Phytochemistry 94 (2013) 123–134
Contents lists available at SciVerse ScienceDirect
Phytochemistry
journal homepage: www.elsevier.com/locate/phytochem
Flower color polymorphism in Iris lutescens (Iridaceae): Biochemical analyses in light of plant–insect interactions
Hui Wang a,1, Lucie Conchou b,1, Jean-Marie Bessière c, Guillaume Cazals d, Bertrand Schatz b,2, ⇑ Eric Imbert a, ,2 a Institut des Sciences de l’Évolution de Montpellier (ISEM), UMR 5554 CNRS-Université Montpellier 2, Bâtiment 22, Université Montpellier 2, place E. Bataillon, 34095 Montpellier Cedex 5, France b Centre d’Ecologie Fonctionnelle et Evolutive (CEFE), UMR 5175 CNRS, 1919 route de Mende, 34293 Montpellier Cedex 5, France c Ecole Nationale Supérieure de Chimie de Montpellier (ENSCM), 8, rue de l’Ecole Normale, 34296 Montpellier Cedex 5, France d Institut des Biomolécules Max Mousseron (IBMM), UMR 5247 CNRS-Université Montpellier 1 et 2, Bâtiment Chimie (17), Université Montpellier 2, place E. Bataillon, 34095 Montpellier Cedex 5, France article info abstract
Article history: We describe a flower color polymorphism in Iris lutescens, a species widespread in the Northern part of Received 13 February 2013 the Mediterranean basin. We studied the biochemical basis of the difference between purple and yellow Received in revised form 13 May 2013 flowers, and explored the ecological and evolutionary consequences of such difference, in particular Available online 18 June 2013 visual discrimination by insects, a potential link with scent emitted and the association between color and scent. Anthocyanins were found to be present in much greater concentrations in purple flowers than Keywords: in yellow ones, but the anthocyanin composition did not differ between color morphs. Likewise, no quan- Flower color polymorphism titative difference in anthocyanin content was found between vegetative tissues of the two morphs. Flo- Iris lutescens ral anthocyanins were dominated by delphinidin 3-O-(p-coumaroylrutinoside)-5-O-glucoside (also called Anthocyanins Floral scent delphanin) and its aliphatic derivatives. Small amounts of delphinidin 3-O-(p-caffeoylrutinoside)-5-O- Terpenoids glucoside and its aliphatic derivatives were also characterized. Based on a description of bumblebees’ Plant–insect interactions (one of the main pollinators of I. lutescens) color perception, purple and yellow flowers of I. lutescens could Color-scent association be visually discriminated as blue and blue-green, respectively, and likely by a wide variety of other insects. The overall chemical composition of the scent produced was not significantly different between morphs, being dominated by terpenoids, mainly myrcene, (E)-b-ocimene and limonene. A slight color- scent correlation was nevertheless detected, consistent with the shared biosynthetic origin of both pig- ments and volatile compounds. Therefore in this species, the difference in the amounts of pigments responsible for flower color difference seems to be the major difference between the two morphs. Poll- inators are probably the main selective agent driving the evolution of flower color polymorphism in I. lutescens, which represents a suitable species for investigating how such polymorphism is maintained. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction polymorphism has been documented for numerous floral traits such as color, scent, size and phenology (Delle-Vedove et al., Ecologists have long been intrigued by the emergence and 2011). While flower color is diverse throughout the angiosperms, maintenance of polymorphism in floral traits (Schaefer et al., variation of flower color within species is uncommon; and species 2004; Weiss, 1995). Indeed, as pollinator visitation rate is generally that show a true, stable, genetically based polymorphism are com- correlated with plant fitness in entomogamous species (i.e. 80% of paratively rare (Kay, 1978; Weiss, 1995). Flower color polymor- plant species, Potts et al., 2010), stabilizing selection mediated by phism has been well described in a limited number of species, associative learning is expected to occur on floral traits, leading such as Ipomoea purpurea (Convolvulaceae, Clegg and Durbin, to low intra-specific variation (Ashman and Majetic, 2006; 2000), Linanthus parryae (Polemoniaceae, Schemske and Bierzych- Dormont et al., 2010; Salzmann and Schiestl, 2007). However, udek, 2001), Mimulus aurantiacus (Scrophulariaceae, Streisfeld and Kohn, 2005), Hesperis matronalis (Brassicaceae, Majetic et al., 2007), ⇑ Corresponding author. Tel.: +33 0 467144910. and the orchids Dactylorhiza sambucina (Gigord et al., 2001) and E-mail address: [email protected] (E. Imbert). Calanthe sylvatica (Juillet et al., 2010). 1 These authors contributed equally to this work and are considered as joint first Flower color is among the most important visual signals in pol- authors. linator attraction (Menzel and Shmida, 1993) and generalist pollin- 2 These authors contributed equally to this work and are considered as joint last ators use color and scent to find flowers when first exploring the authors.
0031-9422/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.phytochem.2013.05.007 124 H. Wang et al. / Phytochemistry 94 (2013) 123–134 world (Chittka and Raine, 2006; Giurfa et al., 1995; Lunau and between composition of pigments and of volatiles in plant species Maier, 1995). Consistently, previous studies suggested that flower with flower polymorphism. Recent reviews consider the flower color polymorphism could be maintained by contrasting pollinator scent-color combination to be an outstanding open question in preferences for particular color morphs. For example, hawkmoth pollination ecology (Raguso, 2008; Rausher, 2008; Schaefer and and hummingbird pollinators impose divergent selection, respec- Ruxton, 2009). tively, on yellow and red morphs of Mimulus aurantiacus (Streisfeld We investigated flower color polymorphism in a rewardless and Kohn, 2007). Pollinators may also exert temporally or spatially species common in the Northern part of the Mediterranean region, fluctuating selection pressures, when their abundance and assem- Iris lutescens, which displays a spectacular purple-yellow flower blage structure are variable (Eckhart et al., 2006; Salzmann and color polymorphism within populations (Fig. 1). Information on Schiestl, 2007). In the food-deceptive orchid Dactylorhiza sambuci- the reproductive biology of this species is presented in the Materi- na, negative frequency-dependent selection mediated by the learn- als and methods section below. We analyzed the biochemical basis ing abilities of pollinators has been observed (Gigord et al., 2001; of color and scent, and explored ecological and evolutionary conse- but see Pellegrino et al., 2005). quences of color differentiation. We aimed to determine whether Three groups of pigments are responsible for coloration in pollinator agents (mainly bumblebees and solitary bees) are poten- plants (Tanaka et al., 2008): the composite group flavonoids (pale tial selective agents acting on the evolution of the flower color yellow to yellow)/anthocyanins (orange to blue), the betalains polymorphism. More specifically, we addressed the following four (yellow to red, found only in the order Caryophyllalles), and the questions: (1) What are the pigmentation differences between the carotenoids (yellow to red). Flavonoids and anthocyanins are the two color morphs? (2) Are the two color morphs discriminated be- major contributors to flower color (Tanaka et al., 2010). In addition tween by insects? (3) What floral scents are emitted by the two to producing color to attract pollinators, all of these pigments func- color morphs? (4) Is there an association between floral scent tion in protecting plants against damage caused by UV and visible and flower color in this species? light (Tanaka et al., 2008). Flavonoids and anthocyanins are known to function in the responses of plants to stress (drought, cold) and 2. Results and discussion in resistance to attack by microbes and herbivores (Chalker-Scott, 1999; Harborne and Williams, 2000a). Carotenoids also play essen- 2.1. What are the pigmentation differences between the two color tial roles in photosynthesis. These pigments can thus affect plant morphs? survival in several ways, and various selection pressures can act indirectly on flower color traits. Selection pressures exerted by a 2.1.1. Quantifying pigment contents range of abiotic factors such as precipitation, soil or temperature The amounts of three classes of pigments, flavonoids (mainly (Dick et al., 2011; Schemske and Bierzychudek, 2001, 2007; War- chalcones, flavones and flavonols), anthocyanins and carotenoids, ren and Mackenzie, 2001) and biotic agents such as herbivores were all significantly different between plants of the two color and pathogens (Frey, 2004) could be involved in maintaining poly- morphs in flowers (petals and sepals, except for flavonoids in se- morphism. Whether pollinators are the main selective agents influ- pals, Fig. 2), while contents of none of them differed significantly encing flower polymorphisms, or whether these polymorphisms in leaves (Fig. 2). The largest difference in flowers was for anthocy- are driven mainly by selective forces other than pollinators, re- anins, which were present in about 18- to 28-fold higher concen- mains under debate, and the answer may depend on the studied trations in petals and sepals, respectively, of the purple morph species (Dormont et al., 2010; Strauss and Whittall, 2006). than in those of the yellow morph (Fig. 2). Not surprisingly, yellow Flower color and scent often function synergistically to attract flowers accumulated more yellowish pigments, carotenoids (high- pollinators (Burger et al., 2010; Chittka and Raine, 2006; Dötterl er contents in both petals and sepals) and flavonoids (in petals et al., 2011; Milet-Pinheiro et al., 2012; Raguso and Willis, 2002, only). The slight difference in content of yellowish pigments in 2005). Furthermore, relationships between color and scent are sepals of the two morphs probably resulted from the pale partly inherent, because the two kinds of traits rely on shared bio- yellow-orange beard present on each sepal in both morphs. The synthetic pathways (Armbruster, 2002; Majetic et al., 2007), implying that any mutation in a gene coding for an enzyme or a regulatory element of these pathways could have an impact on both color and scent emitted. Volatile organic compounds involved in floral scent are dominated by fatty-acid derivatives, (mono- or sesqui-) terpenoids and phenylpropanoids/benzenoids (Knudsen et al., 2006). All the terpenoids (carotenoids, mono- and sesquit- erpenoids) are synthesized from the same precursors – isopentenyl pyrophosphate and dimethylallyl pyrophosphate (Wu, 2008). Phe- nylpropanoid/benzenoid volatile compounds and flavonoid/antho- cyanin pigments both originate from the same phenylpropanoid biosynthetic pathway (Dudareva et al., 2004; Zvi et al., 2008). In both terpenoids and phenylpropanoids, volatiles are early products of the pathway while pigments are produced at later steps. The fact that biosynthetic pathways are shared suggests the potential for correlated responses in color and scent to a single change in bio- chemistry. For instance, in a study manipulating pigmentation in Dianthus caryophyllus petals, Zuker et al. (2002) found that removal of petal pigment by antisense suppression of the flavanone 3- hydroxylase gene in the anthocyanin biosynthetic pathway led to higher emission of an aromatic volatile (methyl benzoate). Although a number of studies have investigated scent composition of different color morphs (reviewed in Dormont et al., 2010), as far Fig. 1. Purple- and yellow-flowered individuals of Iris lutescens. Picture was taken as we know, no research has directly examined correlation by B. Anderson near Montpellier, Languedoc-Roussillon region, France. H. Wang et al. / Phytochemistry 94 (2013) 123–134 125
Fig. 2. Quantification of the three classes of pigments in purple and yellow morphs. Welch two-sample t-test was used to test for a significant difference in concentration of each pigment class in each tissue between purple and yellow morphs (⁄⁄⁄⁄, P < 0.0001; ⁄⁄⁄, P < 0.001; ⁄⁄, P < 0.01; ⁄, P < 0.05; NS, P > 0.05). Standard errors are shown. Sample sizes are as follows (purple, yellow): flavonoids and anthocyanins: N = 15, 15; carotenoids: N = 10, 10. brightly colored beards, together with prominent purple veins in The most abundant anthocyanin (peak 2) accounted for 46.75% of sepals, might serve as a guide to insect visitors. The relative con- the total peak area, and the ESI/MS spectrum showed m/z 919.5 for centration of flavonoids in yellow sepals was significantly greater its molecular ion [M]+, in good agreement with the molecular mass than the relative concentration of flavonoids in purple ones of delphinidin 3-O-(p-coumaroylrutinoside)-5-O-glucoside (delph-
97% ± 0.1% vs. 64% ± 1% (Welch two-sample t-test, P<0.0001). For anin, C42H47O23, 919.251), which has already been characterized leaves, the difference was not significant (P = 0.47). as the major anthocyanin compound in cyanic Iris flowers, including Flavonoid pigments are intermediates (chalcones) or products I. lutescens (Williams et al., 1997). MS2 spectrum also showed the of side branches (flavones and flavonols) of the anthocyanin most typical fragmentation products. Fragment peaks were ob- biosynthetic pathway (as reviewed in Tanaka et al., 2008). In flow- served at m/z 757.4 for a glucoside loss, at m/z 465.2 for the p-cou- ers of the yellow morph, although anthocyanins occurred at signif- maroylrutinoside loss, and at m/z 303.2 for the aglycone icantly lower concentrations than in the purple morph, flavonoid (delphinidin). The identification of peak 2 as delphanin was further pigments were present at greater concentrations than in purple confirmed by comparing the MS2 and ISD-MS2 spectrum data with flowers. This suggests that in yellow flowers, structural genes late the literature (Ichiyanagi et al., 2005; Sadilova et al., 2006). The in the anthocyanin biosynthetic pathway are down-regulated, ESI/MS spectrum of peak 1 gave its molecular ion [M]+ at m/z while the early stages or side branches of the pathway are not 935.5, corresponding to the mass calculated for C42H47O24 affected. The regulation of the pathway is floral tissue-specific, as (935.246). Comparison of MS2 and ISD-MS2 spectrum with those the production of flavonoids and anthocyanins in vegetative tis- of peak 2 showed that peak 1 was attached to a different organic acid sues did not differ between the two color morphs (Fig. 2). Numer- residual with an additional oxidation (detail mass fragmentation ous studies have shown that flavonoids and anthocyanins, pattern see Fig. 4, Table 1). Finally, it was identified as delphinidin especially those in vegetative tissues, can contribute to stress 3-O-(p-caffeoylrutinoside)-5-O-glucoside by comparison with the resistance (Chalker-Scott, 1999; Harborne and Williams, 2000a). literature (Sadilova et al., 2006). Such pleiotropic effects of pigment content on plant survival Peaks 5, 6, 7, 9, 10 were identified as acylated derivatives of peak interfere with pollinator-driven evolution of flower color in some 2. Peaks 5, 6, 7 had the same molecular ions [M]+ at m/z 961.5 species (Rausher, 2008; Strauss and Whittall, 2006; Warren and (C44H49O24, 961.262), and shared the same fragmentation pattern Mackenzie, 2001). In contrast, in I. lutescens, plants of the two in MS2 spectrum, indicating they are isomers. Comparing these morphs are unlikely to show any differences in the level of peaks with peak 2, the only different fragmented ion at m/z 799.4 + resistance to abiotic and biotic factors that could be accounted [MÀC6H10O5] indicated an additional acetylation in the p-cou- for differences in flavonoid and anthocyanin production in maroylrhamnosylglucoside residual. ISD-MS2 analyses of peak 7 vegetative tissues. with transition m/z 653.3?465.2 (C8H12O5) indicated that the acet- ylation could be assigned to the rhamnose. The 3 isomers might dif- 2.1.2. Identification of anthocyanins fer at the acetylation position in the rhamnose (3- or 4-), and also in We investigated whether the identity of anthocyanins also dif- the configuration of coumaric acid. Peaks 9 and 10 were two iso- + fered between purple and yellow flowers by HPLC-DAD-ESI/MS mers with the same molecular ions [M] at m/z 1003.5 2 analyses. Fig. 3 shows the typical HPLC chromatogram monitored (C46H51O25, 1003.272) and the same MS fragmentation pattern. at 540 nm. Ten peaks were detected in petals of both color morphs. Similarly, they were tentatively identified as derivatives of peak 2 Based on the HPLC retention times, UV–visible spectral properties with a butyryl group or two acetyl groups linked to the rhamnose. and ESI/MS spectrometric data, all of the 10 compounds were iden- The two peaks 9 and 10 were isomers probably differing only in the tical in both morphs (Table 1). Consistent with the results for total coumaric acid configuration. Previous studies have concluded that pigment contents, the total concentration of anthocyanins in flow- the trans isomer always occurs in larger amounts, but had lower ers of the purple morph was about 11-fold higher than in those of polarity than the cis isomer (Ando et al., 1999; Zheng et al., 2011). the yellow morph (Table 1). Thus peaks 9 and 10 were assigned to be cis and trans configuration, Anthocyanins in flowers of the purple morph were character- respectively. Peaks 3, 4 and 8 were identified as acylated derivatives ized in MS2 and ISD-MS2 analyses for further identification (Ta- of peak 1. With the same consideration as above, peaks 3 and 4 were ble 1). All of them were observed to have a fragment peak at m/z tentatively characterized as delphinidin 3-O-[acetyl-(cis-p- 303 in MS2 spectra, indicating that they are all delphinidin deriva- caffeoyl)-rhamnosylglucoside)-5-O-glucoside and delphinidin 3- tives. Identification of peaks 1 and 2 was achieved mainly by com- O-[acetyl-(trans-p-caffeoyl)-rhamnosylglucoside)-5-O-glucoside, paring the elution orders, molecular ions and fragmented ions with respectively. Peak 8 was identified as a derivative of peak 1 with a literature data (Ichiyanagi et al., 2005; Sadilova et al., 2006); peaks butyryl group or two acetyl groups in the 3-O-(p-coumaroylrutino- 3–10 were identified tentatively by comparing their MS2 and ISD- side) residual. Sample limitation prevented us from performing 2 MS2 spectra with those for peaks 1 and 2. additional ISD-MS analyses necessary for further identification. 126 H. Wang et al. / Phytochemistry 94 (2013) 123–134
2
100
7
%
6
5
10 1 3 4 9 8
-14 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 min
4
100 7
2 1
3 5 10 % 6 8 9
-47 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 min
Fig. 3. HPLC-DAD chromatograms recorded at 540 nm for purple (above) and yellow (below) petals. Compound identities are listed in Table 1.
Total anthocyanin amount was higher in purple- than in yel- Chittka’s physiological model (Chittka, 1992) allows us to eval- low-flowered I. lutescens, while their composition was identical uate how a given trichromatic insect species perceives colors. We in both morphs, each possessing the same 10 compounds. These applied this model to color perception of the two morphs by bum- results indicate the quantity of anthocyanins rather than different blebee (Bombus terrestris). We chose this insect model species be- anthocyanin components might be a major factor in determining cause it is one of the main pollinators frequently observed to flower color, as in the previously studied I. germanica (Ashtakala visit I. lutescens flowers. It is also one of the best-studied species and Forward, 1971a,b). The anthocyanins in Iris are commonly for behavioral responses to color perception. The maximal spectral based on delphinidin, petunidin and malvidin, modified by a 3-O- sensitivity (kmax, Briscoe and Chittka, 2001) of this insect’s UV pho- rhamnosylglucoside-5-O-glucoside moiety, with p-coumaroyl toreceptor is at k = 328 nm and no significant difference in its rel- acylation at the rhamnose residual (Harborne and Williams, ative excitation (E(U)) was observed between yellow and purple
2000b; Ishikura, 1980; Williams et al., 1997). In the present study, morphs. Both the blue and green photoreceptors (kmax = 428 and besides delphanin (peak 2), the main anthocyanin of Iris flowers, 536 nm, respectively), their relative excitations (E(B) and E(G), we also characterized other aromatic acylated (p-caffeoyl) and ali- respectively) were significantly different between morphs phatic acylated (acetyl, diacetyl or butyryl) delphinidin derivatives (Fig. 5(A) and Table 2). The two morphs also showed significant dif- (peak 1, peaks 3–10), which have not been previously character- ferences in color contrast and in brightness contrast against the ized in the genus Iris. background (Table 2). All the individuals we measured were then given color loci in the hexagonal color perception space 2.2. Are the two color morphs discriminated between by insects? (Fig. 5(B)). All the yellow flowers were perceived as ‘‘blue-green’’, as blue and green photoreceptors were comparably highly excited; Total brightness, i.e. the sum of the percent reflectance between while all the purple flowers were perceived as ‘‘blue’’, as only the 300 and 700 nm (Gomez, 2006), of flowers of the yellow morph blue photoreceptor was highly excited (Fig. 5(B) and Table 2). was 10225.8 ± 2457.1 (mean ± SD), which was significantly higher Therefore, yellow and purple flowers are perceptually different than the purple morph 3700.4 ± 1763.5 (Welch two-sample t-test, and easily discriminated by bumblebees. P < 0.0001). The two morphs showed similar low reflectance in the Color perception varies considerably within insects, ranging ultraviolet portion of the spectrum (300–400 nm), while they be- from mono- to tetrachromatic vision (Briscoe and Chittka, 2001). came more and more different in the 400–500 nm and 500– Consequently, the perception of I. lutescens flowers by bumblebees 650 nm portions (Fig. 5(A)). could be extrapolated to other trichromatic Apidae – the main Table 1 Chromatographic, mass spectral feature and proposed identification of the 10 anthocyanin compounds detected in purple and yellow petals. The bold fragment ions obtained in MS2 were fragmented in ISD-MS2 analyses. Relative abundance of each compound was calculated by taking the area under the peak in the chromatogram and normalizing it to the total area under all the anthocyanin peaks from the purple morph.
+ 2 2 Peak t R (min) [M] (m/z)MSfragment ions (m/z) ISD-MS fragment ions (m/z) Tentative identification Relative concentration (%) Purple Yellow .Wn ta./Pyohmsr 4(03 123–134 (2013) 94 Phytochemistry / al. et Wang H.
1 16.69 935.5 773.3/465.2/303.1 611.3:Y2 Dp 3-O-[p-caffeoyl-Rhap-(1?6)-Glcp]-5-O-Glcp 2.50 1.32 465.2:Y 1 303.1:Y 0 2 17.81 919.5 757.4/465.2/303.2 611.2:Y2, Dp 3-O-[p-coumaroyl-Rhap-(1?6)-Glcp]-5-O-Glcp (Delphanin) 46.75 1.59 595.2:Z2 465.2:Y 1,449.2:Z 1 303.1:Y 0 3 18.11 977.5 815.5/465.2/303.1 – Dp 3-O-[acetyl-(cis-p-caffeoyl)-Rhap-(1?6)-Glcp]-5-O-Glcp 1.58 0.40 4 18.93 977.5 815.4/465.2/303.2 653.4:Y2 Dp 3-O-[acetyl-(trans-p-caffeoyl)-Rhap-(1?6)-Glcp]-5-O-Glcp 2.33 1.64 465.2:Y 1 303.1:Y 0 5 19.21 961.5 799.3/465.1/303.1 465.0:Y1 Dp 3-O-[acetyl-(p-coumaroyl)-Rhap-(1?6)-Glcp]-5-O-Glcp (isomer 1) 6.12 0.47 303.1:Y 0 6 19.56 961.5 799.4/465.2/303.1 465.1:Y1 Dp 3-O-[acetyl-(p-coumaroyl)-Rhap-(1?6)-Glcp]-5-O-Glcp (isomer 2) 10.84 0.47 303.1:Y 0 7 20.08 961.5 799.4/465.2/303.1 653.3:Y2,637.4:Z 2 Dp 3-O-[acetyl-(p-coumaroyl)-Rhap-(1?6)-Glcp]-5-O-Glcp (isomer 3) 25.14 1.45 465.2:Y 1,449.2:Z 1 303.1:Y 0 8 21.28 1019.5 857.4/465.2/303.1 – Unknown (derivative of peak 1 with a butyryl or two acetyl groups in the 3-O-(p-coumaroylrutinoside) residual) 0.43 0.51 9 22.01 1003.5 841.5/465.2/303.1 – Dp 3-O-[diacetyl or butyryl-(cis-p-coumaroyl)-Rhap-(1?6)-Glcp]-5-O-Glcp 0.96 0.41 10 22.78 1003.5 841.4/465.2/303.1 695.4:Y2,679.3:Z 2 Dp 3-O-[diacetyl or butyryl-(trans-p-coumaroyl)-Rhap-(1?6)-Glcp]-5-O-Glcp 3.36 0.52 466.3:Y 1,449.4:Z 1 303.1:Y 0 Total – – – – – 100.00 8.79
+ 2 2 2 2 R:t retention time; [M] : molecular ion; MS : major fragment ions obtained in the MS analyses; ISD-MS : major fragment ions obtained in ISD-MS analyses; Dp: delphinidin; Rhap: rhamnose pyranoside; Glcp: glucose pyranoside. 127 128 H. Wang et al. / Phytochemistry 94 (2013) 123–134
+ Fig. 4. Chemical structures of peaks 1 and 2. Fragment nomenclature commonly applied for O-glycosides was used to show the mass fragmentation pattern of [MÀC6H10O5] (with 3-O-glucoside loss) in ISD-MS2. insect visitors of the species (see Section 4.1) – and probably also to total anthocyanin amount was negatively correlated with E(G) some other trichromatic insects possessing photoreceptors with (Spearman’s rank test, rs = À0.71, P<0.01, N = 13), while positively similar kmax values. Broader extrapolation is not possible, however. correlated with E(B) (Spearman’s rank test, rs = 0.61, P<0.05). This Nevertheless, the fact that the reflectance spectra of the two mor- suggests that if the flowers accumulate more anthocyanins, insect phs were highly contrasted between 500 and 650 nm, we can infer visitors’ green photoreceptors might be relatively less excited that any insect species that possesses at least one photoreceptor whereas blue photoreceptors might be more excited. So for the in- type with a kmax value between 500 and 650 nm should be able sect visitors, the flower becomes more ‘‘blue’’. In addition, in the to discriminate between the two color morphs. This is the case purple morph, total anthocyanin content is negatively correlated for most insect species reviewed in Briscoe and Chittka (2001). with brightness contrast (Spearman’s rank test, rs = À0.61, The two color morphs are thus likely discriminable by a wide vari- P<0.05, N = 13), while it is marginally positively correlated with ety of insects. color contrast (Spearman’s rank test, rs = 0.55, P=0.053). No corre- Since the pollinators can discriminate between color morphs, lation was found between other pigment amounts and color prop- they could potentially exert selection driving flower color poly- erties (data not shown). These findings reinforce the central role of morphism, depending on their behavioral responses to such color anthocyanin concentration in determining flower color of I. difference (innate preference for particular color morph, learning lutescens. process). Previous studies have found that newly emerged apid bees (e.g. Apis mellifera, Bombus terrestris), show a strong bias to- 2.3. What floral scents are emitted by the two color morphs? wards violet and blue (Chittka et al., 2004; Raine and Chittka, 2007). In addition, color preferences interact with preferences con- Among the 80 compounds retained for analyses (see Sec- cerning other floral characteristics such as size. For instance in tion 4.5), 69 and 75 were detected from purple and yellow- bumblebees, search time is negatively correlated with flower size, flowered individuals, respectively (Table 3). 18 compounds could and for large flowers, search time is negatively correlated with col- not be identified but most of them could be assigned to biosyn- or contrast (Spaethe et al., 2001). For I. lutescens, we found that the thetic categories, with more or less precise inferences on their purple morph showed significantly higher color contrast (CC, Ta- chemical structure. ble 2). We therefore predict that bumblebees should show an in- NMDS ordination showed no segregation of the purple and yel- nate preference for the purple morph of I. lutescens. low morphs according to the overall composition of their floral vol- Our results show that flower color difference between the two atile blend (Fig. 6). Consistent with numerous other studies morphs is determined by differences in concentrations of pig- (Salzmann et al., 2007; Dormont et al., 2010), we observed high in- ments, rather than by differences in chemical components. There ter-individual variability, and the stress value was high (21.75%), is color variation within morphs as well, and we found that, in indicating a low goodness of fit of the ordination distances to the the purple morph, the color properties are correlated with pigment observed distances. The PERMANOVA analysis showed that the concentrations. Taking into account bumblebees’ color perception, two color morphs did not differ significantly in their overall
Fig. 5. (A) Mean reflectance (full line) and standard error of the mean (dotted lines) for spectra curves of purple and yellow petals. (B) Color loci in the hexagonal bumblebees’ color perception space. E(U), E(B), E(G): relative excitations of bumblebees’ UV, blue, green photoreceptors. Sample size (purple, yellow): N = 16, 20. H. Wang et al. / Phytochemistry 94 (2013) 123–134 129
Table 2 Summary of color properties of purple and yellow petals. Welch two-sample t-test was used to test for significant differences in color parameters between purple and yellow morphs (⁄⁄⁄⁄, P < 0.0001; ⁄⁄⁄, P < 0.001; ⁄⁄, P < 0.01; ⁄, P < 0.05; NS, P > 0.05). E(U), E(B), E(G): relative excitations of bumblebees’ UV, blue, green photoreceptors; CC, BC: color contrast and brightness contrast of the color against the background. Sample size (purple, yellow): N = 16, 20.
Purple morph Mean (SD, range) Yellow morph Mean (SD, range) P value Sig. E(U) 0.22 (0.06, 0.12/0.31) 0.23 (0.03, 0.20/0.28) 0.55 NS E(B) 0.55 (0.09, 0.42/0.72) 0.38 (0.02, 0.35/0.41) 3.28 Â 10À6 ⁄⁄⁄⁄ E(G) 0.23 (0.04, 0.15/0.27) 0.38 (0.01, 0.35/0.40) 8.11 Â 10À11 ⁄⁄⁄⁄ CC 0.32 (0.14, 0.13/0.58) 0.15 (0.04, 0.08/0.19) 1.28 Â 10À4 ⁄⁄⁄ BC À0.27 (0.11, À0.44/À0.09) 0.09 (0.05, À0.005/0.18) 3.64 Â 10À11 ⁄⁄⁄⁄
volatile blend composition (pseudo-F=0.15, d.f. = 1, P=0.33), 0.49, respectively) nor for monoterpenoids and sesquiterpenoids whereas the population effect was significant (pseudo-F=1.57, when tested separately (P=0.10 and 0.59, respectively). d.f. = 6, P=0.01). The color-by-population interaction was not sig- In addition, a few qualitative differences (i.e. absence of com- nificant (pseudo-F=0.65, d.f. = 2, P=0.87). pounds in one morph compared to the other) were observed be- The scent blend emitted by I. lutescens, regardless of color, was tween the two morphs. Five compounds were found only in the dominated by monoterpenoids (Table 3). The 3 compounds with volatile emissions from purple flowers, while nine other com- the highest relative abundance were myrcene, limonene and (E)- pounds were present only in volatiles from yellow flowers. All b-ocimene. These compounds are among the most frequently these fourteen ‘specific’ compounds were found in low relative found compounds in floral scents, and they are present in ca. 70% proportions (11 compounds < 1% and 3 others ranging from 1% to of the plant families as reviewed by Knudsen et al. (2006). These 3%) and were found only in a few individuals (all of them occurred three monoterpenoids were the only ones found to occur in more in 2–4 individuals). Interestingly, four of the nine compounds than three-fourths of all individuals sampled (Table 3). Four addi- found exclusively in yellow-morph volatiles are phenylpropa- tional compounds occurred in at least half of the individuals: 1 noids/benzenoids and the five compounds specific to purple- monoterpenoid (terpinolene) and 1 sesquiterpenoid ((E)-caryo- morph profiles are terpenoids. In a color-polymorphic orchid, phyllene) in the purple morph only, and 2 monoterpenoids in both Dormont et al. (2010) also found terpenoids in purple morphs that yellow and purple morphs (Table 3). These compounds represent were absent in white morphs. The other four compounds found only 7 out of 69 volatile compounds (8.6%) in the purple morph exclusively in the yellow morph are fatty-acid derivatives. These and 4 out of 75 volatile compounds (6.7%) in the yellow morph. elements indicate that a color-scent correlation does exist in I. Thus, the scent emitted by I. lutescens individuals is based on a lutescens, even if it is not detectable from an analysis of the overall few frequently occurring terpenoids (mainly monoterpenoids) volatile blend. The negative correlations between terpenoid vola- accompanied by other compounds highly variable in their relative tile proportion and carotenoid concentration on the one hand proportions, with most variation accounted for by identity of pop- and between phenylpropanoids/benzenoids volatile diversity and ulation, as observed in Hesperis matronalis (Majetic et al., 2008). As anthocyanin concentration on the other suggest that the enzymes the most abundant components of I. lutescens scent, terpenoids contributing to the yellow-purple differentiation are located may play the major role in pollinator attraction (Knudsen et al., downstream from volatile synthesis but upstream of pigment syn- 2006). thesis (Majetic et al., 2008). Along with color, scent is a crucial trait in the specificity of the attraction of insect visitors. Any difference in the scent produced 2.4. Is there an association between floral scent and flower color in this by the two morphs that would make them discriminable by insects species? (pollinators, florivores) could modify the patterns of color prefer- ences and the selection pressures exerted by these insects on the As proposed by some authors (Dormont et al., 2010; Majetic color polymorphism (Fenster et al., 2004; Raguso et al., 2003). To et al., 2007; Zuker et al., 2002), we hypothesized that color-scent further explore the color-scent correlation, we focused on the most associations in flowers may result from particular biochemical common compounds, i.e. those that are relevant to the definition of interactions, that can be positive or negative depending on the en- an olfactory signature. We focused on the 7 terpenoids that oc- zyme affected by any mutation leading to the purple-yellow differ- curred in more than half of the individuals (see Section 2.3 and Ta- entiation and its location in the biosynthetic pathway (Majetic ble 3). A PERMANOVA analysis of relative proportions of these et al., 2008). Thus, a simple hypothesis would be that a mutation compounds showed the same result as the analysis of the overall located upstream of the biosynthesis of both volatiles and pig- blend (color effect: pseudo-F=1.30, d.f. = 1, P=0.29; population ef- ments would lead to a positive correlation whereas a mutation lo- fect: pseudo-F=1.91, d.f. = 6, P=0.03; interaction: pseudo-F=0.40, cated downstream from volatile synthesis but upstream of d.f. = 2, P=0.88). Additionally, an evaluation of the multivariate pigment synthesis would lead to a negative correlation. dispersion using the BETADISPER procedure reveals that scent pro- Even if the overall scent composition revealed no color-scent files calculated from these 7 compounds were equally variable correlation, the NMDS ordination seems to be structured according within each morph (F = 0.071, d.f. = 1, P = 0.79). Given these statis- to biosynthetic categories of the compounds found in each sample tical patterns, it is likely that insects would not be able to discrim- (Fig. 6, the samples associated with a higher proportion of sesquit- inate between the two color morphs on the basis of these most erpenoids are located higher on axis 1 and lower on axis 2). There- abundant compounds. fore, to go further into the analysis of color-scent correlation in I. lutescens, we compared the total relative proportions of the three main biosynthetic categories detected (terpenoids, phenylpropa- 3. Concluding remarks noids/benzenoids and fatty-acid derivatives) between the two col- or morphs. Purple flowers emitted a higher proportion of Three important results emerge from the present study. First, terpenoids than yellow flowers (Mann–Whitney U-test, P = 0.03). the most significant difference in pigmentation was that the con- The difference between the morphs was not significant for phenyl- centration of anthocyanins was higher in purple flowers than in propanoids/benzenoids and fatty-acid derivatives (P=0.15 and yellow flowers, while anthocyanin composition was the same for 130 H. Wang et al. / Phytochemistry 94 (2013) 123–134
Table 3 Mean relative composition of Iris lutescens floral scent. Mean values are expressed as a percentage relative to total volatile compounds in each sample. For unidentified unassigned compounds, m/z and the relative abundance of the main ions are indicated. An asterisk (Ã) indicates the compounds with occurrence P8 within at least one morph.
Volatile compound RI Purple flowers (N = 16) Yellow flowers (N = 15) Mean (%) SD CV Occurrence Mean (%) SD CV Occurrence Fatty-acid derivatives 4.21 5.01 1.19 12 7.78 10.36 1.33 12 Hexenol 850 0.06 0.09 1.67 5 0.74 1.67 2.26 6 1-Hexanol 862 0.11 0.24 2.17 5 0.57 2.10 3.69 3 (Z)-3-hexenyl acetate 1011 0.30 0.53 1.79 5 1.06 3.00 2.83 5 Unidentified fatty-acid derivative 1 1075 0.15 0.43 2.92 2 0.18 0.63 3.43 2 Unidentified fatty-acid derivative 2 1091 0.03 0.07 2.78 2 0.17 0.30 1.79 6 Methyl octanoate 1123 0.98 3.93 4.00 1 0.26 0.74 2.81 3 Unidentified fatty-acid derivative 3 1139 0.37 1.06 2.90 3 1.09 1.82 1.67 6 n-Nonanol 1160 0.32 0.90 2.81 2 0.33 1.01 3.10 2 Unidentified fatty-acid derivative 4 1245 0.17 0.57 3.35 2 0.35 1.36 3.87 1 Unidentified alkenol 1253 0.00 0.00 – 0 0.21 0.72 3.47 2 n-Decanol 1275 0.00 0.00 – 0 0.12 0.38 3.12 2 Unidentified fatty-acid derivative 5 1842 0.00 0.00 – 0 0.13 0.38 2.86 2 2-Heptadecanone NA 1.65 3.18 1.93 5 1.84 5.35 2.90 4 2-Methyl-2-butenal NA 0.01 0.03 2.59 3 0.20 0.65 3.18 2 Unidentified fatty-acid derivative 6 NA 0.00 0.00 – 0 0.16 0.45 2.76 2 Terpenoids 84.90 12.89 0.15 16 69.94 21.75 0.31 15 Monoterpenoids 64.40 29.02 0.45 16 50.95 27.64 0.54 15 a-Thujene 916 0.32 0.75 2.37 3 0.07 0.26 3.87 1 a-Pinene 925 2.13 5.43 2.56 3 0.00 0.00 - 0 Sabinene 974 1.54 3.46 2.25 3 0.93 3.60 3.87 1 b-Pinene 980 1.60 3.62 2.25 3 0.87 3.38 3.87 1 Myrcene⁄ 996 13.05 14.70 1.13 15 23.00 19.86 0.86 12 d-3-Carene⁄ 1022 0.29 0.37 1.28 9 0.17 0.31 1.80 5 Limonene⁄ 1034 20.07 18.90 0.94 13 10.76 12.75 1.18 12 b-Phellandrene 1035 0.65 1.79 2.74 5 0.18 0.36 1.98 4 (Z)-Beta-ocimene 1041 1.32 2.26 1.71 7 1.13 1.89 1.68 7 (E)-Beta-ocimene⁄ 1050 10.89 19.13 1.76 14 8.43 18.07 2.14 10 c-Terpinene 1061 0.23 0.42 1.84 4 0.13 0.25 1.96 4 Unidentified monoterpene alcohol⁄ 1074 0.95 2.78 2.93 8 0.41 0.64 1.58 8 a-Pinene oxide 1 1077 0.40 0.87 2.19 3 0.28 0.90 3.19 2 Terpinolene⁄ 1087 0.98 1.23 1.26 10 0.38 0.72 1.88 7 a-Pinene oxide 2 1096 0.71 2.00 2.82 2 0.59 1.80 3.07 2 Linalool 1100 3.78 9.17 2.42 7 2.36 5.11 2.16 7 Menthatriene 1131 0.12 0.38 3.02 2 0.03 0.10 3.87 1 (E)-verbenol 1151 0.30 0.84 2.80 2 0.02 0.09 3.87 1 Geraniol 1255 3.00 8.26 2.75 2 0.00 0.00 - 0 Geranial 1271 0.60 1.82 3.05 2 0.00 0.00 - 0 Sesquiterpenoids 20.50 25.05 1.22 12 19.00 31.30 1.65 11 a-Copaene 1380 0.09 0.34 3.61 3 0.00 0.00 - 0 Bergamotene 1389 0.42 0.89 2.14 4 0.16 0.63 3.87 1 (Z)-a-bergamotene 1419 0.36 0.68 1.92 5 0.10 0.39 3.87 1 (E)-caryophyllene⁄ 1430 2.35 4.86 2.07 8 9.49 21.98 2.32 3 Unidentified oxygenated sesquiterpene 1437 0.03 0.13 4.00 1 0.38 0.95 2.51 3 (E)-a-bergamotene 1439 0.70 1.42 2.02 5 0.20 0.77 3.87 1 (Z)-b-farnesene 1447 1.29 2.48 1.93 5 0.32 1.24 3.87 1 (E)-b-farnesene 1458 2.24 4.31 1.93 7 0.69 2.51 3.65 5 a-Humulene 1467 0.06 0.22 3.56 2 0.27 0.62 2.27 3 Caryophyllene 1 1481 0.01 0.06 4.00 1 0.14 0.34 2.39 3 Acoradiene 1487 0.32 0.79 2.44 5 0.08 0.33 3.87 1 Ar-curcumene 1491 0.11 0.26 2.28 4 0.09 0.36 3.87 1 Farnesene 1496 2.75 5.20 1.89 5 0.97 3.75 3.87 1 (Z,E)-a-farnesene 1498 0.11 0.42 3.71 2 0.03 0.06 2.15 3 a-Zingiberene 1505 1.06 3.40 3.22 4 0.02 0.09 3.87 1 (E,E)-a-farnesene 1513 3.06 9.00 2.94 7 1.45 3.99 2.75 5 b-Bisabolene 1518 0.45 0.87 1.93 5 0.10 0.40 3.87 1 Unidentified sesquiterpene 1 1520 0.15 0.31 2.09 4 0.04 0.15 3.87 1 b-Sesquiphellandrene 1533 3.03 5.95 1.96 5 1.32 5.10 3.87 1 Caryophyllene 2 1582 0.02 0.09 4.00 1 0.18 0.56 3.09 2 Caryophyllene-oxide 1592 1.39 4.18 3.01 2 1.94 4.29 2.21 3 Unidentified sesquiterpene 2 1620 0.11 0.41 3.74 2 0.00 0.00 - 0 Caryophylladienol 1653 0.04 0.18 4.00 1 0.23 0.60 2.58 3 Unidentified sesquiterpene alcohol 1 1673 0.06 0.25 4.00 1 0.27 0.63 2.31 3 Unidentified sesquiterpene alcohol 2 1690 0.04 0.18 4.00 1 0.22 0.53 2.44 3 Phenylpropanoids/benzenoids 10.39 13.18 1.27 13 21.57 22.63 1.05 13 Benzaldehyde 963 0.01 0.04 4.00 1 2.43 9.14 3.76 2 2-(4-Methoxyphenyl)ethanol 1117 0.21 0.51 2.46 3 0.63 2.27 3.62 3 Phenylacetonitrile 1141 1.84 4.08 2.22 6 1.57 3.92 2.50 5 Veratrole 1146 0.37 1.46 4.00 1 0.97 2.84 2.92 3 p-Dimethoxybenzene 1165 2.88 11.52 4.00 1 2.65 7.01 2.64 2 3-Phenyl-propanol 1236 0.00 0.00 – 0 0.40 1.20 3.01 2 H. Wang et al. / Phytochemistry 94 (2013) 123–134 131
Table 3 (continued)
Volatile compound RI Purple flowers (N = 16) Yellow flowers (N = 15) Mean (%) SD CV Occurrence Mean (%) SD CV Occurrence p-Anisaldehyde 1261 0.49 1.07 2.21 3 0.13 0.41 3.15 2 Chavicol 1264 1.48 3.75 2.53 4 1.02 2.04 2.00 4 (E)-cinnamaldehyde 1279 0.00 0.00 – 0 0.32 0.97 3.06 2 (Z)-methyl cinnamate 1307 0.04 0.15 4.00 1 0.18 0.54 3.00 2 Unidentified benzenoid 1 1357 0.00 0.00 – 0 1.51 4.86 3.21 3 Eugenol 1358 0.44 1.04 2.35 6 0.21 0.51 2.48 3 Methoxyphenyl-ethanol 1375 0.93 3.72 4.00 1 3.37 9.54 2.83 2 (E)-methyl-cinnamate 1389 1.00 3.57 3.55 2 3.38 9.19 2.72 4 Unidentified benzenoid 2 isomer 1 1411 0.01 0.05 4.00 1 1.13 4.08 3.61 2 Unidentified benzenoid 2 isomer 2 1416 0.00 0.00 – 0 0.36 1.24 3.44 2 (E)-nerolidol 1565 0.57 1.72 3.03 3 1.10 2.05 1.86 4 Unidentified 0.50 1.42 2.85 4 0.71 1.41 2.00 4 41;43/90;67/75;69/60;71/50 1159 0.00 0.00 – 0 0.53 1.11 2.09 4 55;69/75;41/50;43/50;84/40 1184 0.13 0.37 2.81 2 0.15 0.39 2.67 2 84;41/75;55/75;69/60;109/60;134/50 1239 0.34 1.33 3.87 2 0.01 0.05 3.87 1
RI, Kovats retention index (calculated from retention times of C8–C19 n-alkanes, consequently available only for compounds with RI between 800 and 1900); SD, standard deviation; CV, coefficient of variation; occurrence, the number of individuals in which the compound was detected.
Fig. 6. NMDS ordination on relative scent composition of purple and yellow flowers (colored symbols). This analysis computes the two-dimensional locus for each individual (floral scent blends) and each variable (volatile organic compounds). Black symbols represent the mean location of the volatile organic compounds belonging to each biosynthetic category. Black lines represent standard deviations. Unidentified compounds are not represented. both morphs. No pigmentation difference (absolute or relative con- (Delle-Vedove et al., 2011; Dormont et al., 2010; but see Salzmann centrations) was detected in leaves. Thus abiotic factors, herbi- and Schiestl, 2007). I. lutescens is a rewardless species (see Sec- vores (excluding florivores) and pathogens are unlikely to have tion 4.1 below) displaying mainly color differences that are not influenced the evolution of flower color polymorphism in this spe- associated with differences in scent composition (at least for the cies. Secondly, purple- and yellow-flowered I. lutescens could be most abundant compounds) nor level of defense against biotic/abi- visually discriminated by bumblebees, which perceive them as otic factors, suggesting that this species is a good candidate for ‘‘blue’’ and ‘‘blue-green’’, respectively. We suspect that a wide vari- investigating the maintenance of flower color polymorphism. For ety of insects, including florivores, could similarly discriminate be- instance, the scent emitted by I. lutescens was dominated by a tween these color morphs as well. Thirdly, the color polymorphism few frequently occurring terpenoids (the most frequent and di- is not associated with significant difference in the scent emitted verse biosynthetic categories, Knudsen et al., 2006) accompanied between the two morphs in I. lutescens, and what an insect smells by other compounds that occurred less frequently and were highly does not match what it sees, as in several previous cases in orchids variable in their relative proportions, mainly due to a population 132 H. Wang et al. / Phytochemistry 94 (2013) 123–134 effect (Dormont et al., 2010; Majetic et al., 2008). The maintenance yellow and 16 purple individuals, originating from the Puechredon of such polymorphism is often considered to be complex and population. Floral scent extractions were performed in February strongly influenced not only by pollinators but also by other selec- 2010 on 15 yellow- and 16 purple-morph individuals originating tive agents such as herbivores and abiotic factors through selection from all the eight populations. All the statistical analyses were per- acting on other traits correlated with flower color (Dormont et al., formed using R statistical program version 2.10.1 (R Development 2010; Rausher, 2008; Schaefer et al., 2004; Schaefer and Ruxton, Core Team, 2009). 2009). However, our results strongly suggest that pollinators are the most probable selective agent responsible for the maintenance 4.3. Analyses of pigments of this polymorphism, and further experiments, in particular test- ing pollinators’ behavioral responses to such color difference, are The sepals, petals and leaves of each individual were collected now required to formally demonstrate the mechanism(s) responsi- separately, weighed and immediately stored at À80 °C until pig- ble in I. lutescens. ment extraction. Total flavonoids and anthocyanins were extracted with 0.5% HCl in methanol (1.5 mL for each sample), sonicated for 10 min, and kept overnight at 4 °C in darkness (Harborne, 1967; de 4. Materials and methods Rosso and Mercadante, 2007). After centrifugation at 5000 rpm for 10 min, the supernatant was stored at À80 °C. Similarly, total 4.1. Study organism carotenoids were extracted with acetone (1.5 mL) overnight; after centrifugation, the supernatant was dried at room temperature for I. lutescens Lam. (Iridaceae) is a perennial rhizomatous species, 24 h, and further separated with diethyl ether (0.75 mL) and 1% with a range that extends from Spain through France to Italy. NaCl (0.75 mL) in distilled water (Kishimoto et al., 2005). The crude The species occurs in dry places in the Mediterranean region of carotenoid extracts were finally transferred to diethyl ether, dried these countries. It grows 10–30 cm tall, with erect, sword-shaped and stored at À80 °C. leaves and one showy flower at the apex of each shoot. Each flower For quantitative analyses, UV–visible spectra (300–700 nm) of has three pendant, bearded sepals, alternating with three erect pet- the two extracts were recorded on a Jasco J-815 spectropolarimeter als (Fig. 1). Flowers are hermaphroditic but self-incompatible, and (Jasco, Tokyo, Japan). In the acid methanol extracts, the absorbance thus highly pollinator-dependent. Flowers are visited by bumble- at 540 nm and 350 nm was used to quantify total anthocyanins bees (Bombus spp.) and by solitary bee species of several genera (TA) and total flavonoid pigments (TF, including chalcones, flav- (Eucera, Nomada, Andrena). A florivorous beetle (Tropinota hirta, ones and flavonols), respectively, and delphinidin-3-O-rhamnoside Cetoniidae) also commonly visits the species. Flowering occurs in chloride and rutin (both from Extrasynthèse, Genay, France) were early spring, and flowers do not produce any nectar reward for used as standard compounds. In the diethyl ether extracts, the their pollinators. Like many rewardless species, I. lutescens mainly absorbance at 444 nm was used to quantify total carotenoids exploits newly emerged insect pollinators and fruit set in natural (TC) with lutein as a standard compound. The amounts of TA, TF populations is low (Imbert and Schatz, pers. observ.). and TC were then calculated in mg/100 mg fresh weight. Populations of I. lutescens in southern France are polymorphic For identification of anthocyanin compounds in the acidic for flower color without any spatial segregation of morphs. So far methanol extracts, HPLC–DAD-ESI/MS2 analyses were performed as we know, populations are monomorphic in Spain. The two dom- on a Waters alliance 2790 liquid chromatography coupled with a inant French morphs, purple and yellow, are common in most nat- diode array detector and a tandem quadrupole time-of-flight ural populations (on the basis of more than 50 populations (QqTOF) mass spectrometer (all Waters, Milford, USA). Separations studied), with the frequency of the yellow morph ranging from were done on a Waters Symmetry C18 (4.6 Â 250 mm, 5 lm) col- 0.04 to 1, and averaging around 0.6. Blue and white morphs also umn at room temperature. Mobile phases were 0.1% trifluoroacetic rarely occur. Flower color morphs do not differ in flower shape, acid in water (A) and 0.1% trifluoroacetic acid in HPLC-grade aceto- pollen and ovule production or vegetative characteristics, but differ nitrile (B), running a linear gradient from 0% to 50% B over 30 min. in flower size (purple flowers are larger, Imbert et al., in prep.). The flow rate was 0.7 mL/min, and the injection volume was 50 lL. The chromatogram was monitored at 540 nm, and UV–visible 4.2. Population origin of the sampled individuals spectra were recorded from 200 to 600 nm. The mass spectrometer equipped with an electrospray ionization interface was operated in The experiments were carried out on transplanted individuals the positive electrospray ionization mode (ESI(+)). The capillary sampled from eight natural populations and reared in a common voltage was 3 kV and the capillary temperature was set at garden (for details on locations see Supplementary Table 1)in 250 °C. The cone voltage was 20 V. Mass spectra were recorded be- southern France. Populations were chosen based on their accessi- tween m/z 100 and 1500. Selected ions (see Table 1 for detail infor- bility around Montpellier and to represent a range of frequency mation) were then fragmented in MS2 and ISD-MS2 analyses. The of the yellow morph. In 2006, 35 pieces of rhizomes with two collision energy was adjusted to 30 eV in MS2 analyses; in-source leaves each were collected in each of five natural populations. Only dissociation (ISD) was performed by increasing cone voltage to 25 rhizomes were collected in the smallest population (Cazevieil- 50 V to induce fragmentation in atmospheric source conditions, le). Rhizomes were transplanted in 1 L pots with sterile soil. For and ions formed this way were then selected to perform classical two populations (La Clape and Puechredon), mature seeds were MS2. collected in June 2008. Seeds were allowed to germinate in sterile soil and seedlings (39 for La Clape and 15 for Puechredon) were 4.4. Measurements of flower color transplanted into 1 L pots with sterile soil during the following spring. Surveys of the flower color of individual plants during con- The reflectance spectra of petals were measured using an Ava- secutive years has shown that color morphs are stable, suggesting Spec-2048 fiber-optic spectrometer, an AvaLight-DH-S deute- the polymorphism observed has a genetic basis. rium-halogen light source and an FCR-7UV200-2-ME-SR All measurements were performed on individuals under green- reflection probe with optic fiber (all Avantes, Eerbeek, the Nether- house conditions. The individuals were randomly selected when lands). Each spectrum was recorded between 300 and 700 nm, cov- flowers were at the full-bloom stage. Measurements of flower color ering the spectral sensitivity range for most insects (Briscoe and and pigment extractions were performed in February 2012 on 20 Chittka, 2001). H. Wang et al. / Phytochemistry 94 (2013) 123–134 133
The spectrometric data were then analyzed with AVICOL v.6 composition of floral scents emitted by purple- and yellow- (Gomez, 2006). To describe flower color as perceived by insects, flowered individuals was compared using Non Metric Multidimen- Chittka’s physiological model for trichromatic vision was used. sional Scaling (NMDS). The effects of color and geographical origin This model offers a visual representation of color perception (the were tested using Permutational Non-Parametric Multivariate color hexagon) by any trichromatic species (bees, bumblebees...), Analysis of Variance (PERMANOVA, see Anderson, 2001 for details). depending on the photoreceptor spectral sensitivity functions that Both analyses were based on Bray-Curtis distances and performed are introduced into the model (Chittka, 1992). We described I. using the Vegan package for R statistical software (Oksanen et al., lutescens flower color perception by bumblebees by introducing 2011). Because some categories of volatile compounds are pro- spectral sensitivity of the three photoreceptors of Bombus terrestris duced by the same biosynthetic pathways as the main pigment (Skorupski et al., 2007). Additional required parameters are light classes, we tested independently the effect of color on the total source and background. As a light source we used CIE standard illu- proportions of the three main biosynthetic categories of volatile minant D65, which has an irradiance spectrum similar to midday compounds (terpenoids, phenylpropanoids/benzenoids, fatty-acid sun in Western Europe. The background of the flowers basically derivatives) present in our samples using Mann–Whitney U-tests. consists of their green leaves, and thus we used the mean leaf spec- Finally, as only the most occurring volatile compounds may be rel- trum, which was obtained from leaves of 12 randomly chosen indi- evant to the recognition of I. lutescens flowers by pollinators (or viduals including both color morphs, as the reflectance spectrum of other insects), we analyzed the patterns of variation of the relative the background. proportions of the volatile compounds that appeared in at least The color parameters calculated for statistical analyses were the 50% of the samples within at least one of the two color morphs relative excitation values of bumblebees’ UV, blue and green pho- using PERMANOVA and an evaluation of multivariate dispersion toreceptors (E(U), E(B), E(G)), color contrast and brightness con- by BETADISPER procedure (vegan package, Anderson, 2006). trast of the color against the background (CC,BC). Color space coordinates (X,Y) of each individual’s petal spectrum, which were Acknowledgements converted from the relative excitation values of bumblebees’ UV, blue and green photoreceptors, were used to assign to each indi- The authors would like to thank David Carbonell and the staff of vidual a locus in the hexagonal color perception space. the CEFE’s field experiment station for their help in cultivating the irises. We also thank Bruno Buatois for chemical analyses in the la- 4.5. Analyses of floral volatiles bex CEMEB platform PACE (Platform for Chemical Analyses in Ecol- ogy) and Professor Christine Enjalbal for chemical analyses in the Floral volatiles were sampled by solid-phase microextraction IBMM technical platform LMP (Laboratory for Physical Measure- (SPME) as described in Dormont et al. (2010), using 65 lm poly- ments). We are grateful to D. McKey and R. Kassen for their com- dimethylsiloxane/divinylbenzene (PDMS-DVB) fibers (Supelco, Sig- ments on the manuscript. Financial support was provided by ma–Aldrich, Bellefonte, PA, USA). The whole flower was enclosed in Université Montpellier 2 and the CNRS, the French National Center a bag made from polyethylene terephthalate (Nalophan; Kalle Nalo for Scientific Research. Hui Wang acknowledges a grant from the GmbH, Wursthüllen, Germany). The fiber was introduced into the China Scholarship Council. This is publication ISEM-2013-067 of Nalophan bag containing the flower and exposed for 45 min in the Institut des Sciences de l’Évolution, Montpellier. close proximity (2 cm) to flowers. A control bag was also sampled: an SPME fiber was inserted into an empty Nalophan bag, in order to monitor volatiles from the air surrounding the plant. All the Appendix A. Supplementary data samplings were performed between 1:30 pm and 3:30 pm, during the period of maximum activity of insects in the day, and the sam- Supplementary data associated with this article can be found, in ples were then stored at À20 °C until analyses. This method is not the online version, at http://dx.doi.org/10.1016/j.phytochem.2013. destructive and the plant remains intact after removing the bag in 05.007. the end of experiment. Gas chromatography–mass spectrometry (GC–MS) analyses References were carried out using a Shimadzu QP2010plus GC–MS system (Shimadzu, Kyoto, Japan), with a SLB-5MS (30 m  0.25 mm, Adams, R.P., 2007. Identification of Essential Oil Components by Gas 0.25 lm) fused silica capillary column (Supelco). The injection Chromatography/Mass Spectrometry, fourth ed. Allured Publishing Corp, temperature was set to 250 °C. The column temperature was pro- Carol Stream, Illinois, USA. Anderson, M.J., 2001. A new method for non parametric multivariate analysis of grammed as follows: 40 °C held for 2 min, increased at 5 °C/min to variance. Austral Ecol. 26, 32–46. 200 °C, and finally up to 250 °Cat10°C/min and held for 3 min. Anderson, M.J., 2006. Distance-based tests for homogeneity of multivariate Helium was used as carrier gas with the column flow 1.0 mL/ dispersions. Biometrics 62, 245–253. Ando, T., Saito, N., Tatsuzawa, F., Kakefuda, T., Yamakage, K., Ohtani, E., Koshi-ishi, min. The transfer line and the ion source temperature were set M., Matsusake, Y., Kokubun, H., Watanabe, H., Tsukamoto, T., Ueda, Y., to 250 °C and 200 °C, respectively. The potential for electron im- Hashimoto, G., Marchesi, E., Asakura, K., Hara, R., Seki, H., 1999. Floral pact ionization was 70 eV. The spectrometer was used in scan anthocyanins in wild taxa of Petunia (Solanaceae). Biochem. System. Ecol. 27, 623–650. mode between m/z 38 and 350. All the volatile compounds were Armbruster, W.S., 2002. Can indirect selection and genetic context contribute to identified by comparison with the NIST 2005 software mass spec- trait diversification? A transition-probability study of blossom-color evolution tra library (Varian, Palo Alto, CA, USA), online mass spectra libraries in two genera. J. Evol. Biol. 15, 468–486. (Pherobase, El-Sayed, 2011; NIST Chemistry WebBook, Linstrom Ashman, T.-L., Majetic, C.J., 2006. Genetic constraints on floral evolution: a review and evaluation of patterns. Heredity 96, 343–352. and Mallard, 2011), and retention indices found in the literature Ashtakala, S.S., Forward, D.F., 1971a. Pigmentation in iris hybrids: the nature and (Adams, 2007). development of flavonoid pigments and their effect on flower color in cultivated A total of 170 volatile compounds were detected, of which only hybrids of Iris germanica. Can. J. Botany 49, 1965–1973. Ashtakala, S.S., Forward, D.F., 1971b. Pigmentation in iris hybrids: occurrence of 117 were found in at least two individuals of the same color. flavonoid pigments in six cultivars of Iris germanica. Can. J. Botany 49, 1975– Among these compounds, only 80 accounted for more than 1% of 1979. the total scent emitted by at least one individual. The relative pro- Briscoe, A.D., Chittka, L., 2001. The evolution of color vision in insects. Annu. Rev. Entomol. 46, 471–510. portions of these 80 compounds were considered as individual Burger, H., Dötterl, S., Ayasse, M., 2010. Host-plant finding and recognition by visual scent profiles and used for statistical analyses. First, the chemical and olfactory floral cues in an oligolectic bee. Funct. Ecol. 24, 1234–1240. 134 H. Wang et al. / Phytochemistry 94 (2013) 123–134
Chalker-Scott, L., 1999. Environmental significance of anthocyanins in plant stress Milet-Pinheiro, P., Ayasse, M., Schlindwein, C., Dobson, H.E.M., Dötterl, S., 2012. Host responses. Photochem. Photobiol. 70, 1–9. location by visual and olfactory floral cues in an oligolectic bee: innate and Chittka, L., 1992. The colour hexagon: a chromaticity diagram based on learned behavior. Behav Ecol 23, 531–538. photoreceptor excitations as a generalized representation of colour Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., O’Hara, R.B., Simpson, G.L., opponency. J. Comp. Physiol. A 170, 533–543. Solymos, P., Stevens, M.H.H., Wagner, H., 2011. Vegan: community ecology Chittka, L., Raine, N.E., 2006. Recognition of flowers by pollinators. Curr. Opin. Plant package. R Package Version 1.17-11. Available at:
学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，
提供一站式文献检索和下载服务”的24 小时在线不限IP 图书馆。 图书馆致力于便利、促进学习与科研，提供最强文献下载服务。
图书馆导航：
图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具