Color and Scent: How Single Genes Influence Pollinator Attraction

H. SHEEHAN,K.HERMANN, AND C. KUHLEMEIER Institute of Plant Sciences, Altenbergrain 21, CH-3013 Bern, Switzerland Correspondence: [email protected]

A major function of angiosperm is the recruitment of animal pollinators that serve to transfer among conspecific plants. Distinct sets of floral characteristics, called syndromes, are correlated with visitation by specific groups of pollinators. Switches among pollination syndromes have occurred in many plant families. Such switches must have involved coordinated changes in multiple traits and multiple genes. Two well-studied floral traits affecting pollinator attraction are petal color and scent production. We review current knowledge about the biosynthetic pathways for floral color and scent produc- tion and their interaction at the genetic and biochemical levels. A key question in the field concerns the genes that underlie natural variation in color and scent and how such genes affect pollinator preference, reproductive isolation, and ultimately speciation.

Pollination syndromes are sets of floral traits that are been thought of as coevolutionary, with plants also exert- adapted to different guilds of animal pollinators (Faegri ing selection on pollinators (Feinsinger 1983). However, 1979; Fenster et al. 2004). The floral traits that constitute more recent work in traits of attraction, color and scent a particular can be separated into has suggested that the plants have adapted their traits three categories: traits that attract the pollinator, such as to sensory capabilities already present in the pollinators floral color and scent; traits that reward the pollinator, (Schiestl and Do¨tterl 2012), although there may be some such as volume; and efficiency traits, such as the fine-tuning of pollinator sensory preferences (Chittka and positioning of reproductive organs that affect pollen re- Menzel 1992; Raine and Chittka 2007). moval and deposition (Bradshaw et al. 1995). Although it Switches among syndromes are widespread. For in- is debated whether pollination syndromes can accurately stance, in the Solanaceae, hummingbird pollination is predict the type of pollinating visitor to a , the thought to have arisen at least 10 times (Knapp 2010). pollination syndrome concept is a useful way to try to This raises the question of the genetic basis of the evolu- understand floral diversification and to frame questions tion of pollination syndromes. If each of the individual about evolutionary transitions (Waser et al. 1996; Kings- traits is encoded by multiple genes that must coordinately ton and Mc Quillan 2000; Fenster et al. 2004; Ollerton evolve, how is it possible that switches among syndromes et al. 2009; Danieli-Silva et al. 2012). Pollination syn- happen frequently? To answer this question, it is neces- dromes can be considered as a spectrum, with generalist sary to identify the underlying plant genes that are under species that produce flowers that appeal to many pollina- selection by animal pollinators. In their seminal study, tors at one end and flowers that require pollination by Bradshaw et al. (1995, 1998) investigated the genetic specific pollinators at the other end (Fenster et al. 2004). bases for multiple divergent pollination syndrome traits The force behind the convergent evolution of pollina- in the monkey flower Mimulus. In its natural habitat, tion syndromes across plant families is thought to be Mimulus lewisii is pollinated by bumblebees, whereas selection pressure exerted by the particular guilds of pol- M. cardinalis is pollinated by birds. The two species linators (van der Pijl 1960; Faegri 1979; Fenster et al. live in habitats that are altitudinally separated except for 2004). Animals have an innate preference for certain flo- a hybrid zone in which they overlap. Although the species ral attributes and/or learn to associate the flower with are cross-compatible, they are almost completely repro- providing a food source, such as nectar or pollen (Riffell ductively isolated by the different pollinators that they 2013). In cases in which a variant arises that is more attract (Schemske and Bradshaw 1999; Ramsey et al. attractive to a particular pollinator, this will lead to a 2003). To investigate the genetics defining the differenc- higher number of pollinator visits and may result in an es in pollination syndromes between these two species, a increase in the fitness of the plant. This increase in fitness quantitative trait locus (QTL) analysis was performed. It can cause the trait to spread as the variant allele increases was found that all of the pollination syndrome traits in- in frequency in the population. vestigated (which included traits of pollinator attraction, Floral constancy shown by the pollinator causing as- reward, and efficiency) could be explained by one to six sortative mating may then lead to prezygotic isolation and QTLs each (Bradshaw et al. 1995). This implies that there possibly even speciation. This process has traditionally may only be a limited number of genes that determine

Copyright # 2012 Cold Spring Harbor Laboratory Press; all rights reserved; doi: 10.1101/sqb.2013.77.014712 Cold Spring Harbor Symposia on Quantitative Biology, Volume LXXVII 117 118 SHEEHAN ET AL. differences in individual pollination traits and that genes to survive in the wild. Thus, we focus here on natural of large effect may have a central role in the pollinator- color variants and the genes that determine these. determined prezygotic isolation of plant species. So far, the molecular basis of natural variation is Quantitative trait loci analysis provides rough map po- known for only two pollination syndrome traits: petal sitions, and each of the identified loci may still consist of color, specifically anthocyanin-based color, and scent multiple linked genes. Thus, the next logical step is to production. We discuss the types of genes that have identify the genes underlying the QTL. The search for been found, their role in the attraction of different classes “pollination syndrome genes” in plants has been in full of pollinators, and the similarities among species. We progress during the last 10 years. Once isolated, it is pos- focus specifically on anthocyanin-based petal color and sible to determine the number and nature of the genetic volatile benzenoid-derived scent, both of which are changes required to change an individual floral trait. Do derived from phenylalanine (see Fig. 1). We base our genetic changes preferably occur in structural or in regu- discussion around the genus Petunia, in which the bio- latory genes and is there evidence for selection? Inter- chemistry and molecular genetics of the two pathways specific crosses followed by reciprocal introgressions of have been worked out in considerable detail. single loci also provide attractive material for pollinator The genus Petunia comprises 14 species, 11 of which choice studies, both under controlled laboratory condi- show a typical bee pollination syndrome (Stehmann et al. tions and in the natural habitat. They make it possible to 2009). Petunia integrifolia, a representative of the bee determine the extent to which differences in single traits/ clade, has a purple corolla with a wide tube that gives loci condition differences in pollinator visitation and their bees access to the nectar, and stamens and stigma within effect on female and male fitness in the plant. Ultimately, the length of the tube (Fig. 2A) (Stuurman et al. 2004). transgenic lines can determine the effect of single genes. Petunia axillaris possesses a hawkmoth pollination syn- It is important here to make the distinction between drome (Fig. 2B), with white flowers that produce a strong gene polymorphisms that have arisen in nature (natural scent at dusk. Hawkmoths reach the abundant dilute nec- variants) versus those that have arisen in the laboratory tar by extending their probosces through the long, narrow (mutants) or by breeding (varieties). Most mutants and floral tubes (Stuurman et al. 2004). P. integrifolia con- varieties will have reduced overall fitness and are unlikely forms to its pollination syndrome, with visits primarily by

Figure 1. The amino acid phenylalanine is the common precursor of volatile benzenoids, flavonoid pigments, and other secondary metabolites with diverse functions. The positions of major products in the pathway are depicted. Enzymes are shown in red. Black arrows indicate a single enzymatic step; dashed arrows indicate more than one enzymatic step. (4CL) 4-Coumarate CoA-ligase, (C4H) cinnamate 4-hydroxylase, (CHS) chalcone synthase, (DFR) dihydroflavonol reductase, (FLS) flavonol synthase, (PAL) phenylalanine ammonia lyase. POLLINATION SYNDROME GENES 119

Figure 2. Closely related Petunia species attract different animal pollinators: Petunia integrifolia with the solitary bee species Callonychium petuniae (A), Petunia axillaris ssp. axillaris with the hawkmoth Hyles lineata, Uruguay 2008 (B), Petunia exserta lab-line with the hummingbird Hylocharis chrysura, Uruguay 2009 (C). Photographs: Alexandre Dell’Olivo. bee species (Gu¨bitz et al. 2009), whereas P. axillaris is competition with bees, because detection of red flowers pollinated by both nocturnal hawkmoths and diurnal bees by bees may not be as efficient as detection of other (Hoballah et al. 2007). Petunia exserta has bright red colors (Spaethe et al. 2001; Lunau et al. 2011). petals that are reflexed, with the reproductive organs ex- Associative learning is also used by pollinators to deter- serted from the floral tube (Fig. 2C) (Gu¨bitz et al. 2009). mine which particular colored flowers will provide a re- These traits are typical for the hummingbird pollination ward (Bene´ 1941; Miller and Miller 1971; Weiss 1997; syndrome, but detailed field observations of this rare spe- Goyret et al. 2008). In some cases, learned colors may cies are scarce (Lorenz-Lemke et al. 2006). become stronger than innate preferences (Kelber 1997; Goyret et al. 2008). In other cases, it seems that innate color preferences are valued at a similar level (Weiss FLORAL COLOR 1997) or may come into play again when learned colors are not helpful (Gumbert 2000). For instance, in honey- Floral Color Can Influence Pollinator bees, floral constancy has been shown for short bouts of Behavior in Different Ways foraging that could be attributed to associative learning Particular floral colors characterize different pollina- and short-term memory (Raine and Chittka 2007), al- tion syndromes and are associated with particular classes though innate preferences are still relevant because hon- of pollinators. This must be conditioned in part by the eybees are thought to learn innately preferred colors more perceptive abilities of the pollinators. All insects can see rapidly than noninnately preferred colors (Giurfa et al. into the ultraviolet (UV) but have limited ability to see 1995). Color patterns can also influence the attractiveness into the red end of the light wavelength spectrum (Menzel of the flower and increase the efficiency of pollination. For and Backhaus 1991). Pollinating birds have four photo- example, nectar guides are patterns of contrasting colors, receptors with peak sensitivities in the UV, blue, green, often comprising “bulls-eye” spots at the mouth of the and red light ranges (Herrera et al. 2008). The association floral tube. Nectar guides increase pollinator efficiency of floral colors with particular pollinators is also likely to by helping insects to orient on the flower and guide them be affected by innate color preferences held by the polli- to the food source more quickly than they would without nators. Innate preferences have been shown in many spe- (Waser and Price 1981, 1983; Dinkel and Lunau 2001). cies to initially attract pollinators or aid recognition of The question of whether an alteration in floral color flowers (Giurfa et al. 1995; Kelber 1997; Weiss 1997; alone can affect pollinator attraction was addressed in Gumbert 2000; Goyret et al. 2008). For instance, bee further work on Mimulus (Schemske and Bradford pollination syndromes are associated with blue and yel- 1999; Bradshaw and Schemske 2003). The bumblebee- low flowers (Faegri 1979). Bees have photoreceptors for pollinated M. lewisii possesses pink flowers, whereas blue, green, and UV light (for review, see Chittka and hummingbird-pollinated M. cardinalis has red flowers. Raine 2006) and both bumblebees and honeybees show The difference in color can be attributed to one locus, innate preferences for blue wavelengths (Giurfa et al. yellow upper (yup), from which the dominant allele 1995; Gumbert 2000). However, innate preferences do from M. lewisii prevents carotenoid accumulation, where- not always match the pollination syndrome. Hawkmoths as the homozygous recessive yup allele from M. cardi- display an innate preference for blue despite the preva- nalis enables carotenoid accumulation (Bradshaw and lence of white, UV-absorbing flowers being pollinated by Schemske 2003). hawkmoths (Goyret et al. 2008). Other pollinators do not Reciprocal introgressions of the yup locus into the pa- seem to have innate preferences at all. For instance, the rental lines resulted in a yellow-orange M. lewisii line and hummingbird pollination syndrome is associated with a dark-pink M. cardinalis line. Small differences in six red flowers; however, hummingbirds do not have an in- other traits important for pollinator visitation were pre- nate preference for red (Miller and Miller 1971). Rather, sent but were probably not functionally significant (Brad- the color association may have evolved in avoidance of shaw and Schemske 2003). The differences in color 120 SHEEHAN ET AL. altered the preferences of the pollinators, decreasing the The type of anthocyanins produced can be influenced attraction of the species to its original pollinator and in- by the presence/absence as well as substrate specificities creasing the attractiveness to its noncognate pollinator. of the biosynthetic enzymes. P. integrifolia has highly This shows that color likely has an important role in decorated anthocyanins, such as malvidin-3-( p-cou- maintaining the reproductive isolation of these two Mim- maroyl)-rutinoside-5-glucoside (Fig. 3), whereas P. ulus species. It also indicates that a change at a single exserta produces simple anthocyanins, such as delphini- locus can alter the pollinator class attracted to a plant. din-3-rutinoside and cyanidin-3-rutinoside (Fig. 3) Although at this time the underlying gene (or genes) (Ando et al. 1999). The low affinity of P. hybrida DFR has not been identified, it seems likely that the yup locus for dihydrokaempferol and dihydromyricetin results in has a role in the regulation or biosynthesis of carotenoid the production of mainly delphinidin-based anthocyanins compounds. (Gerats et al. 1982; Tornielli et al. 2009). This is likely Can a change in pollinator preference between two attributable to the specific amino acid sequence in the species be pinpointed to a difference in one gene? To putative substrate-binding region of DFR (Johnson et al. study the role of flower color in pollinator attraction, we 2001). Competition for substrates may occur between the have used Petunia as a fitting model system. In Petunia, different branches of the anthocyanin pathway and/or the contrasting color differences in the pollination syn- between the anthocyanin and other branches of the flavo- dromes are characterized by anthocyanin-based petal noid pathway. For instance, studies have shown that when colors. The anthocyanin pathway is one of the best-char- a dominant flavonol ( fl) locus is present in P. hybrida, acterized metabolic pathways in plants and much of the the anthocyanin biosynthetic genes likely cannot com- biochemical and molecular-genetic studies were per- pete against the flavonol biosynthetic gene FLS for the formed in Petunia (Holton and Cornish 1995; Koes common substrate dihydroquercetin, resulting in a lower et al. 2005; Tornielli et al. 2009). production of cyanidin-derived anthocyanins (Gerats et al. 1982).

The Anthocyanin Pathway Is a Useful Model in which to Study Differences in Floral Color Regulation of the Anthocyanin Pathway Is by a Conserved Group of Transcription Factors Anthocyanins, a class of flavonoid, are the predominant floral pigment in angiosperms (Winkel-Shirley 2001). Floral pigmentation mutants in Petunia have revealed a Other flavonoids also serve as floral pigments: aurones number of genes that activate anthocyanin structural produce yellow flowers in some species (e.g., see Schwinn genes. AN2, AN1, and AN11 code for a R2R3-MYB tran- et al. 2006), whereas flavonols are important for absorp- scription factor (R2R3-MYB), a basic-helix-loop-helix tion in the UV light range and can act as copigments in (bHLH) transcription factor, and a WD40-repeat (WDR) the presence of anthocyanins. Other components, such as protein, respectively. These proteins act in a complex the pH of the vacuoles in which the anthocyanins are (MYB-bHLH-WDR) to coordinately up-regulate the ex- stored, can also alter the light-absorption properties of pression of late anthocyanin biosynthetic genes, from DFR anthocyanin compounds (Grotewold 2006). onward (Beld et al. 1989; Quattrocchio et al. 1993). An- The anthocyanin biosynthetic pathway is depicted in other, as-yet-unidentified group of regulators must be re- Figure 3. Excellent reviews have been written on antho- sponsible for expression of the early biosynthetic genes cyanin biosynthesis and its regulation in Petunia and because they are expressed independently (Tornielli et al. other species (Koes et al. 2005; Grotewold 2006; Quat- 2009). It is proposed that this separate regulation provides trocchio et al. 2008; Tornielli et al. 2009). Here, we give a a mechanism for the diversification of color while allow- brief overview of the pathway and focus on the main genes ing other classes of flavonoids, with other roles in the that influence color. The early enzymes of the path- plant, to be produced in tissues in which anthocyanin pig- way convert phenylalanine to dihydroflavonols (dihydro- ments are not present (Quattrocchio et al. 2008). kaempferol, dihydroquercetin, and dihydromyricetin), the A number of closely related R2R3-MYBs confer pig- precursors for both anthocyanins and UV-absorbing fla- mentation in specific tissues. The R2R3-MYB AN2 is ex- vonols. Flavonol synthase (FLS) specifically catalyzes the pressed mainly in the petal limb in which it is the major synthesis of the flavonols, whereas dihydroflavonol re- regulator of anthocyanin production, whereas AN4 is pri- ductase (DFR) is the first dedicated enzyme in the antho- marily responsible for pigmentation in the anthers and cyanin branch of the pathway. Anthocyanidin synthase floral tube (Quattrocchio et al. 1998, 1999; Kroon 2004). (ANS) establishes the conjugated double-bond structure DEEPPURPLE (DPL)/MYBB expression is correlated that makes the molecule appear colored to the human eye. with flower tube venation, whereas another R2R3-MYB, The three main classes of anthocyanidins are orange-red PURPLEHAZE (PHZ), regulates anthocyanin production pelargonidins, red-blue cyanidins, and blue-purple del- in light-exposed tissue in floral buds (Albert et al. 2011). phinidins. Further modification occurs by glucosyltrans- A number of experiments that use ectopic and heterolo- ferases, acyltransferases, and rhamnosyltransferases, and gous expression in mutants show that, in many cases, methyltransferases produce “redder” compounds: peoni- these R2R3-MYBs can compensate for the loss of function dins from cyanidins, and petunidins and malvidins from of native R2R3-MYBs, producing pigmentation in tissues delphinidins. in which they are expressed (Quattrocchio et al. 1998; POLLINATION SYNDROME GENES 121

Figure 3. The flavonoid pathway leading to the production of anthocyanins and flavonols in Petunia hybrida. Enzymes are shown in red. Light-red enzyme names and gray arrows leading to the pelargonidins and myricetin indicate enzymatic reactions that do not occur in Petunia hybrida, owing to an inability of the enzyme to use the substrate. (30AMT and 3050AMT) Anthocyanidin 30 and 3050 O- methyltransferase, (3GT and 5GT) anthocyanidin 3 and 5 glucosyltransferase, (4CL) 4-coumarate CoA-ligase, (AAT) anthocyanidin 3-rutinoside acyltransferase, (ANS) anthocyanidin synthase, (ART) anthocyanidin 3-glucoside rhamnosyltransferase, (C4H) cinna- mate 4-hydroxylase, (CHI) chalcone isomerase, (CHS) chalcone synthase, (DFR) dihydroflavonol reductase, (F30H and F3050H) flavonoid 30 and 3050 hydroxylase, (F3H) flavanone 3-hydroxylase, (FLS) flavonol synthase, (Glc) glucose, (PAL) phenylalanine ammonia lyase, (Rha) rhamnose. 122 SHEEHAN ET AL.

Spelt et al. 2000; Albert et al. 2011). Specific spatial species, including AN2 from Petunia, show homology expression of the R2R3-MYBs thus determines the pro- with this subfamily (see Fig. 3 in Cooley et al. 2011). duction of anthocyanin compounds in only certain floral tissues (Tornielli et al. 2009). In contrast, bHLH and WDR genes have wider expres- Regulatory and Structural Genes Contribute sion profiles and also affect other processes in the plant. to Natural Variation in Color In Petunia, AN1 is expressed in pigmented tissues and is required for anthocyanin production, but it also has other In the previous two sections, we discussed the major functions besides pigmentation that influence vacuolar biosynthetic and regulatory genes in anthocyanin produc- pH and morphology of the seed coat (Spelt et al. 2000, tion. Natural variation in these genes may form the basis 2002). Another bHLH gene, JAF13, also has a role in for color differences between species and variants. Most anthocyanin pigmentation, although its exact role is of these genes have been inactivated by mutations in unclear (Spelt et al. 2000). The WDR gene AN11 is ex- P. hybrida. However, for some of these, mutational pressed in all tissues, even in those that are nonpigmented changes will not only result in shifts in petal color but (de Vetten et al. 1997). Its role has not been precisely also affect other pathways and, thereby, may negatively characterized but it also may regulate vacuolar pH, and affect overall fitness (Coberly and Rausher 2003). So, plants deficient in AN11 have defects in the epidermal which of these genes are likely targets of natural selection cells of seeds and flowers as well as pigmentation loss in during shifts in pollination syndromes? This section gives both tissues (Quattrocchio 1994; Zenoni et al. 2011). Re- an overview of the kind of molecular changes that have pressors of anthocyanin production have also been char- been attributed to causing color changes in natural pop- acterized. The R3-MYB transcription factor MYBx is ulations of plant species. It also considers whether color known to repress anthocyanin biosynthesis, possibly by variants that have arisen and spread in the population competing with the R2R3-MYBs for the promoters of could be attributed to pollinator behavior. This informa- target genes (Kroon 2004). tion is summarized in Table 1. Examples are given in The MYB-bHLH-WDR complex comprises the main which a color polymorphism has been linked to a specific classes of proteins involved in anthocyanin regulation gene and the investigators have shown this by transgenic across all species studied thus far (Koes et al. 2005; Gro- experiments or by genetic studies showing cosegregation tewold 2006). Readers are referred to recent reviews for of the gene with flower color. Studies of particular inter- tables of these homologous regulators in different species est are discussed in greater detail below. (Koes et al. 2005; Petroni and Tonelli 2011; Davies et al. The first gene shown to be involved in a naturally oc- 2012). In dicotyledonous plants, the MYB-bHLH-WDR curring color transition was the R2R3-MYB transcription complexup-regulateslateanthocyaninbiosyntheticgenes; factor AN2 (Wijsman 1983; Quattrocchio et al. 1999). As however, the starting point of the “late” genes differs described above, AN2 regulates the later part of the antho- among species. In Petunia, F3050H is regulated along cyanin biosynthetic pathway. The gene is active in the with DFR and subsequent genes (Brugliera et al. 1999), purple flowers of P. integrifolia but is inactive in the white whereas in Ipomoea, the early biosynthetic genes CHS, flowers of P. axillaris (Quattrocchio et al. 1999; Hoballah CHI, and F3H may be partly regulated by the MYB- et al. 2007). The role of AN2 and color in pollinator attrac- bHLH-WDR complex but are also independently regu- tion was shown by transforming a functional AN2 allele lated (Morita et al. 2006). This regulatory separation con- into the P. axillaris genetic background that produced a trasts with the monocotyledonous plant maize, in which P. axillaris line with anthocyanin-pigmented flowers. Pol- all anthocyanin biosynthetic genes are regulated together linator choice tests showed that the hawkmoth Manduca by the MYB-bHLH-WDR complex (Mol et al. 1998). sexta (one of P. axillaris’ natural pollinators) preferred the In other species, it is also likely that bHLH and WDR white P. axillaris parent, whereas the bumblebee pollina- proteins are involved in additional processes unrelated to tor Bombus terrestris showed a preference for the trans- pigmentation while interacting with R2R3-MYBs that genic line (Hoballah et al. 2007). This experiment showed provide specific functions (Koes et al. 2005). In Antirrhi- that the modification of a single gene can have a major num and Ipomoea, it has also been shown that different effect on pollinator behavior. The loss of function of AN2, MYBs regulate anthocyanin production in specific floral and thus color, has happened multiple times, as shown by and vegetative tissues (Morita et al. 2006; Schwinn et al. different mutations in populations of P. axillaris (Hobal- 2006). In Arabidopsis, four highly related R2R3-MYB lah et al. 2007). Although it is debated whether the loss of sequences similarly regulate late anthocyanin biosyn- AN2 function represented the first step in the divergence of thetic genes. These genes are not involved in floral an- these two species (Hoballah et al. 2007; A. Lorenz- thocyanin production because Arabidopsis has white Lemke, pers. comm.), the loss of color constitutes a nec- flowers and reproduces by self-fertilization; one of these essary step in the transition from a bee pollination syn- R2R3-MYBs, PAP1, is necessary for anthocyanin pro- drome to a moth pollination syndrome in P. axillaris. duction in young seedlings, whereas the function of The genus Mimulus contains a yellow, carotenoid-col- the other genes is unknown (Gonzalez et al. 2008). This ored clade, in which red, anthocyanin-colored flowers are clade of Arabidopsis genes is classified as subfamily 6 of thought to have independently arisen five times (Cooley the R2R3-MYB transcription factor family (Stracke et al. and Willis 2009). Two of these gains of red color have 2001). Anthocyanin-implicated R2R3-MYBs from many been elucidated and have each been attributed to a single, POLLINATION SYNDROME GENES 123

Table 1. Genetic polymorphisms associated with natural variants in floral anthocyanin-derived color and benzenoid floral volatiles in species that are pollinated by animal pollinators Gene/s Taxon (locus) Gene family Mutation Molecular polymorphism Role in pollinator preference Reference Antirrhinum ROSEA1, R2R3-MYB R Causal polymorphisms not Field tests with A. majus Schwinn et al. ROSEA2 (subfamily yet identified in natural NILs differing in VENOSA 2006;Whibley (rosea); 6) accessions. Differences in and ROSEA1 activity et al. 2006; VENOSA expression and target showed that bumblebee Shang et al. (venosa) specificity of these genes pollinators preferred 2011 contribute to variation in at flowers with anthocyanin least six species of pigmentation to those Antirrhinum. without. Hybrid zone with yellow- flowered A. m. var. stri- atum and pink-flowered A. m. var. pseudomajus shows cline in flower color and genotype of ROSEA1 indicating selection- maintaining color differ- ence. Iochroma F3050H CYP75A S Deletion of a F3050H gene in Color is not associated with Smith et al. I. gesnerioides compared selection by functional 2008; Smith to I. cyaneum leads to pollinator groups. and Rausher pelargonidin-derived 2011 anthocyanins instead of cyanidin-derived anthocyanins. DFR DFR S Difference in substrate specificity between DFR from I. gesnerioides and I. cyaneum. Ipomoea F30H (P) CYP75B S A loss-of-function mutation Zufall and purpurea in the F30H gene causes pp Rausher 2003 individuals to be unable to produce cyanidin-derived anthocyanins. Ipomoea CHS-D (A) CHS S A loss-of-function mutation Visitation frequency by Coberly and purpurea in the CHS-D gene causes mainly bumblebee pol- Rausher 2003; aa individuals to be unable linators to white aa Fehr and to produce anthocyanins in individuals does not differ Rausher 2004 flowers and also affects from pigmented AA flavonoid production in individuals. other tissues. Ipomoea IpMYB1 R2R3-MYB R A loss-of-function mutation White-flowered individuals Epperson and purpurea (W ) (subfamily in IpMYB1 is correlated are visited less by bum- Clegg 1987; 6) with the down-regulation blebee pollinators when Tiffin et al. of multiple anthocyanin rare and self at a higher rate 1998; Chang biosynthetic genes in ww than plants with et al. 2005 individuals and loss of pigmented flowers. pigmentation. Ipomoea F30H CYP75B R Causal polymorphism not Des Marais and quamoclit yet identified. A cis-reg- Rausher 2010 ulatory change to F30H affects expression of the gene and influences flux between cyanidin-derived anthocyanins and pelar- gonidin-derived anthocyanins. Mimulus MYB R2R3-MYB R Causal polymorphism not Selection on color loci Streisfeld and aurantiacus (subfamily yet identified. A cis-or maintains geographic Kohn 2005, 6) trans-regulatory change in differentiation of yellow- 2007; Streis- R2R3-MYB influences and red-flowered races, feld and color in a complex fashion and hawkmoths and hum- Rausher in yellow- and red-flow- mingbirds show strong 2009a ered races of M. auran- preferences for the differ- tiacus. ent races, respectively. (Continued) 124 SHEEHAN ET AL.

Table 1. (Continued) Taxon Gene/s Gene family Mutation Molecular polymorphism Role in pollinator preference Reference (locus) Mimulus McMYB1, R2R3-MYB R Causal polymorphism not Pollinator visitation rate by Cooley et al. cupreus McMYB2, (subfamily yet identified. The M. Bombus dahlbomii to 2008, 2011 McMYB3 6) cupreus pla1 locus and, in M. cupreus in the popu- (pla1) particular, McMYB2 ex- lation of central Chile was pression are correlated low compared to other with higher expression of Mimulus species at the anthocyanin biosynthetic same study site. Discrim- genes compared to alleles ination was not associated from the yellow morph of with flower color. Popu- M. cupreus. lation at this site is thought to be maintained by selfing. Mimulus MvMYB4, R2R3-MYB R Causal polymorphism not Pollinator visitation by Cooley et al. luteus MvMYB5 (subfamily yet identified. The M. l. single, generalist pol- 2008, 2011 (pla2) 6) variegatus allele of linator in the population in MvMYB5 correlated with central Chile did not show higher expression of an- discrimination between thocyanin biosynthetic red M. l. variegatus and genes compared to the yellow M. l. luteus. Some allele from yellow Ml. assortative mating may luteus. occur owing to preferences of individual pollinators. Petunia AN2 R2R3-MYB R Functional gene regulates Introduction of functional Quattrocchio (subfamily production of antho- AN2 into white P. axillaris et al. 1999; 6) cyanins in the corolla of partially restored antho- Hoballah et al. P. integrifolia. At least five cyanin production. The 2007 independent loss-of-func- pigmented, transgenic line tion mutations in the cod- showed fourfold less feed- ing region contribute to ings by Manduca sexta and down-regulation of antho- threefold more visits by cyanin biosynthetic genes Bombus terrestris com- in the corolla of P. pared to wild type in pol- axillaris. linator-choice assays in controlled conditions. Petunia ODO1 R2R3-MYB R Causal polymorphisms not NILs that differed only in Klahre et al. (subfamily yet identified. A cis- floral scent production 2011 undefined) regulatory change is were bred into the genetic suggested to cause ex- backgrounds of P. axillaris pression differences. and P. exserta. Manduca sexta moths showed a significant preference for scented lines over non- scented lines. No prefer- ence was displayed when confronted with conflict- ing visual and olfactory cues. Phlox MYB R2R3-MYB R Causal polymorphisms not Sympatric populations of Hopkins and drummondii (subfamily yet identified. A cis-reg- P. drummondii with Rausher 2011, 6) ulatory change affects ex- P. cuspidata show dark red 2012; pression of the MYB, flower color; allopatric Hopkins et al. which is correlated with populations are light blue. 2011 the expression level of Common garden-field anthocyanin biosynthetic experiment showed re- genes. duced hybridization be- tween light- and dark- colored plants caused by color-intensity constancy by pollinators. F3050H CYP75A RAcis-regulatory change No evidence for discrim- affects expression of ination by color hue, F3050H and influences despite previous study color hue. showing selection on the “red” allele. The Mutation column refers to the type of mutation that has occurred: either affecting the coding region of a structural (S) gene or affecting regulatory (R) sequences (either a cis-regulatory element or the coding sequence of a transcription factor). Where available, information has been included about the role of the genetic polymorphism in pollinator preference and/or selection by pollinators. (AN2) Anthocyanin2, (CHS) chalcone synthase, (CYP) cytochrome P450 monooxygenase, (DFR) dihydroflavonol reductase, (F30H and F3050H) flavonoid 30 and 3050 hydroxylase, (NIL) near isogenic line, (ODO1) odorant1. POLLINATION SYNDROME GENES 125 dominant gain-of-function mutation in loci containing with current knowledge, for maintaining the species R2R3-MYBs of subfamily 6. The locus, petal lobe an- (Hopkins and Rausher 2012), despite the previous finding thocyanin 1 ( pla1), is associated with red color in the of a selective sweep on the allele causing red color (Hop- species M. cupreus, whereas the locus pla2 produces kins et al. 2011). This highlights the necessity to compare red pigmentation in M. luteus variegatus (Cooley and population genetics studies with tests of pollinator prefer- Willis 2009; Cooley et al. 2011). ence and fitness differences in the field. The pla1 locus may also be associated with the produc- In Ipomoea, a different mechanismhasbeen shown to be tion of anthocyanin-pigmented corolla banding and calyx responsible for the production of red, bird pollinated flow- spotting in M. guttatus (Lowry et al. 2012). Each locus ers. The ancestral color state is believed to be blue-purple contains duplicated homologous R2R3-MYBs and the flowers which are pollinated by bees, with red-flowered loci themselves are likely to be related by a previous hummingbird pollinated species having arisen four times duplication event (Cooley et al. 2011). For both loci, it independently (Streisfeld and Rausher 2009b). Ipomoea was shown that the level of anthocyanins is correlated quamoclit represents one member of a clade of red-flow- with the coordinate up-regulation of both early and late ered species. In this species, Des Marais and Rausher anthocyanin biosynthetic genes, indicating that the R2R3- (2010) showed that downregulation of the biosynthetic MYBs function as regulators of the anthocyanin pathway gene, F30H, at least partly as a result of mutation in a cis- (albeit there may be some differences to other dicotyled- regulatory element, resulted in flux in the pathway being onous species). The exact role of each R2R3-MYB in in- directed from cyanidin-derived anthocyanins to pelargo- fluencing the difference in color between the yellow nidin-derived anthocyanins. This regulation is tissue-spe- ancestors and derived red species has not yet been eluci- cific, occurring in the floral tissue but not in the vegetative dated. However, this example of parallel evolution shows tissue. The investigators speculate that this change char- the central role of R2R3-MYB factors in determining acterizes the other species in the I. quamoclit clade, and color differences in Mimulus. Similarly, two trans-acting that it is similar to what has occurred in another red-flow- regulatory loci are implicated in determining color differ- ered lineage of Ipomoea, I. horsfalliae. In this species, ences between parapatrically distributed red- and yellow- F30H is also downregulated, but the causal mechanism flowered races of M. aurantiacus, a species from a separate has not been found (Streisfeld and Rausher 2009b). The Mimulus clade (Streisfeld and Rausher 2009a). These rac- investigators postulate that F30H could be acommon target es are likely under selection by their hawkmoth and hum- of mutations to transition from bee pollination to hum- mingbird pollinators who show strong preferences for mingbird pollination; many red-flowered angiosperm spe- flower color (Streisfeld and Kohn 2007). Conversely, little cies which have gone through atransition from blue-purple evidence was found to suggest that flower color in flowers are colored by pelargonidin-derived anthocyanins M. cupreus, M. l. variegatus, and a yellow-flowered spe- (Rausher 2008; Streisfeld and Rausher 2009b). It remains cies M. l. luteus influences pollinator behavior (Cooley to be established whether the transition to red flowers in et al. 2008). An alternative explanation for the occurrence Ipomoea is as a result of selection from pollinators. of red-flowered species is that selection on other traits has In all the above examples, color changes have involved caused anthocyanin pigmentation to arise and it may be a changes to the regulation of the anthocyanin pathway— pleiotropic effect of the biosynthetic pathways control- either regulatory genes (trans-regulatory changes) or the ling pigmentation (Cooley and Willis 2009). targets of regulatory genes, regulatory elements (cis-reg- A member of the R2R3-MYB subfamily 6 has also been ulatory changes). However, examples of changes to color shown to be responsible for a gain in red color in Phlox.In caused by mutations in structural genes are also present. this example, the change in the R2R3-MYB works in con- The common morning glory Ipomoea purpurea has pop- cert with a cis-regulatory change to the F3050H gene. ulations that are highly polymorphic for color. All of Phlox drummondii has light blue flowers similar to anoth- these polymorphisms have been attributed to variation er species, P. cuspidata. Where these species are found in in four loci (Epperson and Clegg 1988). The genetic basis sympatry, P. drummondii produces red flowers. The level for three of these loci has been elucidated and two of them of anthocyanins produced by the red flowered P. drum- result from mutations to biosynthetic genes. For the P mondii is higher than that in P. cuspidata, and this is locus, a mutation in the coding region of the F30H gene correlated with a higher expression of R2R3-MYB in produces a truncated protein that cannot use dihydroquer- P. drummondii (Hopkins and Rausher 2011). The up-reg- cetin and, thus, cannot produce cyanidin-derived antho- ulation of R2R3-MYB is caused by an as-yet-unidentified cyanins (Zufall and Rausher 2003). Flux is redirected change in a cis-regulatory element of the P. drummondii from the cyanidin branch of the biosynthetic pathway to allele. The change in hue is attributed also to a cis-regu- the pelargonidin branch, thereby producing red flowers in latory difference acting on F3050H, meaning that flux in a similar way as to that described for I. quamoclit.The the pathway is redirected from the malvidin branch of the A locus produces a class of white-colored variants of pathway to the cyanidin branch (Hopkins and Rausher I. purpurea that cannot produce anthocyanins owing to a 2011). Field experiments showed that the “color intensi- transposon insertion in the CHS-D gene (Johzuka-Hisa- ty” R2R3-MYB locus maintains reproductive isolation of tomi et al. 1999). This loss-of-function mutation in the the red-flowered P. drummondii from P. cuspidata by CHS-D gene means that neither anthocyanins nor other decreasing interspecific pollen flow. Surprisingly, the flavonoid compounds can be produced throughout the F3050H locus affecting hue does not seem to be important, whole plant. These mutants show decreased fitness and 126 SHEEHAN ET AL. this is likely attributable to a deficiency in flavonoid com- from diverse primary metabolites (Dudareva et al. 2006). pounds that function to ameliorate heat stress (Coberly and Most floral volatiles belong to fatty acid derivatives, terpe- Rausher 2003). The third locus represents a regulatory noids, or benzenoids. The number and concentration of change controlled by IpMYB1, another member of the compounds provide species-specific olfactory cues R2R3-MYB subfamily 6, and is also responsible for pro- (Raguso et al. 2003; Hoballah et al. 2005; Raguso 2008), ducing a class of white-flowered variants (Chang et al. and the combination with visual stimuli allows searching 2005). In this example, white-flowered variants do not pollinators to use different sensory channels to identify appear to be under selection by pollinators (for review, target flowers. see Clegg and Durbin 2003). Bats, bees, butterflies, hummingbirds, and moths are able to sense floral scents and have been shown to dis- criminate among differences (von Helversen et al. The Role of Ultraviolet Color May Be Linked 2000; Kunze and Gumbert 2001; Wright et al. 2002, to Anthocyanin Pigmentation 2005; Oˆ mura and Honda 2005; Kessler et al. 2008; Zvi Insects, pollinating birds, and bats have photoreceptors et al. 2012). Recently, it was shown that the hawkmoth for UV light (Menzel and Backhaus 1991; Winter et al. Manduca sexta processes olfactory stimuli through two 2003; Herrera et al. 2008), and many flowering plants olfactory channels, allowing it to coordinate an innate absorb in the UV. UV-absorbing nectar guides that help preference with olfactory learning (Riffell et al. 2012). to orient insects with regard to the source of nectar have Hawkmoths use odor cues to locate target plants over been known for some time (Daumer 1958; Jones and long distances (Raguso and Willis 2002a), which may Buchmann 1974; Penny 1983). Recent work in Mimulus explain the high odor emission rates in flowers displaying has shown that UV-absorbing petals can be important for moth-pollination syndromes. For instance, moth-polli- the attraction of bees and other pollinators in a natural nated Clarkia breweri flowers emit strong floral scents habitat and also reaffirmed that these petals can be im- compared to flowers of the closely related bee-pollinated portant for bee orientation (Rae and Vamosi 2012). C. concinna (Raguso and Pichersky 1995). However, the UV-absorbing floral pigments may be useful for detec- importance of olfactory and visual stimuli for the attrac- tion of flowers by night-active insects. Night-active tion of specific pollinator guilds differs (Raguso and Wil- hawkmoths generally visit white-flowered plants, such lis 2005; Goyret et al. 2007; Klahre et al. 2011). as Petunia axillaris. These flowers’ maximal reflectance seems optimal for the flower to be perceived at low-light intensity. However, most white flowers absorb UV light The Biosynthesis of Volatile and, thus, do not reflect all wavelengths perceived by the Benzenoid Compounds insect’s photoreceptors and do not take advantage of all available photons (Kevan et al. 1996). Interestingly, hawk- Petunia x hybrida cv Mitchell is a popular model sys- moths have been shown to display true color vision at light tem to elucidate the biosynthesis of floral volatile benze- intensities as low as dim starlight (Kelber et al. 2002). noids, and many of the structural and regulatory genes Therefore, hawkmoths probably use color vision to detect have been identified in Petunia. For other odor com- UV-absorbing flowers against a background of “dull” fo- pounds, the reader can refer to excellent reviews of liage that reflects to a low degree across all wavelengths Pichersky and Gershenzon 2002; Dudareva et al. 2006. (Kevan et al. 1996). Phenylalanine-derived volatiles come in three groups In Petunia, flavonols are the main pigments contribut- that are identified by the number of carbon atoms in their ing to the UV absorbance of the flower. The absence of side chains (Fig. 4; all volatile benzenoid compounds are anthocyanins and the presence of flavonols in the P. axil- highlighted in purple). In the C6 –C2 group, the key en- laris flower may be attributable to selection by hawk- zyme phenylacetaldehyde synthase (PAAS) catalyzes moths, but alternative roles of flavonols in plant defense the decarboxylation of phenylalanine to phenylacetalde- against pathogens or abiotic stresses should also be con- hyde, which can subsequently be reduced, oxidized, or sidered (Gould and Lister 2005). Another possibility is esterified. that the presence of flavonols is a consequence of selection The C6 –C3 and C6 –C1 pathways both start from trans- for loss of anthocyanin biosynthesis, because down-regu- cinnamic acid, which is synthesized from phenylalanine lation of DFR would redirect metabolic flux toward FLS by the activity of phenylalanine ammonia lyase (PAL). and flavonol biosynthesis (Fig. 1). Molecular genetics in Entry into the C6 –C1 pathway requires shortening of the combination with behavioral experiments can test such C3 side chain of trans-cinnamic acid by two carbons, hypotheses. which can occur via both a CoA-dependent b-oxidative pathway analogous to the catabolism of fatty acids (Hert- weck et al. 2001; Moerkercke et al. 2009; Klempien FLORAL SCENT et al. 2012; Qualley et al. 2012) and a CoA-independent non-b-oxidative pathway (Boatright et al. 2004; Orlova Floral Volatiles Act as Species-Specific Long- et al. 2006). Distance Signals to Attract Pollinators The conversion of trans-cinnamic acid into p-coumaric Floral scent bouquets may consist of as many as 20–60 acid leads into the C6 –C3 branch. The specific reaction different volatiles (Knudsen et al. 2006), synthesized that gives rise to coniferyl alcohol is as-yet unknown. POLLINATION SYNDROME GENES 127

Figure 4. The benzenoid pathway leading to the production of volatile compounds in scented Petunia. Volatile compounds emitted by scented Petunia are highlighted in purple. Enzymes shown in red indicate known steps in Petunia; unknown enzymatic steps are indicated by dashed arrows. Multiple dashed arrows indicate multiple unknown steps. Transcriptional regulators are highlighted in yellow. (4CL) 4-Coumarate CoA-ligase, (ADT) arogenate dehydratase, (BALDH) benzaldehyde dehydrogenase, (BSMT) (S-adeno- syl-L-methionine) benzoic acid/salicylic acid carboxyl methyltransferase, (BPBT) benzoyl-CoA:benzylalcohol/2-phenylethanol ben- zoyltransferase, (C4H) cinnamate 4-hydroxylase, (CFAT) coniferyl alcohol acyltransferase, (CHD) cinnamoyl-CoA hydratase- dehydrogenase, (CM) chorismate mutase, (CNL) cinnamate CoA-ligase, (DAHPS) 3-deoxy-D-arabino-heptulosonate-7-phosphate synthase, (EGS) eugenol synthase, (EPSPS) 5-enolpyruvylshikimate-3-phosphate synthase, (IGS) isoeugenol synthase, (KAT1) 3- ketoacyl-CoA thiolase, (PAAS) phenylacetaldehyde synthase, (PAL) phenylalanine ammonia lyase. 128 SHEEHAN ET AL.

Coniferyl alcohol acyltransferase (CFAT) acetylates con- upstream enhancer region to activate the expression of iferyl alcohol to produce coniferyl acetate. This serves as ODO1 (Van Moerkercke et al. 2011). EOBII also acti- the substrate for isoeugenol synthase (IGS) and eugenol vates PAL and IGS promoters (Spitzer-Rimon et al. synthase (EGS), which catalyze the formation of isoeu- 2010). Silencing of EOBII down-regulates ODO1 and genol and eugenol, respectively (Clark 2009). Some P. genes of the shikimate and phenylpropanoid pathways. axillaris emit neither eugenol nor isoeugenol but pro- As with ODO1, it has a rhythmic expression pattern, duce high levels of dihydroconiferyl acetate. In P. a. par- but the transcript levels peak in the early morning hours odii S7, EGS is functional, but the gene encoding IGS and are lowest during the late afternoon. The difference in contains a frameshift mutation leading to its inactivity. phasing of the ODO1 and EOBII rhythms might be ex- Unexpectedly, this does not lead to higher eugenol levels plained by time-delayed activation of ODO1 expression. (Koeduka et al. 2009). But EOBII is expressed in early flower development Because all benzenoid volatiles derive from phenylal- when ODO1 is not. Functional analyses of ODO1 pro- anine, substrate competition may regulate the fluxes moter fragments have provided evidence that at least one through the C6–C2,C6–C1, and C6 –C3 pathways. Such additional factor is needed for high ODO1 promoter ac- substrate competition may be determined by the relative tivity (Van Moerkercke et al. 2012). This as-yet-uniden- abundance and enzymatic properties of the branch-spe- tified factor might partly explain the interaction of EOBII cific enzymes. Indeed, there is evidence that a higher and ODO1 despite their different expression patterns and affinity for phenylalanine by PAL compared to PAAS the finding that overexpression of EOBII did not result in causes a higher flux through the C6–C1 and C6–C3 significant higher ODO1 levels (Spitzer-Rimon et al. branches of the pathway (Maeda et al. 2010). 2010). The third identified transcription factor, PhMYB4, is also an R2R3-MYB transcription factor. It negatively regulates C4H (Colquhoun and Clark 2011; Colquhoun The Regulation of Floral Scent Production et al. 2011), indirectly controlling the emission of isoeu- The emission of floral volatile benzenoids from the genol and eugenol by precursor supply. It was hypothe- epidermal cells of P. axillaris and P. hybrida cv Mitchell sized to “fine-tune” the bouquet of floral fragrance by petals begins at anthesis and ceases after pollination affecting the balance between the C6 –C1 and C6 –C3 (Negre et al. 2003; Underwood et al. 2005). Emission branches of the pathway. follows a circadian rhythm that peaks at dusk (Verdonk et al. 2005; Clark 2009). Together, this indicates precise spatial (Van Moerkercke et al. 2012), developmental, and Interaction Between Color and Scent temporal regulation of the pathway. So far, three tran- at the Biochemical Level scriptional regulators of scent production, ODORANT1 (ODO1), EMISSION OF BENZENOIDS II (EOBII), and Interaction between the branches of the benzenoid and PhMYB4, have been identified. Expression of the R2R3- flavonoid pathways might occur owing to pleiotropic reg- MYB ODO1 peaks before the onset of volatile emission ulation or to metabolic competition for the common pre- and its transcript levels are low the next morning. ODO1 cursor, phenylalanine. Redirection of metabolic flux activates the promoter of the gene encoding EPSPS, an from one branch to the other has been shown in Dianthus enzyme of the shikimate pathway, suggesting that ODO1 caryophyllus (Zuker et al. 2002). These investigators is involved in the petal-specific up-regulation of phenyl- showed that suppression of F30H led to white-flowered alanine as the precursor for volatile production (Verdonk mutants that produced higher amounts of methylbenzoate et al. 2005). Accordingly, down-regulation of ODO1 in compared to colored control plants. In contrast, the intro- transgenic P. hybrida cv Mitchell plants strongly reduced duction of the Arabidopsis R2R3-MYB transcription fac- benzenoid emission (Verdonk et al. 2005). The expres- tor PRODUCTION OF ANTHOCYANIN PIGMENT 1 sion levels of genes encoding DAHPS, EPSPS, PAL, and (PAP1; a member of the R2R3-MYB subfamily 6) into CM were decreased, but, interestingly, the late-acting P. hybrida enhanced both branches of the phenylpropa- genes encoding BSMT and BPBT were up-regulated. Ex- noid pathway, leading to the production of color and pression of the structural genes BSMT, BPBT, together scent. This argues against metabolic competition but sup- with PAAS and IGS, precedes volatile emission and has ports the notion that PAP1 has a pleiotropic effect on both been detected at high levels during the light period and pathways (Zvi et al. 2006, 2008). Similarly, 35S:PAP1 low levels during the dark period (Colquhoun et al. 2010). transgenic Rosa hybrida showed increased pigmentation However, the rhythmic transcript accumulation of at least as well as eugenol accumulation (Zvi et al. 2012). A more BSMT and PAAS seems not to cause the rhythmic emis- general argument against metabolic competition is the sion of the floral volatiles, based on constant activities temporal separation of the two pathways. Flavonoids are detected over a 24-h period (Kolosova et al. 2001). In- produced during growth and development of the flower, stead, oscillations of early precursor pools and ODO1 whereas benzenoids are only produced when the flowers expression have been suggested to control the rhythmic open (Verdonk et al. 2005). emission of floral fragrance. However, multiple examples contradict the idea of a The second R2R3-MYB transcription factor, EOBII, pleiotropic regulator of scent and color. The introduction binds to a specific MYB-binding site located in an of the key regulator of anthocyanin biosynthesis, AN2, POLLINATION SYNDROME GENES 129 into P. axillaris altered flower color but had no effect on the difference in expression level of the two alleles is a scent production (Hoballah et al. 2007), suggesting that cis-acting effect (Klahre et al. 2011). The actual polymor- this Petunia PAP1 homolog has no effect on volatile phism, potentially a sequence change in the ODO1 pro- emission. Silencing ODO1 and thus decreasing shikimate moter, has not yet been determined. The gene(s) un- precursor compounds for benzenoid production did not derlying the chromosome II QTL are presently unknown. lead to decreased flavonoid production (Verdonk et al. However, repeated backcrossing resulted in an unscented 2005). In addition, EOBII expression in transgenic Nico- P. axillaris NIL as well as a scented P. exserta NIL and tiana tabacum increased PAL transcript levels but CHS made it possible to study hawkmoth attraction to scent in and F30H expression levels and anthocyanin levels were otherwise undistinguishable genetic backgrounds. Both not affected (Spitzer-Rimon et al. 2010). Because pleio- in the P. axillaris and P. exserta backgrounds, the moths tropic effects have been observed when foreign trans- had a strong preference for the scented lines over the genes were expressed from a strong viral promoter, the nonscented lines. Interestingly, when the insects were evidence in favor of regulatory interactions between the given the choice between the red, scented P. exserta two pathways should be interpreted with caution. NIL and the white, nonscented P. axillaris NIL, a situa- tion in which they had to choose between conflicting visual and olfactory cues, they displayed no preference. Behavioral Experiments Test the Importance This suggests that, at least under the artificial conditions of Color and Scent in Pollinator Attraction of a wind tunnel experiment, color and scent appear to be Classical behavioral studies have tried to disentangle equivalent cues (Klahre et al. 2011). the importance of olfactory and visual stimuli for polli- nator attraction using artificial flowers and cues (Raguso and Willis 2002b; Balkenius and Kelber 2006; Balkenius CONCLUSIONS et al. 2006; Goyret et al. 2007). To disentangle visual and olfactory cues in complex flowers, Hirota et al. (2012) Pollination syndromes consist of specific traits, such as generated F1 and F2 hybrids between the butterfly-polli- floral color, scent, floral morphology, and nectar volume nated daylily and the hawkmoth-pollinated nightlily, each and composition. The correlation between floral traits differing in floral color and scent. In an F2 progeny, seg- and pollinator classes across the angiosperms suggests regating for both floral traits, neither butterflies nor that this convergent evolution is driven by pollinator-me- hawkmoths showed significant preferences for overall diated selection (Fenster et al. 2004). However, other scent emission. However, butterflies preferentially visit- contributing factors must also be considered. Rausher ed red- or orange-colored flowers, resembling the daylily (2008) discusses this in depth, pointing out that the few color phenotype, whereas moths preferred the yellow- examples investigated thus far show selection on color colored flowers, representing the nightlily color pheno- changes but that the selective agent cannot necessarily be type. Thus, the visual stimuli were more important for attributed to pollinators. pollinator attraction. It should be noted, however, that To attract a new pollinator, coordinated changes in additional floral traits might have been segregating in a multiple traits are required. This raises important ques- F2 population that were not considered in this study. tions about the presumably complex molecular genetics In Petunia, we set out to construct near isogenic lines of shifts in pollination syndromes. This review focuses on (NILs) through repeated backcrossing of hybrids among the molecular basis of the pollination syndrome traits, species with different pollination syndromes. The aim was color and scent. For both traits, substantial knowledge to obtain plants that differ genetically only in single rele- has been acquired about the biochemical pathways and vant loci, and phenotypically only in single traits, for in- transcriptional regulators. A central question in the ecol- stance, petal color or scent production. This makes it ogy and evolution of pollination syndromes is which of possible to dissect the interactions between visual and the trait-determining genes were selected in nature to olfactory stimuli in complex flowers. Such plants can be achieve a shift in pollinator attraction. Altogether, how studied under controlled conditions but can also be intro- many genes were involved and how did they change in a duced into the natural habitats. They can thus be exposed to coordinated fashion? For color and scent, both structural the varietyof natural pollinators visiting during the season, and regulatory genes contribute to major phenotypic and fitness across generations can be determined. In addi- changes, although more regulatory changes are represent- tion, NILs are attractive genetic material for the isolation ed in the limited number of examples found so far (Table of the genes underlying shifts in pollination syndromes. 1). Genes of small effect would in most cases have es- Using this approach, we crossed the red-flowered, non- caped detection, but the fact that major genes were found scented P. exserta with the white, scented P. axillaris and at all is of great significance to the debate about the role of mapped two major QTL for scent production in the F2 major mutations in adaptive evolution (Bradshaw and population to chromosomes II and VII. The scent QTL on Schemske 2003). chromosome VII covered the position of ODO1 (Stuur- The first transcription factor involved in natural varia- man et al. 2004; Klahre et al. 2011). ODO1 is highly tion in petal color was AN2 and R2R3-MYBs are the only expressed in P. axillaris but is present at low levels in regulatory genes represented so far (Table 1). Subsequent P. exserta. In the F1 hybrid, most of the ODO1 transcript discovery of genes underlying natural variation has relied was produced from the P. axillaris allele, indicating that on candidate gene approaches that are inevitably biased 130 SHEEHAN ET AL. toward R2R3-MYBs. We note that no genes causing natural environmentally regulated to control complex floral and veg- variation in pollination syndromes haveyet been identified etative pigmentation patterning. Plant J 65: 771–784. Ando T, Saito N, Tatsuzawa F, Kakefuda T, Yamakage K, de novo. Changes consist of alterations in the coding re- Ohtani E, Koshi-ishi M, Matsusake Y, Kokubun H, Watanabe gion of the R2R3-MYBs themselves or changes in cis-reg- H, et al. 1999. Floral anthocyanins in wild taxa of Petunia ulatory elements. Other regulatory changes concern (Solanaceae). Biochem Syst Ecol 27: 623–650. putative cis changes or changes to the coding sequences Balkenius A, Kelber A. 2006. Color preferences influences odor of biosynthetic genes, for instance, F30H or F3050H genes. learning in the hawkmoth, Macroglossum stellatarum. Natur- wissenschaften 93: 255–258. Such changes have effects on the types of anthocyanins Balkenius A, Rose´n W, Kelber A. 2006. The relative importance produced by adjusting flux through branches of the bio- of olfaction and vision in a diurnal and a nocturnal hawkmoth. synthetic pathways. Duplication also had a role in the J Comp Physiol A 192: 431–437. diversification of color, with many of the described exam- Beld M, Martin C, Huits H, Stuitje AR, Gerats AGM. 1989. Flavonoid synthesis in Petunia hybrida: Partial characteriza- ples involving duplication of R2R3-MYBs. Duplication is tion of dihydroflavonol-4-reductase genes. Plant Mol Biol 13: thought to be a significant factor influencing phenotypic 491–502. novelty in plants (Flagel and Wendel 2009). Bene´ F. 1941. Experiments on the color preference of black- R2R3-MYBs are also involved in the regulation of vo- chinned hummingbirds. Condor 43: 237–242. latile benzenoid biosynthesis. The R2R3-MYBs EOBII Boatright J, Negre F, Chen X, Kish CM, Wood B, Peel G, Orlova I, Gang D, Rhodes D, Dudareva N. 2004. Understanding in and ODO1 regulate scent biosynthesis in different species vivo benzenoid metabolism in Petunia petal tissue. Plant and there is evidence that ODO1 underlies natural varia- Physiol 135: 1993–2011. tion in scent production. However, both ODO1 and EOBII Bradshaw HD, Schemske DW. 2003. Allele substitution at a primarily affect early steps in the pathway and alterations flower color locus produces a pollinator shift in monkey- flowers. Nature 426: 176–178. to these genes might be expected to have pleiotropic ef- Bradshaw HD, Wilbert SM, Otto KG, Schemske DW. 1995. fects. It seems likely, therefore, that additional regulators Genetic mapping of floral traits associated with reproductive of natural variation in scent remain to be discovered. isolation in monkeyflowers (Mimulus). Nature 376: 762–765. Coordinated changes in multiple traits could be Bradshaw HD, Otto KG, Frewen BE, McKay JK, Schemske brought about by genetic changes in master genes that DW. 1998. Quantitative trait loci affecting differences in flo- ral morphology between two species of monkeyflower (Mim- affect multiple traits or by genetic linkage of multiple ulus). Genetics 149: 367–382. regulators. There is little convincing evidence thus far Brugliera F, Barri-Rewell G, Holton TA, Mason JG. 1999. Iso- for the involvement of coregulation of scent and color lation and characterization of a flavonoid 30-hydroxylase in natural variants despite regulators being found that cDNA clone corresponding to the Ht1 locus of Petunia hybrida. Plant J 19: 441–451. influence early, shared steps of the phenylpropanoid path- Chang S-M, Lu Y, Rausher MD. 2005. Neutral evolution of way. Linkage of genes in the genome may have occurred the nonbinding region of the anthocyanin regulatory gene in Mimulus and Aquilegia (Bradshaw et al. 1995; Hodges IpMYB1 in Ipomoea. Genetics 170: 1967–1978. et al. 2002), but much more detailed genetic and physical Chittka L, Menzel R. 1992. The evolutionary adaptation of flow- maps, as well as a better understanding of the genetic and er colors and the insect pollinators’ color vision. J Comp Physiol A 171: 171–181. biochemical basis of the traits, will be necessary to ad- Chittka L, Raine NE. 2006. Recognition of flowers by pollina- dress this issue. tors. Curr Opin Plant Biol 9: 428–435. Undoubtedly, the rapid advances in whole-genome se- Clark DG. 2009. Benzenoids dominate the fragrance of Petunia quencing will accelerate the identification of the genes flowers. In Petunia: Evolutionary, developmental and physi- ological genetics (ed. Gerats T, Strommer J), pp. 51–70. that cause natural variation. Functional analysis of such Springer, New York. genes, at the level of the gene, trait, and population, will Clegg MT, Durbin ML. 2003. Tracing floral adaptations from provide new insight not only into plant–pollinator inter- ecology to molecules. Nat Rev Genet 4: 206–215. actions but also in reproductive isolation and ultimately Coberly LC, Rausher MD. 2003. Analysis of a chalcone syn- speciation. thase mutant in Ipomoea purpurea reveals a novel function for flavonoids: Amelioration of heat stress. Mol Ecol 12: 1113– 1124. Colquhoun TA, Clark DG. 2011. Unraveling the regulation of ACKNOWLEDGMENTS floral fragrance biosynthesis. Plant Signal Behav 6: 378–381. Colquhoun TA, Verdonk JC, Schimmel BC, Tieman DM, Un- We thank Sarah Robinson, Korinna Esfeld, Arne Jung- derwood BA, Clark DG. 2010. Petunia floral volatile benze- wirth, and Ulrich Klahre for critically reading the manu- noid/phenylpropanoid genes are regulated in a similar man- script and for insightful comments; Roman Ko¨pfli for ner. Phytochemistry 71: 158–167. expert assistance with illustrations; and Alexandre Del- Colquhoun TA, Kim JY, Wedde AE, Levin LA, Schmitt KC, l’Olivo for photography. This work was supported by Schuurink RC, Clark DG. 2011. PhMYB4 fine-tunes the flo- ral volatile signature of Petuniahybrida through PhC4H. grants from the National Centre for Competence in Re- J Exp Bot 62: 1133–1143. search “Plant Survival,” the Swiss National Science Cooley AM, Willis JH. 2009. Genetic divergence causes parallel Foundation, and the University of Bern. evolution of flower color in Chilean Mimulus. New Phytol 183: 729–739. Cooley AM, Carvallo G, Willis JH. 2008. Is floral diversifica- tion associated with pollinator divergence? Flower shape, REFERENCES flower color and pollinator preference in Chilean Mimulus. Albert NW, Lewis DH, Zhang H, Schwinn KE, Jameson PE, Ann Bot 101: 641–650. Davies KM. 2011. Members of an R2R3-MYB transcription Cooley AM, Modliszewski JL, Rommel ML, Willis JH. 2011. factor family in Petunia are developmentally and Gene duplication in Mimulus underlies parallel floral POLLINATION SYNDROME GENES 131

evolution via independent trans-regulatory changes. Curr Gumbert A. 2000. Color choices by bumble bees (Bombus ter- Biol 21: 700–704. restris): Innate preferences and generalization after learning. Danieli-Silva A, de Souza JMT, Donatti AJ, Campos RP, Vicen- Behav Ecol Sociobiol 48: 36–43. te-Silva J, Freitas L, Varassin IG. 2012. Do pollination syn- Herrera G, Zagal J, Diaz M, Ferna´ndez M, Vielma A, Cure M, dromes cause modularity and predict interactions in a Martinez J, Bozinovic F, Palacios A. 2008. Spectral sensitiv- pollination network in tropical high-altitude grasslands? ities of photoreceptors and their role in color discrimination Oikos 121: 35–43. in the green-backed firecrown hummingbird (Sephanoides Daumer K. 1958. Blumenfarben, wie sie die Bienen sehen. sephaniodes). J Comp Physiol A 194: 785–794. Zeitschrift Physiol 41: 49–110. Hertweck C, Jarvis AP, Xiang L, Moore BS, Oldham NJ. 2001. Davies KM, Albert NW, Schwinn KE. 2012. From landing lights A mechanism of benzoic acid biosynthesis in plants and bac- to mimicry: The molecular regulation of flower coloration and teria that mirrors fatty acid b-oxidation. ChemBioChem 2: mechanisms for pigmentation patterning. Funct Plant Biol 784–786. 39: 619–638. Hirota SK, Nitta K, Kim Y, Kato A, Kawakubo N, Yasumoto Des Marais DL, Rausher MD. 2010. Parallel evolution at mul- AA, Yahara T. 2012. Relative role of flower color and scent on tiple levels in the origin of hummingbird pollinated flowers in pollinator attraction: Experimental tests using F1 and F2 hy- Ipomoea. Evolution 64: 2044–2054. brids of daylily and nightlily. PLoS ONE 7: e39010. de Vetten N, Quattrocchio F, Mol J, Koes R. 1997. The an11 Hoballah ME, Stuurman J, Turlings TCJ, Guerin PM, Conne´t- locus controlling flower pigmentation in Petunia encodes a able S, Kuhlemeier C. 2005. The composition and timing of novel WD-repeat protein conserved in yeast, plants, and ani- flower odor emission by wild Petunia axillaris coincide with mals. Genes Dev 11: 1422–1434. the antennal perception and nocturnal activity of the pollina- Dinkel T, Lunau K. 2001. How drone flies (Eristalis tenax L., tor Manduca sexta. Planta 222: 141–150. Syrphidae, Diptera) use floral guides to locate food sources. J Hoballah ME, Gu¨bitz T, Stuurman J, Broger L, Barone M, Man- Insect Physiol 47: 1111–1118. del T, Dell’Olivo A, Arnold M, Kuhlemeier C. 2007. Single Dudareva N, Negre F, Nagegowda DA, Orlova I. 2006. Plant gene-mediated shift in pollinator attraction in Petunia. Plant volatiles: Recent advances and future perspectives. Crit Rev Cell 19: 779–790. Plant Sci 25: 417–440. Hodges SA, Whittall JB, Fulton M, Yang JY. 2002. Genetics of Epperson BK, Clegg MT. 1987. Frequency-dependent variation floral traits influencing reproductive isolation between Aquile- for outcrossing rate among flower-color morphs of Ipomoea gia formosa and Aquilegia pubescens. Am Nat 159: S51–S60. purpurea. Evolution 41: 1302–1311. Holton TA, Cornish EC. 1995. Genetics and biochemistry of Epperson BK, Clegg MT. 1988. Genetics of flower color poly- anthocyanin biosynthesis. Plant Cell 7: 1071–1083. morphism in the common morning glory (Ipomoea pur- Hopkins R, Rausher MD. 2011. Identification of two genes purea). J Hered 79: 64–68. causing reinforcement in the Texas wildflower Phlox drum- Faegri K. 1979. The principles of pollination ecology by K. Fae- mondii. Nature 469: 411–414. gri and L. van der Pijl. Pergamon Press, Oxford; New York. Hopkins R, Rausher MD. 2012. Pollinator-mediated selection Fehr C, Rausher MD. 2004. Effects of variation at the flower- on flower color allele drives reinforcement. Science 335: color A locus on mating system parameters in Ipomoea pur- 1090–1092. purea. Mol Ecol 13: 1839–1847. Hopkins R, Levin DA, Rausher MD. 2011. Molecular signatures Feinsinger P. 1983. Coevolution and pollination. In Coevolu- of selection on reproductive character displacement of flower tion (ed. Futuyma J, Slatkin M), pp. 283–310. Sinnauer, color in Phlox drummondii. Evolution 66: 469–485. Sunderland. Johnson ET, Ryu S, Yi H, Shin B, Cheong H, Choi G. 2001. Fenster CB, Armbruster WS, Wilson P, Dudash MR, Thomson Alteration of a single amino acid changes the substrate spe- JD. 2004. Pollination syndromes and floral specialization. cificity of dihydroflavonol 4-reductase. Plant J 25: 325–333. Annu Rev Ecol Evol Syst 35: 375–403. Johzuka-Hisatomi Y, Hoshino A, Mori T, Habu Y, Iida S. 1999. Flagel LE, Wendel JF. 2009. Gene duplication and evolutionary Characterization of the chalcone synthase genes expressed in novelty in plants. New Phytol 183: 557–564. flowers of the common and japanese morning glories. Genes Gerats AGM, Vlaming P, Doodeman M, Al B, Schram AW. Genet Syst 74: 141–147. 1982. Genetic control of the conversion of dihydroflavonols Jones CE, Buchmann SL. 1974. Ultraviolet floral patterns as into flavonols and anthocyanins in flowers of Petunia functional orientation cues in hymenopterous pollination sys- hybrida. Planta 155: 364–368. tems. Anim Behav 22: 481–485. Giurfa M, Nu´n˜ez J, Chittka L, Menzel R. 1995. Color prefer- Kelber A. 1997. Innate preferences for flower features in the ences of flower-naive honeybees. J Comp Physiol A 177: hawkmoth Macroglossum stellatarum. J Exp Biol 200: 827– 247–259. 836. Gonzalez A, Zhao M, Leavitt JM, Lloyd AM. 2008. Regulation Kelber A, Balkenius A, Warrant EJ. 2002. Scotopic color vision of the anthocyanin biosynthetic pathway by the TTG1/ in nocturnal hawkmoths. Nature 419: 922–925. bHLH/MYB transcriptional complex in Arabidopsis seed- Kessler D, Gase K, Baldwin IT. 2008. Field experiments with lings. Plant J 53: 814–827. transformed plants reveal the sense of floral scents. Science Gould KS, Lister C. 2005. Flavonoid functions in plants. In 321: 1200–1202. Flavonoids chemistry, biochemistry and applications (ed. An- Kevan P, Giurfa M, Chittka L. 1996. Why are there so many and derson ØM, Markham KR), pp. 397–441. CRC Press, Boca so few white flowers? Trends Plant Sci 1: 252. Raton, FL. Kingston AB, Mc Quillan PB. 2000. Are pollination syndromes Goyret J, Markwell PM, Raguso RA. 2007. The effect of decou- useful predictors of floral visitors in Tasmania? Austral Ecol pling olfactory and visual stimuli on the foraging behavior of 25: 600–609. Manduca sexta. J Exp Biol 210: 1398–1405. Klahre U, Gurba A, Hermann K, Saxenhofer M, Bossolini E, Goyret J, Pfaff M, Raguso R, Kelber A. 2008. Why do Manduca Guerin PM, Kuhlemeier C. 2011. Pollinator choice in Petunia sexta feed from white flowers? Innate and learnt color pref- depends on two major genetic loci for floral scent production. erences in a hawkmoth. Naturwissenschaften 95: 569–576. Curr Biol 21: 730–739. Grotewold E. 2006. The genetics and biochemistry of floral Klempien A, Kaminaga Y, Qualley A, Nagegowda DA, Wid- pigments. Annu Rev Plant Biol 57: 761–780. halm JR, Orlova I, Shasany AK, Taguchi G, Kish CM, Cooper Gu¨bitz T, Hoballah ME, Dell’Olivo A, Kuhlemeier C. 2009. BR, et al. 2012. Contribution of CoA ligases to benzenoid Petunia as a model system for the genetics and evolution of biosynthesis in Petunia flowers. Plant Cell 24: 2015–2030. pollination syndromes. In Petunia: Evolutionary, developmen- Knapp S. 2010. On “various contrivances”: Pollination, phylog- tal and physiological genetics (ed. Gerats T, Strommer J), pp. eny and flower form in the Solanaceae. Philos Trans R Soc B 29–49. Springer, New York. Biol Sci 365: 449–460. 132 SHEEHAN ET AL.

Knudsen J, Eriksson R, Gershenzon J, Sta˚hl B. 2006. Diversity Qualley AV, Widhalm JR, Adebesin F, Kish CM, Dudareva N. and distribution of floral scent. Bot Rev 72: 1–120. 2012. Completion of the core b-oxidative pathway of benzoic Koeduka T, Orlova I, Baiga TJ, Noel JP, Dudareva N, Pichersky acid biosynthesis in plants. Proc Natl Acad Sci 109: 16383– E. 2009. The lack of floral synthesis and emission of isoeu- 16388. genol in Petunia axillaris subsp. parodii is due to a mutation Quattrocchio F. 1994. Regulatory genes controlling flower pig- in the isoeugenol synthase gene. Plant J 58: 961–969. mentation in Petunia hybrida. Vrije Universiteit, Amsterdam. Koes R, Verweij W, Quattrocchio F. 2005. Flavonoids: A color- Quattrocchio F, Wing JF, Leppen HTC, Mol JNM, Koes RE. ful model for the regulation and evolution of biochemical 1993. Regulatory genes controlling anthocyanin pigmentation pathways. Trends Plant Sci 10: 236–242. are functionally conserved among plant species and have dis- Kolosova N, Gorenstein N, Kish CM, Dudareva N. 2001. Re- tinct sets of target genes. Plant Cell 5: 1497–1512. gulation of circadian methyl benzoate emission in diurnally Quattrocchio F, Wing JF, Van der Woude K, Mol JNM, Koes R. and nocturnally emitting plants. Plant Cell 13: 2333–2347. 1998. Analysis of bHLH and MYB domain proteins: Species- Kroon AR. 2004. Transcription regulation of the anthocyanin specific regulatory differences are caused by divergent evo- pathway in Petunia hybrida. Vrije Universiteit, Amsterdam. lution of target anthocyanin genes. Plant J 13: 475–488. Kunze J, Gumbert A. 2001. The combined effect of color and Quattrocchio F, Wing J, van der Woude K, Souer E, de Vetten N, odor on flower choice behavior of bumble bees in flower Mol J, Koes R. 1999. Molecular analysis of the anthocyanin2 mimicry systems. Behav Ecol 12: 447–456. gene of Petunia and its role in the evolution of flower color. Lorenz-Lemke AP, Ma¨der G, Muschner VC, Stehmann JR, Plant Cell 11: 1433–1444. Bonatto SL, Salzano FM, Freitas LB. 2006. Diversity and Quattrocchio F, Baudry A, Lepiniec L, Grotewold E. 2008. The natural hybridization in a highly endemic species of Petunia regulation of flavonoid biosynthesis. In The science of flavo- (Solanaceae): A molecular and ecological analysis. Mol Ecol noids (ed. Grotewold E), pp. 97–122. Springer, New York. 15: 4487–4497. Rae JM, Vamosi JC. 2012. Ultraviolet reflectance mediates pol- Lowry DB, Sheng CC, Lasky JR, Willis JH. 2012. Five anthocy- linator visitation in Mimulus guttatus. Plant Species Biology anin polymorphisms are associated with an R2R3-MYB clus- doi:10.1111/j.1442-1984.2012.00375.x. ter in Mimulus guttatus (Phrymaceae). Am J Bot 99: 82–91. Raguso RA. 2008. Wake up and smell the roses: The ecology Lunau K, Papiorek S, Eltz T, Sazima M. 2011. Avoidance of and evolution of floral scent. Annu Rev Ecol Evol Syst 39: achromatic colors by bees provides a private niche for hum- 549–569. mingbirds. J Exp Biol 214: 1607–1612. Raguso R, Pichersky E. 1995. Floral volatiles from Clarkia Maeda H, Shasany AK, Schnepp J, Orlova I, Taguchi G, Cooper breweri and C. concinna (Onagraceae): Recent evolution of BR, Rhodes D, Pichersky E, Dudareva N. 2010. RNAi sup- floral scent and moth pollination. Plant Syst Evol 194: 55– pression of arogenate dehydratase1 reveals that phenylalanine 67. is synthesized predominantly via the arogenate pathway in Raguso RA, Willis MA. 2002a. Hawkmoth pollination in Ari- Petunia petals. Plant Cell 22: 832–849. zona’s Soronan Desert: Behavioral responses to floral traits. Menzel R, Backhaus W. 1991. Color vision in insects. In Vision In Evolution and ecology taking flight: Butterflies as model and visual dysfunction: The perception of color (ed. Gouras systems (ed. Boggs CL, et al.), pp. 43–65. University of Chi- P), pp. 262–293. Macmillan, New York. cago Press, Chicago. Miller RS, Miller RE. 1971. Feeding activity and color prefer- Raguso RA, Willis MA. 2002b. Synergy between visual and ence of ruby-throated hummingbirds. Condor 73: 309–313. olfactory cues in nectar feeding by naive hawkmoths, Man- Moerkercke AV, Schauvinhold I, Pichersky E, Haring MA, duca sexta. Anim Behav 64: 685–695. Schuurink RC. 2009. A plant thiolase involved in benzoic Raguso RA, Willis MA. 2005. Synergy between visual and ol- acid biosynthesis and volatile benzenoid production. Plant J factory cues in nectar feeding by wild hawkmoths, Manduca 60: 292–302. sexta. Anim Behav 69: 407–418. Mol J, Grotewold E, Koes R. 1998. How genes paint flowers and Raguso RA, Levin RA, Foose SE, Holmberg MW, McDade LA. seeds. Trends Plant Sci 3: 212–217. 2003. Fragrance chemistry, nocturnal rhythms and pollination Morita Y, Saitoh M, Hoshino A, Nitasaka E, Iida S. 2006. Iso- “syndromes” in Nicotiana. Phytochemistry 63: 265–284. lation of cDNAs for R2R3-MYB, bHLH and WDR transcrip- Raine NE, Chittka L. 2007. The adaptive significance of sensory tional regulators and identification of c and ca mutations bias in a foraging context: Floral color preferences in the conferring white flowers in the japanese morning glory. Plant bumblebee Bombus terrestris. PLoS ONE 2: e556. Cell Physiol 47: 457–470. Ramsey J, Bradshaw HD Jr, Schemske DW. 2003. Components Negre F, Kish CM, Boatright J, Underwood B, Shibuya K, Wag- of reproductive isolation between the monkeyflowers Mimu- ner C, Clark DG, Dudareva N. 2003. Regulation of methyl- lus lewisii and M. cardinalis (Phrymaceae). Evolution 57: benzoate emission after pollination in snapdragon and 1520–1534. Petunia flowers. Plant Cell 15: 2992–3006. Rausher MD. 2008. Evolutionary transitions in floral color. Int J Ollerton J, Alarco´n R, Waser NM, Price MV, Watts S, Cranmer Plant Sci 169: 7–21. L, Hingston A, Peter CI, Rotenberry J. 2009. A global test of Riffell JA, Lei H, Abrell L, Hildebrand JG. 2013. Neural basis of the pollination syndrome hypothesis. Ann Bot 103: 1471– a pollinator’s buffet: Olfactory specialization and learning in 1480. Manduca sexta. Science 399: 200–204. Oˆ mura H, Honda K. 2005. Priority of color over scent during Schemske DW, Bradshaw HD. 1999. Pollinator preference and flower visitation by adult Vanessa indica butterflies. Oecolo- the evolution of floral traits in monkeyflowers (Mimulus). gia 142: 588–596. Proc Natl Acad Sci 96: 11910–11915. Orlova I, Marshall-Colo´n A, Schnepp J, Wood B, Varbanova M, Schiestl FP, Do¨tterl S. 2012. The evolution of floral scent and Fridman E, Blakeslee JJ, Peer WA, Murphy AS, Rhodes D, olfactory preferences in pollinators: Coevolution or pre-exist- et al. 2006. Reduction of benzenoid synthesis in Petunia flow- ing bias? Evolution 66: 2042–2055. ers reveals multiple pathways to benzoic acid and enhance- Schwinn K, Venail J, Shang Y, Mackay S, Alm V, Butelli E, ment in auxin transport. Plant Cell 18: 3458–3475. Oyama R, Bailey P, Davies K, Martin C. 2006. A small family Penny JHJ. 1983. Nectar guide color contrast: A possible rela- of MYB-regulatory genes controls floral pigmentation inten- tionship with pollination strategy. New Phytol 95: 707–721. sity and patterning in the genus Antirrhinum. Plant Cell 18: Petroni K, Tonelli C. 2011. Recent advances on the regulation of 831–851. anthocyanin synthesis in reproductive organs. Plant Sci 181: Shang Y, Venail J, Mackay S, Bailey PC, Schwinn KE, Jameson 219–229. PE, Martin CR, Davies KM. 2011. The molecular basis for Pichersky E, Gershenzon J. 2002. The formation and function of venation patterning of pigmentation and its effect on pollina- plant volatiles: Perfumes for pollinator attraction and defense. tor attraction in flowers of Antirrhinum. New Phytol 189: Curr Opin Plant Biol 5: 237–243. 602–615. POLLINATION SYNDROME GENES 133

Smith SD, Rausher MD. 2011. Gene loss and parallel evolution active in the epidermal cells of petals. J Exp Bot 63: 3157– contribute to species difference in flower color. Mol Biol Evol 3171. 28: 2799–2810. Verdonk JC, Haring MA, Van Tunen AJ, Schuurink RC. 2005. Smith SD, Ane´ C, Baum DA. 2008. The role of pollinator shifts Odorant1 regulates fragrance biosynthesis in Petunia flowers. in the floral diversification of Iochroma (Solanaceae). Evolu- Plant Cell 17: 1612–1624. tion 62: 793–806. von Helversen O, Winkler L, Bestmann HJ. 2000. Sulphur-con- Spaethe J, Tautz J, Chittka L. 2001. Visual constraints in forag- taining “perfumes” attract flower-visiting bats. J Comp Phys- ing bumblebees: Flower size and color affect search time and iol A 186: 143–153. flight behavior. Proc Natl Acad Sci 98: 3898–3903. Waser NM, Price MV. 1981. Pollinator choice and stabilizing Spelt C, Quattrocchio F, Mol JNM, Koes R. 2000. Anthocyanin1 selection for flower color in Delphinium nelsonii. Evolution of Petunia encodes a basic helix-loop-helix protein that 35: 376–390. directly activates transcription of structural anthocyanin Waser NM, Price MV. 1983. Pollinator behavior and natural genes. Plant Cell 12: 1619–1631. selection for flower color in Delphinium nelsonii. Nature Spelt C, Quattrocchio F, Mol J, Koes R. 2002. Anthocyanin1 of 302: 422–424. Petunia controls pigment synthesis, vacuolar pH, and seed Waser NM, Chittka L, Price MV, Williams NM, Ollerton J. coat development by genetically distinct mechanisms. Plant 1996. Generalization in pollination systems, and why it mat- Cell 14: 2121–2135. ters. Ecology 77: 1043–1060. Spitzer-Rimon B, Marhevka E, Barkai O, Marton I, Edelbaum Weiss MR. 1997. Innate color preferences and flexible color O, Masci T, Prathapani NK, Shklarman E, Ovadis M, Vain- learning in the pipevine swallowtail. Anim Behav 53: 1043– stein A. 2010. EOBII, a gene encoding a flower-specific reg- 1052. ulator of phenylpropanoid volatiles’ biosynthesis in Petunia. Whibley AC, Langlade NB, Andalo C, Hanna AI, Bangham A, Plant Cell 22: 1961–1976. The´baud C, Coen E. 2006. Evolutionary paths underly- Stehmann J, Lorenz-Lemke A, Freitas L, Semir J. 2009. The ing flower color variation in Antirrhinum. Science 313: 963– genus Petunia.InPetunia evolutionary, developmental and 966. physiological genetics (ed. Gerats A, Strommer J), pp. 1–28. Wijsman HJW. 1983. On the interrelationships of certain species Springer, New York. of Petunia II. Experimental data: Crosses between different Stracke R, Werber M, Weisshaar B. 2001. The R2R3-MYB gene taxa. Acta Bot Neerlandica 32: 97–107. family in Arabidopsis thaliana. Curr Opin Plant Biol 4: Winkel-Shirley B. 2001. Flavonoid biosynthesis. A colorful 447–456. model for genetics, biochemistry, cell biology, and biotech- Streisfeld MA, Kohn JR. 2005. Contrasting patterns of floral and nology. Plant Physiol 126: 485–493. molecular variation across a cline in Mimulus aurantiacus. Winter Y, Lopez J, von Helversen O. 2003. Ultraviolet vision in Evolution 59: 2548–2559. abat.Nature 425: 612–614. Streisfeld MA, Kohn JR. 2007. Environment and pollinator-me- Wright G, Skinner B, Smith B. 2002. Ability of honeybee, diated selection on parapatric floral races of Mimulus auran- Apis mellifera, to detect and discriminate of varieties tiacus. J Evol Biol 20: 122–132. of canola (Brassica rapa and Brassica napus) and snapdragon Streisfeld MA, Rausher MD. 2009a. Altered trans-regulatory flowers (Antirrhinum majus). J Chem Ecol 28: 721–740. control of gene expression in multiple anthocyanin genes Wright G, Lutmerding A, Dudareva N, Smith B. 2005. Intensity contributes to adaptive flower color evolution in Mimulus and the ratios of compounds in the scent of snapdragon flow- aurantiacus. Mol Biol Evol 26: 433–444. ers affect scent discrimination by honeybees (Apis mellifera). Streisfeld MA, Rausher MD. 2009b. Genetic changes contrib- J Comp Physiol A 191: 105–114. uting to the parallel evolution of red floral pigmentation Zenoni S, D’Agostino N, Tornielli GB, Quattrocchio F, Chiu- among Ipomoea species. New Phytol 183: 751–763. sano ML, Koes R, Zethof J, Guzzo F, Delledonne M, Stuurman J, Hoballah ME, Broger L, Moore J, Basten C, Kuh- Frusciante L, et al. 2011. Revealing impaired pathways in lemeier C. 2004. Dissection of floral pollination syndromes in the an11 mutant by high-throughput characterization of Petu- Petunia. Genetics 168: 1585–1599. nia axillaris and Petunia inflata transcriptomes. Plant J 68: Tiffin P, Miller RE, Rausher MD. 1998. Control of expression 11–27. patterns of anthocyanin structural genes by two loci in the Zufall RA, Rausher MD. 2003. The genetic basis of a flower common morning glory. Genes Genet Syst 73: 105–110. color polymorphism in the common morning glory (Ipomoea Tornielli G, Koes R, Quattrocchio F. 2009. The genetics of flower purpurea). J Hered 94: 442–448. color. In Petunia evolutionary, developmental and physiolog- Zuker A, Tzfira T, Ben-Meir H, Ovadis M, Shklarman E, Itzhaki ical genetics (ed. Gerats T, Strommer J), pp. 269–299. H, Forkmann G, Martens S, Neta-Sharir I, Weiss D, et al. Springer, New York. 2002. Modification of flower color and fragrance by antisense Underwood BA, Tieman DM, Shibuya K, Dexter RJ, Loucas suppression of the flavanone 3-hydroxylase gene. Mol Breed HM, Simkin AJ, Sims CA, Schmelz EA, Klee HJ, Clark DG. 9: 33–41. 2005. Ethylene-regulated floral volatile synthesis in Petunia Zvi MMB, Spitzer B, Vainstein A. 2006. Navigating the network corollas. Plant Physiol 138: 255–266. of floral scent production. Acta Hortic 714: 143–154. van der Pijl L. 1960. Ecological aspects of flower evolution: 1. Zvi MMB, Negre-Zakharov F, Masci T, Ovadis M, Shklarman Phyletic evolution. Evolution 14: 403–416. E, Ben-Meir H, Tzfira T, Dudareva N, Vainstein A. 2008. Van Moerkercke A, Haring MA, Schuurink RC. 2011. The tran- Interlinking showy traits: Co-engineering of scent and scription factor EMISSION OF BENZENOIDS II activates color biosynthesis in flowers. Plant Biotechnol J 6: 403– the MYB ODORANT1 promoter at a MYB binding site spe- 415. cific for fragrant Petunias. Plant J 67: 917–928. Zvi MMB, Shklarman E, Masci T, Kalev H, Debener T, Shafir S, Van Moerkercke A, Galva´n-Ampudia CS, Verdonk JC, Haring Ovadis M, Vainstein A. 2012. PAP1 transcription factor en- MA, Schuurink RC. 2012. Regulators of floral fragrance pro- hances production of phenylpropanoid and terpenoid scent duction and their target genes in Petunia are not exclusively compounds in rose flowers. New Phytol 195: 335–345.