Specialization of flowers: is floral orientation an overlooked first step?
Article first published online: 7 MAY 2009
© The Authors (2009). Journal compilation © New Phytologist (2009)
Special Issue: Plant adaptation - following in Darwin's footsteps
Volume 183, Issue 3, pages 502–506, August 2009
How to Cite
Fenster, C. B., Armbruster, W. S. and Dudash, M. R. (2009), Specialization of flowers: is floral orientation an overlooked first step?. New Phytologist, 183: 502–506. doi: 10.1111/j.1469-8137.2009.02852.x
- Issue published online: 17 JUL 2009
- Article first published online: 7 MAY 2009
At the very inception of pollination biology, both floral symmetry and floral presentation received attention from a functional perspective (Sprengel, 1793). Sprengel suggested that floral symmetry (radial or bilateral) dictates insect movement in the flower, such that bilateral symmetry results in more directed and predictable insect movement. By contrast, floral presentation (where ‘vertical’ is facing upwards; ‘horizontal’ is roughly parallel to the ground; and ‘pendant’ is facing downwards; Fig. 1a–c, respectively), hereafter referred to as floral orientation, was thought to be more closely associated with protection from the elements and inflorescence architecture. Sprengel emphasized that nectar and pollen in horizontal and pendant flowers on spikes are protected from the rain, whereas flowers facing horizontally also provide a more visible display when packaged in a spike or raceme than if projected upwards (for example, Lamiaceae). Flower orientation as an adaptive response to abiotic factors has been documented, for example protection from rain (Tadey & Aizen, 2001; Huang et al., 2002; Aizen, 2003; Ushimaru et al., 2006; Sun et al., 2008), and the regulation of heat load in the flower, as a consequence of either water conservation (Patiño et al., 2002) or, in the case of alpine plants, as a pollinator reward (Hocking & Sharplin, 1965; Kevan, 1975; Totland, 1996) or enhanced environment for pollen germination (Galen & Stanton, 2003). However, how floral orientation might directly influence the approach of a pollinator to a flower has received little attention, despite the historical and contemporary interest in the latter in relation to floral symmetry.
The recognition that zygomorphy dictates pollinator movement within a flower, and therefore potentially pollination precision (Sprengel, 1793, Darwin, 1885), has led to considerable attention being devoted to the exploration of its functional and evolutionary significance (for example, Gomez et al., 2006). Reviews have discussed the relationship of symmetry patterns with plant diversification (Donoghue et al., 1998; Endress, 1999; Sargent, 2004) and plant–pollinator interactions (Stebbins, 1974; Giurfa et al., 1999; Fenster et al., 2004). Curiously, early workers recognized the potential adaptive significance of the horizontal orientation of flowers as a mechanism to ensure consistent pollinator directionality in approach to, and behaviour on, the flower. Robertson (1888a,b,c) recognized that a pollinator will approach an upward-facing flower from any direction, whereas a pollinator can approach a horizontally oriented flower essentially from one direction and in one position [the pollinator is upright, as pointed out by Sprengel (1793)]. Indeed, Ushimaru and Hyodo (2005) demonstrated that floral orientation can strongly influence pollinator directionality in the zygomorphic flowers of Commelina communis. Pollinators were more likely to contact the anthers and simultaneously touch the stigmas when C. communis was in its normal horizontal position than when the flowers were tethered vertically towards the sky. This occurs because the pollinators approach the flower from many different directions when tethered upright, and thus the zygomorphic flowers only impose directionality of pollinator movement when facing horizontally, their natural position (see also Berg, 1960). Many plant species pollinated by hovering pollinators also present their flowers horizontally or in a pendant fashion, suggesting that this floral orientation may have adaptive significance in terms of pollination precision.
The question remains, however, whether floral orientation alone, independent of floral symmetry, influences pollinator directionality and the consistency of pollinator movement. Directionality of pollinator movement to and within a flower may ultimately contribute to a plant's overall reproductive success in terms of efficient export and receipt of pollen (Inouye et al., 1994). This is because consistent directionality of the pollinator's movement allows the opportunity for the plant to respond evolutionarily by clustering its reproductive parts within the flower. This clustering may increase both the accuracy and precision of pollen transfer (sensu Armbruster et al., 2004; Hansen et al., 2006). Accuracy is increased by the better correspondence of pollen placement with the position of stigma contact with the pollinator, and vice versa. Clustering of stamens or stigmas results in contact with the pollinator in a smaller area and lowers the variance in pollen placement and stigma contact, hence increasing the precision of pollination. Because this is advantageous only if pollinators position themselves consistently as they enter or land on a flower, a more thorough understanding of the adaptive floral features that enforce consistent directionality in the approach of a pollinator to a flower is needed.
In this article, we investigate the role of floral orientation, vertical vs. horizontal vs. semi-pendant, in determining the direction of approach and orientation of a pollinator to radially symmetrical flowers. Hence, we investigate how floral orientation alone may contribute to more precise pollination by making the pollinator's approach and orientation to a flower more consistent. We measured the directionality of pollinator visits to artificial flowers (see Fenster et al., 2006), made to resemble hummingbird-pollinated Silene virginica (Caryophyllaceae; Fenster and Dudash, 2001), in response to different floral orientations. We present evidence that radial or actinomorphic flowers that are presented horizontally promote consistent, restricted directionality of movement and positioning by hummingbirds compared with vertically presented flowers.
Materials and Methods
Silene virginica has radially symmetrical, horizontally oriented flowers that are red, scentless and tubular with relatively copious and dilute nectar (Reynolds et al., 2009). They are almost exclusively pollinated by the ruby-throated hummingbird, Archilochus colubris (Fenster & Dudash, 2001; Reynolds et al., 2009). We induced hummingbirds to participate in a choice experiment by providing a hummingbird feeder in an open area surrounded by forest, and then removing the feeder and presenting artificial flowers in an array. The feeder was oriented close to the ground, similar to the plants in nature, somewhat analogously to flowers with their corolla tube openings oriented horizontally. The experiments were conducted at Mountain Lake Biological Station (37°22′32″N latitude, 80°31′20″W longitude). The construction of the artificial flowers has been described in Fenster et al. (2006). To ensure that nectar did not influence the choice of flowers, all artificial flowers were filled with 200 µl of 23% sugared water, c. 10–15 times the amount of nectar normally found in S. virginica flowers, but of the same sucrose concentration. To ensure that differences in nectar reward did not influence hummingbird approach, the experiment was halted and all artificial flowers were refilled at the very first sign that the nectar was depleted from any artificial flower in the array.
The experiment consisted of 12 artificial flowers arranged 0.4 m apart with three on each side of the perimeter of a square. The flowers, schematically illustrated in Fig. 1, were oriented vertically (floral tube of the flower facing the sky; Fig. 1a), horizontally (flower parallel to the ground; Fig. 1b) and semi-pendantly (flower oriented c. 45° downward from the horizontal; Fig. 1c). The position of a flower within each side of the array was random. Thus, there were four replicates (one replicate per side of the array) for each floral orientation in a given array and trial. After each observation period or trial, the position of a particular flower orientation was re-randomized for each side of the array. The visitation patterns of hummingbirds were observed for four observation periods or trials of 20 min each in one afternoon in 2005, and for five afternoons, two 60-min observation periods per day, in 2007. In 2005, 10 hummingbird individuals were observed in the array at a single instance, and up to four hummingbirds were observed in a single instance in 2007. Thus, hummingbird approach to the artificial flowers represents the behaviour of at least 10–14 different ruby-throated hummingbirds across 2 yr. Ideally, hummingbirds should have been identified by individual, with each bird used as a replicate for testing the effect of flower orientation on bird approach direction. Consequently, our inference on how floral orientation affects all hummingbird approach decisions depends on our assumption that each visit represents a sample of behaviours that all hummingbirds would exhibit if presented with the different floral orientation treatments, perhaps resulting in an inflation of the Type I error rate.
Because pollinators generally prefer to remain upright throughout their visit, approaches are expected to be from any direction in the hemisphere above a vertically oriented flower, but most likely in any of the four compass directions on the plane perpendicular to the tube opening (Fig. 1a, A and C) (as opposed to ‘dive-bombing’ from above: Fig. 1a, B). The direction of visitation to the vertical flowers was quantified in the context of the arrangement of the array. Thus, a hummingbird's approach to a vertical flower was quantified as 0° (south), 90° (east), 180° (north) and 270° (west), where each of the angles represents directionality independent of the position of the observer. If the hummingbird approached the flower at an intermediate angle, it was assigned to the closest major category of direction it represented. The assignment of directionality to only four cardinal directions makes our results conservative, as we have reduced the possible directionality options to only four. For horizontally or semi-pendantly oriented flowers, we noted the position of the hummingbird's forehead relative to the petal in the 12 o’clock or 90° position, again corresponding to an upright approach to the flower by the hummingbird (Fig. 1).
The extent of consistent directionality by the hummingbird visitor to a particular floral orientation was assessed by χ2 analysis, where random orientation of the hummingbird was the expected frequency (all four approach angles equally used by the hummingbirds).
Across the 2 yr of the study, a total of 471 hummingbird visits was observed and, of these, 224, 122 and 125 visits were to vertically oriented, horizontally oriented and semi-pendant flowers, respectively. The smaller numbers of observations to horizontally oriented and semi-pendant flowers was not a result of lower overall visitation rates to these treatments. Rather, because consistent directionality was observed (> 99% of all visits were in the same direction) for the horizontal and semi-pendant artificial flowers, observer attention was focused on the approach behaviour of the hummingbirds to the vertically oriented flowers, resulting in their greater sample size.
Of the 224 visits to vertically oriented flowers, all were on the plane perpendicular to the flower tube entrance (bird position portrayed in Fig. 1a, A and C). The visits came from all compass directions with 61, 51, 55 and 57 hummingbird visits approaching the flowers from the 0°, 90°, 180° and 270° angles, respectively. This represents a random directional approach to the upright flower by the hummingbirds (χ2 = 0.2143, d.f. = 3, P = 0.9505). As a consequence, the birds would have contacted fertile parts and corolla from any side. By contrast, all 122 visits to the horizontal flowers were by hummingbirds that approached the flower en face and hovered in a corresponding upright position, and 124 of 125 visits to the semi-pendant flowers were also en face and upright. These birds contacted the flowers in a consistent fashion and would have contacted the same side or part of the sexual parts and corolla if the flowers themselves were oriented consistently relative to vertical (as is nearly always the case in real flowers). In the one exceptional visit, the hummingbird approached the semi-pendant flower with its wings tilted towards the sky and ground. This exceptional approach would have resulted in contact with the sexual parts of the flower at a different location on the head relative to the upright visits to the horizontal and semi-pendant flowers.
Bilateral symmetry in flowers is commonly viewed as enhancing pollination accuracy and precision (Armbruster et al., 2004) by ‘forcing the pollinator to occupy a certain position’ (Faegri & van der Pijl, 1979: p. 62; see also Darwin, 1885). This perspective is reflected in the observation that radially symmetrical flowers often have their anthers and stigmas diffusely distributed in the flower because pollinator approach to the flower can be from more than one direction (Neal & Anderson, 2005). In this article, we demonstrate that the pattern of floral symmetry is not the sole factor influencing pollinator approach. A simple change in flower orientation from vertical to horizontal can dramatically change pollinator approach and orientation relative to the floral parts from random to directional. The important consequence is that pollinator approach to a flower is now consistent and predictable. When a hummingbird approaches a vertical-facing flower, it can approach on a plane orthogonal to the tube entrance from any direction. By contrast, for a horizontal or semi-pendant flower, the hummingbird is prevented from approaching from the back and sides of the flower, and is oriented consistently in front of the flower by its predilection to remain upright (Warrick et al., 2005; also evident from our personal observations of animals nearly always flying in an upright position). Consequently, the hummingbird approaches the flower in one direction, en face, or on the same plane as the flower tube opening, and always with its head in the 12 o’clock position. As a result, it will consistently contact certain floral parts, for example, the upper petals with its forehead and the lower petals with its chin. If the sexual parts were in the middle, they would probably contact the forehead consistently, as the bird usually approaches the flower from just slightly below the horizontal plane. Clearly, the next step is to verify that pollen transfer precision is increased by going from the vertical to the horizontal position.
If a pollinator's movement and orientation are consistent, natural selection should favour a corresponding shift in the position of the reproductive parts of the flower, such that more pollen is removed and deposited per visit by the pollinator, that is, increasing the accuracy of pollination (Armbruster et al., 2004). Therefore, the reproductive parts should also evolve to contact the pollinator either on the dorsal surface (forehead) or the ventral surface (chin), in concert with an evolutionary transition from vertical- to horizontal-facing flowers. In fact, S. virginica exhibits radially symmetrical petals with zygomorphic placement of its reproductive parts, such that its stigmas are clustered in the centre of the floral tube and its anthers emerge at the 12 o’clock position as predicted. This arrangement is referred to as moderate zygomorphy (Neal et al., 1998; Endress, 1999), and it has been hypothesized to enhance the accuracy of pollen transfer, as anthers and/or stigmas can be clustered in parts of the flower that are consistently contacted by the pollinator (Vogel, 1996). Clustering of reproductive parts decreases the variance of placement, hence simultaneously increasing the precision and accuracy of pollination. We hypothesize that the consistent upright approach by hummingbirds to horizontally oriented flowers imposes selection on the clustering of reproductive parts within the flower. By contrast, vertical flowers need to place their anthers and stigmas diffusely throughout the flower or risk ‘losing out’ on approaches from some directions. For example, the closely related, completely radially symmetrical and vertically oriented flowers of S. caroliniana present their reproductive parts throughout the entire circumference of the vertical floral tube, as predicted. The comparison of these two closely related species prompts us to hypothesize that the orientation of flowers is indeed the first evolutionary step towards the evolution of zygomorphy. Horizontal orientation sets the selective stage for the evolution of slight zygomorphy in sexual parts, followed by the evolution of full zygomorphy. It should be possible to test this hypothesis with phylogenetic comparative approaches in species-rich groups that exhibit repeated origins of zygomorphy, for example Boraginaceae, Solanaceae and Lamiales (Reeves & Olmstead, 2003).
Large-bee-pollinated and horizontally oriented flowers are typically packaged in racemes or spikes (Sprengel, 1793), which impose much greater directionality of bee movement within the inflorescence relative to other floral arrangements, as bees typically approach a raceme or spike inflorescence from the bottom and subsequently forage upwards (Jordan & Harder, 2006). Thus, for large-bee-pollinated species with horizontal flowers arranged in a spike or raceme, consistent directionality of pollinator movement can be expected at both the flower and inflorescence level. We note that an inflorescence architecture that facilitates movement by walking or crawling between flowers would diminish the constancy of pollinator approach.
Downward-facing or semi-pendant artificial flowers also impart strong directionality to pollinator movement. However, downward-facing flowers are thought to be at a disadvantage, because they are less easily seen by pollinators (Sprengel, 1793). There is limited but intriguing evidence from both insects and birds of reduced visitation rates to downward-facing flowers compared with vertically oriented flowers, as expected (Fulton & Hodges, 1999; Giurfa et al., 1999; Ushimaru & Hyodo, 2005; Ushimaru et al., 2006). If upward-facing flowers are more likely to be seen by pollinators (at least in flat habitats), increased pollen transfer accuracy and precision associated with imposed pollinator directionality may offset the attractiveness disadvantage.
Although the results of this study were anticipated by earlier workers (Sprengel, 1793; Robertson, 1888a,b,c), we are not aware of any experimental work that has confirmed our intuition of how flower orientation can consistently direct pollinator movement independent of floral symmetry. Clearly, more attention should be focused on floral orientation and its role in the enhancement of plant fitness through directing pollinator visitation behaviour to a flower. The manipulation of floral orientation effects on pollinator directionality should be tested across a wide arrange of floral morphologies, beyond the tubular flowers tested here, and for other pollinators, including hovering and nonhovering visitors.
The authors are grateful to R. Reynolds and C. Williams for help in the initial stages of this project, E. Nagy, H. Wilbur and E. Brodie for logistical support and encouragement, P. Endress, J. Hereford, S. Marten-Rodriguez, M. Rausher, R. Reynolds H. Wilbur and two anonymous reviewers for comments on previous versions of the manuscript, and funding from NSF DEB 0108285 to C. Fenster and M. Dudash.
- 2003. Down-facing flowers, hummingbirds and rain. Taxon 52: 675–680. .
- 2004. Floral integration and modularity: distinguishing complex adaptations from genetic constraints. In: PigliucciM, PrestonK, eds. Phenotypic integration: studying the ecology and evolution of complex phenotypes. Oxford, UK: Oxford University Press, 23–49. , , , .
- 1960. The ecological significance of correlation pleiades. Evolution 14: 171–180. .
- 1885. The various contrivances by which orchids are fertilized by insects. London, UK: John Murray.
- 1998. Phylogeny and the evolution of flower symmetry in the Asteridae. Trends in Plant Sciences 3: 311–317. , , .
- 1999. Symmetry in flowers: diversity and evolution. International Journal of Plant Sciences 160 (6 Suppl.): S3–S23. .
- 1979. The principles of pollination ecology. Oxford, UK: Pergamon. ,
- 2004. Pollination syndromes and floral specialization. Annual Review of Ecology, Evolution and Systematics 35: 375–403. , , , , .
- 2006. Nectar reward and advertisement in hummingbird-pollinated Silene virginica (Caryophyllaceae). American Journal of Botany 93: 1800–1807. , , , .
- 2001. Spatiotemporal variation in the role of hummingbirds as pollinators of Silene virginica (Caryophyllacaeae). Ecology 82: 844–851. , .
- 1999. Floral isolation between Aquilegia formosa and Aquilegia pubescens. Proceedings of the Royal Society London B 266: 2247–2252. , .
- 2003. Sunny-side up: flower heliotropism as a source of parental environmental effects on pollen quality and performance in the snow buttercup, Ranunculus adoneus (Ranunculaceae). American Journal of Botany 90: 724–729. , .
- 1999. Floral symmetry and its role in plant–pollinator systems. International Journal of Plant Sciences 160 (6 Suppl.): S41–S50. , , .
- 2006. Natural selection on Erysimum mediohispanicum flower shape: insights into the evolution of zygomorphy. American Naturalist 168: 531–548. , , .
- 2006. On adaptive accuracy and precision in natural populations. American Naturalist 168: 168–181. , ,
- 1965. Flower basking by arctic insects. Nature 206: 215. ,
- 2002. Why does the flower stalk of Pulsatilla cernua (Ranunculaceae) bend during anthesis? American Journal of Botany 89: 1599–1603. , ,
- 1994. A model and a lexicon for pollen fate. American Journal of Botany 12: 1517–1530. , , , .
- 2006. Manipulation of bee behavior by inflorescence architecture and its consequences for plant mating. American Naturalist 167: 496–509. , .
- 1975. Sun-tracking solar furnaces in high arctic flowers: significance for pollination and insects. Science 189: 723–726. .
- 2005. Are ‘mating systems’‘breeding systems’ of inconsistent and confusing terminology in plant reproductive biology? Or is it the other way around? Plant Systematics and Evolution 250: 173–185. , .
- 1998. Floral symmetry and its role in plant–pollinator systems: terminology, distribution and hypotheses. Annual Review of Ecology, Evolution and Systematics 29: 245–373. , ,
- 2002. The ecological role of orientation in tropical convolvulaceous flowers. Oecologia 130: 373–379. , ,
- 2003. Evolution of the TCP ene family in Asteridae: cladistic and network approaches to understanding regulatory gene family diversification and its impact on morphological evolution. Molecular Biology and Evolution 20: 1997–2009. , .
- 2009. Pollinator specialization and pollination syndromes of three related North American Silene. Ecology, in press. , , , , , .
- 1888abc. Zygomorphy and its causes. I–III. Botanical Gazette 13: 146–151, 203–208, 224–230.
- 2004. Floral symmetry affects speciation rates in angiosperms. Proceedings of the Royal Society London B 271: 603–608. .
- 1793. Das entdeckte geheimniss der natur im bau und in der befruchtung der blumen. Berlin: Vieweg. .
- 1974. Flowering plants: evolution above the species level. Cambridge, MA, USA: Harvard University Press. .
- 2008. Multifunctional bracts in the Dove tree Davidia involucrata (Nyssaceae: Cornales): Rain protection and pollinator attraction. American Naturalist 171: 119–124. , , , .
- 2001. Why do flowers of a hummingbird-pollinated mistletoe face down? Functional Ecology 15: 782–790. , .
- 1996. Flower heliotropism in an alpine population of Ranunculus acris (Ranunculaceae): effects on flower temperature, insect visitation, and seed production. American Journal of Botany 83: 452–458.
- 2005. Why do bilaterally symmetrical flowers orient vertically? Flower orientation influences pollinator landing behaviour. Evolutionary Ecology Research 7: 252–260. ,
- 2006. Flowers adaptively face down-slope in 10 forest-floor herbs. Functional Ecology 20: 585–591. , ,
- 1996. Christian Konrad Sprengel's theory of the flower: the cradle of floral ecology. In: LloydDG, BarrettSCH, eds. Floral biology: studies on floral evolution in animal-pollinated plants. (Transl. P. Haase.) New York, NY, USA: Chapman & Hall, 44–62.
- 2005. Aerodynamics of the hovering hummingbird. Nature 435: 1094–1097. , , .