American Journal of Botany 92(10): 1624–1631. 2005.
CHEMISTRY
AND GEOGRAPHIC VARIATION OF FLORAL
SCENT IN
YUCCA FILAMENTOSA (AGAVACEAE)1
GLENN P. SVENSSON,2,5 MICHAEL O. HICKMAN, JR.,2 STEFAN BARTRAM,3 WILHELM BOLAND,3 OLLE PELLMYR,4 AND ROBERT A. RAGUSO2 2 Department of Biological Sciences, University of South Carolina, Coker Life Sciences Building, 700 Sumter St., Columbia, South Carolina 29208 USA; 3Max-Planck-Institute for Chemical Ecology, Hans Kno¨ll Strasse 8, D-07745 Jena, Germany; and 4Department of Biological Sciences, University of Idaho, P.O. Box 443051, Moscow, Idaho 83844-3051 USA
We identified volatiles from the floral headspace of Yucca filamentosa using gas chromatography and mass spectrometry and analyzed floral scent composition and variation among populations pollinated by different yucca moth species. Twenty-one scent compounds were repeatedly identified and most could be categorized into two major classes: (1) homoterpenes derived from the sesquiterpene alcohol nerolidol and (2) long chain aliphatic hydrocarbons. Two biosynthetic pathways are thus responsible for the majority of floral volatiles in Y. filamentosa. The homoterpene E-4,8-dimethylnona-1,3,7-triene, which is released systemically by higher plants upon herbivory, was the most abundant compound. Two di-oxygenated compounds not previously reported as floral compounds also were detected. No differentiation in floral scent was observed between populations pollinated by different yucca moths, nor was there any correlation between chemical distance and geographic distance among populations. The total release rate of volatiles differed significantly among populations, but not between populations with different pollinators. The combination of unique compounds and low variation in the fragrance blend may reflect highly selective attraction of obligate pollinators to flowers. The observed lack of differentiation in floral scent can putatively explain high moth-mediated gene flow among sites, but it does not explain conservation of odor composition across populations with different pollinators. Key words: Agavaceae; floral scent; geographic variation; pollination mutualism; pollinator specificity; Prodoxidae; Tegeticula; Yucca filamentosa.
A central component in the evolution of plant–pollinator associations is the production of sensory signals by flowers to attract pollinators. Plants use a variety of sensory cues to signal their presence and to guide animals to their flowers, including visual, olfactory, acoustic, and thermal information (Raguso, 2004). Volatile chemical compounds are critical for the attraction of insects in many pollination systems, and floral fragrances have been suggested to be a key factor in the diversification of angiosperms and pollinating insects (Pellmyr and Thien, 1986). However, the potential to study pollinatormediated selection on floral traits, such as fragrance, is often restricted when plants are pollinated by more than a single species, making it hard to distinguish what sensory signals drive coevolutionary processes in plants and the associated pollinators. This problem can be limited by analyzing systems that involve exclusive pollinators. Several studies have recently addressed the importance of floral sensory cues for attraction of exclusive pollinators in insect–plant associations involving deception by resource mimicry. In addition to elaborate morphological structures for pollen transfer, the odors mediating such interactions have been shown to be highly specific. For example, Australian deceptive orchids exclusively attract male wasps for pollination by mimicking the sex pheromone of conspecific females Manuscript received 25 February 2005; revision accepted 29 June 2005. The authors thank Gold Head State Park, Ocala National Forest, Forty Acre Rock Heritage Preserve, Heggie’s Rock Preserve, and the Archbold Biological Station for permission to perform odor collection. This research was supported by NSF grant DEB-0317217 to R.A.R., a Wenner-Gren Foundations postdoctoral fellowship to G.P.S., NSF grants DEB-0075944 and DEB-0075803 to O.P., and a Howard Hughes Medical Institute fellowship to M.O.H. 5 Author for correspondence (e-mail:
[email protected]) present address: Department of Ecology, Lund University, So¨lvegatan 37, SE-223 62 Lund, Sweden. 1
(Schiestl et al., 2003), and the compound mediating the interaction (2-ethyl-5-propylcyclohexan-1,3-dione) is unique to that system. Similarly, the Mediterranean dead-horse arum solely attracts carrion blowflies for pollination by producing the same blend of sulfur-containing compounds that is emitted from bird carcasses, which constitute important oviposition sites for these insects (Stensmyr et al., 2002). Obligate pollination mutualisms, in which seed parasites are exclusive pollinators of their host, also offer suitable models to study pollinator-mediated selection on floral scent. Such associations include fig wasps on figs (Weiblen, 2002), yucca moths on yuccas (Pellmyr, 2003), senita moths on senita cacti (Fleming and Holland, 1998), and Epicephala moths on Phyllanthaceae plants (Kato et al., 2003; Kawakita and Kato, 2004). Mechanisms facilitating the attraction of exclusive pollinators to the host flowers are crucial for the reproduction and survival of both participants in these associations. Thus, the sensory signals produced by plants in obligate pollination mutualisms are predicted to be highly specific, similar to what has been observed for resource mimicry systems (Raguso, 2003). Recent research has revealed the importance of odor cues in the fig–fig-wasp mutualism. Female wasps are attracted only to odors produced by figs of their own host and only to odors from receptive figs (Hossaert-McKey et al., 1994; Ware and Compton, 1994; Gibernau et al., 1998; Grison-Pige´ et al., 2002a). Although volatiles from headspace collection of receptive figs have now been chemically identified for more than 20 species of Ficus (Grison et al., 1999; Grison-Pige´ et al., 2002b), the specific compounds (or blends thereof) responsible for wasp attraction remain unknown in nearly all species. In contrast to the fig–fig-wasp association, the role of host odors in the second classic case of obligate pollination mutualism, involving yucca moths and yucca plants, has been far less studied. Yuccas (Agavaceae) are found in arid areas in
1624
October 2005] TABLE 1.
ET AL.—FLORAL SCENT OF
YUCCA FILAMENTOSA
1625
Sites and sample sizes of Yucca filamentosa populations with different yucca moth pollinators in southern USA in the present study. Site (state)
Nags Head (NC) Forty Acre Rock (SC) Due West (SC) Heggie’s Rock (GA) Eglin (FL) Jesup (GA) Gold Head (FL) Ocala (FL) Inverness (FL) Lake Placid (FL) a
SVENSSON
Latitude
35913009 34940005 34920000 33932030 30944007 31942029 29949034 29908000 28950009 27914000
Pollinator (ref.)a
Longitude
N N N N N N N N N N
75941026 80931040 82923016 82915005 86921020 81944032 81958013 81931000 82919049 81924000
W W W W W W W W W W
T. T. T. T. T. T. T. T. T. T.
yuccasella (1) yuccasella (1) yuccasella (1) yuccasella (1) yuccasella (2) cassandra (2) cassandra (3) cassandra (3) cassandra (3) cassandra (3)
N
7 8 8 6 9 9 11 10 9 10
(1) G. Svensson, personal observation, (2) Segraves et al. (2004), and (3) Pellmyr (1999).
North and Central America and rely exclusively on yucca moths (Tegeticula and Parategeticula) for pollination (Pellmyr, 2003). In the same way, the larvae of yucca moths can feed only upon maturing yucca seeds. A female yucca moth uses special maxillary appendages for pollen collection and deposition (Pellmyr and Krenn, 2002). She oviposits into the ovary and deposits pollen into the stigmatic cavity of a flower, thereby ensuring the presence of developing seeds for her offspring. All reproductive behaviors of moths (mating inside flowers, pollination, and oviposition) in all but one species (Powell and Mackie, 1966) take place after sunset when yucca flowers are open and fragrant. As nocturnal insects commonly rely on odor cues for mate and host location, yucca moths are predicted to use olfaction to find flowering yuccas. Here we report on the first step in a program to elucidate the role of floral odors in the yucca–yucca-moth mutualism, using the capsular-fruited Yucca filamentosa as a model. The plant is native to the southeastern coast of the United States, but has been naturalized to most parts of the eastern USA during the last two centuries (e.g., Pammel, 1925). Populations of Y. filamentosa on the Florida peninsula rely on the endemic Tegeticula cassandra for pollination, whereas populations in the northern parts of its range are pollinated by the distantly related T. yuccasella (Pellmyr, 1999). The aims of this study were to chemically characterize the floral scent of Y. filamentosa and to analyze whether populations pollinated by different yucca moths differ in floral scent, i.e., to test for pollinatorbased specialization of floral scent in this plant. Three alternative hypotheses were tested regarding the variation in floral scent observed among populations: (1) Variation among populations is structured based on pollinator species. (2) Variation among populations is at least partly explained by distance between populations. (3) Populations do not vary significantly in scent composition within the study range. MATERIALS AND METHODS The plant—Yucca filamentosa L. is native to coastal areas in the southeastern United States. From a basal rosette, 50–500 white, bell-shaped flowers are produced on a 1–3 m tall paniculate inflorescence. Flowers open in overlapping sequence along the inflorescence for 2 to 3 weeks during late May to mid July. Individual flowers generally are open and attractive to moths during a single night. Volatile collection—The floral scent of Y. filamentosa was collected using a dynamic headspace technique (Raguso and Pellmyr, 1998). Sampling was conducted from 2000–2400 hours, when maximum odor is released and yucca
moths are most active. Before the collection, the young flowers on each plant were counted to estimate mass-standardized release rates of volatile compounds (described later). A polyvinylacetate bag (406 3 444 mm) was wrapped around the inflorescence and sealed with a plastic tie. A glass cartridge (7 mm i.d.) filled with 100 mg of Super Q (Alltech Associates, State College, Pennsylvania, USA) adsorbent was connected to the bag. During odor collection, air was passed through the filter at a rate of 200 mL per min, using a PAS-500 personal air sampler (Supelco, Bellefonte, Pennsylvania, USA). After collection, the filter was eluted with 3 mL of hexane and the extracts were stored at 2188C. Empty bags were used as ambient controls to check for possible contaminants emitted from a bag itself. Prior to chemical analysis, extracts were concentrated to 75 mL under N2 and 5 mL of 0.03% toluene was added as an internal standard to each sample to enable estimation of release rates of compounds. Gas chromatography mass spectrometry (GC–MS) of compounds in floral extracts—Extracts were first analyzed using a Shimadzu GC-17A gas chromatograph equipped with a DB-5 column (30 m 3 0.32 mm i.d., 1 mm film thickness), linked to a Shimadzu QP5000 mass spectrometer. Helium was used as carrier gas at a velocity of 43 cm/s and injector temperature was 2708C. Column temperature was maintained at 508C for 2 min after injection and then linearly increased to 2758C at a rate of 108C/min. To obtain high resolution GC-MS data, additional analyses were conducted using an Agilent 6890 gas chromatograph (Agilent Technologies, Palo Alto, California, USA), equipped with an Alltech EC-5 column (30 m 3 0.25 mm i.d., 0.25 mm film thickness), and linked to a TOF mass spectrometer (GCT Micromass, Manchester, UK). Helium was used as a carrier gas at a flow of 1 mL/min, and injector temperature was set to 2608C. Column temperature was 508C and after injection, was increased to 1208C at a rate of 108C/min, then to 1908C at 38C/min, and to 2758C at 308C/min, which was maintained for 2 min. Compounds were identified by comparing mass spectra and retention times with those of reference compounds as well as with mass spectra in different computer libraries. Geographic variation in floral scent—Volatiles were collected from 10 populations of Y. filamentosa—five pollinated by T. yuccasella and five by T. cassandra (Table 1)—all within the native range of the species (Trelease, 1902) and established before human-mediated dispersal in the past two centuries. Twenty-one scent compounds were reliably detected from all individuals of all populations (see Results, Floral fragrance chemistry). For each population, the mean proportion of total scent comprised by each compound was calculated. Also, the coefficient of variation (CV 5 standard deviation 3 100/mean) for each compound was calculated, using arcsine square-root transformed ratios to better fit normal distribution of data. A principal component analysis (PCA) was performed on arcsine square-root transformed proportions of scent compounds. Each variable was scaled to unit variance before the analysis. To check for differences in the composition of the floral odor signal among sites, Euclidean distances were calculated between individuals based on z scores of the relative abundance of each compound. Mean Euclidean distances among populations were then used to analyze differentiation of floral scent. A Kruskal-Wallis nonparametric test was used to analyze if
1626
AMERICAN JOURNAL
OF
BOTANY
[Vol. 92
TABLE 2. Mean percentage and coefficient of variation (CV) of compounds emitted from flowers of Y. filamentosa at different sites. Asterisks (*) show compounds for which mass spectra and retention times were compared to those of reference compounds. N 5 no. of plants. Site (N) Lake Placid (10) Compound
Monoterpenes 1. b-myrcene* Homoterpenes 2. E-4,8-dimethyl-1,3,7-nonatriene* 4. Unknown homoterpene 1 6. Unknown homoterpene 2 7. C11-alcohol 11. Nerolidol* Unknowns 3. Unknown 5. Unknown 10. Unknown 14. Unknown
m/z m/z m/z m/z
67 67 66 66
1 (C11H18O) 2 (C11H18O) 1 (C11H14O2) 2
Aliphatic hydrocarbons 8. Pentadecene 9. Pentadecane 12. Hexadecene 13. Hexadecane 15. Heptadecadiene 16. 1-heptadecene* 17. Heptadecane* 18. Octadecene* 19. Octadecane* 20. Nonadecene* 21. Nonadecane*
%
CV
Inverness (9) %
CV
Ocala (10) %
Gold Head (11) CV
%
CV
Jesup (9) %
CV
0.5
118.6
1.3
13.5
0.4
31.8
0.7
45.6
0.6
12.6
33.4 0.6 3.9 23.5 0.4
27.1 15.1 17.1 13.5 76.1
30.3 0.3 1.7 17.0 0.8
19.6 22.0 14.3 7.0 51.9
45.6 0.1 1.7 19.8 0.2
9.7 18.6 21.7 13.9 9.7
24.3 0.8 6.1 16.4 0.3
18.9 22.5 18.6 8.4 43.1
43.7 0.2 3.4 18.9 0.3
11.3 13.2 11.4 13.5 25.6
0.3 0.6 5.7 0.6
28.1 18.8 48.6 53.2
0.5 0.9 4.6 0.2
29.6 20.1 30.3 27.0
0.2 0.4 2.8 ,0.1
26.1 23.8 30.6 26.6
0.6 1.9 5.0 0.3
38.5 30.0 27.0 45.6
0.2 1.0 3.5 0.1
17.9 17.9 18.6 22.7
0.4 2.1 0.4 0.6 2.5 15.4 4.5 0.2 0.2 3.1 0.6
45.3 16.1 48.8 28.0 28.8 15.6 28.7 69.1 68.4 37.6 93.0
0.7 3.6 0.7 1.8 4.7 15.2 8.0 0.6 0.5 1.9 0.9
27.2 13.1 33.9 19.0 22.1 15.3 15.8 47.6 22.3 24.5 29.7
0.4 2.2 0.3 0.5 2.8 15.1 5.1 0.3 0.1 1.3 0.4
19.4 9.1 22.1 19.0 28.7 22.3 22.3 55.4 13.4 18.0 24.4
0.7 4.2 0.6 1.0 4.6 19.6 7.9 0.4 0.3 3.0 0.8
31.7 16.1 28.7 19.0 17.9 11.5 28.4 54.7 26.6 17.7 19.1
0.2 1.6 0.2 0.4 2.7 13.1 5.6 0.2 0.2 1.7 2.1
13.2 19.3 13.0 23.0 24.4 13.8 29.1 29.2 24.6 19.4 86.4
Note: Trace (tr.) indicates amounts less than 0.1% of total. the floral odor blend differed more between populations pollinated by different yucca moths than between those pollinated by the same moth species. Also, mean Euclidean distances were plotted against geographic distances among populations to test for isolation by distance in floral scent. Finally, the total release of compounds from a plant was quantified by the formula: peak area of compound 1O peak area of IS 2 · amount of IS x
Emission rate 5
no. flowers on plant h of sampling 3 extract volume after addition of IS,
where IS 5 internal standard. Total release rates of compounds were compared between populations using one-way analysis of variance and between populations with different pollinators using an unpaired t test on pooled data. Vouchers collected from each population were deposited at the A. C. Moore Herbarium at the University of South Carolina (USCH), USA.
Fig. 1. Gas chromatogram of volatiles in headspace sample from flowers of Yucca filamentosa. Compound numbers are the same as in Table 2.
RESULTS Floral fragrance chemistry—GC-MS analysis revealed 29 compounds in headspace collections of Y. filamentosa, with 21 compounds repeatedly found in sufficient abundance to allow quantification in all individuals of all populations; these were used in further statistical analysis (Fig. 1, Table 2; information about the remaining compounds is found in Appendix S1 [see Supplemental Data with online version of this article]). Most compounds could be classified according to two major biosynthetic pathways: (1) homoterpene hydrocarbons derived from the sesquiterpene alcohol nerolidol, or (2) aliphatic alkenes and alkanes. The homoterpene E-4,8-dimethylnona1,3,7-triene was produced in highest relative amounts in all populations (14.2–82.5%), followed by a C11 alcohol (2.5– 31.6%) and 1-heptadecene (3.4–27.1%) (Table 2). The remaining compounds typically comprised ,10% of the total sample. Alkanes and n-alkenes with 15–19 carbons appeared as regularly spaced peak doublets in GC runs (Fig. 1). Two compounds with a prominent signal at m/z 66 in the mass spectrum were identified as di-oxygenated compounds with unknown structures. Ambient controls generally contained few compounds in small amounts, and bag contaminants did not have a major influence on the analysis. Geographic variation in floral scent—The composition of floral odor blends in populations of Y. filamentosa is shown in Fig. 2. The five principal components with eigenvalues .1 explained 68.3% of the total variation in floral fragrance data, based on 21 compounds. On PC1, most hydrocarbons had similar positive loadings, whereas E-4,8-dimethylnona-1,3,7-tri-
October 2005] TABLE 2.
SVENSSON
ET AL.—FLORAL SCENT OF
YUCCA FILAMENTOSA
1627
Extended.
Site (N) Eglin (9) %
Heggie’s Rock (6) CV
Due West (8) %
CV
Forty Acre Rock (8)
0.8
26.1
1.3
26.8
0.9
12.9
0.9
32.0
0.8
45.0
tr.-2.1
42.7 0.1 3.1 19.0 1.1
17.5 25.3 18.6 11.9 67.0
26.4 0.5 6.2 12.0 0.4
16.2 23.9 23.5 15.6 48.4
59.6 0.1 1.6 14.2 1.0
12.9 14.6 18.1 23.7 31.3
46.2 0.2 2.9 16.2 0.2
22.2 23.7 35.9 23.7 35.6
36.0 0.5 5.6 17.4 0.4
20.4 31.3 32.0 30.7 70.0
14.2–82.5 tr.-1.4 0.7–10.1 2.5–31.6 tr.-3.9
0.2 0.7 2.9 ,0.1
21.8 22.6 36.4 103.5
0.6 1.8 4.6 0.2
34.3 31.9 21.6 30.8
0.1 0.3 2.0 0.1
24.1 30.0 30.1 33.9
0.3 0.6 4.0 0.2
21.6 37.7 38.7 37.5
0.3 1.2 3.7 0.2
56.5 37.7 45.7 43.5
tr.-1.3 0.2–3.6 0.5–15.3 tr.-2.2
0.5 2.1 0.5 0.7 2.2 14.2 5.6 0.4 0.3 2.0 0.5
32.7 26.5 26.4 39.3 29.9 24.4 21.4 44.5 26.1 18.9 23.6
0.8 5.4 0.8 1.2 5.0 19.3 7.2 0.8 0.5 3.4 0.6
26.9 14.5 25.5 17.8 11.9 8.2 7.5 32.0 27.2 26.8 22.5
0.2 1.7 0.1 0.3 1.7 8.5 3.3 0.1 0.1 2.9 0.4
23.2 20.7 66.0 18.5 23.3 16.7 19.1 47.2 21.7 18.3 30.3
0.3 3.6 0.3 0.4 3.5 9.6 4.9 0.2 0.3 3.4 1.0
44.3 33.1 55.9 29.2 15.0 22.6 17.0 58.5 41.4 30.0 33.3
0.3 2.1 0.2 0.4 3.0 17.2 4.6 0.4 0.5 3.5 0.8
74.6 22.8 71.5 12.9 18.3 15.9 24.0 31.1 52.8 18.4 34.5
tr.-1.4 0.6–7.6 tr.-1.4 0.1–2.5 0.4–7.3 3.4–27.1 1.4–14.2 tr.-1.6 tr.-1.8 0.6–6.9 tr.-13.2
DISCUSSION This study provides a first step in elucidating floral fragrance chemistry and the role of pollinator-mediated specialization of floral scent in the yucca–yucca-moth mutualism. Two main classes of compounds can be distinguished in floral extracts of Y. filamentosa: homoterpenes putatively derived from the sesquiterpene alcohol nerolidol (Ga¨bler et al., 1991;
CV
%
CV
Range all populations (%)
%
ene was the only compound with negative loading (Table 3). In contrast, nearly all of these compounds loaded negatively on PC2, whereas both unique m/z 66 compounds showed high positive loadings on PC2. A score plot of the first two principal components revealed considerable overlap in the odor blend among sites (Fig. 3). Euclidean distances were almost identical between populations with the same and with different pollinators (x2 5 0.002, df 5 2, P . 0.10, Fig. 4), indicating minimal differentiation of the floral scent in Y. filamentosa based on pollinator. In addition, no correlation between Euclidean distance and geographical distance was detected (r 5 0.020, P . 0.10, Fig. 5). Most populations had similar coefficients of variation for compounds, with the possible exception of Lake Placid, which had slightly higher values (Table 2). The total emission rate of compounds differed significantly between individual populations (F 5 3.61, df 5 9, P , 0.001), ranging from 0.7 mg/ flower/h at Nags Head to 5.8 mg/flower/h at Eglin (Fig. 6). However, no significant difference in release rate of compounds was observed when comparing populations with different pollinators (T. yuccasella: 2.7 6 0.4 mg/flower/h; T. cassandra: 1.8 6 0.4 mg/flower/h, t 5 1.69, df 5 85, P 5 0.09).
%
Nags Head (7)
CV
Degenhardt and Gershenzon, 2000), and 15–19 carbon aliphatic alkenes and alkanes. Thus, two biosynthetic pathways are responsible for producing the majority of floral scent constituents in this yucca. Most compounds found in extracts have earlier been reported as part of the floral odor bouquet in other angiosperms (see Knudsen et al., 1993). The homoterpene E4,8-dimethylnona-1,3,7-triene was the most abundant compound in extracts and has previously been identified in floral odor extracts of Y. filamentosa (Ga¨bler et al., 1991). This compound, better known as a nearly universal component of volatile blends released from vegetative parts in higher plants upon herbivore attack, functions as a kairomone to attract natural enemies of various insect herbivores (reviewed by Pare´ and Tumlinson, 1999; Dicke and van Loon, 2000). Its function in Y. filamentosa is apparently different, because it is emitted constitutively from undamaged floral tissue. Two di-oxygenated compounds with unknown structure, which have never been reported as floral volatiles in angiosperms, were also identified in Y. filamentosa floral headspace. The emission of unique floral compounds may reflect a ‘‘private’’ communication channel that facilitates selective attraction of exclusive pollinators, similar to what has been observed in sexually deceptive orchids (Schiestl et al., 2003). In fact, one of these compounds has been shown to trigger strong responses from antennae of T. cassandra in preliminary electrophysiological recordings (G. Svensson and R. Raguso; University of South Carolina, unpublished data). The structural elucidation and synthesis of this compound, as well as behavioral assays to test its function as a pollinator attractant, will be critical steps towards understanding the chemical ecology of the yucca–yucca-moth association. Also, hypotheses re-
1628
AMERICAN JOURNAL
OF
BOTANY
[Vol. 92
Fig. 2. Mean chemical composition of the floral odor blend in Yucca filamentosa from five populations pollinated by Tegeticula cassandra (c) and five populations pollinated by T. yuccasella (y). Compound numbers are the same as in Table 2.
garding possible canalization of the floral odor signal in yuccas to facilitate attraction of obligate pollinators have to be tested rigorously by altering the relative ratios of electrophysiologically active compounds and by testing off-ratio blends not produced in natural populations to check the attraction of moths to such mixtures (e.g., Schiestl, 2004). Yucca populations pollinated by T. cassandra and T. yuccasella, respectively, did not differ in floral scent. The former species is known to feed only on Y. filamentosa, except for a single site in central Florida, where Y. aloifolia is a host (Pellmyr, 1999), whereas T. yuccasella utilizes several closely related Yucca species elsewhere in its range. It is currently not possible to infer the ancestral host of T. yuccasella. Whereas mtDNA-based analyses suggested a close relationship between the two pollinator species (Pellmyr and Leebens-Mack, 2000), morphological (Pellmyr, 1999) and genomic data from AFLP (D. Althoff, K. Segraves, O. Pellmyr, University of Idaho, and J. Leebens-Mack, Pennsylvania State University, unpublished data) instead suggest that they are part of two groups that
independently radiated onto yuccas. In the case of Y. filamentosa, there is very limited overlap of pollinator ranges, with some areas on the northern Florida peninsula having low density of T. yuccasella individuals amidst dominant T. cassandra populations. Additional phylogeographic analyses of the moths are required to determine whether this reflects an ongoing range expansion by either species or a stable long-term contact zone. Regardless, there is clearly no variation in floral scent composition between populations pollinated by T. cassandra and T. yuccasella that would indicate longstanding dependence of these two pollinators outside the contact region. One possibility would be high levels of gene flow that would limit effects of local adaptation or drift. Indeed, a study by Massey and Hamrick (1998), using allozyme data, revealed high genetic diversity within and relatively weak genetic structuring among populations of Y. filamentosa, with on average only 17% of the total variation partitioned between populations. This can theoretically explain similarity among populations that share a
October 2005]
SVENSSON
ET AL.—FLORAL SCENT OF
YUCCA FILAMENTOSA
1629
TABLE 3. Loading of the first two principal components of the 21 compounds used in the principal components analysis. Compound numbers are the same as in Table 2. Compound
PC1
PC2
Total variation explained by each factor (%)
37.3
10.5
Monoterpenes 1. b-Myrcene Homoterpenes 2. E-4,8-dimethyl-1,3,7-nonatriene 4. Unknown homoterpene 1 6. Unknown homoterpene 2 7. C11-alcohol 11. Nerolidol Unknowns 3. Unknown 5. Unknown 10. Unknown 14. Unknown
m/z m/z m/z m/z
67 67 66 66
1 (C11H18O) 2 (C11H18O) 1 (C11H14O2) 2
Aliphatic hydrocarbons 8. Pentadecene 9. Pentadecane 12. Hexadecene 13. Hexadecane 15. Heptadecadiene 16. 1-heptadecene 17. Heptadecane 18. Octadecene 19. Octadecane 20. Nonadecene 21. Nonadecane
0.10
20.39
20.32 0.25 0.27 0.01 0.11
20.18 0.22 0.10 0.49 0.07
0.30 0.28 0.22 0.16
0.04 0.02 0.33 0.41
0.25 0.27 0.27 0.28 0.27 0.00 0.04 0.22 0.21 0.19 0.09
20.17 20.21 20.09 20.16 20.11 0.12 20.08 20.17 20.24 0.09 0.00
pollinator species, but not between pollinator-differentiated populations, so a more complex explanation is required. One hypothetical mechanism would be recent invasion with replacement by one of the extant pollinator species, which would most likely be T. yuccasella because of its oligophagous habit
Fig. 3. Score plot of the first two principal components (PC) based on 21 compounds in floral odor extracts of Yucca filamentosa from 10 populations. PC1 explained 37.3% and PC2 10.5%, respectively, of the total variation observed.
Fig. 4. Median Euclidean distances of floral scent between populations of Yucca filamentosa with the same or with different pollinators.
outside the native range of Y. filamentosa. This hypothesis can be tested with phylogeographic tools. The homogeneity among populations with a shared pollinator is consistent with the prediction of extensive pollen-mediated gene flow among host populations by dispersing female moths. Analyses on moth-mediated pollen dispersal in yuccas are limited, however, due to the difficulties in observing such events. Marr et al. (2000) used fluorescent dye to track pollen transfers by T. yuccasella within populations of Y. filamentosa. Most pollen was transferred to flowers of the same plant or to neighboring plants, and the number of pollen transfers declined rapidly with distance from the source. Although the study showed restricted dispersal in moths, the method used could not detect long-range dispersals, and therefore the frequency of such events remains unknown. The mean release rate of floral compounds differed greatly
Fig. 5. Relationship between Euclidean distance and geographic distance between populations of Yucca filamentosa.
1630
AMERICAN JOURNAL
Fig. 6. Total (mean 6 SE) emission of 21 volatile compounds from populations of Yucca filamentosa pollinated by different yucca moths (white bars 5 T. cassandra, black bars 5 T. yuccasella).
among populations, but was not attributable to pollinator specificity. This variation could be due to both intrinsic and extrinsic factors, such as phenotypic variation in floral odor emission or differences in ambient temperature during scent collection at different sites. Our estimations of release rates controlled for the different numbers of flowers per plant. Such data may be uninformative when analyzing interpopulation variation in floral scent emission, because of the great variation in total number of flowers produced and flowering phenology expected also among inflorescences within populations. Thus, inflorescences within a population will differ greatly in fragrance emission at any given night, which in turn may influence the attraction of moths for pollination. Because yuccas are exclusively pollinated by yucca moths, plants are predicted to produce highly specific floral odors to facilitate the attraction of pollinators, either by emitting unique compounds or by using strongly canalized odor blends (Raguso, 2003). Although intraspecific variation in the floral odor blend has been observed for many angiosperms, extensive screening for geographic variation in floral scent composition has been conducted in only a few species, e.g., Magnolia kobus (Magnoliaceae) (Azuma et al., 2001), Geonoma macrostachys (Arecaceae) (Knudsen, 2002), and Silene latifolia (Caryophyllaceae) (Do¨tterl et al., 2005). A comparison of the total variation in floral odor (based on CV values) found in the present study with that of G. macrostachys shows no support for strong canalization of the floral scent in Y. filamentosa, when compared with a species that is pollinated by a diverse insect fauna. Although variation in scent composition among populations of M. kobus and S. latifolia was dramatic, we can-
OF
BOTANY
[Vol. 92
not directly compare our data with those of these studies, because coefficients of variation were not presented. A more critical approach, however, would be to analyze the variation within the subset of compounds that constitutes the attractive signal of a flower—to extract ‘‘signal’’ from ‘‘noise’’ in the fragrance blend (Raguso, 2003). So far such analysis has only been conducted for the sexually deceptive orchid Ophrys sphegodes, for which Ayasse et al. (2000) observed that the relative ratios of electrophysiologically active compounds in floral extracts varied less than nonactive ones, indicating strong selection on pollinator-attracting compounds in this highly specialized pollination system. The scarcity of data on geographic variation in floral scent limits the possibility to test for canalized floral odor blends in highly specialized pollination systems, such as the yucca–yucca-moth mutualism, compared to systems in which plants are pollinated by a diverse insect fauna. In addition, variation in the floral scent has not been analyzed in genera that are closely related to yuccas and have low pollinator specificity. Thus, the hypothesis of canalization of the floral odor in yuccas as a consequence of their mutualism with yucca moths cannot yet be tested within a phylogenetic framework. In conclusion, we have chemically characterized the floral scent in Y. filamentosa and analyzed the floral odor blend between populations that rely on different, distantly related yucca moth species for their sexual reproduction. Although the odor blend did not differ between populations with different pollinators, novel compounds and strong canalization of the floral odor blend may indicate mechanisms in Y. filamentosa, as well as other species of yuccas, to facilitate the attraction of exclusive pollinators to flowers. The association between yuccas and yucca moths thus offers an attractive model system to study pollinator-mediated selection on floral traits as well as the sensory basis for host-specificity in insects. LITERATURE CITED AYASSE, M., F. P. SCHIESTL, H. F. PAULUS, C. LO¨FSTEDT, B. S. HANSSON, F. IBARRA, AND W. FRANCKE. 2000. Evolution of reproductive strategies in the sexually deceptive orchid Ophrys sphegodes: how does flowerspecific variation of odor signals influence reproductive success? Evolution 54: 1995–2006. AZUMA, H., M. TOYOTA, AND Y. ASAKAWA. 2001. Intraspecific variation of floral scent chemistry in Magnolia kobus DC. (Magnoliaceae). Journal of Plant Research 114: 411–422. DEGENHARDT, J., AND J. GERSHENZON. 2000. Demonstration and characterization of (E)-nerolidol synthase from maize: a herbivore-inducible terpene synthase participating in (3E)-4,8-dimethyl-1,3,7-nonatriene biosynthesis. Planta 210: 815–822. DICKE, M., AND J. J. A. VAN LOON. 2000. Multitrophic effects of herbivoreinduced plant volatiles in an evolutionary context. Entomologia Experimentalis et Applicata 97: 237–249. DO¨TTERL, S., L. M. WOLFE, AND A. JU¨RGENS. 2005. Qualitative and quantitative analyses of flower scent in Silene latifolia. Phytochemistry 66: 203–213. FLEMING, T. H., AND J. N. HOLLAND. 1998. The evolution of obligate pollination mutualisms: senita cactus and senita moth. Oecologia 114: 368– 375. GA¨BLER, A., W. BOLAND, U. PREISS, AND H. SIMON. 1991. Stereochemical studies on homoterpene biosynthesis in higher plants; mechanistic, phylogenetic, and ecological aspects. Helvetica Chimica Acta 74: 1773– 1789. GIBERNAU, M., M. HOSSAERT-MCKEY, J. FREY, AND F. KJELLBERG. 1998. Are olfactory signals sufficient to attract fig pollinators? Ecoscience 5: 306–311. GRISON, L., A. A. EDWARDS, AND M. HOSSAERT-MCKEY. 1999. Interspecies
October 2005]
SVENSSON
ET AL.—FLORAL SCENT OF
variation in floral fragrances emitted by tropical Ficus species. Phytochemistry 52: 1293–1299. GRISON-PIGE´, L., J.-M. BESSIE`RE, AND M. HOSSAERT-MCKEY. 2002a. Specific attraction of fig-pollinating wasps: role of volatile compounds released by tropical figs. Journal of Chemical Ecology 28: 283–295. GRISON-PIGE´, L., M. HOSSAERT-MCKEY, J. M. GREFF, AND J. M. BESSIE`RE. 2002b. Fig volatile compounds—a first comparative study. Phytochemistry 61: 61–71. HOSSAERT-MCKEY, M., M. GIBERNAU, AND J. E. FREY. 1994. Chemosensory attraction of fig wasps to substances produced by receptive figs. Entomologia Experimentalis et Applicata 70: 185–191. KATO, M., A. TAKIMURA, AND A. KAWAKITA. 2003. An obligate pollination mutualism and reciprocal diversification in the tree genus Glochidion (Euphorbiaceae). Proceedings of the National Academy of Sciences, USA 100: 5264–5267. KAWAKITA, A., AND M. KATO. 2004. Obligate pollination mutualism in Breynia (Phyllanthaceae): further documentation of pollination mutualism involving Epicephala moths (Gracillariidae). American Journal of Botany 91: 1319–1325. KNUDSEN, J. T. 2002. Variation in floral scent composition within and between populations of Geonoma macrostachys (Arecaceae) in the western Amazon. American Journal of Botany 89: 1772–1778. KNUDSEN, J., L. TOLLSTEN, AND L. G. BERGSTRO¨M. 1993. Floral scents—a checklist of volatile compounds isolated by head-space techniques. Phytochemistry 33: 253–280. MARR, D. L., J. LEEBENS-MACK, L. ELMS, AND O. PELLMYR. 2000. Pollen dispersal in Yucca filamentosa (Agavaceae): the paradox of self-pollination behavior by Tegeticula yuccasella (Prodoxidae). American Journal of Botany 87: 670–677. MASSEY, L. K., AND J. L. HAMRICK. 1998. Genetic diversity and population structure of Yucca filamentosa (Agavaceae). American Journal of Botany 85: 340–345. PAMMEL, L. H. 1925. The extension of the yucca moth. Science 61: 414– 415. PARE´, P. W., AND J. H. TUMLINSON. 1999. Plant volatiles as a defense against insect herbivores. Plant Physiology 121: 325–331. PELLMYR, O. 1999. Systematic revision of the yucca moths in the Tegeticula yuccasella complex (Lepidoptera: Prodoxidae) north of Mexico. Systematic Entomology 24: 243–271.
YUCCA FILAMENTOSA
1631
PELLMYR, O. 2003. Yuccas, yucca moths, and coevolution: a review. Annals of the Missouri Botanical Garden 90: 35–55. PELLMYR, O., AND L. B. THIEN. 1986. Insect reproduction and floral fragrances: keys to the evolution of the angiosperms? Taxon 35: 76–85. PELLMYR, O., AND H. W. KRENN. 2002. Origin of a complex key innovation in an obligate insect–plant mutualism. Proceedings of the National Academy of Sciences, USA 99: 5498–5502. PELLMYR, O., AND J. LEEBENS-MACK. 2000. Reversal of mutualism as a mechanism for adaptive radiation in yucca moths. American Naturalist 156: S62–S76. POWELL, J. A., AND R. A. MACKIE. 1966. Biological interrelationships of moths and Yucca whipplei. University of California Publications in Entomology, no. 42. University of California Press, Berkeley, California, USA. RAGUSO, R. A. 2004. Flowers as sensory billboards: progress towards an integrated understanding of floral advertisement. Current Opinion in Plant Biology 7: 434–440. RAGUSO, R. A. 2003. Olfactory landscapes and deceptive pollination: signal, noise and convergent evolution in floral scent. In G. J. Blomquist and R. G. Vogt [eds.], Insect pheromone biochemistry and molecular biology: the biosynthesis and detection of pheromones and plant volatiles, 631– 650. Elsevier Academic Press, Amsterdam, Netherlands. RAGUSO, R. A., AND O. PELLMYR. 1998. Dynamic headspace analysis of floral volatiles: a comparison of methods. Oikos 81: 238–254. SEGRAVES, K. A., AND O. PELLMYR. 2004. Testing the out-of-Florida hypothesis on the origin of cheating in the yucca–yucca moth mutualism. Evolution 58: 2266–2279. SCHIESTL, F. P. 2004. Floral evolution and pollinator mate choice in a sexually deceptive orchid. Journal of Evolutionary Biology 17: 67–75. SCHIESTL, F. P., R. PEAKALL, J. G. MANT, F. IBARRA, C. SCHULZ, S. FRANKE, AND W. FRANCKE. 2003. The chemistry of sexual deception in an orchid–wasp pollination system. Science 302: 437–438. STENSMYR, M. C., I. URRU, I. COLLU, M. CELANDER, B. S. HANSSON, AND A. M. ANGIOY. 2002. Rotting smell of dead-horse arum florets. Nature 420: 625–626. TRELEASE, W. 1902. The Yucceae. Annual Reports, Missouri Botanical Garden 13: 27–133. WARE, A. B., AND S. G. COMPTON. 1994. Responses of fig wasps to host plant volatile cues. Journal of Chemical Ecology 20: 785–802. WEIBLEN, G. D. 2002. How to be a fig wasp. Annual Review of Entomology 47: 299–330.