea Yellow H. caribaea
Bracts Joules Bracts Joules Bracts Joules
Contact zone present
Salisbury Loop 4.3 ± 0.2 (51) 2128 6.0 ± 0.2 (52) 2138 7.3 ± 0.4 (31) 2312
Mt. Diablotin 4.8 ± 0.1 (34) 2376 6.3 ± 0.2 (76) 2245 7.4 ± 0.3 (45) 2344
Freshwater Lake 3.6 ± 0.1 (92) 1782 5.5 ± 0.2 (55) 1960 6.3 ± 0.2 (66) 1996
Morne Trois Pitons 3.5 ± 0.6 (26) 1732 7.0 ± 0.3 (30) 2494 7.9 ± 0.2 (32) 2503
Contact zone absent
Layou River Absent 5.2 ± 0.3 (41) 1853 4.6 ± 0.4 (14) 1457
Central Forest Absent 8.3 ± 0.3 (46) 2958 7.2 ± 0.5 (13) 2280
Table 3.

Number of bracts [mean ± SE (n, where n is the number of plants)] and joules per inflorescence of H. caribaea and the red-green and green morphs of H. bihai at four reserves where H. caribaea was common or rare on St. Lucia, West Indies. Calculations of energy rewards in joules were based on measured nectar concentrations of 25, 27.5, and 29.5% sucrose and 24-hour volumes of 98.3, 88.8, and 88.9 μl for H. caribaea, the red-green morph, and the green morph, respectively.

Reserve Green H. bihai Red-green H. bihaiH. caribaea
Bracts Joules Bracts Joules Bracts Joules
H. caribaea common
Quilesse 3.9 ± 0.1 (863) 1900 4.1 ± 0.5 (10) 1843 9.6 ± 0.2 (161) 4304
Barre de l'Isle 3.7 ± 0.1 (171) 1803 4.2 ± 0.5 (26) 1887 9.4 ± 0.2 (150) 4125
H. caribaea rare
Des Cartiers 4.3 ± 0.1 (610) 2095 5.0 ± 0.2 (186) 2247 8.0 ± 0.4 (18) 3787
Forestière 3.5 ± 0.1 (261) 1705 4.6 ± 0.2 (103) 2067 Absent

The complete reversal between St. Lucia and Dominica in floral dimorphism of the two Heliconia species together with an increase in the length and curvature of flowers of the red H. caribaea morph to match the bills of females provides a natural experiment on the two islands for detecting coadaptation between bill morphology of E. jugularis and the size and shape of Heliconia flowers. The corresponding decrease in bract numbers and energy rewards of the red morph of H. caribaea on Dominica and increase in bract numbers and energy rewards of the red-green morph of H. bihai on St. Lucia also suggest that sexual differences in body size of E. jugularis, and not just bill morphology, are maintained by feeding specialization on the different Heliconia food plants. Because male E. jugularis defend the same food plants in the breeding and nonbreeding periods, both natural and sexual selection may play roles in the evolution of their larger size. Nonetheless, we believe that the ecological differences in the Heliconia food plants drive the feeding and mating strategies as well as the sexual differences in size, morphology, and behavior of E. jugularis.

Sexual dimorphism in bill length and body size is common in hummingbirds, and in many species the pattern of dimorphism in bill length is opposite the pattern of dimorphism in body size, as in E. jugularis (10, 11). In many of these species, however, bill dimorphism is tied to sexual dichromatism in plumage, with brighter males having shorter bills and duller females having longer bills (10). These differences in bill morphology and plumage may be associated with sexual differences in dominance and resource use: Dominant males claim access to nectar-rich dense patches of short flowers and subordinate females are left to forage at nectar-poor, scattered resources (10). Longer bills may allow females to feed from a broader range of flower lengths than males (10). Studies of other hummingbird species (1214) and E. jugularis support this hypothesis, but indicate that sexual differences in plumage are not always linked to sexual differences in bill morphology, resource use, or dominance.

Dimorphisms of bill and body size in E. jugularis are among the most extreme of any hummingbird species, and their evolution may be the result of the paucity of competing species on islands relative to mainland hummingbird communities (15). Such “ecological release” has been observed in many island species, including hummingbirds, although its expression does not always involve sexual dimorphisms (16, 17). In hummingbirds, sexual dimorphism has been hypothesized to reflect interactions between social systems and feeding ecology (10, 11); our studies of E. jugularis support this hypothesis. Whether other species of hummingbirds with less extreme sexual dimorphisms exhibit similar tight coadaptations with their food plants requires further study, but given the widespread occurrence of sexual differences in bill length and body size within this group of birds, the role of feeding ecology in the evolution of sexual differences may be more common than has been previously thought.

Supporting Online Material

www.sciencemag.org/cgi/content/full/300/5619/630/DC1

Materials and Methods

References and Notes

View Abstract

Navigate This Article