Diversity and distribution of floral scent.
Abstract Introduction Collection Methods and Materials Chemical Classification Plant Names and Classification Floral Scent at Different Taxonomic Levels Population-Level Variation Species-and Genus-Level Variation Family-and Order-Level Variation Floral Scent and Pollination Biology Floral Scent Chemistry and Biochemistry Floral Scent and Evolution Floral Scent and Phylogeny Acknowledgments Literature Cited Appendex I: Distribution of Floral Scent Compounds at the Level of Plant Family Appendix II: List of Taxa Appendix: III: List of References from which the Data Included in Appendices I and II are Extracted
In 1993, one of us published a checklist of floral scent constituents collected with headspace techniques (Knudsen et al., 1993). Since then a large number of papers dealing with floral odors in many different groups of seed plants have been published. The number of known compounds present in floral scents has increased from 730 to 1719, and an updated checklist has been requested for a long time. The present review is intended to meet this demand. However, apart from adding to the list of known compounds, we also felt the need to summarize the present knowledge of floral scent in a broader context. Advances in phylogenetic research during the last decade have produced a much better understanding of relationships of angiosperms at the family level and below (e.g., Soltis et al., 1999; Hilu et al., 2003). Consequently, the distribution of various scent compounds is here compared to phylogenies of angiosperms orders based on recent analyses of DNA sequence data. In addition, new techniques and research have provided much needed insight into the evolution of floral scent at lower taxonomic levels and its role in evolutionary processes at the population level, and the chemistry of floral scent in general. A general scheme with references to different kinds of investigations with regard to floral scent is presented in Table I.
The present review is based on a total of 268 papers including data on floral scent composition. Many of these are cited by Knudsen et al. (1993); others were found through searches in Biological Abstracts (BA; http://www.biosis.org) and Chemical Abstracts (CA; http://www.cas.org). The last searches in BA and CA were made in October 2002. Some studies that have come to our attention at a later date have also been cited, but are not included in the Appendices.
Only studies in which headspace techniques in a broad sense were used to collect the floral scent are included in the present review. This delimitation was made in order to better reflect the world of scent that flower visitors and pollinators actually encounter, for example, during their search for food or a site for breeding, mating, egg laying, or floral scent collecting. This means that studies using other methods, such as solvent extraction or steam distillation, are not included because the compounds detected may not necessarily be released from intact flowers under natural conditions.
A relational database, SCENTbase, was developed in FileMaker Pro 5.5 to handle all compound, taxon, and reference data, and to generate the lists. In addition, a separate database was created to analyze various properties of the data at different chemical and taxonomic levels, and to export matrices for further phylogenetic analyses. A limited version of SCENTbase is searchable online at http://www2.botany.gu.se/SCENTbase.html.
Collection Methods and Materials
The preferred method to collect floral scent is through dynamic headspace adsorption onto various artificial porous polymers and activated charcoal. This method is usually easy to apply in the field by enclosing flowers or floral structures in glass vials or polyacetate bags, and then creating a stream of air with a battery-operated pump. The samples are extracted by a solvent in the field, or stored until extracted in the laboratory, and then analyzed by coupled gas chromatography/mass spectrometry. Thermal desorption, in which the sample is desorbed directly onto a gas chromatographic column using heat, is less commonly employed. Recently, static headspace adsorption or solid-phase microextraction (SPME) with direct thermal desorption has been employed in studies of floral scents either alone or as a supplement to traditional dynamic headspace sampling. Detailed reviews of collection methods have been published by Raguso and Pellmyr (1998) and Agelopoulus and Pickett (1998). It is important to keep in mind that methodological differences among the reviewed studies may be responsible for inconsistencies in the compiled information. If present, they should only affect the accuracy of interspecific comparisons, which are not the focus of this review.
The total number of investigated species and subspecific taxa is 991, representing 90 families and 38 orders of seeds plants. Two of the orders are gymnosperms, with a total of eight species investigated. Although representatives of most of the orders recognized by the APG II (2003) have been analyzed, the usually broad circumscriptions of these orders and the estimated total number of angiosperm species clearly indicate that the sample is still very small. In addition, the sample is very uneven. Orchidaceae is by far the best-collected family, followed by several families (e.g., Araceae, Arecaceae, Cactaceae, and Nyctaginaceae) with a reasonably large proportion of taxa sampled (Table II). Most families, however, remain poorly sampled or still not investigated with regard to floral scent chemistry.
In this checklist, our goal has been to classify floral scent compounds on the basis of their mode of biosynthesis, because this should be of greatest value for making comparisons with phylogenetic hypotheses. Although the broad outlines of biosynthesis are known for most large groups of floral volatiles, the actual routes by which most individual compounds are formed are still uncertain. Thus, it is especially difficult to arrange compounds biosynthetically at lower levels of classification. We divide floral scent constituents into seven major compound classes based on inspection of their structures and knowledge of the major pathways of plant secondary metabolism. These classes are aliphatics, benzenoids and phenylpropanoids, C5-branched compounds, terpenoids (including monoterpenes, sesquiterpenes, diterpenes, and irregular terpenes), nitrogen-containing compounds, sulfur-containing compounds, and a class of miscellaneous cyclic compounds. Within classes, subdivisions have been made primarily on the basis of chain length or skeletal type and secondarily according to functional groups (for details, see the legend of Appendix I). While this classification may not accurately reflect biosynthetic considerations, it does align compounds by their chemical properties and help in locating them in the list.
The accuracy and thoroughness with which individual chemical compounds have been identified varies among studies. Sometimes the precise isomers present were not determined, and often the enantiomeric composition was not resolved. The specific information on isomers and enantiomers reported in the original literature has been preserved in the checklist. However, when these details are not specified, the information is also missing here. It is important to be aware of this problem when comparing distributions of individual compounds among taxa. For example, "lilac alcohol," which is listed under monoterpenes/acyclic/alcohols, may include one or more of the four possible isomers A, B, C, and D.
Plant Names and Classification
No taxonomic evaluations of the usage of names have been made by us. Consequently, a certain taxon may occur under two or more different names. However, since we focus on the distribution of chemical compounds or compound classes at the level of plant family and plant order, such mistakes do not interfere with the general patterns observed.
We follow a recent classification of the angiosperms (APG II, 2003). However, to retain as much information as possible, we have adopted the narrower circumscription options in that work, recognizing Agavaceae, Amaryllidaceae, Fumariaceae, Hemerocallidaceae, Hyacinthaceae, Linnaeaceae, Rhizophoraceae, Ruscaccae, and Valerianaceae as separate families.
The phylogenetic trees are based on those presented by APG II (2003).
Floral Scent at Different Taxonomic Levels
Studies of floral scent variation at individual or population levels including more than just a few individuals are still scarce. However, some studies have been made on selected species of Apiaceae (Tollsten & Ovstedal, 1994), Arecaceae (Knudsen, 2002), Caryophyllaceae (Dotterl et al., 2005), Magnoliaceae (Azuma et al., 2001), Mimosaceae (Pettersson & Knudsen, 2001), Orchidaceae (Whitten & Williams, 1992; Moya & Ackerman, 1993; Tollsten & Bergstrom, 1993), Fumariaceae (Olesen & Knudsen, 1994) and Pyrolaceae (Knudsen, 1994). In general, species relying mainly on one class of pollinator have higher similarity among populations than more generalized species pollinated by several orders of insects (Knudsen, unpubl.). Furthermore, the variation in floral scent composition may be clinal, with similarity among populations being negatively correlated with distance (Knudsen, 2002). However, this picture may be obscured by species with deceptive pollination systems in which variation both within and between populations is extreme, disrupting the associative learning of visiting insects (Moya & Ackerman, 1993). This may also be the case if floral scent types rarely or never influence pollinator behavior and reproductive success. Scent types caused by random genetic drift could then be maintained in natural populations (Azuma et al., 2001).
SPECIES- AND GENUS-LEVEL VARIATION
Many studies have shown that the floral scent composition usually differs among closely related species. This has been shown in works on Nicotiana (Raguso et al., 2003), Narcissus (Dobson et al., 1997), and Silene (Jurgens et al., 2002), as well as on various genera of Arecaceae (Knudsen, 1999a; Knudsen et al., 2001), Lecythidaceae (Knudsen & Moil, 1996), Magnoliaceae (Thien et al., 1975), Nyctaginaceae (Levin et al., 2001), and Orchidaceae (Gregg, 1983; Whitten & Williams, 1992; Kaiser, 1993; Barkman et al., 1997). In addition, floral scent composition may vary as much among genera within a family as among species of a given genus. Thus, unless monospecific, taxa above the species level usually cannot be characterized by a distinct floral scent profile.
FAMILY- AND ORDER-LEVEL VARIATION
Most compound classes are present in most orders of flowering plants (Figs. 1, 2), suggesting that the distribution of floral scent compounds is not phylogenetically constrained, at least not at this broad scale. Monoterpenes are found in all orders, and three, four and six orders out of 38 lack aliphatics, benzenoids and sesquiterpenes respectively. However, in most of the orders lacking these classes only one genus has been investigated and the result thus likely reflects lack of sampling. Diterpenes are rare in floral scents and are only reported from four species belonging to three plant orders, reflecting their low volatility.
[FIGURES 1-2 OMITTED]
C5-branched compounds are lacking in 12 orders, based on information from one or a few genera from each order. The same patterns were found for irregular terpenes and miscellaneous compounds, suggesting low sampling frequency rather than true absence.
Nitrogen- and sulfur-containing compounds are present in 63% and 39% of the plant orders, respectively. However, orders lacking nitrogen-containing compounds are all sampled at low frequency, while orders sampled at both high and low frequency lack sulfur-containing compounds. The latter indicates that the occurrence of volatile sulfur-containing compounds is probably restricted within seed plants.
Twelve compounds occur in more than 50% of the families investigated (Table III), and these should probably, even with investigation of additional taxa, be regarded as the most common compounds in floral scents. Most of these are monoterpenes, that is, limonene (71%), (E)-[beta]-ocimene (71%), myrcene (70%), linalool (70%), [alpha]-pinene (67%), [beta]-pinene (59%); followed by benzenoids, that is, benzaldehyde (64%), methyl 2-hydroxybenzoate (57%), benzyl alcohol (56%), 2-phenyl ethanol (54%); a sesquiterpene, caryophyllene (52%); and an irregular terpene, 6-methyl-5-hepten-2-one (52%). The widespread distribution of these substances suggests that they may have other roles in flowers in addition to pollinator attraction (see Floral Scent and Evolution).
Floral Scent and Pollination Biology
During the exploration of the function of floral scents in pollination, it is necessary to sort out which chemicals constitute the actual signals and which are historical and/or biosynthetic artifacts (Raguso, 2001). Historical and biosynthetic artifacts may contain important phylogenetic information, either alone or in combination. However, some compounds, such as thio- and isothiocyanates, may be both a signal and a historical artifact.
Sulfur-containing compounds are found mainly in plants pollinated by bats and carrion flies. The latter group of pollinators is restricted mainly to the Araceae and a few other families of basal angiosperms. Thus, the presence of sulfur-containing compounds may be constrained phylogenetically, or these compounds may have developed across plant orders as a way of attracting a group of pollinators with similar scent preferences, as has been suggested for microbats in the New World tropics (Knudsen & Tollsten, 1995; Helversen et al., 2000).
Floral Scent Chemistry and Biochemistry
The constituents of floral scent comprise a large variety of generally lipophilic plant products with molecular masses less than 300. By definition, all constituents have high enough vapor pressures at atmospheric pressure and normal growth temperatures to allow significant rates of release into the air. The main classes of floral volatiles are the same as those reported to be released from other parts of the plant. Vegetative organs also release many different volatile constituents, especially after herbivore damage (Pichersky & Gershenzon, 2002). However, the overall diversity of vegetative volatiles is less than that found in floral scent. When headspace collections are made from intact plants, it is in fact difficult to readily distinguish between volatiles released from flowers and those released from nearby vegetative organs. Nevertheless, the sites of synthesis and storage of floral volatiles seem to be consistently associated with the floral organs themselves. Several genes and enzymes involved in the biosynthesis of floral scent compounds are expressed locally in the tissues of the petals, stigma, and style (Dudareva & Pichersky, 2000; Kolosova et al., 2001).
In the last few years, the biochemistry and molecular biology of floral scent formation have become major topics of research in several laboratories. Studies have centered around three species, Antirrhinum majus (Scrophulariaceae), Clarkia breweri (Onagraceae), and Rosa hybrida (Rosaceae), whose floral scents are dominated by terpenoid and benzenoid compounds. The biosynthesis of other classes of floral scent compounds has not yet been rigorously examined at the gene and enzyme levels, but the general routes can be inferred from investigations on similar or identical compounds in vegetative tissues.
Among the major categories of floral volatiles listed in Appendix I, the aliphatics are probably biosynthesized predominantly from fatty acids. For example, the abundant C6 and C9 aldehydes and alcohols are formed via lipoxygenase-catalyzed degradations of linolenic and linoleic acids (Hatanaka, 1999). Other compounds arise by oxidation of the double bonds or chain-shortening via [beta]-oxidation.
The class of benzenoids and phenylpropanoids is formed starting from the phenylpropanoid pathway, which begins with the deamination of phenylalanine. Biosynthesis of the benzenoids (C6-C1), the most widespread members of this group in floral scent, thus requires the loss of one or two carbon atoms from a phenylpropanoid precursor (C6-C3). However, the steps of this process are not yet clear (Jarvis et al., 2000). Alternatively, methyl 2-hydroxybenzoate (methyl salicylate) and similar benzenoids could arise from an intermediate of the shikimate pathway prior to phenylalanine (Wildermuth et al., 2001). Several late steps in the formation of benzenoid esters and ethers have been well characterized (Murfitt et al., 2000; D'Auria et al., 2002; Lavid et al., 2002).
The C5-branched compounds are probably derived from the amino acids valine, leucine, and isoleucine, but there is only a little direct evidence to date (Rowan et al., 1996). Although the basic isopentane carbon skeleton of this group is the same as that of the terpenoids, the oxidation patterns present make it very unlikely that any are terpenoid-derived.
The terpenes themselves are formed from C5 isopentanoid building blocks synthesized by the mevalonate or methylerythritol 4-phosphate (MEP) pathways (Gershenzon & Kreis, 1999; Rodriguez-Concepcion & Boronat, 2002). After assembly of the C5 units into prenyl diphosphate precursors, enzymes known as terpene synthases catalyze the formation of the basic terpene skeletons of monoterpenes (C10), sesquiterpenes (C15), or diterpenes (C20) (Bohlmann et al., 1998). The most well-studied enzyme of floral scent biosynthesis, linalool synthase, is a terpene synthase that converts geranyl diphosphate to linalool (Dudareva & Pichersky, 2000). The initial terpene synthase products, such as linalool, can be further modified to form other floral volatiles (Burkhardt & Mosandl, 2003; Shalit et al., 2003). The irregular terpenoids include cleavage products of carotenoids, such as the widespread ionones, and derivatives of smaller terpenoids (Winterhalter & Rouseff, 2002). For example, one member of the latter group, the widely-occurring C11 volatile, (E)-4,8-dimethyl-1,3,7-nonatriene, is believed to be formed by oxidative cleavage of a sesquiterpene (C15) precursor (Boland & Gabler, 1989).
Both the nitrogen- and sulfur-containing floral scent compounds are derived from amino acid metabolism. Indole, the most widely distributed member of these groups, is also one of the few whose biosynthesis is understood in plants. It is formed by direct cleavage of the tryptophan precursor, indole-3-glycerol phosphate. (Frey et al., 2000). The miscellaneous cyclic compounds are collectively of uncertain biosynthetic origin, although some are undoubtedly derivatives of fatty acids or amino acids.
In summary, the constituents of floral scent are drawn from nearly all of the major pathways of plant secondary metabolism. As these pathways are present in all plants, where they make a wide range of pigments, membrane constituents, cell wall components, hormones, and other signaling compounds, it is perhaps not surprising that the principal classes of floral scent compounds are so widely distributed in seed plants.
Floral Scent and Evolution
The primary function of floral scent in flowering plants is to attract and guide pollinators (Dobson, 1994; Raguso, 2001; Metcalf, 1987; Robacker et al., 1988; Williams, 1983). However, additional functions may be ascribed to the presence of volatile chemicals in flowers (reviewed by Pichersky & Gershenzon, 2002), including defense and protection against abiotic stresses. These additional functions may help explain some of the abundance and variety of different constituents detected. The possibility that flowers are chemically well defended against herbivores and pathogens is not surprising. By producing pollen and ovules for the next generation, flowers have a very high fitness value to the plant and must be protected accordingly. In addition, the attraction of insects for pollination could increase the risk of herbivory on floral structures, and floral tissues may therefore require relatively more protection from enemies. Representatives of all of the major classes of floral volatiles have been shown to have toxic or deterrent activity against microbes and herbivores (e.g., De Moraes et al., 2001; Friedman et al., 2002; Hammer et al., 2003). Certain floral volatiles could also have a physiological role of providing resistance to abiotic stress. For example, some of the monoterpenes found in abundance in flowers have also been shown to ameliorate high temperatures and reduce damage caused by oxidative stress (Delfine et al., 2000; Loreto et al., 2004). The most abundant floral sesquiterpene, caryophyllene, is very reactive with ozone (Bonn & Moortgat, 2003).
The defensive and physiological functions of floral volatiles may well predate the origin of the angiosperms (Pellmyr & Thien, 1986, Thien et al., 2000), and floral scent, especially that of pollen, may therefore constitute an ancient trait, already present in preangiosperms (Dobson & Bergstrom, 2000). A likely scenario suggests that some insects overcame the repellence of the floral chemicals and that pollination in early angiosperms was based on a meshing of the sexual life cycle of insects with that of plants, in which volatile floral chemicals served as mediating cues for rendezvous and mating sites (the flower) and food (primarily microspores) for pollinating insects. This hypothesis is supported by several observations: (1) floral scent is present in most extant basal angiosperms and in some nonangiospermous seed plants of Gnetales, Cycadales, and Pinales; (2) floral scent compounds similar to many general herbivore repellents are present; (3) all groups of ancient insects involved in pollination share phytophagy on nonangiosperms as an ancestral condition; and (4) a high percentage of flowers function as mating sites in extant basal angiosperms (Thien et al., 2000, 2003; Bernhardt et al., 2003).
In addition, thermogenesis may also have existed in preangiosperms, since this phenomenon occurs in cycads and is particularly common in basal angiosperms (Thien et al., 2000, 2003). Most thermogenic plants are beetle- or fly-pollinated. In beetles, heat produced by the plant may help to regulate body temperature. Because of this, Seymour and Schultze-Motel (1997) suggested that beetle pollination coevolved with thermogenesis and floral scent production in plants. Recently, Seymour et al. (2003) showed that heat is a reward to Cyclocephala colasi (Scarabaeidae: Cyclocephalini) beetles visiting Philodendron solimoesense (Araceae), the energy requirements being from 2.0 to 4.8 times lower inside the heated floral chamber than outside.
The correlation of high angiosperm diversity and the adoption of biotic pollination is undisputed. However, the mechanism or mechanisms responsible for this diversity are still not understood. Most likely a number of mechanisms, for example, coevolution of insect herbivores and plant chemical defenses and coevolution of seed dispersing animals and plants, have worked concomitantly to produce the diversity of animals and plants that we see today (see Gorelick, 2001, for a review and additional hypotheses).
Pollinators show different scent and color preferences, which may result in a certain degree of flower constancy. In combination with divergence in floral traits related to pollination (scent, color, flowering phenology, reward), these factors have led to reproductive isolation in present-day angiosperms (e.g., Gregg, 1983; Groth et al., 1987; Whitten & Williams, 1992; Dobson et al., 1997; Knudsen 1999a). The selective pressures probably have been especially strong in sympatric, coflowering species (Knudsen, 1999b; Schiestl & Ayasse, 2002). On the other hand, floral scents have converged in chemical composition even across plant orders in species sharing a suite of morphological and phenological characters adapting them to pollination by one particular group of pollinators (pollination syndromes), for example, by moths or bats (Miyake et al., 1998; Levin et al., 2001; Kaiser, 1993; Kaiser & Tollsten, 1995; Knudsen & Tollsten, 1993, 1995; Bestmann et al., 1997; Helversen et al., 2000; Raguso et al., 2003), or production has ceased, as in hummingbird-pollinated species in the neotropics (Knudsen et al., 2004). This contradiction suggests that floral scent evolution is influenced by several factors and that floral scent is best defined as a mosaic product of biosynthetic pathway dynamics, phylogenetic constraints, and balancing selection due to pollinator and florivore attraction (Raguso, 2001).
Floral Scent and Phylogeny
Seed plants are still poorly sampled with regard to floral scent compounds. In addition, the available information is also very uneven, with some groups being well sampled (e.g., 42% of all taxa investigated so far are orchids) and others poorly sampled or not at all (e.g., Ebenaceae, Marantaceae). Consequently, the present information on the distribution of floral scent compounds is of limited use for broad phylogenetic reconstructions. Nonetheless, during the course of this work a few analyses of orders and families with compound classes classified into broad as well as less-restricted groups were made (with 30 and 68 characters, respectively). For example, aliphatics were classified into groups with 8 or 25 characters based on the number of carbon atoms present. Both analyses resulted in the >25,000 shortest trees producing largely unresolved strict consensus cladograms, corroborating the supposition that floral scent data are of little value for phylogenetic estimates at high taxonomic levels. It is also evident that the few groups that do appear in the consensus cladograms are mere artifacts of the uneven data set. Order and family-level analyses of the same group characters using the APG tree (APG II, 2003) as a constraint exposed high levels of homoplasy, with a majority of the characters having consistency indices between 0.05 and 0.15.
A few studies have used the distribution of floral scent compounds for reconstructing plant phylogenies and to infer the evolutionary history of ecological relationships (Azuma et al., 1997, 1999; Williams & Whitten, 1999; Lindberg et al., 2000; Barkman, 2001; Levin et al., 2003). The outcome of most of these studies is that only the outermost branching pattern is consistent with phylogenetic trees obtained using either morphological or DNA sequence data. This outcome indicates that, in general, floral scent chemicals may be too evolutionarily labile to be useful for phylogenetic inference (Williams & Whitten, 1999; Barkman, 2001). However, this does not exclude the possibility that some floral scent chemicals are patterned phylogenetically at lower taxonomic levels (Levin et al., 2003; Barkman, 2001). Furthermore, all these studies have been based on all identified compounds in the floral scent blends, and it is possible that a selection of compounds may produce alternative interpretations.
|Printer friendly Cite/link Email Feedback|
|Author:||Knudsen, Jette T.; Eriksson, Roger; Gershenzon, Jonathan; Stahl, Bertil|
|Publication:||The Botanical Review|
|Date:||Jan 1, 2006|
|Previous Article:||The evolution of gender-biased nectar production in hermaphroditic plants.|
|Next Article:||Diversity and distribution of floral scent.|