Botanical Journal of the Linnean Society (1981), 83: 57-84.
Pigment distribution, light reflection and cell structure in petals
Department of Botany and Microbiology, University College, Swansea SA2 8PP
Botanical Research Institute, Private Bag X101, Pretoria 0001, South Africa

Accepted for publication September 1980

Petal structure and the distribution of pigments in petals were studied in relation to the functional anatomy of petals and the ways in which petals absorb and reflect light. We examined 201 species from 60 angiosperm families. Anthocyanins, betalains and ultraviolet-absorbing flavonoids were normally confined to the epidermal cells, occurring in solution in the vacuole; carotenoids were found in the epidermis and in smaller quantities in the mesophyll, normally in chromoplasts. In a few species, mainly blue-flowered members of the Boraginaceae and Liliaceae-Scilleae, anthocyanins were confined to the mesophyll.

Six basic kinds of petal epidermis anatomy were found, sometimes in combination; papillate (112 species) and multiple-papillate (13 species), in which the conical-papillate form of the cells traps incident light and scatters emergent light, with surface striations aiding these functions in many cases; reversed-papillate (4 species), multiple reversed-papillate (29 species), lenticular (32 species) and flat (11 species), all with surface striations in some cases. Light is usually reflected from petals mainly by an aerenchymatous unpigmented reflective mesophyll; in certain species this is replaced by a reflective layer of starch grains in the upper mesophyll.


The chemistry, biochemistry and taxonomic distribution of petal pigments have been intensively studied for many years, and an immense and increasing body of data has been amassed on these topics (Harborne 1967, 1975, 1977). There remains a surprising lack of data on the distribution of these pigments within the cells and tissues of living petals; indeed little is known of the range of cell structure in the petals of angiosperms. There is a corresponding lack of understanding of the relationships between petal structure and pigment distribution, and of the ways in which petals absorb and reflect light and thus acquire their characteristic colours, brightness and textures. Little progress has been made in this field since the work of Exner & Exner (1910), who investigated the relationships between petal structure and function in a common kind of petal with a conical-papillate, pigment-containing epidermis above an aerenchymatous mesophyll, and also investigated the less common Ranunculus kind of petal. Their conclusions, which are discussed below, were largely correct but were limited to certain aspects of petal function and structure. Exner & Exner's work is surprisingly little-known today.

The remarkable and distinctive surface structures of the outer cell walls of the epidermis of petals have recently been investigated with the SEM in major plant families such as the Asteraceae (Baagoe, 1977a, b, 1978), Fabaceae (Stirton, 1980), Orchidaceae (Ehler, 1976) and Asclepiadaceae (Ehler, 1975). An attempt has also been made to establish a universal terminology for describing the different types and patterns of surfaces found in petals, leaves and seeds (Barthlott & Ehler, 1977). However, there have been few recent studies of the internal structure of petals, and there appears to be little discussion on the mechanisms by which petals reflect light whether externally or internally. Although the functions and general features of the two main kinds of reflective tissue in petals were correctly described by Exner & Exner in 1910, their work has been forgotten and is ignored in recent publications concerned with the reflective properties of petals (e.g. Stickland, 1974; Kevan, 1978).

Observations that anthocyanins are confined to the epidermis of the petals are scattered through the literature (e.g. Hildebrand, 1863; Blank, 1947; Arditti & Fisch, 1977). A smaller number of workers has observed that this was also true for ultraviolet-absorbing flavonoids (Caldwell, 1971; Stewart, Norris & Asen, 1975; Arditti & Fisch, 1977) and in some instances for carotenoids (Paech, 1955; Arditti & Fisch, 1977). The published data are sufficient to justify assertions of the kind made by Exner & Exner (1910) that petal anthocyanins are known to be localized in the epidermis.

There are similarly scattered observations of the occurrence of petal epidermises with a conical-papillate cell structure (e.g. Müller, 1893; Exner & Exner, 1910; Parkin, 1928; Knoll, 1938; Stewart et a!. 1975; Roberts & Humphreys, 1980) but published data on the nature and distribution of this type of epidermis are sparse. Apparently the best source of published information on the structure of the petal epidermis is the survey made by Schubert (1925), who investigated more than 330 species from 51 families and gave some data on pigment distribution in the petals of 120 species. Schubert was mainly concerned with the gross morphology of the petal epidermis, especially the occurrence of papillate cells. He hardly commented on surface striations, and his observations of pigment distribution were confined to visible pigments in 30 families; he made few observations of the internal structure of petals and failed to obtain satisfactory petal sections for many species. We have investigated pigment distribution and petal structure in 201 species from 60 angiosperm families, including 41 of the families studied by Schubert (1925).


Fresh, fully-open (lowers were obtained from wild plants or cultivated plants of known origin. Standard methods of fixation and preparation (Hall, Skerret & Thomas, 1978; Daoud, 1980) were used for all work with the SEM, but fresh unfixed petals were used in all other investigations, either unmounted, or mounted in water. Fresh petal sections were cut by hand with a sharp razor blade and examined immediately. Ultraviolet microphotographs were made by transmitted light with a Schott UG1 or UG5 filter inserted into the optical system of the microscope to exclude visible wavelengths; a high-temperature electronic flash provided the source of ultraviolet light, and Kodak Tri-X or Ilford FP4 film was used (Daoud, 1980). The localization of ultraviolet-absorbing pigments in the petal was determined either by ultraviolet microphotography of fresh sections, or by irrigating fresh sections mounted in water with ammonium hydroxide solution; most flavones and flavonols then become visible as yellow pigments, because their absorption spectra are shifted bathochromically in alkaline solutions. In several cases the ultraviolet-absorbing pigments were also extracted and isolated by chromatography, and provisionally identified by ultraviolet spectroscopy with a Unicam 8000 recording spectrophotometer, using standard techniques (Mabry, Markam & Thomas, 1970; Daoud, 1980).


Petal pigments

Our data for the distribution of petal pigments are compared with the data of Schubert (1925) in Table 1 and summarized in Table 2. Our results show that, in the wide range of families investigated, both anthocyanins and ultraviolet-absorbing flavonoids appear to be substantially confined to the epidermis in the majority of petals. However, there are certain exceptions to this general rule, notably blue-flowered members of the Boraginaceae and Liliaceae-Scilleae in which the blue anthocyanin pigments occur mainly or entirely in the mesophyll. We have also observed that anthocyanins, yellow flavones and flavonols, and the ultraviolet-absorbing flavones and flavonols of 'white' (insect-yellow; Kay, 1979) petals normally appear to be present in solution, evenly dispersed through the vacuoles of the epidermal cells, and are rapidly lost from damaged and marginal cells in preparations mounted in water (Figs 1-4). This is probably also true of ultraviolet-absorbing flavonoids in carotenoid-containing epidermal cells (Daoud, 1980). We have not observed any apical concentrations of ultraviolet-absorbing pigment granules in living cells of the type reported in the papillate epidermis of freeze-dried petals by Brehm & Krell (1975). These could have been artifacts produced during the process of freeze-drying, in the manner discussed by Hall, et al. (1978).

Figures 1-4. Sections and surface views of fresh petals mounted in water, showing the shape of the papillate epidermal cells and the restriction of water-soluble ultraviolet-absorbing compounds to the epidermal cells. Fig. 1 Saxifraga rosacea, in visible light, longitudinal section, x 150. Fig. 2. Saxfraga rosacea in ultraviolet light of 330-400 nm, x 150. Fig. 3. Leucanthemum vulgare, in visible light, edge of lamina showing surface and side views of papillate cells of upper epidermis, x 210. Fig. 4. Leucanthemum vulgare, in ultraviolet light of 330 400 nm. x 210.
Close-up, papilate surface of Black Iris above, 'Ville d'Ettelbruck' rose below.

Structure and functions of the petal epidermis

Although the data of Schubert (1925) on the shape of the outer cell walls of the petal epidermis are extensive, he gave little information on the occurrence of surface striations on the outer cell wall, and his observations of the shape of the inner cell wall are of uneven quality. Of the 201 species examined by us (Table 2), 45 had been investigated or were closely related to species investigated by Schubert. Although our results generally agree, the occasional shortcomings of his observations are clear. His failure to observe any of the many cases in which the epidermal cells are multiple-papillate is particularly striking. Multiple-papillate cells which do not appear to have been reported previously, are longitudinally elongated, and may bear a row of acute papillae on the outer face with a corresponding row of lenticular projections on the inner face, as in many Caryophyllaceae and Cistaceae (Figs 5-8), or may have a row of lenticular projections on both faces, as in Anagallis arvensis ( Figs 7, 8), or may have a relatively flat outer face with a row of lenticular or papillate projections on the inner face (reversed multiple-papillate) as in Crocus species and some Caryophyllaceae and Papaveraceae (Figs 9, 10). Schubert's failure to observe cells of any of these multiple-papillate kinds may have been the result of his reliance on transverse sections, in order to observe multiple-papillate cells, which are normally of considerable length, one must prepare longitudinal sections of the petal because in transverse sections the multiple-papillate cells are cut and consequently collapse.

Figures 5-10. Longitudinal sections Figs 5, 7, 9 and surface views (Figs 6, 8, 10) of fresh petals mounted in water, showing multiple-papillate and reversed multiple-papillate epidermal cells. Figs 5, 6. Cistus albidus, x 420 (acute multiple-papillate). Figs 7, 8. Anagallis arvensis, x 480 (rounded multiple-papillate) Figs 9, 10. Papaver dubium, x 420 (reversed multiple-papillate).

Schubert observed and figured a very large number of cases in which the petal epidermis was of the common, singly conical-papillate kind, as in Viola (Fig. 6), and he discussed extensively, the possible relationships between structure and function in this kind of epidermis. His view that the papillae functioned as footholds for pollinating insects in certain specialized cases (e.g. Convallaria majalis and Polygonatum species) may he correct. His dismissal of the theory, of Hiller (1884) that the petal epidermis functioned as a water-storage tissue is convincing, but his own conclusion that the major function of the papillate epidermis was to act as an energy-concentrating light-absorbing tissue, in which metabolism was enhanced as a result of increased temperature, has little evidence to support it. A similar function had already been proposed for the papillate epidermises of some shade-leaves (Solereder, 1908: Haberlandt, 1914. Papillate epidermises also occur in secretory organs including floral nectaries) and on the receptive surface of the stigma, but it is clear that the functions of these papillate epidermises differ from those of the surfaces of petals, although secretory cells may sometimes occur on the petal lamina (Loomis & Croteau, 1973). The surface morphology of the petal epidermis may also act directly, as a specific tactile or visual recognition stimulus in certain cases, as Kullenberg (1961) has suggested for Ophrys and Stirton (1980) for Fabaceae. The papillate epidermis of the petals of Rosa species may aid flower opening (A. V. Roberts, personal communication); in the light of this suggestion it seems that the concertina-like structure of the multiple-papillate petal epidermises of Cistaceae and Papaveraceae may aid the rapid petal expansion that is characteristic of these families, while similar multiple-papillate or multiple reversed-papillate petal epidermises in other families may, function in a similar way during diurnal opening and closing, as for example in Anagallis, many Caryophyllaceae-Alsinoideae, Crocus and Oenothera (Table 2). Our observations strongly support the conclusion of Exner & Exner (1910) that the primary function of the papillate epidermis of petals is to act as a light-trap for incident light and, in conjunction with the reflective mesophyll, to guide incident light through the pigments contained in the epidermal cells and to return it to the exterior by a combination of external reflection, refraction and internal reflection.

Table 1. Localization of pigments within the petal: a comparison between the survey of 120 species in 30 families made by
Schubert (1925) and the present survey of 201 species in 60 families. Schubert did not investigate UV-absorbing pigments.
The number of families in each category is shown in parentheses.

  Schubert Present survey
  Anthocyanin Carotenoid Anthocyanin Carotenoid UV-absorbing pigment
Substantially confined to epidermis 64 (21) 20 (11) 73 (38) 10 (7) 83 (34)
Also clearly present in some mesophyll cells -- 20 (11) 8 (7) 20 (16) 24 (18)
Mainly or exclusively in mesophyll 10 (5) 5 (3) 8 (3) -- --

Figures 11, 12. Petals of Viola tricolor. Fig. 11. Longitudinal section, x 510, mounted in water, showing the shape of the conical-papillate cells of the upper epidermis and the epidermal localization of pigment. Fig. 12. Petal surface, x 160 in incident light, showing the surface reflections from the tips of the conical-papillate cells of the upper epidermis.

Reflection and absorption of light by petals

A typical papillate petal epidermis absorbs incident light very efficiently, in that almost all incident light enters the epidermal cells, with only a small proportion being reflected directly from the outer walls of the papillae towards an observer. Direct surface reflection is usually confined to the tips of the papillae, which thus appear as bright spots in photographs of petal surface made by incident light (Fig. 12), especially, when a filter which transmits only, within the absorption range of petal pigment is placed over the camera lens Knoll, 1938). In contrast, smooth-surfaced petals, including those of Crocus species and others of the reversed multiple-papillate kind show strong surface reflections in such photographs (Figs 13-15). The reflectance spectra of papillate petals are noteworthy for their strong absorption within the absorption range of the petal pigments, again showing that very little light is reflected directly, from the outer surface; almost all the light that is reflected from the petal has passed through the pigments contained within the epidermal cells.

Figures 13, 14. Flowers of Crocus sieberi var. sieberi, x 0.8. Fig. 13. In visible light. Fig. 15. In ultraviolet light of 330-400 nm. (Kay, 1979). The petals (perianth segments) contain an ultraviolet-absorbing flavonoid in the epidermal tells; Fig. 14 shows the strong surface reflections from the flat oilier faces of the reversed multiple-papillate cells of the upper epidermis.

Figure 15. Petal of Crocus vernus in visible light, x 160. Note strong surface reflections from upper epidermis.

If the aerenchymatous mesophyll of a typical petal is exposed by stripping off part of the epidermis, it normally appears white or almost white in colour to the human eye: it reflects ultraviolet light as well as visible light and is thus insect-white, in the sense of Kevan (1978) (Figs 16, 17). Its true colour thus differs from that of normal 'white' petals, which with rare exceptions (Kay, 1979) contain ultraviolet-absorbing pigments, usually flavones or flavonols (Roller, 1956: Harborne, 1967; Daoud, 1980) in the epidermis (Tables 1,2) and are thus insect-yellow (Kevan, 1978). If the epidermis is stripped from small areas of such ultraviolet-absorbing 'white' (insect yellow) petals to expose the ultraviolet-reflecting mesophyll (insect white), both parts appear white to the human eye but contrast very sharply in photographs taken with a filter that transmits ultraviolet but not visible light (Figs 16, 17). This corresponds to the visible contrast between the exposed white mesophyll and the intact coloured lamina when part of the epidermis is stripped from petals with anthocyanin pigmentation (Exner & Exner, 1910).

Optical properties of papillate cells

The data in Table 2 show that there are several different kinds of petal anatomy, which may represent different adaptive complexes. The most common kind of petal epidermis is the simple conical-papillate kind, with or without striated outer walls. The striking, but apparently previously unobserved multiple-papillate kind (Figs 5-l0, which we have so far found only in Caryophyllaceae, Cistaceae, Hypericaceae, Onagraceae and Primulaceae appears to have very similar optical properties. The inner faces of the papillate epidermal cells are usually convex-lenticular in petals with anthocyanin or ultraviolet-absorbing flavonoid pigments in clear solution, but are often more or less flat in petals containing carotenoid pigments; the carotenoids normally occur in a basal layer of chromoplasts, with a secondary apical group in some cases. The papillate cells range in height and shape from exceptionally tall and acute papillae like those of Dianthus barbatus, Arnebia echioides and Primula vulgaris to short stumpy papillae, as in Dryas octopetala and Brassica oleracea; the latter are clearly transitional to the lenticular type of epidermal cell. The shape of the papillate cells varies widely, and the optical properties associated with different shapes must differ. The most common form of papillate cell is subconical in form with a rounded apex, slightly concave outer walls and a somewhat expanded convex base Figs 1, 2, 11). A distinctive kind of epidermis is found in the Geraniaceae, in which the papillae are usually low and wide with strongly concave outer walls and a strongly concave base. Another unusual kind occurs in some Saxifraga species, in which the tip of the papilla forms a convex-ended projection, which is cylindrical or even distinctly waisted below. The range of variation in Saxfraga is unusual in including both species with smooth outer papilla walls, as is S. cebennensis, S. hypnoides and S. rosacea all of which have projecting papilla tips and species with striated papilla walls, as in S. spathularis. Within a family it is more usual for the outer cell-walls of the petal epidermis to be uniformly smooth (as for example in the Caryophyllaceae and Geraniaceae) or uniformly striated (as in the Asteraceae, Brassicaceae and Rosaceae). However, the Lamiaceae, Primulaceae, Scrophulariaceae and some other families include both smooth and striated forms, and further work may show that both forms of cell-wall occur in families in which only one form has been reported. The prominence of the striations ranges from weak so that they, are only just detectable, as in Senecio cruentus, to very strong and conspicuous, as in several Apiaceae and Viola species Figs 18-21).

Figures 16, 17. Lamina of ray-floret of Leucanthemum vulgare with part of the ultraviolet-absorbing upper epidermis stripped to reveal the ultraviolet-reflecting mesophyll, x 10. Fig. 16. In visible light. Fig. 17. In ultraviolet light of 330-400 nm.

A comparison of the optical geometry of a papillate epidermis with that of a flat epidermis (Fig. 22) shows that a flat epidermis will reflect light that strikes it at a shallow angle, whereas a papillate epidermis will absorb the light over the greater part of its surface. Experiments carried out with cell models have confirmed this. Petals reflect light more or less strongly, but this reflection takes place mainly from the surfaces of mesophyll cells at cell wall/air interfaces (except in Ranunculus species and a few similar cases). These processes were described by Exner & Exner (1910), but they did not consider the optical effects produced by the bases of the papilla cells, nor did they consider the optical geometry of the mesophyll. and they were apparently unaware of the existence of reversed-papillate and multiple-papillate petal epidermis cells.

Figures 18-21. SEM surface views of the upper epidermis of petals, showing typical patterns of striations on lenticular cells (Figs 18 & 19) and papillate cells (Figs 20 & 21). Fig. 18. Prunus spinosa, x400. Fig. 19. Scilla bifolia, x510. Fig. 20. Veronica chamaedrys x400. Fig. 21. Dactylorhiza fuchsii, x 600.

If one considers the influence of reflection and refraction in conical-papillate epidermal cells on reflected light emerging from the mesophyll (Fig. 23), it is clear that the convex base (if present) will refract this light and that the converging outer cell walls will reflect it. In this way the light will pass through a more uniform length of pigment-containing solution than would be the case in flat epidermal cells. Furthermore, the loss of emergent light by surface reflection from the base of the epidermis directing it back into the mesophyll will be reduced by the action of the convex base (Fig. 23 ). The same mechanisms will scatter the light that emerges from papillate petals, so that the effects of unidirectional light sources (e.g. sunlight) are reduced and the brightness of the petal remains relatively constant irrespective of the angle from which it is viewed. Goniophotometric measurements of the light reflected from different kinds of petal and experiments with detached epidermises and with cell models have confirmed this. These measurements and observations have also shown that internal reflection to the exterior from the convex base of the papilla is only a minor effect in typical conical-papillate epidermises, but is much more important in reversed-papillate and reversed multiple-papillate epidermises, in which the internal papillae directed towards the mesophyll correspond to the convex base of a normal papilla (Fig. 24). Reversed-papillate epidermises appear to differ from normal papillate epidermises in a number of optical features and produce strikingly different optical effects which can be seen most clearly when (as is the case in many Caryophyllaceae) the two kinds occur in different areas of the same petal and contribute to petal patterning.

Figure 22. Light absorption and reflection by a papillate petal surface (left) and a flat petal surface (right).

In a simple reversed-papillate epidermal cell, the outer face is relatively weakly convex but the inner face is semi-ovoid or rounded-conical, so that the cell resembles a normal papillate cell, but with the papilla facing inwards towards the mesophyll instead of outwards. Schubert (1925) found this kind of anatomy in several Verbascum species; we have also found it in Campanula and Roscoea. The optical properties of a simple reversed-papillate petal epidermis are similar to those of the much more frequent multiple reversed-papillate kind; fairly strong surface reflections from the relatively flat outer face (e.g. in Campanula, Crocus, Papaver, Tulipa and Verbascum; Figs l3-l5) are common in both kinds and may be adaptive. The papillate inner face of the epidermal cell will act as a light-trap both for light reflected from the mesophyll and for light transmitted from below, and it will also reflect some externally incident light by total internal reflection, in all cases guiding the light through the pigment contained within the cell.

In the lenticular kind of petal anatomy, the outer faces of the epidermal cells, and often also the inner faces, are convex. In some respects this kind of petal anatomy is intermediate between the papillate and reversed-papillate kinds, and sometimes, as in Campanula, lenticular epidermises lacking striations intergrade with the reversed-papillate kind. However, lenticular epidermises are more commonly associated with the normal papillate kind, either forming the lower epidermis on petals with a papillate upper epidermis, or forming the inner part of a mainly papillate upper epidermis. Lenticular cells also fairly commonly form the whole of the upper or both petal surfaces (Tables 2, 3).

Figure 23. Paths of light rays emerging from the reflective mesophyll of a petal with conical-papillate epidermal cells with convex bases. (A refractive index of 1.35 is assumed: the intercellular spaces in the mesophyll are air-filled.) Figure 24. Internal reflection of light falling vertically on a multiple reversed-papillate epidermal cell. The cell is assumed to have a uniform refractive index of 1.35. with the intercellular spaces completely air-filled. More than 60% of vertically incident light will be reflected by this mechanism in cells of this basal configuration under these conditions, but the amount of light that is reflected by this mechanism decreases rapidly as the angle of incidence diverges from the vertical.

Optical functions of striations

The optical geometry, of a smooth lenticular petal surface does not enable it to function as a light-trap, and smooth (unstriated) lenticular petal surfaces usually show fairly strong surface reflections, as in Campanula rotundifolia and Malva sylvestris. Lenticular cell models also show this effect. Lenticular petal surfaces with striations on the outer cell walls contrastingly show much weaker surface reflections (e.g. Cyclamen persicum, Hebe species, Philadelphus coronarius, Sambucus nigra) and we consider that an important optical function of the striations is to act as a light-trapping structure; on lenticular epidermises, where they are often strongly developed (Figs 18, 19) they compensate to some extent for the absence of light-trapping papillae, and on papillate epidermises they supplement the light-trapping action of the papillae: in both cases acting in the same manner as the papilla reflection and refraction followed by internal reflection (Fig. 22). The striations on the side-walls of papillae normally run from base to apex (Figs 20, 21), thus giving a greater efficiency of light absorption than would be the case for latitudinal striations. In some species of Galium (as in Galium aparine, G. cruciatum and G. verum; Table 2) striations are replaced by small regularly arranged domed projections of the outer cell-wall, resembling a miniature version of a papillate petal epidermis. In Galium aparine the upper petal epidermis is thus doubly papillate, with papillate cells the outer walls of which are themselves minutely papillate.

A second probable optical function of striations is to scatter emergent light, thus further increasing the constancy of petal brightness (when viewed from a distance) regardless of the direction of viewing and the angle of incident light (see above). However, they may also guide emergent light to some extent, and the possibility that insects may react to the fine patterning or other optical effects produced by the striations should be borne in mind and investigated further. Longitudinal striations, orientated towards the base of the petal, are fairly common (such striations occur in Oenothera species, Ornithogalum umbellatum, Setcreasea purpurea and Trifolium repens, for example). The striations do not appear to be capable of producing any structural colour effects (Daoud 1980).

Structure and function of petal mesophyll

The reflective mesophyll cells that are associated with multiple reversed-papillate petal epidermis cells are usually extremely regular in structure, usually being elongated four-faced cells with a row of small evenly, spaced projections (resembling small papillae) on each face, one face being directed towards the upper and another towards the lower epidermis. Projections from adjacent cells meet at their tips, and the intercellular spaces are air-filled. In the Caryophyllaceae and Cistaceae the mesophyll is composed of two to six layers of such cells in the examples we have investigated, but three or four layers are most common. The mesophyll cells in petals with other kinds of epidermis are sometimes similar in structure to this kind (e.g. in Chrysanthemum, which has a papillate epidermis, and in Lampranthus, which has a flat epidermis) but are often rather irregular, especially below the hypodermal layer. In nearly all cases, however, the petal mesophyll is a more or less open aerenchymatous tissue with copious air-filled intercellular spaces. The basic mechanism by which the mesophyll reflects light by a combination of refraction and external and internal reflection was described by Exner & Exner (1910), but very little is known of the optical properties of the elements of the mesophyll. The efficiency of the mesophyll as a reflector varies in different species, and the appearance of the mesophyll when it is exposed by peeling off the epidermis also varies; the regular mesophyll of Chrysanthemum coronarium, for example, has a shiny appearance (unidirectional reflection whereas Pelargonium zonale mesophyll shows diffuse reflection. These appearances may however be artifacts.

Table 3. Cell shape and outer cell wall structure in the upper petal epidermis of 201 species from 60 families. A few
species combine more than one kind of anatomy. The number of families in each category is shown in parentheses

Cell shape Outer cell wall structure
Striated Smooth
Papillate 74 (28) 38 (19)
Multiple-papillate 1 (1) 12 (4)
Lenticular 21 (14) 11 (8)
Multiple-lenticular -- 2 (1)
Reversed-papillate 1 (1) 3 (2)
Multiple reversed-papillate 16 (7) 13 (8)
Multiple reversed-lenticular -- 2 (2)
Flat 6 (6) 5 (4)

In petals with carotenoid pigmentation, carotenoid-containing chromoplasts commonly occur in the mesophyll, at lower concentrations than in the epidermis but still in significant quantities that colour the tissue. Dissolved pigments (anthocyanins, betalains and ultraviolet-absorbing flavonoids are, however, normally absent from the mesophyll, or occur only in insignificant quantities, in most of the cases that we have examined, but there are some surprising exceptions to this rule. Schubert (1925) was the first to observe and report petals in which the visible anthocyanin pigmentation was actually confined to the mesophyll, and absent from the epidermis (several members of the Boraginaceae). Most of the cases that we have observed of this type have blue anthocyanin complexes in the subepidermal layer of mesophyll cells; the epidermis appears to contain no pigment, but in fact contains ultraviolet-absorbing flavonoids (as for example in Echium, Myosotis and other blue-flowered members of the Boraginaceae, and in Scilla tubergeniana and Chionodoxa cretica in the Liliaceae-Scilleae). The petals of the Boraginaceae appear intensely blue and it is clear that a substantial proportion of the incident light must pass through the blue pigment contained within the outer mesophyll cells in these species. These mesophyll cells are morphologically different from the corresponding mesophyll cells in species with unpigmented mesophyll, in that pronounced projections are confined to their inner faces instead of occurring on all faces. It seems possible that in these cases the absence of projections may be correlated with decreased surface reflection and hence increased absorption of incident light by the pigment-containing outer mesophyll cells. Thus, the surface projections of mesophyll cells, unlike the papillae of the petal epidermis, may aid surface reflection.

The reflective mechanism in the petals of most yellow-flowered Ranunculus species is different and is very distinctive. Here light is reflected by starch-grains within the upper mesophyll cells, and light reflection does not require intercellular air-spaces, although these may be present and may supplement reflection. The reflective starch layer of Ranunculus petals was first described by Schimper (1885) and Möbius (1885) and was independently rediscovered by Exner & Exner (1910). Reflective starch layers of this kind are rare; we have observed a similar layer in Anemone species, but we have found air-containing reflective mesophyll lacking starch in Ranunculus peltatus and other white-flowered species in the genus, and also in Caltha palustris. Most yellow-flowered Ranunculus species have an unusual flat epidermis, but Ranunculus gramineus has a papillate epidermis above the reflective starch layer, as do Anemone heldreichii and A. nemorosa. Parkin (1928, 1931, 1935) made an extensive study of the morphology and taxonomic distribution of the starch-containing layer in the petals of Ranunculus and related genera, but rather surprisingly did not appear to understand its function as a reflector, which is independent of the unusual glossy petal surface of many Ranunculus species. Parkin found starch layers in all of the 33 yellow-flowered Ranunculus species that he investigated, but in only two out of nine white-flowered Ranunculus species. He also found a strong starch layer in the closely related genus Oxygraphis, and weak starch layers in Adonis and Callianthemum, but did not investigate genera outside the tribe Ranunculeae.

General discussion

Knowledge of petal structure and function is still far from complete. The range of structures that occur in nature and their distribution among different taxonomic and ecological groups are poorly known. The relationships between the pigmentation, structure and surface micromorphology of petals and their functions are still poorly understood. Much more work is needed, with implications for several fields of biology.

Petal structure and function is of great importance in pollination biology, and analyses of the relationships that may exist between the adaptive complexes of petals and the biology of the whole flower or inflorescence may be very productive. For example, our preliminary SEM examination of the petal surfaces of typical bee-, bird-, bat-, moth-, fly- and butterfly-pollinated flowers in the Polemoniaceae suggests that there is a broad correlation between petal surfaces, anthocyanin characteristics (Harborne & Smith, 1978) and pollinating agents (Grant & Grant, 1965).

The whole question of the relationships between papillate and reversed-papillate petals, which appear to represent different adaptive complexes, is of great interest, as is the problem of the relative rarity of multiple-papillate petal epidermises. The possibility that other adaptive complexes may exist, especially among tropical plants (for example the translucent orchid perianths described by Exner & Exner, 1910) requires investigation. Scarcely anything is known, in precise terms, of the structure and function of the petal mesophyll, and here again it is possible that different adaptive complexes exist among petal mesophylls, and that other reflective mechanisms exist in addition to the two basic kinds that are already known.

Further work is also required on the nature and consequences of pigment localization in petals. Until recently, the localization of compounds in intracellular fluid compartments, such as vacuolar sap and the cytosol, were established by histochemical and spectrophotometric analyses. But since 1977, it has become possible to analyse these compartments more directly, by isolating intact vacuoles and a fraction enriched in cytosol, or by quantitatively comparing the contents of vacuoles and entire protoplasts (Wagner, l979. The value of such a direct approach is seen in Wagner's recent study in which he reported, among other topics. the vacuolar/extravacuolar distribution of anthocyanins in petal protoplasts and showed that vacuoles isolated from Tulipa and Hippeastrum petals contained all or essentially all of their anthocyanin pigment. Despite several difficulties discussed by Wagner (1979) this method of analysis may be extremely valuable in investigating pigment distribution in delicate petal tissues quantitatively, at the intracellular level.

A better understanding of the relationships between petal structure, pigmentation, function and appearance is of potential importance both in evolutionary studies and for plant breeders concerned with the production of new cultivars of ornamental plants. Desirable attributes of petal structure could be identified and then searched for among existing cultivars and wild populations of related taxa. Genotypically determined attributes of petal structure could then be combined by the usual genetical techniques on a rational basis, in the same ways in which petal pigment genotypes are combined. Related processes may occur in wild populations as a result of natural selection. Plant breeders should be able to produce new or extreme kinds of petal structure that are at a selective disadvantage in wild populations but may be desirable in ornamental plants; it might he possible to produce new petal structure forms by techniques which are analogous to the production of new pigment varieties by the merging of different biochemical pathways (Straw, 1956; Smith & Levin, 1963: Harborne, 1978).

Finally, the structure and surface micromorphology of petals provide engineers, designers, architects and artists with patterns and structures that may be new and may, in some cases have important applications in optics, solar energy capture and materials science.