Evolutionary Biology

, Volume 40, Issue 2, pp 246–260

Becoming Different But Staying Alike: Patterns of Sexual Size and Shape Dimorphism in Bills of Hummingbirds

Authors

    • Department of Ecology, Evolution and Organismal BiologyIowa State University
  • Dean C. Adams
    • Department of Ecology, Evolution and Organismal BiologyIowa State University
Research Article

DOI: 10.1007/s11692-012-9206-3

Cite this article as:
Berns, C.M. & Adams, D.C. Evol Biol (2013) 40: 246. doi:10.1007/s11692-012-9206-3

Abstract

Hummingbirds are known for their distinctive patterns of sexual dimorphism, with many species exhibiting sex-related differences in various ecologically-relevant traits, including sex-specific differences in bill shape. It is generally assumed that such patterns are consistent across all hummingbird lineages, yet many taxa remain understudied. In this study we examined patterns of sexual size and sexual shape dimorphism in bills of 32 of 35 species in the monophyletic Mellisugini lineage. We also compared patterns of bill size dimorphism in this group to other hummingbird lineages, using data from 219 hummingbird species. Overall, the presence and degree of sexual size dimorphism was similar across all hummingbird lineages, with the majority of Mellisugini species displaying female-biased sexual size dimorphism, patterns that remain unchanged when analyzed in a phylogenetic context. Surprisingly however, we found that sexual dimorphism in bill shape was nearly absent in the Mellisugini clade, with only 3 of the 32 species examined displaying bill shape dimorphism. Based on observations in other hummingbird lineages, the lack of sexual shape dimorphism in Mellisugini is particularly unusual. We hypothesize that the patterns of sexual size dimorphism observed here may be the consequence of differential selective forces that result from competition for ecological resources. We further propose that an influential mechanism underlying shape dimorphism is competition and niche segregation. Taken together, the evolutionary changes in patterns of sexual shape dimorphism observed in Mellisugini suggest that the evolutionary trends of sexual dimorphism in the Trochilidae are far more dynamic than was previously believed.

Keywords

Sexual dimorphismGeometric morphometricsSexual size dimorphismSexual shape dimorphismHummingbirdMellisugini

Introduction

Understanding the origin of phenotypic diversity is a major focus of evolutionary research, and patterns of sexual dimorphism represent a particularly intriguing component of this diversity. Sexual dimorphism is exhibited throughout the animal kingdom (see e.g., Butler et al. 2000) and is displayed in a myriad of ways, including sex-specific behavior and vocalizations (Potter et al. 2005), sexual differences in body size (Fairbairn 1997), coloration (Stuart-Fox and Ord 2004), and other morphological traits (Berns and Adams 2010; Butler et al. 2007; Worthington et al. 2012). Darwin (1871) drew attention to these patterns and suggested that morphology can vary between the sexes due to the action of sexual selection operating in one or more ways, which subsequently enhances sex-specific fitness in relation to reproduction (Darwin 1871; Jones and Ratterman 2009).

A number of mechanisms have been proposed to explain the evolution of sexual dimorphism. For example, sexual selection can generate sex-specific differences as the sexes evolve in distinct directions that maximize their own reproductive success (Darwin 1871; Andersson 1994). Alternatively, ecological mechanisms such as competition for resources, may exert distinct selective forces on the sexes, resulting in the evolution of sexual dimorphism (Selander 1972; Hedrick and Temeles 1989; Slatkin 1984). Here, intraspecific competition in species-poor communities may allow divergent selection between the sexes (rather than between species) to result in niche segregation between males and females (i.e., intersexual niche packing: sensu Butler et al. 2000). A third possible mechanism that may enhance sexual dimorphism in some species is the influence of sex-specific divergence in response to environmental gradients, where males and females exhibit differential responses to the same environmental selective pressures (Hendry et al. 2006). For instance, in guppies (Poecilia reticulata), open canopy sites resulted in selection for smaller heads and distended abdomens in females but not in males, whereas both sexes in high flow sites had small heads and deeper caudal peduncles (Hendry et al. 2006). Likewise, weaker latitudinal clines in male houseflies (Musca domestica) but not in females suggest sex-specific responses to food resource abundance which may vary clinally (Alves and Belo 2002).

Hummingbirds (family Trochilidae) offer a unique opportunity to study patterns of sexual dimorphism and elucidate the underlying mechanisms responsible for these patterns. This monophyletic lineage of 338 currently recognized species (McGuire et al. 2007, 2009; Birdlife.org 2012) has been a model taxon for the study of sexual dimorphism, as these species exhibit patterns such as dimorphic plumage (Darwin 1871; Bleiweiss 1992, 1997) sex-specific behavior (Feinsinger and Colwell 1978, Kodric-Brown and Brown 1978), and ornamentation (Evans et al. 1995; Zusi and Gill 2009), among others. Several hypotheses have been proposed to explain the evolution of this sexual dimorphism in this group. For example, the remarkable ornamentation that males sometimes exhibit, such as the tail ‘boots’ of the hummingbird Ocreatus underwoodii (Booted Racket-tail) or tail length in Phaethornis superciliosus (Long-tailed Hermit) may be due to ‘Fisherian runaway selection’, where females prefer males that exhibit more extreme phenotypes within a population, which over evolutionary time become increasingly exaggerated despite the potential fitness costs to the males themselves (Andersson 1994). Further, differential plumage coloration between the sexes in some species is thought to be the result of females selecting males that exhibit superior characteristics (i.e., the ‘good genes’ hypothesis: sensu Darwin 1871).

Hummingbirds also exhibit sexual dimorphism in both body size and bill morphology. For example, in hummingbirds bill morphology is under strong selection pressures due to its role in foraging, and differential foraging among species is thought to be a major cause of diversification in trophic structures among species (Darwin 1871; see also Feinsinger 1978; Brown and Bowers 1985; Collins and Paton 1989; Bleiweiss 1998). Additionally, males and females of some species forage differently (Carpenter et al. 1991; Temeles and Kress 2003) and many of these species also display sexual dimorphism in bill size and bill curvature (Temeles et al. 2010). Recently, quantitative support for Darwin’s hypothesis is found in some species where there is a direct link between sexual differences in bill morphology in Eulampis jugularis (Purple-throated Carib) and patterns of sex-specific foraging. Here, males and females of this species forage on distinct resources, and the bill morphology of each sex is correlated with the morphology of the flower that each feeds from (the ‘ecological causation hypothesis’ sensu Temeles et al. 2010). Similar patterns have been observed in other clades of hummingbirds, where differences in bill sexual size and shape dimorphism have a direct link to feeding ecology (Temeles et al. 2010). Together, these findings suggest that, not only is sexual dimorphism in bill morphology the evolutionary result of various ecological processes and sex-specific adaptations to foraging (Temeles et al. 2010), but also that bill size and shape dimorphism is common throughout hummingbirds.

Much of the prior work on sexual dimorphism in Trochilidae has focused on species in tropical lineages known to display sexual dimorphism, and have concentrated largely on dimorphism in bill size and bill shape (curvature: see Temeles et al. 2010; Rodríguez-Flores and Stiles 2005). Based on these studies, it has generally been assumed that sexually-dimorphic patterns in bill shape are consistent across Trochilidae due to the tight coevolutionary link between bill morphology and flower morphology in this group, yet little work on sexual dimorphism has been performed in taxa outside of a few focal lineages (e.g., Bleiweiss 1999; Temeles et al. 2005). For example, a few studies have examined sexual dimorphism in bill morphology in select species in the Mellisugini clade, finding sexual size dimorphism of the bill in Selasphorus rufus (Temeles and Roberts, 1993), S. scintilla and S. flammula (Stiles 1983). However the prevalence of such patterns in bill morphology remains largely unknown, as most of the 35 species in the Mellisugini clade have yet to be examined. Recently, it was observed that two sister species within the Mellisugini clade, Archilochus alexandri and A. colubris, differ in their patterns of dimorphism in bill morphology (Berns and Adams 2010), where both species display significant bill size dimorphism, but only A. colubris exhibits bill shape dimorphism. These results were surprising, as prior work on Trochilidae in other clades suggested that shape dimorphism in bill curvature is common (bill curvature is one aspect of bill shape: Temeles et al. 2010).

Given these findings, we conducted a broader study of species in the Mellisugini clade with the purpose of addressing the following questions. First, is sexual dimorphism in bill size common across species in the Mellisugini clade? Based on previous studies, we predicted that the majority of species in the Mellisugini clade would exhibit primarily female-biased (that is, the bill of females is larger than males) bill size dimorphism. Second, is the presence and pattern of sexual shape dimorphism in the Mellisugini clade concordant with those found across all Trochilidae? Studies of other clades have found sexual curvature dimorphism to be common, therefore we predicted the same pattern would be present in the Mellisugini clade. We addressed these questions in a phylogenetic context using both linear measurements and landmark-based geometric morphometric techniques to quantify sexual size and shape dimorphism in the bill morphology in 32 of the 35 Mellisugini species (McGuire et al. 2009). We then compared patterns of bill size dimorphism in the Mellisugini lineage to those found in other Trochilidae lineages, using a dataset collected from prior studies published in the literature (219 species).

Methods

Specimen Information

We measured 1,347 hummingbirds representing 32 of the 35 species in the Mellisugini lineage, obtained from 14 museum collections (see “Appendix”). All specimens included in this study were adults of known sexes, and a roughly equal proportion of males and females of each species were examined (722 males, 625 females; see “Appendix”).

Morphometrics

We obtained digital images of the left-lateral side of the head and bill of each specimen using a Nikon DXM-1200 digital camera mounted on a Nikon SMZ 1500 stereomicroscope (a Nikon D-90 was used to collect images obtained at museums, as the stereomicroscope system was not easily transportable). Each photograph included a ruler in order to account for size in the analyses, and the birds were placed in the same position on a modeled clay surface. From these images, two sets of data were obtained. First, the exposed culmen was measured on each specimen (culmen length: CL: Fig. 1a; e.g., Bleiweiss 1999; Colwell 2000) and was treated as a measure of bill size for each individual. Culmen length and bill centroid size were highly correlated (r = 0.97), so only CL is used here as a measure of size. To place our findings in a broader phylogenetic context, we combined these linear measurements with additional bill length data that we obtained from the literature, yielding a total dataset of CL from 219 hummingbird species (see “Appendix”) across the family Trochilidae. Second, bill shape was obtained from each image using landmark-based geometric morphometric methods (Rohlf and Marcus 1993; Adams et al. 2004). These methods allow for quantification of shape after the effects of non-shape variation (position, orientation, and scale) have been mathematically held constant. For this approach we first digitized the locations of 10 biologically homologous landmarks from the images of each specimen. Additionally, we included 15 sliding semilandmarks along the boundary curve of each bill to represent its shape and curvature (see Berns and Adams 2010) using TPSDig 2 (Rohlf 2010).
https://static-content.springer.com/image/art%3A10.1007%2Fs11692-012-9206-3/MediaObjects/11692_2012_9206_Fig1_HTML.gif
Fig. 1

a Exposed culmen representing bill length. b Landmark-based geometric morphometrics. Open circles designate biologically homologous landmarks and filled circles represent sliding semilandmarks. c Representative individual from the species with extreme bill curvature (Calothorax lucifer), and d from the species with the straight bills (Archilochus colubris)

Together, the 25 landmarks and semilandmarks were subjected to a Generalized Procrustes Analysis (GPA: Rohlf and Slice 1990). This least-squares procedure translates all specimens to the origin, scales them to unit centroid size, and optimally rotates them to minimize the total sums-of-squares deviations of the landmark coordinates from all specimens to the average configuration. During this procedure, semilandmarks were allowed to slide along their tangent directions (Bookstein et al. 1999; Gunz et al. 2005) so as to minimize Procrustes distance between specimens (see e.g., Serb et al. 2011). After superimposition, the aligned Procrustes shape coordinates describe the location of each specimen in a curved space related to Kendall’s shape space (Rohlf 1999). These were then projected orthogonally into a linear tangent space yielding Kendall’s tangent space coordinates (Dryden and Mardia 1993, 1998; Rohlf 1999), which were treated as a set of shape variables for further analyses of shape variation and covariation (e.g., Adams et al. 2007; Adams and Nistri 2010; Adams 2010).

Patterns of Sexual Size Dimorphism

We performed two sets of analyses to assess patterns of sexual dimorphism of bill size. First, to determine whether bill size and size dimorphism differed among species in the Mellisugini clade, we used our culmen length data and a two-factor analysis of variance (ANOVA); where variation in bill size (culmen length) was explained by species, sex, and their interaction. A significant species × sex term would imply that there was a significant difference in the degree of sexual size dimorphism among species. Due to the large differences in variance within the species × sex groups, we also ran ANOVAs for each species separately to determine whether sexual dimorphism was present in each species.

Second, we performed a family-wide analysis of bill size dimorphism, using the bill size measurements of the 32 species in the Mellisugini clade as above, as well measurements of an additional 187 species obtained from literature (Temeles et al. 2010; Colwell 2000; Bleiweiss 1999). In total, this analysis included 219 of 338 hummingbird species (Birdlife.org), or 65 % of the current diversity of the group. From these measurements, the average male and female bill sizes (\( \overline{CL}_{M} ,\overline{CL}_{F} \)), were obtained. Next, measures of sexual size dimorphism were estimated as the Lovich-Gibbons ratio (Lovich and Gibbons 1992), which is found as: (\( \overline{CL}_{F} /\overline{CL}_{M} - 1 \))*(1) for species where the female is the larger sex, and (\( \overline{CL}_{M} /\overline{CL}_{F} - 1 \))*(−1) when the male is the larger sex (see Stephens and Wiens 2009; Temeles et al. 2010). Using these size-dimorphism ratios, we performed an ANOVA to determine whether the seven major hummingbird clades for which we had information (sensu McGuire et al. 2009) differed in their patterns of sexual size dimorphism. In addition, we performed a phylogenetic ANOVA (Garland et al. 1993) on the same ratio data to account for non-independence due to shared evolutionary history. For this approach, the evolutionary relationships among species were based on the current molecular phylogeny for hummingbirds (McGuire et al. 2007). Finally, these analyses were repeated on the simple ratio of male to female bill size (\( \overline{CL}_{M} ,\overline{CL}_{F} \)), from which statistically similar results were obtained (results not shown).

Patterns of Sexual Shape Dimorphism

To determine whether bill shape and shape dimorphism differed among species in the Mellisugini clade we performed a two-factor multivariate analysis of variance (MANOVA), where variation in bill shape was explained by species, sex, and species × sex interaction. A significant species × sex term would imply that there were differences in the degree of sexual shape dimorphism among species. As with the ANOVA, we ran a separate permutational-MANOVA with 9,999 iterations (Anderson 2001) for each species to examine the degree of sexual shape dimorphism within each species. We then calculated vectors of sexual shape dimorphism for each species as the difference between male and female means (see Berns and Adams 2010), and used a permutation procedure with 9,999 iterations to determine whether the degree of sexual shape dimorphism differed between species (for details see Adams and Collyer 2007, 2009; Collyer and Adams 2007). Here, the observed sexual shape dimorphism for each species was quantified as the multivariate vector connecting male and female means for each species. The length of this vector (i.e., Euclidean distance) corresponds to the magnitude of sexual shape dimorphism exhibited by each species, which were compared using the permutation procedure above to determine the extent to which the degree of sexual shape dimorphism differed among species (see Berns and Adams 2010).

Visualization

Finally, to visualize patterns of bill shape variation, we performed a principal components analysis using the full set of Kendall’s tangent space coordinates and plotted the first two PCs, which described the largest amount of shape variation. In this principal components plot we also included vectors connecting male and female means for species that displayed significant sexual shape dimorphism (see “Results” below). We then generated thin-plate spline deformation grids (Bookstein 1991) for phenotypic means of males and females to graphically depict differences in bill shapes for these species, and to facilitate biological interpretation of the observed shape differences within and between them. All analyses were conducted in R version 2.11.1 (R Development Core Team 2010), TpsSpline (Rohlf 2005) and TPSRelW (Rohlf 2004).

Results

Patterns of Sexual Size Dimorphism

Using a two-factor ANOVA on the Mellisugini data, we found significant differences in bill size among species, between the sexes, as well as a significant species × sex interaction term (Table 1A). The latter term revealed that the degree of sexual size dimorphism differed significantly among species. When sexual size dimorphism was examined for each species separately, we found that the majority of species (26) displayed significant sexual size dimorphism (Table 1B). Interestingly, we identified considerable variation in the degree of size dimorphism exhibited among species in this group, with a 27-fold difference in the degree of sexual size dimorphism between the species with the least size dimorphism (Selasphorus sasin: sexual size dimorphism = 0.0045), and the most dimorphic species (Calothorax pulcher: sexual size dimorphism = 0.12257).
Table 1

(A) Statistical results from ANOVA on culmen length for 32 species of hummingbird in the Mellisugini lineage. (B) ANOVA comparing male versus female culmen lengths for each of 32 species of Mellisugini hummingbirds

(A) Effects

Mean squares

Approx. F

df

P

Species

493.06

393.3193

31

<0.001

Sex

240.74

192.0392

1

<0.001

Species × sex

2.66

2.1192

31

<0.001

Residuals

1.25

 

1,283

 

(B) Species

Mean squares

Approx. F

df

P

Ratio

Archilochus alexandri

33.165

19.070

1

<0.001

0.038

Archilochus colubris

62.972

125.479

1

<0.001

0.119

Atthis Heliosa

7.102

20.873

1

<0.001

0.086

Calliphlox amethystina

1.144

1.283

1

0.269

0.031

Calliphlox bryantae

3.936

6.626

1

0.017

0.053

Calliphlox evelynae

1.616

1.012

1

0.324

−0.008

Calliphlox mitchelli

6.236

9.152

1

0.005

0.069

Calothorax lucifer

2.958

2.585

1

0.119

0.032

Calothorax pulcher

25.477

10.599

1

0.003

0.123

Calypte anna

7.319

5.563

1

0.025

0.086

Calypte costae

4.664

5.128

1

0.032

0.051

Chaetocercus astreans

7.459

19.981

1

<0.001

0.082

Chaetocercus bombus

2.580

6.552

1

0.023

0.064

Chaetocercus heliodor

8.158

7.389

1

0.011

0.074

Chaetocercus jourdanii

5.939

6.592

1

0.016

0.067

Chaetocercus mulsanti

15.844

13.059

1

0.001

0.083

Doricha eliza

2.581

5.664

1

0.033

0.041

Doricha enicura

9.781

2.660

1

0.118

0.084

Mellisuga helenae

7.454

11.726

1

0.002

0.093

Mellisuga minima

5.451

14.684

1

0.001

0.091

Microstilbon burmeisteri

9.615

24.023

1

<0.001

0.047

Myrmia micrura

7.351

7.384

1

0.014

0.088

Myrtis fanny

0.293

0.197

1

0.660

−0.013

Rhodopis vesper

34.673

3.506

1

0.071

0.080

Selasphorus flammula

4.499

7.610

1

0.009

0.062

Selasphorus platycercus

11.400

8.092

1

0.009

0.074

Selasphorus rufus

14.766

36.033

1

<0.001

0.190

Selasphorus sasin

0.065

0.054

1

0.817

0.005

Stellula calliope

2.443

6.788

1

0.016

0.059

Selasphorus scintilla

4.896

5.838

1

0.021

0.056

Thaumastura cora

7.753

17.005

1

<0.001

0.060

Tilmatura dupontii

3.503

8.499

1

0.007

0.057

Size dimorphism ratios for each species (expressed as Lovich-Gibbons ratios) are shown. Significant effects are shown in bold

When magnitudes of sexual size dimorphism were examined across the entire hummingbird family, we found no difference in the degree of size dimorphism exhibited among clades within Trochilidae (F7,211 = 0.632; P = 0.7292), suggesting that all groups displayed a similar degree of size dimorphism. Similarly, no differences among clades were identified when the phylogenetic relationships among species were taken into consideration (F7,114 = 0.2143; Prand = 0.9714). While all species displayed similar magnitudes of size dimorphism in their bills, overall the majority of species (156 of 219) were female-biased (Fig. 2). Consistent with prior results, some lineages (e.g., Phaethornithinae) displayed male-biased size dimorphism for the majority of their species (e.g., Temeles et al. 2010: Fig. 2). However, this was not the case in the Mellisugini clade, where 92 % of the species examined (24 of 26) exhibited female-biased size dimorphism (Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs11692-012-9206-3/MediaObjects/11692_2012_9206_Fig2_HTML.gif
Fig. 2

Histograms displaying the variation of sexual size dimorphism within each clade of hummingbirds. Frequency is on the Y-axis and bill size dimorphism (expressed as the Lovich-Gibbons ratio) is on the X-axis. Those to the left of 0 are male-biased

Patterns of Sexual Shape Dimorphism

Using a factorial MANOVA, we found that all factors (species, sex, species × sex) explained significant proportions of variation in bill shape (Table 2A). The significant interaction term implied that patterns of sexual shape dimorphism differed among species, and when this was examined separately in each species, we found that only three of 32 species displayed significant shape dimorphism: Archilochus colubris, Selasphorus scintilla, and Mellisuga minima (Table 2B). Thus, in stark contrast to patterns of size dimorphism, and in contrast to patterns of shape dimorphism observed in other hummingbird lineages, very little shape dimorphism is exhibited in the Mellisugini lineage. When patterns of sexual shape dimorphism were compared among these three species, M. minima exhibited a significantly greater degree of shape dimorphism (Euclidean distance) than the other two species (DM minima = 0.054, DA colubris = 0.026, DS scintilla = 0.041, P = 0.001), whereas A. colubris and S. scintilla did not differ in the amount of shape dimorphism displayed (P = 0.19). Further, the direction of shape dimorphism in morphospace also differed between M. minima and both A. colubris and S. scintilla, (P = 0.001, Fig. 3). Thin-plate spline deformation grids revealed that in these species, females have longer, more curved bills at both the tip and main body of the bill relative to the mean, while males have straighter and shorter bills and M. minima has the largest magnitude of sexual shape dimorphism (Fig. 3). Thus, the significant sexual shape dimorphism in these species can be generally characterized as females having more curved bills, while males have relatively straighter bills.
Table 2

(A) MANOVA analysis of bill shape (Kendall’s tangent space coordinates) for 32 species in the Mellisugini clade. (B) Results of permutational-MANOVA for those species displaying significant sexual shape dimorphism (A. colubris, S. scintilla and M. minima)

(A) Effects

Pillai’s trace

Approx. F

df

P

Species

208.95

160.167

35, 1,167

<0.001

Sex

470.97

361.009

1,085, 37,107

<0.001

Species × sex

2.24

1.715

1,085, 37,107

<0.001

(B) Species

Mean squares

Approx. F

P

Archilochus colubris

0.0137

2.944

0.04

Selasphorus scintilla

0.0146

3.005

0.032

Mellisuga minima

0.0208

4.965

0.006

Significant effects shown in bold

https://static-content.springer.com/image/art%3A10.1007%2Fs11692-012-9206-3/MediaObjects/11692_2012_9206_Fig3_HTML.gif
Fig. 3

Principal components plot of bill shape data based on Kendall’s tangent space coordinates. Here, all individuals are shown, as well as the male and female means for the three species displaying significant sexual shape dimorphism (A. colubris, S. scintilla and M. minima). Lines connecting symbols represent the magnitude of sexual shape dimorphism in each of the 3 species. Thin-plate spline deformation grids of the average female and male bill shape relative to the mean are also presented and have been scaled to a factor of 2 to enhance biological interpretation

Discussion

For centuries, evolutionary biologists have been interested in the phenotypic differences between the sexes and the effect that these patterns have on organismal diversity. Many studies have identified patterns of sexual size and sexual shape dimorphism in hummingbird bill morphology, particularly in species from tropical lineages. However, none have used a phylogenetic context to study sexual shape dimorphism within an entire hummingbird lineage, nor compared patterns of sexual size dimorphism across all hummingbird clades in a phylogenetic context. Based on existing studies of sexual dimorphism in other species, it is generally assumed that the presence of sexual size and sexual shape dimorphism will be found throughout all hummingbird clades. However, while this appears to be the case for sexual size dimorphism of hummingbird bills, our results show that shape dimorphism in the Mellisugini lineage does not follow this general pattern.

Using bill size data for 219 species of Trochilidae, we found that the majority of the species in the Mellisugini clade exhibited sexual size dimorphism in similar magnitudes as was observed in other groups (Temeles et al. 2010; Colwell 2000; Bleiweiss1999). Our findings also revealed that the Mellisugini lineage is decidedly female-biased in bill sexual size dimorphism, with 92 % of the species examined (24 of 26) displaying larger bills in females as compared to males. Thus, the patterns of sexual size dimorphism in Mellisugini are concordant with what is observed in other lineages within Trochilidae except Phaethornithinae, which is male-biased (Fig. 2; also Bleiweiss 1999; Temeles et al. 2010).

The presence and magnitudes of sexual size dimorphism observed here may be the consequence of differential selective forces that result from competition for ecological resources. For example, Bleiweiss (1999) hypothesized that feeding ecology may influence the evolution of male- or female-biased sexual size dimorphism in hummingbirds. In hummingbirds, a longer bill allows feeding from a wide variety of resources whereas a short bill is more efficient in a patch of the same flower. In species where the males are the dominant sex and defend territories, females tend to have longer bills, allowing them to feed from more dispersed resources, whereas males tend to have shorter bills, increasing foraging efficiency in small territories (Bleiweiss 1999). By contrast, in species with lekking mating systems males no longer hold territories, and must instead compete with other hummingbirds. In this case, the male bill tends to be longer to allow them to feed from a wider variety of flowers due to competitive forces, whereas females feed from small patches outside the lekking grounds and have smaller bills better suited to feeding in small patches. Thus in both cases, there is a direct relationship between sexual social interactions, foraging, and bill morphology (see Bleiweiss 1999). Our results are concordant with Bleiweiss’ hypothesis, as species in the Mellisugini clade do not lek, and the majority display female-biased sexual size dimorphism. By contrast, species in the Phaethornithinae clade exhibit primarily male-biased sexual size dimorphism of bill morphology, and the majority of these species indeed have lekking behavior. It is therefore reasonable to hypothesize that the presence of female-biased sexual size dimorphism in the Mellisugini clade is a reflection of the non-lekking and male territorial behavior in these species, which subsequently affects differential foraging between males and females of these species.

The most surprising result of our study is that in stark contrast to other Trochilidae lineages, only three species (A. colubris, S. scintilla and M. minima) in the Mellisugini exhibited significant sexual shape dimorphism in bill morphology. Trochilidae are generally considered to be dimorphic in both bill size and bill shape, as the majority of species in some lineages display differences in bill curvature (e.g., Phaethornithinae: Temeles et al. 2010; Stiles 1995). Further, individual species in many other groups, such as E. jugularis in the Polytmini clade (Temeles et al. 2005) and Oreotrochilus estella in the Lophornithini clade (Bleiweiss 1999) have also been shown to exhibit significant sexual dimorphism of both bill size and bill curvature. Together, these patterns have been interpreted as evidence that hummingbirds generally display sexual shape dimorphism in their bill morphology. However, in contrast to this general pattern, we found that nearly all species in the Mellisugini clade (29 of 32 species examined) displayed no significant dimorphism in bill shape.

One possible explanation for these patterns is that hummingbird species in species-poor environments may have increased intraspecific competition, as the lack of interspecific competitors would allow the sexes to utilize distinct niches that would otherwise be occupied by congenerics. Indeed, this scenario would facilitate niche separation between the sexes, which would provide possible divergent selection on bill morphology between the sexes. While this hypothesis may hold true for A. colubris, which breeds allopatric with respect to other hummingbird species, neither S. scintilla nor M. minima live in species-depauperate communities. Therefore, this hypothesis alone is insufficient to explain the few instances of sexual shape dimorphism displayed in Mellisugini. Interestingly however, S. scintilla is the smallest hummingbird in its range (Wood 1983) and M. minima is the second smallest hummingbird species in Trochilidae (Bird 2004). Thus, niche segregation may still be a major mechanism driving the evolution of sexual dimorphism in these taxa, as these tiny hummingbird species may not have selective pressures because larger species are not able to feed from the resources the smallest hummingbirds can feed from.

Finally, it is of interest to examine the observed patterns in the Mellisugini in light of their phylogenetic placement within the Trochilidae. Phylogenetically, Mellisugini are a recently diverged lineage nested deeply within Trochilidae (Fig. 4), and are part of a radiation that includes the evolution of several species of neotropical migrants with nearly all species within this lineage displaying little to no sexual shape dimorphism in their bills. By contrast, lineages more basal to the Mellisugini display strong patterns of bill shape dimorphism, as well as bill size dimorphism. Therefore, the available data suggest the hypothesis that both bill size dimorphism and bill shape dimorphism arose early in the diversification of Trochilidae, and that the lack of sexual shape dimorphism presently displayed in the Mellisugini lineage is a derived trait.
https://static-content.springer.com/image/art%3A10.1007%2Fs11692-012-9206-3/MediaObjects/11692_2012_9206_Fig4_HTML.gif
Fig. 4

Current phylogeny of Trochilidae indicating presence of sexual size dimorphism throughout all Trochilidae, presence of sexual shape dimorphism throughout all Trochilidae and the loss of sexual shape dimorphism in the Mellisugini clade

Taken together, the evolutionary changes in patterns of sexual size and shape dimorphism observed in Mellisugini suggest that the trends of sexual dimorphism in the Trochilidae are far more varied than was previously believed. It is possible that a combination of environmental and evolutionary factors leads to these patterns of sexual dimorphism, and further analyses examining phenotypic and environmental variation in light of phylogenetic history may reveal further insight into the underlying mechanisms driving the evolution of sexual dimorphism in the bills of hummingbirds.

Acknowledgments

We thank S.M. Binz, M.B. Manes, A.W. Worthington, N.M. Valenzuela, A.N. Pairett, A. Alejandrino, A.J. Krause, A. Kraemer, R. Literman, Badenhorst, G. Rivera, A. Kaliontzopoulou, J.M. Serb, N.A. Gidaszewski, N. Navarro, and one anonymous reviewer for comments, and E.J. Temeles, R.M. Zink, A. Cardini and one anonymous reviewer made valuable suggestions on previous versions of the manuscript. We thank the United States National Science Foundation for partial financial support through grants DEB-1118884 (to DCA) and National Science Foundation Graduate Research Fellowship DGE0751279 (to CMB), the Society for the Study of Evolution Rosemary Grant Award (to CMB) and American Museum of Natural History Collections Grant (to CMB). Finally, we thank the many institutions (standard abbreviation given in parentheses), curators and collection managers who provided specimens, especially P. Sweet and G. Barrowclough (AMNH), S. Rogers (CM), C. Dardia and K. Botswick (CUMV), J. Woods (DMNH), D. Willard (FMNH), K. Garrett (LACM), J. Trimble (MCZ), C. Witt (MSB), C. Cicero (MVZ), P. Unitt (SDNHM), J. Hinshaw (UMMZ), C. Angle and D. James (NMNH), R. Corado (WFVZ), and K. Zyskowski (YPM), and K. Roe (NHM at ISU).

Copyright information

© Springer Science+Business Media New York 2012