Floral scent in a whole-plant context: moving beyond pollinator attraction


*Correspondence author. E-mail: rar229@cornell.edu

The natural world is awash with odours – blends of volatile organic compounds (VOCs) – whose chemical structures and composition convey information to organisms endowed with the sensory capacity to perceive them. Pheromones and other chemical signals provide specific information that facilitates kin recognition, mate acquisition, social interactions and defence across the tree of life (Meinwald & Eisner 2008). Even more general volatile cues, such as CO2, convey context-dependent information about habitat and host or food quality, illness and microbial decay (Guerenstein & Hildebrand 2007). The terrestrial fossil record teems with evidence of chemical communication and warfare, in the form of animal olfactory appendages (Strausfeld & Hildebrand 1999; Labandeira 2002) and plant essential oil glands (Fahn 2002; Krings, Taylor & Kellogg 2002). This evidence, combined with our knowledge of cycads, water lilies and other ‘living fossils’(Thien, Azuma & Kawano 2000; Terry et al. 2007), attests to the antiquity of volatile communication between plants, their mutualists and antagonists.

Pleiotropic links between floral advertisement and plant defence

Over 20 years ago, Pellmyr & Thien (1986) framed a provocative hypothesis about the origins of floral scent. They first noted that basal angiosperms and cycads almost universally emit potent odours (also present in their leaves) that deter generalist herbivores, and are pollinated by primitive insect groups that use flowers as rendezvous sites for mating and nursery sites for larval development on floral tissues. These authors therefore proposed that incidental transfer of pollen by weevils, primitive moths and other florivores provided the opportunity for selective pressures to co-opt and fine-tune defensive floral emissions into the first pollinator attractants. Recent studies of the nursery pollination-like mutualism between Epichloë fungal endophytes and Botanophila flies provide independent support for this scenario, as the volatile fly attractant is a compound with known antifungal properties (Steinebrunner et al. 2008). The putative defensive origins of volatile floral attractants embody the dynamic tension between plant defence and reproductive ecology – the ‘survivorship’ and ‘fecundity’ components of Darwinian fitness – and highlight the need for studies that consider the functional ecology of volatile communication within a whole-plant context.

Modern flowering plants use a broader spectrum of sensory signals to advertise to pollinators (Raguso 2004), including bright pigments and distinct patterns and shapes that form dense visual displays. The whole-plant approach has become de rigeur for researchers studying the functional ecology of such traits. Over a decade ago, Strauss & Armbruster (1997) edited a special feature in Ecology that explored the impact of natural enemies (herbivores, flower or seed predators, pathogens) on plant–pollinator interactions, and examined pleiotropic relationships between plant defence and allocation to reproduction. At least two ground-breaking themes emerged: (1) pollinator-mediated selection on floral traits often pales in comparison to the effects of enemies and (2) historical associations with specific enemies may pre-adapt or constrain the evolution of floral phenotype (e.g. nectar toxins or floral pigments) in certain lineages because of pleiotropic effects on defence pathways (Strauss & Armbruster 1997). This paradigm shift resulted in a wealth of studies on the impact of herbivory and/or defence chemistry on floral phenotype (Armbruster 2002; Irwin et al. 2003; Frey 2005), and set the stage for further synergism between the fields of pollination ecology and plant defence. Unfortunately, this potential has not been realized for the volatile components of plant reproductive and defensive phenotypes. The field of tri-trophic interactions among plants, herbivores and carnivores (e.g. parasitoid wasps) has witnessed explosive growth in the past decade (Dicke 2009), due in part to the maturation of chemical techniques enabling the rigorous analysis of plant VOC emissions induced by natural enemies (Tholl et al. 2006). At the same time, pollination ecology has moved in directions that have de-emphasized the importance of floral display and sensory biology in mediating plant–pollinator dynamics (see Ollerton & Waser 2006). A growing number of studies (e.g. Jordano, Bascompte & Olesen 2003) utilize the mathematical tools of food webs to explore community level patterns of plant–pollinator interactions. Such studies tend to favour methodological, historical, phenological and demographic explanations for why certain plants are visited by certain pollinators (Vázquez et al. 2009). Remarkably, there remains almost no commerce between these related fields; there is a clear need to bridge the ‘chemistry gap’ between studies of pollination ecology and volatile-mediated plant defence.

A volatile subject: variation in floral messages and their fitness consequences

If plant VOCs represent an ancient chemical language, what do we know about the vocabulary and dialect of floral scent? Advances in chemical analysis have taught us much about its orthography. At least 1700 floral VOCs with diverse chemical structures and properties have been identified, although a small number of compounds (e.g. limonene, benzyl alcohol) are present in most fragrance blends (Knudsen et al. 2006). Similarly, molecular, biochemical and neurobiological approaches have revealed the etymology and syntax of floral VOCs, from their biosynthetic origins (Gang 2005; Dudareva et al. 2006) to their coding and perception by insects (Carlsson & Hansson 2006; Smith, Wright & Daly 2006). We know less about the processes that shape variation in the information content, complexity and dynamic range of fragrance; the nuanced messages and conspiratorial intrigues conveyed through the language of scent. For example, why do night-blooming Datura flowers emit more than 70 compounds when only nine are required to attract their hawkmoth pollinators (Riffell et al. 2009)? Why must epiphytic Stanhopea orchids broadcast milligrams of fragrance to attract male euglossine bees, when terrestrial, sexually deceptive Ophrys orchids can attract male andrenid bees with a microgram ‘whisper’ (see Dötterl & Vereecken 2009)?

The diversity of potential messages conveyed by floral scent remains poorly explored, perhaps due to the assumption (discussed by Dobson 2006) that its primary message (pollinator attraction) is ‘come hither’. Although their lower concentrations and ephemeral emissions make them harder to study than floral odours, plant VOCs elicited by herbivore damage are already better understood than floral scent from a communication standpoint, and have been more rapidly integrated into the concept structure of their field, the study of plant defence and tri-trophic interactions. Induced plant VOCs attract parasitoids and carnivores to combat the herbivores responsible for plant damage below and above ground (Rasmann et al. 2005; Dicke 2009), discourage further attacks by herbivores (De Moraes, Mescher & Tumlinson 2001) and prime defensive responses in distant shoots of the same plant (Heil & Silva-Bueno 2007). Interestingly, when plant VOCs and defence compounds are included in the kinds of network analyses favoured by pollination ecologists, they are found to contribute to the structure and strength of interaction webs between trophic levels (Bukovinszky et al. 2008; Poelman et al. 2009).

The diverse and specific messages conveyed by herbivore-induced plant VOCs suggest the need to examine the fitness effects of volatile blends outside the context of pollinator attraction. Two recent studies highlight the potential of floral VOCs to convey other kinds of messages. Terry et al. (2007) demonstrated that Macrozamia cycads modulate pollen flow by dramatically increasing scent concentration in conjunction with a thermogenic burst, expelling pollen-feeding thrips pollinators from the male cones with hot, toxic odours. The dose makes the poison in this system, whereas the same VOCs, at one thousand fold lower concentrations, attract the evicted, pollen-bearing thrips to female cones. Kessler, Gase & Baldwin (2008) used gene-silencing techniques to experimentally manipulate the production of attractive (benzyl acetone) and repellent (nicotine) nectar VOCs in flowers of a self-compatible tobacco, Nicotiana attenuata. The experimental manipulations fundamentally altered the quantitative and qualitative interactions between tobacco flowers and their visitors by shifting the residence times and inter-plant movement of pollinators, and by blocking florivory by caterpillars and nectar larceny by carpenter bees. Indeed, both benzyl acetone and nicotine were required to optimize male (pollen export) and female (capsule size) fitness functions in N. attenuata. These studies indicate that ‘move along’ is as important a message as ‘come hither’, that floral VOCs are shaped by selection from enemies as well as mutualists, and that fragrance can play direct and indirect roles in plant gene flow and mating strategies.

Scope and impact of the Special Feature

Each invited paper in this Special Feature addresses at least one of the lacunae highlighted above: the need to explore the full range of fitness consequences of floral VOC variation, and the need to understand how plants use floral VOCs to consolidate their reproductive and defensive imperatives. The first three papers discuss the nature of signal variation in floral scent and its fitness consequences. Wright & Schiestl (2009) lead off with a behaviourally oriented paper that poses a rhetorical question: if flower colour is sufficient to attract generalized pollinators (social bees) to generalized flowers, why should flowers produce scent at all? The answers they provide boil down to receiver bias and signal reliability, and have direct consequences for floral constancy and disruptive selection on floral phenotype.

In the second paper, Ashman (2009) asks a more specific question: what happens to floral scent when male and female functions are separated? Bateman’s Rule predicts that when female fertility is resource limited, males should compete more vigorously for mates (via pollinator attraction) than females. Do gender dimorphic plants express such different strategies in floral VOC composition or emission rates? Furthermore, how does intersexual mimicry, in which flowers of one sex lack nectar or pollen rewards, affect the timing and composition of floral scent? Ashman uses the available data to explore such questions by contrasting hypotheses and predictions that cover multiple levels of analysis.

Whitehead & Peakall (2009) then build upon themes from the preceding papers by examining the impact of scent-mediated pollinator movement and gene flow on population genetic structure. In cases where floral VOCs function as specific pollinator attractants, does scent composition affect species boundaries, by either facilitating or suppressing hybridization and by attracting pollinators whose movement patterns determine levels of outcrossing? The authors’ discussion features research on sexually deceptive Ophrys and Chiloglottis orchids, for which population-level variation in floral VOCs has profound consequences for gene flow across species boundaries.

The last three papers of the Special Feature describe how floral volatiles fit into a larger, whole-plant context. Having already explored how fragrance is used (or not) in deceptive flowers that mimic the scents of nectar, rotting flesh or female insects, we now read about carnivorous plants and the extent to which they use ‘floral’ odours to trap flower-visiting insects in modified leaves. Jürgens, El-Sayed & Suckling (2009) collected VOCs from a spectrum of passive to active trapping pitchers, sundews and fly traps, and provide proof of concept (also see Di Giusto et al. 2008) that olfactory floral mimicry is a viable mode of prey attraction for carnivorous plants. Avenues for further study include the potential for pollinator–prey conflict: do differences in the VOCs of flowers and traps allow carnivorous plants to avoid eating their pollinators?

In the fifth paper, we switch from offence to defence as Willmer et al. (2009) describe how plants prevent ants from pilfering their flowers and harassing their pollinators. Ants can be restrained by physical barriers (floral fences and moats), bribed (and distracted) by extra-floral nectar or chemically deterred by floral VOCs such as E,E-α-farnesene, a pollen odour in flowers of the obligate myrmecophyte, Vachellia (Acacia) seyal-fistula. The authors combine chemical analysis with detailed behavioural and comparative studies, discuss the relative differences amongst co-blooming African acacias in VOC composition and ant deterrence.

The final paper confronts the issue of conflicting selective pressures between herbivores and pollinators on floral chemistry. Kessler & Halitschke’s (2009) study addresses the prediction that induced plant defences compromise pollination, either by drawing resources away from floral display and reward, or by repelling pollinators through the systemic production of deterrent compounds in flowers. The authors used wild tomato species and controlled levels of herbivory to test these ideas, contrasting whole plant and floral VOCs with non-volatile defensive phenolics, and using visits by bumble bees to gauge the relative attractiveness of wounded vs. control plants.

A number of themes emerging from this Special Feature suggest the potential for conceptual growth. For example, several authors discuss the importance of signal honesty for pollinator learning, constancy and intersexual mimicry. How might herbivore damage affect the relationship between floral advertisements and rewards (see Theis, Kesler & Adler 2009)? In a similar vein, phylogenetic and environmental factors contribute to the inherent noisiness of floral VOCs (see Majetic, Raguso & Ashman 2009); to what extent does such variation impact a plant’s interactions with mutualists and enemies? These papers also contain some provocative ideas, such as Wright and Schiestl’s unorthodox view of floral constancy, which emphasizes the salience of olfactory learning over traditional arguments for constrained short-term memory, and Kessler and Halitschke’s suggested link between herbivory and the breakdown of self-incompatibility in tomato flowers. It should be clear from the papers collected in this Special Feature that floral volatiles are relevant to the entire spectrum of plant–pollinator–herbivore interactions, that population genetic structure and community composition may in part be driven by odour-mediated processes, and that current methods are sufficient for non-chemists to incorporate plant VOCs into their own research programmes.