ABSTRACT

Heliconia (Heliconiaceae), a distinctive and ecologically important genus within the Zingiberales, comprises approximately 200 species, most of which are native to the Neotropics. There, heliconias are primarily pollinated by hummingbirds. Heliconia-hummingbird pollination systems are increasingly well understood. To make research within these systems more valuable and to make any confident statements regarding co-evolution, they must be studied within a phylogenetic context. DNA sequence data will be used to reconstruct the phylogeny of Heliconia subgenus Heliconia. Morphological floral characteristics will then be studied along the phylogeny to make inferences about co-evolution with hummingbirds.

INTRODUCTION

Heliconia subgenus Heliconia was chosen as a study group because it contains two species that are part of an intricate and well-studied pollination system. Heliconia bihai and H. caribaea are two species native to the Lesser Antilles islands, where their primary pollinator is the purple-throated carib hummingbird (Eulampis jugularis). Male and female hummingbirds show sexual dimorphism in beak shape. This dimorphism corresponds to the flowers they visit. This ecological association was correlated with certain trends in floral evolution. We seek to determine if similar trends exist within the entire subgenus, suggesting pollinator specialization may be a major driving force behind speciation.

MATERIALS AND METHODS

PHYLOGENY RECONSTRUCTION

• DNA was extracted from leaf material.
• Five molecular loci from two genomes were amplified via the polymerase chain reaction and subsequently sequenced: ETS, ITS (nuclear ribosomal), rpb2 (nuclear), psbA-trnH and trnLF (chloroplast).
• Aligned sequences were analyzed using three different phylogeny reconstruction methods: unweighted parsimony (PAUP*), maximum likelihood (PAUP*), and Bayesian inference (mrbayes).
• Support for topologies was determined using bootstrap analysis (ML and parsimony) and posterior probabilities (Bayesian).

MORPHOLOGICAL ANALYSES

• Nine quantitative measurements relating to pollination biology were taken from 255 FAA-fixed, ethanol stored Heliconia flowers representing 38 ingroup taxa and two outgroup taxa.
• Color data was ascertained from photographs taken in the field.
• Statistical correlations between characters were determined using regression statistics using the mean of each species.
• Ancestral states were reconstructed using squared change parsimony.

RESULTS

PHYLOGENY RECONSTRUCTION

A five-marker (3798 base pair) molecular phylogeny of 28 Heliconia subgenus Heliconia species and four representative outgroup Heliconia species was constructed. Relationships were estimated using Bayesian inference of phylogeny as implemented in the program mrbayes. Values at nodes represent support in the form of posterior probabilities.

TAXONOMY

Sections within subgenus Heliconia as defined by current taxonomy (based on morphology) are highlighted. Section Tortex is nearly monophyletic, as is section Heliconia; Section Farinosae is polyphyletic, splitting into two main clades. Overall, itseems that morphology and molecular data are not contradictory and further molecular work will aid in the revision of Heliconia taxonomy.

FLOWER COLOR

Flower color is normally regarded as a trait that can be easily modified between and even within species, and thus is not expected to display a high degree of phylogenetic signal. However, distal flower color [i.e., the color of the flower outside of the bract and visible to pollinators] is more conserved in Heliconia than most characters. This high degree of signal suggests that floral color is under stabilizing selection and is not a character driving speciation events.

TRAIT CORRELATION
The ancestral state reconstruction of corolla length shows an encouraging trend: corolla length tends to increase upon speciation events (10 of 13 steps on this tree). The ecological implication of this result is that speciation tends to be associated with pollinator specialization as hummingbirds with short bills are excluded from visiting more derived, longer flowers. The ancestral state reconstruction of nectar chamber length shows that evolutionary changes tend to shorten this organ. This is also indicative of an ecological specialization because pollinators will stay faithful to the heliconias they specialize on without the reward of copious amounts of nectar.
These results are similar to the population-level relationship between corolla length and nectar chamber length observed in H. bihai and H. caribaea, where the pollination system is highly specialized and well-documented. A plot of nodal values of ancestral state reconstructions for nectar chamber length (y-axis) and corolla length (x-axis) the negative relationship between the two characters throughout the evolution of Heliconia subgenus Heliconia.

CONCLUSION

The results of this study have given insight into relationships between species of Heliconia that have previously been hypothesized based only on morphological data. Furthermore, this study has built on our knowledge of Heliconia floral evolution and may have important implications for the study of co-evolution with hummingbird pollinators. The phylogenetic perspective suggests that certain extant members of Heliconia subgenus Heliconia may have become more specialized in their pollinator interactions than from when they split from a common ancestor.

FUTURE DIRECTIONS

A project centered around the reconstruction of a nine-gene molecular phylogeny for the entire genus Heliconia is being carried out as a joint project between UC Berkeley and the Smithsonian. The phylogeny, as well as further analysis of the morphological data collected for Heliconia subgenus Heliconia, will be the basis for an undergraduate honors thesis. Floral data may be collected for the entire genus.

Additionally, results from this study provide many jumping points for studying the process of evolution within Heliconia and between plant and pollinators in a broader sense. Undoubtedly, the authors will pursue at least some of them in the future .

ACKNOWLEDGMENTS

I’d like to thank John Kress and Chelsea Specht for being wonderful mentors to me throughout this project. And, of course, this research project would not have been possible without generous funding provided by the NMNH Office of the Director and the Nathan and Violet David Foundation.