The hidden language of flowering plants: floral odours as a key for understanding angiosperm evolution?


We live in a world filled with flowering plants and insects, with both groups showing enormous species diversity. However, explaining this diversity has been a major challenge. One reason suggested for such species diversity is that angiosperms have established mutualistic relationships with different pollinator types, leading to co-evolution of both groups (Stebbins, 1970; Thien et al., 2000). Chemical communication via floral fragrances has been shown to play a central role in attracting pollinators (Knudsen et al., 2006), and it is reasonable to hypothesize that floral volatiles played a key role in early angiosperm evolution (Gottsberger, 1988; Thien et al., 2000). This is supported by observations that strong floral fragrances are a characteristic feature of many extant taxa from basal angiosperm lineages and that they play an important role for pollinator attraction in most species in these groups (Ervik & Knudsen, 2003; Silberbauer-Gottsberger et al., 2003). Unfortunately, compared with other plant groups (e.g. orchids) our knowledge on the odour chemistry of basal angiosperm lineages is still limited (Knudsen et al., 2006). Nevertheless, recent publications on the floral odours of basal groups have revealed interesting insights, for example, into the diversity of floral fragrances (Bernhardt et al., 2003; Goodrich et al., 2006; Teichert, 2008). The same is true for the paper on floral fragrances of Asimina (pawpaw) and Deeringothamnus species (Annonaceae) by Katherine Goodrich and Robert Raguso in this issue of New Phytologist (pp. 457–469). First, the data on 10 species of the North American groups of Annonaceae fill an important gap in our knowledge, as Asimina and Deeringothamnus are the only temperate genera of this otherwise tropical family (Doyle & Le Thomas, 1997). Second, the fragrances they found, typical fermentation odours, are extremely unusual among angiosperm flowers. Four of the Asimina species, with wine red-coloured flowers (‘maroon-phenotype’), emit odours that resemble the scent of decaying organic matter. The scent blends comprise chemicals that are similar to the emissions of baker's yeast (e.g. acetic acid, ethyl acetate, ethanol, 3-methyl-1-butanol; see Goodrich et al., 2006) but they also contain compounds such as amino acid-derived aldoximes and nitriles. The scent composition suggests that these flowers might attract saprophilous flies and beetles, but more detailed investigations of the pollination biology of these species are needed. The four Asimina species and the two Deeringothamnus species with white flowers emit floral volatiles that indicate a generalist pollination system. The floral scent of D. pulchellus was dominated by sweet-smelling benzenoid compounds normally found in flowers pollinated by moths. But there are two more striking findings in this paper: the high diversity of chemicals emitted from the flowers (Goodrich & Raguso detected 272 compounds in 11 species of only two genera), and the complex spatial and temporal odour patterns for female and male ontogenetic stages. Their study is also a fascinating example illustrating how complex and dynamic floral odours in basal angiosperms can be.

We are starting to understand that plants are quite flexible in their odour chemistry and that this flexibility is linked with their evolutionary success.

Generalist vs specialist modes of pollination in basal angiosperm groups and the role of beetles

The hypothesis that beetles pollinated the early angiosperms dates back to Faegri & van der Pijl (1979, p. 51) and their ‘mess and soil’ pollination mode concept in basal angiosperm taxa – unspecialized (primitive) flowers utilize unspecialized (primitive) insects (beetles) as pollinators. Flowers with a ‘mess and soil’ pollination mode often bear many anthers and deposit large amounts of pollen over the body surface of flower visitors while not offering nectar or lipids as a reward (Faegri & van der Pijl, 1979). It is true that beetles are a dominant pollinator group in many magnoliids, including Annonaceae, with up to nine beetle families involved in pollination (Bernhardt, 2000). However, beetles are interestingly not the dominant group of flower visitors in the first three branches of the angiosperm phylogenetic tree (ANITA grade, with ∼201 species) and it now seems that they represent a derived syndrome in the ANITA-grade plants (Thien et al., 2009). Furthermore, the magnoliids are dominated by beetle pollination and are regarded as specialized systems where flowers are specifically adapted to beetle pollinators (Silberbauer-Gottsberger et al., 2003). According to Thien et al. (2009), the affiliations of ANITA-grade species with their flower visitors suggest that Diptera are the strongest candidates as the first pollinators of early angiosperms.

With more than 2500 species, Annonaceae are one of the most diverse families of the basal groups of flowering plants and the family has played an important role in discussions of the origin and evolution of angiosperms (Doyle & Le Thomas, 1997; Gottsberger, 1999). Annonaceae flowers usually contain no nectar but the flowers provide pollinators with nutritious floral tissues, pollen, and in some species, shelter inside a pollination chamber. In many, particularly night-active, Annonaceae species, thermogenesis enhances floral odour emission (Gottsberger, 1999). It has been estimated that c. 90% of the Annonaceae species studied so far seem to be adapted to beetle pollination (Teichert, 2008) with a differentiation into species pollinated by small beetles and large beetles (Silberbauer-Gottsberger et al., 2003). While many of the species in the ancestral genera of the family (e.g. Anaxagorea) are pollinated by small beetles, pollination by large beetles is a specialized condition within the family and was probably acquired late in its evolution (Gottsberger, 1999; Silberbauer-Gottsberger et al., 2003). However, other pollinator groups interacting with Annonaceae, such as thrips, flies, cockroaches and perfume-collecting euglossine bees (Gottsberger, 1999; Silberbauer-Gottsberger et al., 2003; Teichert, 2008 and references therein), also provide evidence for other specialized pollination systems within the Annonaceae.

Differentiation of Annonaceae pollination systems in odour space

Before the present study carried out by Goodrich & Raguso, floral fragrances of only a few, mostly Neotropical, Annonaceae had been analyzed (Jürgens et al., 2000; Teichert, 2008; Teichert et al., 2008). Based on the floral scent data of six Annonaceae species with small beetle pollination syndrome, which emit a wide range of different compounds in different compound classes, Jürgens et al. (2000) wondered whether the flower scent reflects the opportunistic behavior of the flower visitors. It is possible that Nitidulidae beetles, which normally live on, and eat, rotten bark and fruits, are attracted by a wide range of compounds that all might be indicative of fruits and/or rotten bark (Jürgens et al., 2000).

Although there are only limited data available on the floral volatiles of Annonaceae, visualization of the combined data sets in odour space gives some evidence that different pollination types are represented by clearly separated odour clusters (Fig. 1). The analysis was based on nine different genera of the Annonaceae (21 species), including the data of Goodrich & Raguso (this issue of New Phytologist), Jürgens et al. (2000), Teichert et al. (2008), Teichert (2008) and unpublished data (A. Jürgens & G. Gottsberger). The analysis revealed that floral fragrances reflect both phylogenetic constraints/relationships and the pollination biology of the species (Fig. 1), also showing that unrelated species, associated with similar pollinator types, occupy the same odour space.

Figure 1.

Visualization of floral fragrance similarities and (hypothesized) pollinators of 21 Annonaceae species. Nonmetric multidimensional scaling was based on Bray–Curtis similarities of 150 identified scent components. Asimina and Deeringothamnus species (triangles) and Annona glabra are from Goodrich & Raguso (2009) (open circle) (this issue of New Phytologist). The ‘maroon-phenotype’ of Asimina flowers (closed triangles) and the ‘white-phenotype’ (open triangles) are indicated. Unonopsis stipitata and Anaxagorea prinoides are from Teichert et al. (2008) and Teichert (2008); Anaxagorea brevipes, A. dolichocarpa, Xylopia aromatica, X. benthamii, Rollinia insignis and R. mucosa are from Jürgens et al. (2000); Guatteria foliosa and Duguetia asterotricha are from A. Jürgens & G. Gottsberger (unpublished data). The two-dimensional stress value is 0.15; ANOSIM Global R = 0.776; P < 0.01.

For species with sweet fruity odour blends, a green–yellow colouration, and relatively small pollination chambers that are mainly pollinated by small beetles (Anaxagorea, Duguetia, Guatteria, Rollinia and Xylopia; Silberbauer-Gottsberger et al., 2003), the odour space is relatively large. These species probably represent different chemical strategies by using hydrocarbon esters (Anaxagorea, Guatteria), naphthalene (Rollinia), benzenoid compounds (Xylopia), or monoterpenoids (Duguetia, Rollinia) to attract small beetles. Furthermore, the differences in odour chemistry might also reflect the taxonomic heterogeneity of the pollinating beetles. Although mainly pollinated by Nitidulidae, other beetles could play a role as additional pollinators (e.g. Staphylinidae, Chrysomelidae) and, in Xylopia aromatica, thrips also seem to play an important role. However, these are all subsumed under the ‘small beetle pollination syndrome’.

Close to the small beetle pollination syndrome we find Annona glabra, a species pollinated by relatively small Chrysomelidae or Curculionidae beetles (Gottsberger, 1999) where Goodrich & Raguso found 3-pentanyl acetate and several monoterpenoids as major scent compounds. The spicy floral odour of Unonopsis stipitata shows similarities in the odour composition with that of A. glabra and emits many of the same monoterpenoids. However, U. stipitata is pollinated by perfume-collecting male euglossine bees and its scent is dominated by several monoterpenoids, particularly trans-carvone oxide, which was only present in this species (Teichert, 2008). Interestingly, trans-carvone oxide has been found several times in euglossine-pollinated plants of other unrelated plant families (e.g. orchids) and it has been suggested that this compound might be the key attractant for euglossine bees (Whitten et al., 1986).

The Asimina and Deeringothamnus species analyzed by Goodrich & Raguso form a cluster separate from the other genera. Within this cluster we find a chemical differentiation of Asimina flowers with the ‘maroon-phenotype’, emitting fermented/decaying scents (suggesting a mimicry-based pollination strategy), from another cluster with flowers of the ‘white-phenotype’, emitting pleasant scents (suggesting honest signaling and a reward-based pollination strategy) (Fig. 1). It is known that fermenting odours elicit strong physiological and behavioral responses in fruit flies and some beetles (Stensmyr et al., 2003) and it seems likely that scent is mainly responsible for pollinator attraction in Asimina species with yeasty scent.

Floral fragrances as a key innovation promoting pollinator shifts and diversification in flowering plants

Stebbins (1970) wrote that ‘more specialized vectors are attracted to flowers by a variety of stimuli, of which scent may be even more important than either shape or colour’. Advances in our understanding of floral volatiles and their role in pollination have been enormous in the last 20 yr with the emergence of technological breakthroughs for floral volatile analysis. In 1993, Knudsen et al. compiled floral scent data on 441 species, and listed 700 compounds; in a recent update, Knudsen et al. (2006) listed 991 species and 1791 compounds. A rough calculation shows that the average number of new (identified) compounds per species has increased from 1.6 (in 1993) to 1.8 (in 2006). Part of the increase can be attributed to the development of more sensitive technologies for sampling plant volatiles and better ways of compound identification (e.g. via database systems). Nevertheless, it seems that a plateau regarding the chemical diversity of angiosperm flowers has not yet been reached. Odour diversity of the Annonaceae species investigated so far seems to be relatively high compared with that of all angiosperms listed in Knudsen et al. (2006). The Annonaceae analyzed so far, representing 2.2% (22 species of 991) of the angiosperms, comprise 8.4% (150 compounds of 1791, not including the many unidentified compounds listed by Goodrich & Raguso) of the total odour diversity found in flowering plants.

A key question for the coming decade will be whether angiosperm diversification is correlated with a potential for rapid changes in the floral scent composition. We are starting to understand that plants are quite flexible in their odour chemistry and that this flexibility is linked with their evolutionary success. The scent composition of flowers can be very complex, sometimes containing > 50 compounds per sample. In some cases, single key odour signals (like trans-carvone oxide in Unonopsis stipitata) might be responsible for attracting specific pollinators (Teichert et al., 2008); however, the common situation in most flowering plants is that several compounds together attract a diverse array of different flower visitors. Not all are necessarily effective pollinators and other floral features often function as filters. In the Annonaceae, other factors, such as the size of the pollination chamber, floral colour and the reward system, need to be considered to understand the evolution of this group. In situations with multiple flower visitors and multiple volatiles emitted from a flower, compound X might be an attractant for pollinator A only, whereas compound Z might be an attractant for pollinator A plus B (or B only), etc. Thus, floral scent can be seen as a multichannel signal, and small evolutionary changes in the scent composition might open (or close) signal channels for flower visitors that might affect the plant's fitness. Furthermore, insects are often opportunistic in their behavior, and the insect, as a signal receiver, may respond to a range of different chemicals. In such a scenario different chemical strategies may evolve, attracting the same type of pollinator but exploiting different aspects of its olfactory spectrum.


I would like to thank Karl Duffy, Steve Johnson, Adam Shuttleworth and Taina Witt for valuable comments and discussion on the manuscript.