Evidence for Ecological Causation of Sexual Dimorphism in a Hummingbird

See allHide authors and affiliations

Science  21 Jul 2000:
Vol. 289, Issue 5478, pp. 441-443
DOI: 10.1126/science.289.5478.441


Unambiguous examples of ecological causes of animal sexual dimorphism are rare. Here we present evidence for ecological causation of sexual dimorphism in the bill morphology of a hummingbird, the purple-throated carib. This hummingbird is the sole pollinator of two Heliconia species whose flowers correspond to the bills of either males or females. Each sex feeds most quickly at the flower species approximating its bill dimensions, which supports the hypothesis that floral specialization has driven the evolution of bill dimorphism. Further evidence for ecological causation of sexual dimorphism was provided by a geographic replacement of oneHeliconia species by the other and the subsequent development of a floral dimorphism, with one floral morph matching the bills of males and the other of females.

Sexual dimorphism in size and morphology is widespread in animals. Charles Darwin drew attention to these differences and offered three explanations for their evolution that were based on mechanisms of sexual selection, fecundity selection, and ecological causation (for example, resource partitioning) (1). Although empirical studies have demonstrated that the first two mechanisms operate in natural populations (2), unambiguous examples of ecological causation of sexual dimorphism are absent from the literature, an exception being some mosquito species, in which the mouthparts of males are adapted for drinking nectar and the mouthparts of females for drinking blood (3). A major obstacle has been the difficulty of showing that sexual dimorphisms are due to differences in resource use and ecology, because sexual differences in the size of feeding structures scale positively with body size in many animal species (4). The positive relationship between body size and feeding morphology makes it unclear whether ecological differences between the sexes are the cause or the consequence of sexual dimorphism (4).

Darwin was aware of the problem of attributing sexual dimorphisms to ecological causes, and in suggesting divergence in food use as a cause, he noted that such sexual differences should be confined to the feeding apparatus (1). The example Darwin used to illustrate this putative cause of sexual dimorphism was the New Zealand huia (Neomorpha acutirostris), now extinct. The sexes were similar in body size and plumage, but the bill of the male was short, thick, and straight, whereas the bill of the female was longer, slender, and decurved (5). The second example Darwin used to illustrate sexual dimorphism due to differences in food use was hummingbird bills. The relationship between sexual differences in the bills of hummingbirds and patterns of flower visitation, however, has received little attention in studies of natural hummingbird populations. Here we present evidence for ecological causation of sexual dimorphism in the bill morphology of the purple-throated carib hummingbird Eulampis jugularis from the island of St. Lucia, West Indies.

Several features of E. jugularis make it an excellent candidate for studies of food-based hypotheses for the evolution of sexual dimorphism. First, although the wings and body masses of males average 8.6 and 25%, respectively, larger than those of females, the bills of females are on average more than 30% longer than those of males (6), which is one of the most extreme bill dimorphisms of any hummingbird (7). Moreover, the bills of females are curved downward at an approximately 30o angle, whereas the bills of males are much straighter and are curved downward at only a 15o angle (Fig. 1). Second, mapping sexual dimorphism in wing and bill length onto a time-calibrated DNA-hybridization–based phylogeny yields no consistent pattern among E. jugularisand its closest relatives, which suggests that behavioral and ecological factors have had some role in the evolution of sexual dimorphisms within this group (8). Third, the expression of sexual dimorphism in E. jugularis is inconsistent with patterns of sexual selection for larger male size, where bill length should scale positively with male size rather than negatively. Hence,E. jugularis fulfills Selander's criterion that the only reliable evidence for ecological causation of sexual dimorphism is a modification of feeding structures in a direction that is inconsistent with sexual selection and is greater than would be expected on the basis of body size differences alone (4, 5).

Figure 1

Sexual dimorphism in bill length and curvature of male (left) and female (right) purple-throated caribs, E. jugularis. The sexes are monomorphic in plumage, and males are 25% heavier than females. Nonetheless, the bills of females are 30% longer and 100% more curved than the bills of males.

Our fieldwork (May through June 1999) encompassed periods when the birds were actively nesting and rearing young. We censused understory food plants within 100 m of either side of trails passing through four rainforest reserves (9). The only understory food plants available during the months of May and June were a red-bracted Heliconia caribaea and an endemic green-bracted H. bihai (10). To determine whether the sexes of E. jugularis differed in their use of these twoHeliconia species, we conducted watches at “dense” and “sparse” patches of H. caribaea and H. bihaiin Quilesse Reserve (11). Eulampis jugularis was the sole pollinator of H. caribaea and H. bihai. Males were associated with dense patches of H. caribaea, which they defended against intruding conspecifics, although they occasionally fed at flowers of H. bihai on the periphery of their territories. Females intruded into dense patches of H. caribaea that were defended by males, and they also trapline-foraged in sparse patches of H. caribaea andH. bihai and dense patches of H. bihai, which they occasionally defended. Censuses of the reserves support the results of patch watches: 15 of 15 males, but only 7 of 18 females, were observed feeding in patches of H. caribaea(P < 0.001; χ2 = 13.75, df = 1).

To examine the relationship between flower use and bill dimensions, we measured flower lengths and curvatures of H. caribaeaand H. bihai in the three reserves having bothHeliconia species (Table 1). At all three sites, the flowers ofH. caribaea were significantly shorter and straighter than were the flowers of H. bihai [P < 0.05;t tests with sequential Bonferroni adjustments (12)]. Differences in nectar production and concentration between the two species were not significant (P > 0.05;t tests with sequential Bonferroni adjustments).

Table 1

Flower lengths and curvatures (mean ± SE) ofH. caribaea and H. bihai from three St. Lucian rainforest reserves. The number of flowers measured is shown in parentheses.

View this table:

The striking correspondence between flower lengths and curvatures shown in Table 1 and hummingbird bill lengths and curvatures suggests that the bills of males are specialized for feeding from flowers of H. caribaea, whereas the bills of females are specialized for feeding from flowers of H. bihai. If bills of each sex are specialized for feeding from a certain species of flower, each sex should have faster feeding times at the flower species most approximating its bill size and shape and should have slower feeding times at the flower species least approximating its bill size and shape (13, 14). The feeding times of females were significantly shorter at the longer, more curved flowers of H. bihai (mean ± SE = 3.4 ± 0.4 s) than at the shorter, straighter flowers of H. caribaea(mean ± SE = 4.4 ± 0.8 s; pairedt test, t = 2.32, P = 0.034,n = 6 females, 278 feeding visits). In contrast, the feeding times of males (mean ± SE = 2.7 ± 0.5 s) were significantly shorter than the feeding times of females (3.8 ± 0.8 s) at flowers of H. caribaea (pairedt test; t = 4.53, P = 0.023,n = 3 males and females at the same patches, 143 feeding visits). Because males were so overwhelmingly in association withH. caribaea, we observed too few visits by males to H. bihai for use in statistical comparison. Although nectar volumes were not controlled in these observations, the feeding time estimates support the hypothesis that the bills of males and females are specialized for feeding from flowers of H. caribaea andH. bihai, respectively.

Additional support for the hypothesis of floral specialization by males and females comes from Forestière Reserve, which lacks H. caribaea. At this site, we found a red-and-green–bracted morph ofH. bihai (15). At Forestière, we observed males feeding at, and defending, flowers of the red-green– but not the green-bracted H. bihai. The red-green morph had significantly shorter and straighter flowers than the green-bracted morph [P < 0.05, t tests with Bonferroni adjustments (Table 2)], but the two Heliconias did not differ in nectar production or concentrations.

Table 2

Flower lengths and curvatures (mean ± SE) of the green- and red-green–bracted morphs of H. bihai at Forestière Reserve, St. Lucia. The number of flowers measured is shown in parentheses.

View this table:

We interpret the red-green–bracted morph at Forestière Reserve as an ecological replacement of the red-bracted H. caribaea specialized on by male purple-throated caribs. We also recorded the red-green morph at the other three reserves. In support of the hypothesis of ecological replacement, the red-green morph increased in frequency where the red-bracted morph of H. caribaea was rare, and vice versa (Table 3) (P < 0.001, χ2 = 70, df = 6) (16). Additional support for this interpretation is provided by a comparison of flowers at Des Cartiers, where the red-green morph was common, and at Quilesse, where it was rare (Table 3). At Des Cartiers, we also observed males defending patches of the red-green morph. At this reserve it had significantly straighter flowers (25° ± 1° of curvature, n = 23 flowers) than the green-bracted morph (32° ± 1° of curvature, n = 18 flowers;P < 0.05, t test with Bonferroni adjustments). Standing crops and concentrations of nectar did not differ between the two flower morphs. In contrast, at Quilesse, the red-green morph was quite rare and consisted of one small patch used exclusively by female purple-throated caribs but not by males. No significant differences in flower curvature, nectar standing crop, or nectar concentration were recorded between the red-green– and the green-bracted H. bihai, although the red-green morph's flowers were significantly longer (47 ± 0.6 mm, n = 10 flowers) than flowers of the green-bracted H. bihai (43 ± 0.4 mm,n = 28 flowers; P < 0.05, ttest), which is consistent with expectations based on the exclusive use by females of the red-green morph at this site.

Table 3

Frequencies of H. caribaea and the green and red-green morphs of H. bihai, at four St. Lucian rainforest reserves (numbers refer to plants with bracts).

View this table:

Our study provides unambiguous evidence that sexual differences in bill length and bill curvature are associated with sexual differences in Heliconia use by male and female E. jugularis. The variation in the red-green morph's flower morphology among reserves, however, raises the question of which party in the association has evolved to match the other. The bill morphology of E. jugularis differs significantly between islands, which suggests that bill dimorphism of E. jugularis on St. Lucia is not merely a consequence of past adaptation and evolution (17). The two Heliconia species are the major food plants of E. jugularis from January to July, during the period before, during, and after breeding (18,19). Selection might be particularly intense during this period because the birds are constrained to the use of theseHeliconia plants owing to nesting and mating activities. These data suggest that both the birds and the flowers may be targets of reciprocal selection.

Because males and females were defending territories against conspecifics, food competition between the sexes is the most likely explanation for sexual differences in resource use. Nonetheless, sexual selection also may be involved, because the same floral resource that males defend for food also serves for mate attraction (20). Larger size has been shown to be an advantage in contests for territories both within and between hummingbird species (21, 22), and we suggest that both food and mate competition may have selected for larger male body size and partitioning of Heliconia species on the basis of patch reward [H. caribaea inflorescences bear two to three times as many flowers as H. bihai (20)]. Once resource partitioning on the basis of patch rewards was established, natural selection may then have acted on the bill dimensions of males and females. Controlled studies involving experimental manipulation of patch rewards and measurements of feeding times of males and females with known bill sizes and shapes will be necessary to determine the relative roles of these mechanisms underlying sexual differences in flower use (2).

  • * To whom correspondence should be addressed. E-mail: ejtemeles{at}


View Abstract

Navigate This Article