Angew. Chem. Int. Ed. Engl. 30: 654-672 (1991)
The Chemistry of Rose Pigments
Conrad Hans Eugster and Edith Märki-Fischer

8. Anthocyanins in Roses

8.1. General Considerations

Fig. 6. The first intentional crossing product of European garden roses of the foetida group. Top left: 'Soleil d'Or' (1:6.5); Pernet-Ducher, 1900, small shrub, ancestor of most modern yellow and orange garden roses. Right: 'Friesia' (= 'Korresia', 'Sunsprite', 1:4.5; Kordes, 1977, an example of a modern, nonfading bush rose. Botton left: 'Lady Penzance' (1:2.5; shrub, Penzance, 1894). Middle: 'Gottfried Keller' (bed rose, H. Müller, Weingarten/Pfalz, 1894). 'Lady Penzance' and 'Gottfried Keller' had practically no influence on the development of modern yellow roses. Their content of carotenoids and anthocycnins is readily apparent. All four rose varieties contain carotenoids from late biogenetic steps. The additional content of anthocyanins is readily apparent for all varieties with the exception of 'Friesia'.

Anthocyanins are more important than carotenoids for the colors of roses, because their light absorption extends over a greater part of the spectrum. At short wavelengths, the orange almost coincides with that of the carotenoids, while at long wavelengths the color extends to a pale mauve, The properties of anthocyanins could hardly be more different from those of the carotenoids; they are hydrophilic and are generally dissolved in the cell sap. Analytical studies on rose anthocyanins began as early as 1915 with the classical work of Willstätter and Nolan on the isolation of cyanin (66) from dried "flores rosae gallicae rubrae".[56] In 1934, pelargonin (62) was isolated for the first time from the new, orange scarlet variety 'Gloria Mundi' (de Ruiter, 1929);[57] this was followed in 1961 by peonin (71) from R. rugosa and its hybrids[58] as well as the 3-glucosides 63 and 68.[9a] For systematic analytical studies of hundreds of hybrids and species, see Refs. [9a, 59]. Our own HPLC studies of roughly 160 varieties and botanical roses have led to the identification of five to six further anthocyanins and anthocyanidins.[60] The structures of the anthocyanins so far identified in rose petals are shown in Figure 7.

Only glucose is used for glycosidation, and the 3,5-diglucosides are the dominant pigments. All the others are minor components and play only an occasional color-determining role in pink flowers. Whether other glycosides occur in the nonidentified trace components is not yet known. They may be important for breeding. The absence of delphinidin derivatives (e.g., 75) is most important, since the hope of a blue rose vanishes with them.

A qualitative relationship exists between the intensity and saturation of petal coloration and the anthocyanin content (Table 5; see also the older analyses in Ref. [61]).

Less colored precursors of anthocyanins such as chalcones (79) and hydrates (78) (see Scheme 5), which can re-form anthocyanins on acid treatment, are not present in significant amounts, since there is no visible deepening of the color during extraction of pink roses. The noncorrelation between flower color and the light absorption of the flavylium cations (Fig. 12) responsible will be discussed in Section 8.5.

Table 5. Spectrophotometric determination of anthocyanins.
Rose variety Flower color Anthocyanin Content [a]
'Mister Lincoln' (Swim & Weeks, 1964) dark red 66 (little 67, 68) 0.6
Papa Meilland' (Meilland, 1963) dark red 66 0.3
R. rugosa rubra violet-pink 71 (little 72, 73) 0.25
'Veilchenblau' (Schmidt, 1909) lilac with white 66 0.1
'Better Times' (Hill, 1934) cherry red 66 (+ ?) 1.3-1.8 [b]
'Maria Callas' (Meilland, 1965) dark pink 66 (little 67) 0.1
'Queen Elizabeth' (Lammerts, 1954) light pink 66 (little 72) 0.03
[a] % of fresh weight, calculated with e = 37 000 for 66 and 25 000 for 71 in 1% trifluoroacetic acid in MeOH/H20 1:1.
[b] Content in epidermis cells [61].
Fig. 7. Anthocyanins identified in rose flowers. The flavylium cations stable at low pH are shown.

8.2. Pelargonin in Roses

Pelargonin (62; see Fig. 8) did not appear in roses in color-determining concentrations until 1929. In other garden plants, such as pelargoniums and geraniums, it had been known for years. However, analysis shows that pelargonin is hardly ever formed on its own, but is accompanied in almost all scarlet roses by cyanin (66). 'Gloria Mundi' is one of the very few roses that contains almost pure pelargonin (Fig. 8). Since then, the unexpected appearance of 62 in roses has been traced by many authors (cf. Ref. [62]) to a mutation. Now, however, analyses show[9a, f] that even very old varieties contain significant amounts of 62; examples are R. centfolia muscosa (moss rose, appeared before 1750), 'Königin von Dänemark' (Booth, 1816, alba hybrid?), 'Général Jaqueminot' (Roussel, 1853, hybrid perpetual), 'Crimson Rambler' (Japan, pre 1890, multiflora hybrid), and 'Soleil d'Or'. In contrast to the views of Refs.[9 a, d], we also found pelargonin in botanical roses, such as R. pendulina, R. willmottiae, R. rugosa, R. pomifera, and R. gallica versicolor. These are species that originate in quite different parts of the world and are members of different botanical sections of the subgenus Eurosa. This can be explained by referring to the general biosynthetic scheme for flavonoids, modified according to Ref. [63] (Scheme 4 and Fig. 9). It shows that there is no immediate connection between pelargonin, cyanidin, and delphinidin, but that they represent different branches from a series of precursors. Pelargonidin only occurs if the reduction of dihydroflavonol A to flavandiol A is not inhibited, while the hydroxylation of flavanone A to flavanone B is blocked. How the formation of flavonol A is correlated to pelargonidin is not yet known.

CybeRose note: 'Independence' is descended from 'Miss Edith Cavell', which, like 'Gloria Mundi', was a sport from 'Orléans Rose'.

With pelargonin one also always finds some of the 3-glucoside (callistephin, 63; see Fig. 7). The 5-glucoside 64 is very rare; it occurs, for instance, in 'Super Star' (Tantau, 1960). We assume that pelargonidin also arose spontaneously in other breeding strains not related to 'Gloria Mundi' and its relatives, such as 'Independence' (= 'Kordes Sondermeldung', Kordes, 1951).

8.3. Cyanin in Roses

Cyanin (66) is the most important pigment in red roses. Often it is accompanied by smaller quantities of chrysanthemin (68). We have only found a few varieties where the ratio 68:66 is greater than 1, for example, in the old bourbon rose 'Souvenir de la Malmaison' (Béluze, 1843) and in the climbers 'François Juranville' (Barbier, 1906) and 'Dorothy Perkins' (Jackson and Perkins, 1901). As the cherry red flowers of the latter variety show, high levels of chrysanthemin can yield surprising red shades. We found the 5-glucoside 69 quite often, but always in low amounts. Cyanidin (70) and the 3,7-diglucoside (67) are rare. The well-known paradox that cyanin is present in both red and purple flowers is also true for roses; the mauve-colored 'Blue Moon' (= 'Mainzer Fastnacht', Tantau 1964) and 'Veilchenblau' (Fig. 10) both contain almost pure cyanin. For a possible explanation, see Section 8.5. A direct synthesis of cyanidin from caffeoyl-CoA in roses cannot be excluded, as it has not been studied (see Scheme 4).

8.4. Peonin in Roses

The earlier assumption that peonin (71) is typical for R. rugosa and rugosa hybrids must today be modified. More recent studies[9a, 60, 64] show a broad distribution in the sections Cinnamoneae, namely, R. arkansana, R. acicularis, and R. x dulcissima (all from North America), and R. moyesii, R. multibracteata, and R. sweginzowii macrocarpa (all from China), the Caninae (R. canina, R. glauca, and R. pomifera;  all European), and the Pimpinellifoliae (R. foetida bicolor!). The mysterious 'Crimson Rambler' and 'Königin von Dänemark' also contain unexpectedly high levels of peonin, which throws new light on their unknown pedigrees.

Fig. 8. 'Gloria Mundi' (1:2.5; de Ruiter, 1929); the first garden rose in which mainly pelargonin is responsible for the color.
Scheme 4. Flavonoid biogenesis, modified from [63]: , major pathways in roses; →, minor pathways; - - →, not yet detected in roses. The abbreviations refer to Figure 7. Pel = compounds of the type 62-65. Cy = 66-70. Pae = 71-74. Del = 75. glc = D-glucose.


Fig.9. Key to Scheme 4.


Fig. 10. Roses with nearly pure cyanin in their petals. Top left: The apothecary's rose (1:3.5; R. gallica officinalis; shrub rose), widely used medicinally from the Middle Ages to the end of the 19th century Middle: 'Duchesse de Montebello' (shrub, Laffay). Right: 'Veilchenblau' (climber, Schmidt. 1909). Bottom left: 'Bonfire' (rambler, Turbat, 1928). The flower color is primarily dependent on the amount of cyanin as well as on copigmentation and (presumably) the fine details of petal structure.

Pure peonin is not found even in R. rugosa rubra; it is accompanied by much cyanin. The richest source of almost pure peonin is the old shrub rose 'Hansa' (Schaum and Van Tol, 1905; Fig. 11). Its mauve-red flowers show what can be achieved by high levels of pure peonin! Rarely have we found the 3- or 5-glycosides 72 (oxycoccicyanin) and 73, or free peonidin (74).

The Austrian copper (R. foetida bicolor) contains surprising amounts of peonin, as already mentioned.[9a, 60] If its progeny contain anthocyanins—which is not always the case—they always include peonin. An example is 'Lady Penzance' (Penzance, 1894), a shrub rose bearing coppery red flowers with a yellow center (see Fig. 6). However, a connection between rugosa type and the peonin content of the flowers need not exist. Thus, analysis of the flowers of 'Conrad Ferdinand Meyer' (H. Müller, 1899) shows them to be free of peonin.

Peonin is probably derived from cyanin.[63] Whether the methylation occurs at the level of the cyanidin or only at its glucosides is unknown. A direct route starting from activated ferulic acid (3-O-methylcaffeoyl CoA) is also possible.

8.5. The Stabilization of the Anthocyanin Chromophore in Roses

8.5.1. General Considerations

The sap of the anthocyanin-containing epidermis cells of the cerise-red 'Better Times' has a pH of 3.7-4.2, as determined by spectrometric microtechniques.[61c] Within three days, this rises to 4.4-4.5, and the petals take on a slightly bluish tinge. In this pH range, pure flavylium ions are mostly transformed into the colorless hydrates 78 or the yellow chalcones 79 (Scheme 5), since the intermediate quinone methides 77 are so electrophilic that they do not survive in aqueous solution.[65] Hence, the question of stabilization arises for all anthocyanin pigments in plant organs, and a great many studies have been devoted to the subject, especially recently.[66] The interpretations given are selectively true for certain plants and to various extents: complex formation between anthocyanins and flavonoids with metal ions,[66a, 66b] intramolecular stacking,[66a, 66b] association (copigmentation) with the related chalcone,[67] stacking by self-association ,[66] copigmentation with flavonols and related compounds.[66] Which of these are true for roses?

Fig. 11. Peonin-dominated flower colors. Left: 'Frau Dagmar Hartopp' (1:2.3; small shrub, Hastrup, ca. 1914). Right: 'Hansa' (1:3; shrub rose, Schaum and van Tol, 1905).

Metal complexes with anthocyanins and flavonoids have not yet been detected. These would be most likely for lilac or mauve varieties. In any case, studies of 'Veilchenblau' showed that the isolable, intense blue-violet fractions are very unstable and do not correspond[60] to the types described by Goto et al.[66] Intramolecular stacking appears when the glycosyl residues on the anthocyanidin (disaccharides!) are esterified to cinnamic acids. This shields the pyrylium ring from nucleophilic attack at C(2) and hence protects the anthocyanin chromophore. In roses, neither disaccharide glycosides nor esters of phenylpropanoid acids have been found, so that this form of stabilization must be disregarded. Copigmentation with the related chalcone, which is in equilibrium with the anthocyanin, leads in vitro, in 1:1 ratio, to significant stabilization, with bathochromic and hyperchromic effects.[67] This stabilization mechanism was suggested only recently and cannot be excluded for roses. However, it would lead to a great increase in the extinction on acidification, due to a shift of the equilibria towards the flavylium cation (Scheme 5), and this has not yet been observed.

Scheme 5. R=H, OH, OCH3

Stacking by self-association has been demonstrated in in vitro experiments by deviation from the Lambert-Beer relationship and by strong exciton couplings in the circular dichroism spectrum. Al3⊕ salts may also play a role here.[68] This phenomenon could be important in roses with high anthocyanin content, but proof is lacking.

8.5.2. Copigmentation in Roses

From our current knowledge, copigmentation is the most important process for stabilization of rose anthocyanin chromophores. It consists of a loose association between a flavylium cation or quinone methide and a flavonol or related compound and is only observed in aqueous medium. To date, no such loose complexes have been isolated. This association has several consequences: shifting of the anthocyanin equilibria in Scheme 5 towards 76 and/or 77; stabilization of the light absorption in the visible even in the physiological pH range; a shift in the visible maximum towards longer wavelengths; an increase in the extinction compared to a solution without copigment (see Fig. 12). Copigmentation effects are proportional to the concentration of the copigment. Initial studies of the copigments in 'Better Times' revealed a series of flavonol glycosides,[61c] including kaempferol-3-O-ß-D-glucoside, a xyloside, quercitrin (84), and a quercitrin glucoside, arabinoside, and glucuronoside.

Fig. 12. Copigmentation effect for peonin: Absorptions of the pigment from Rosa rugosa in solution
(• • • •, in CH3OH/O.2% HCl) and of its diffuse reflectance from fresh petals (- - - -).

Recent studies on 'Veilchenblau' and 'Papa Meilland' have confirmed the wealth of flavonol glucosides present.[13k, 60a] Compounds 80-84 were identified in both these varieties (Fig. 13). They also occur in numerous white, yellow, and red roses in quite different proportions and mixed with other, not yet identified flavonoids. Spireoside (82) displays an especially strong copigmentation effect. At pH 4.7 and in a cyanin/spireoside ratio of 1:1, λmax is shifted by 18 nm to longer wavelengths; at a 1:2 ratio the shift is as great as 27 nm.[69] More detailed studies of roses with respect to their flavonol glycosides and other, as yet unknown copigments will be most worthwhile.

9. Tannins in Rose Petals

As many rose lovers know, chewing a rose petal produces an astringent effect in the mouth. The assumption that tannins occur in roses is an old one. Tannins have even been considered as possible stabilizers of the anthocyanin chromophore.[70] A more detailed study has now revealed that rose flowers contain a great many ellagitannins.[13k] These are esters of gallic acid (3,4,5-trihydroxybenzoic acid) and its oxidation products with monosaccharides.[71] Compounds 85-89 were identified in 'Papa Meilland' and compound 86 in 'Veilchenblau' (Fig. 14). Ellagitannins are widespread and occur in high concentrations in all rose varieties. Many still await identification.

Fig. 14. Ellagitannins identified in rose petals.
Fig. 13. Flavonol glycosides identified in roses.

The presence of ellagitannins offers the first rational explanation for the long-standing medicinal use of dried rose petals. The apothecary's rose (R. gallica officinalis), in particular, has been used to treat diarrhoea, bleeding, and all kinds of inflammations.[72]

Compounds 85-89 display very little copigment effect. Though anthocyanins do not bleach in their presence, they lose their distinct maxima in the visible.

The trihydroxylated benzene ring of gallic acid is most remarkable! If this hydroxylation were to take place on anthocyanins to yield, for example, 75, then the path to blue roses would lie open. It follows, therefore, the gallic acid and the flavonoids are synthesized by different routes.


Fig. 15. Modern orange-red roses in which the brightness of the pigments is increased by the presence of carotenoids. The most important anthocyanin is pelargonin. Left: 'Ville de Zürich' (1:3.8; bush rose, Gaujard, 1967). Right: 'Orangeade' (1:3.7; bush rose, McGredy, 1959).

10. Orange Roses

In recent decades, the breeding of orange roses has been strongly promoted. These modern varieties give particularly bright colors when planted in large beds; their shades lie somewhere in the overlap region between yellow and red. The almost uniform coloration is remarkable in these roses and is quite different from that of the yellow-red, two-tone roses. In the latter, carotenoids and anthocyanins show quite clear local differences in concentration. A good example is the Austrian copper (R. foetida bicolor), which has carotenoids on the outside and anthocyanins on the inside of the petals. In modern orange garden roses, this differentiation is absent. Pigment analysis shows that the saturation of the yellow cannot be explained by the presence of high levels of pelargonin (62): the new orange roses contain easily detectable quantities of carotenoids and high levels of anthocyanins (Table 6, Fig. 15). Even though the carotenoids are concentrated in the base of the petal, they are easily detectable in the blade by reflection spectroscopy; in vivo, 'Alexander' shows a reflection maximum at 425/445 nm, typical of carotenoids. Two further examples demonstrate how strongly carotenoids influence the color: flowers of the old rose 'Daily Mail Rose' (= 'Mme Edouard Herriot', Pernet-Ducher, 1913) show almost uniform areas of a strong, coppery orange, and the same is true for the bright orange rose 'Louis de Funè' (Meilland, 1984). To our surprise, the anthocyanin content of both of these comprises practically pure cyanin, and the coppery orange is due to a mixture with carotenoids.

Table 6. Anthocyanins and carotenoids in orange-red roses.

Variety Anthocyanins Carot. [a] N [b] MC [c]
'Sarabande'(Meilland, 1957) 62 + 63 (ca. 1:1) 1.5 16 7, 10, 14
'Super Star' (Tantau, 1960) 62 + 66 (ca. 3:2) 1.6 21 7, 13, 22-24
'Orange Bunny' (Meilland, 1980) 62 + 66 (ca. 2:1) 1.0 17 7, 10, 14
[a] Carotenoids: mg per 100 g dry weight. [b] N = number of carotenoids identified. [c] Mc = main carotenoids.

The physiological conditions that allow such an intimate mixture of pigments with such different solubility characteristics, and yet give a uniform coloration, are unknown.

A development is therefore in progress which, according to measurement of color and analytic findings, has not yet reached a conclusion. This is shown by a comparison between modern orange varieties of garden roses and the nasturtium (Tropaeloum majus), an old garden plant. Its flowers also contain the usual carotenoids and anthocyanins.[73] Nevertheless, its orange attains a higher saturation of the yellow than has yet been found in any comparable variety (Table 7).

Table 7. Measurement of color (CIELAB).

Flowers L* [a] + a* [b] + b* [c]
'Super Star' 51-55 56-58 48-55
'Orange Bunny' 54 58 53-55
'Alexander' 56 54 51
Nasturtium (orange) 52-53 55-57 80-83
L*, measure of lightness; a*, measure of color saturation on the red-green
axis; b*, measure of color saturation on the yellow-blue axis.

11. Pigments of Rosehips

Rosehips come in many different colors, shapes. and sizes. The flesh is colored a lively yellow, orange, pink, red, brown, or deep black by carotenoids and anthocyanins. Chemical studies on these pigments began in 1913. many years before carotenoid structures were known.[74] It is striking that they only concerned the hips of botanical roses; perhaps. especially in more recent times, this has been due to an interest in their possible content of provitamin A1.[75] The published results are difficult to evaluate, because of often inadequate purification and identification procedures. Table 8 contains some relevant, recent results.[12, 76] The dominance of lycopene (5) and β,β‑carotene (7) can be seen, also occasionally of rubixanthin (10) and lutein (14). In contrast to rose flowers, epoxides are almost completely absent in R. pomifera only 3-4% of the total content consisted of epoxides, including the stereoisomeric mutatoxanthins (17, 18), which are only found in traces in flowers if at all.

Table 8. Carotenoid consent of rosehips.

Rose Content [a] N [b] Main carotenoids [c] Ref.
R. canina 1.2-2.5 14 5, 7, 10 [76a]
  133 [d] 11 5, 7, 10 [76b]
R. moyesii 22.4 10 5, 7, 10, 11(?), 13, 14 [76a]
R. rubrifolia 8.8 11 5, 7, 10 [76a]
R. rubiginosa 162 [d] 8 5, 10 [76b]
R. pomifera 6.2 43 [e] 5, 7, 10, 13 [13b]
'American Pillar' (van Fleet, 1902) 2 17 5, 7, 14 [12]
'Golden Wings' (Shepherd, 1956) 6.1 15 7, 14 [12]
'Sarabande' (Meilland, 1957) 5.4 16 7 [12]
[a] mg carotenoids per 100 g fresh weight [b] N = number of carotenoids identified
[c] HC = main carotenoids. [d] mg in dried rosehip flesh. [e] Analyzed by HPLC.

The anthocyanins in rose hips have hardly been studied at all. The conspicuously black hips of R. pimpinellifolia contain almost pure chrysanthemin (68), according to Demina[77] and demonstrate in extremis how important it is to identify the copigments that are presumably responsible for the black color.

12. The Dream of a Blue Rose

Many breeders and even more rose fanciers have dreamed of a blue rose to the disapproval of more traditional rose lovers. However, violet-red roses have existed for a long time; for example, Tuscany' (gallica hybrid, pre-1600), 'Cardinal de Richelieu' (Laffay, 1840), and several others from the hybrid perpetuals, such as 'Reine des Violettes' (Millet-Malet, 1860). The roses tending most to violet are, however, 'Veilchenblau', a climbing rose with large inflorescences full of small, simple, lilac-colored individual flowers, and 'General Stefanik' (J. Böhm, 1931), a dwarf hybrid perpetual. In the former, 'Crimson Rambler' may have given rise to the violet color, in the latter R. rugosa. The hybrid polyantha 'Baby Faurax' (Lille, 1924) appeared later, as did the pale lilac tea hybrid 'Blue Moon' and many others. No breakthrough has been achieved to date.

The anthocyanins detected so far leave little hope for breeding a blue rose by conventional methods. With peonidin (74) as a basis, a rose with sky-blue flowers like those of Ipomoea coerulea (morning glory) would be possible if the requirement for intramolecular pigmentation (stacking with disaccharides and caffeic acid esters[78]) were fulfilled. With cyanidin (70), one can fall back on a comparison with cornflowers. Here, too, suitable glycosidation, acylation, and complex formation with cations would be required (see Section 8.5.1).

In Section 9 it was mentioned that rose petals are full of gallic acid esters. Could this hydroxylation activity be redirected to ring B of cyanidin? Otherwise, only the gene-technological path remains, with introduction of suitable genes from Petunia, for example. A far-fetched idea? One must not forget that a successful breeder could expect to receive several millions in licence fees for a good blue rose.[79]

13. Epilog

Our studies of rose pigments bring some more light into the long history of rose breeding. Roses are still full of unsolved puzzles, however. We would like to prompt young researchers to take them on with further scientific methods. Roses are also of great economic importance.[80] Moreover, our investigations have revealed gaps in the color range of current varieties. In yellow roses, the degree of the saturation of the color of the foetida group has still not been reached; an improvement in the lycopene, β,β‑carotene, or rubixanthin content could produce quite new shades (compare tomatoes, corollas of narcissus with about 1.5% β,β-carotene, and rosehips); an increase in the content of chrysanthemin (68) would afford purer and more powerful red shades.

The analysis of rose carotenoids has furnished more insight into the genealogy of garden roses than bare the anthocyanins, where the small number of structures is a disadvantage. The identification of trace components, especially flavonol glycosides and other copigments, could help. Above all, the copigments must 5e studied more intensively. To date, breeders have always oriented themselves by the anthocyanin colors expressed and have ignored (or have been unable to consider) the copigments that are required for stabilization of the color and which are also passed on genetically. Scientifically, a study of carotenoid-degrading enzymes is also long overdue; roses constitute a readily available starting material, and well-developed procedures for analyzing the cleavage products are known.


Eugster & Märki-Fischer: Carotenes (1991)