Journal of Comparative Physiology A (2019) 205:363–372

https://doi.org/10.1007/s00359-018-01313-1

ORIGINAL PAPER

Vividly coloured poppy flowers due to dense pigmentation and strong

scattering in thin petals
Casper J. van der Kooi1 · Doekele G. Stavenga2

Received: 17 October 2018 / Revised: 19 December 2018 / Accepted: 24 December 2018 / Published online: 28 January 2019
© The Author(s) 2019

Abstract
The flowers of poppies (Papaveraceae) exhibit bright colours, despite their thin and floppy petals. We investigated the optical
properties of flowers of Papaver rhoeas, P. dubium, Meconopsis cambrica and Argemone polyanthemos using a combined
approach of anatomy, spectrophotometry and optical modelling. The petals of Papaver flowers are composed of only three
cell layers, an upper and lower epidermal layer, which are densely filled with pigment, and an unpigmented mesophyll layer.
Dense pigmentation together with strong scattering structures, composed of serpentine cell walls and air cavities, cause the
striking poppy colours. We discuss how various aspects of the optical signal contribute to the flower’s visibility to pollinators.

Keywords Papaver · Pollination · Anthocyanin · Bee vision · Reflectance

Introduction Lee 2007; Vignolini et al. 2012; van der Kooi et al. 2014).
Pigments absorbing in a specific wavelength range filter
Flowers have been called sensory billboards (sensu Raguso the light flux travelling inside the petal, giving the flower a
2004), because they feature numerous traits to entice pollina- certain hue. For example, blue-absorbing carotenoids create
tors. The bewildering diversity in floral colours is considered yellow colours, and blue–green-absorbing anthocyanins cre-
to have evolved with respect to the visual perception of their ate red colours. Though the chemistry, molecular synthesis
pollinators (e.g., Barth 1991; Chittka and Menzel 1992; Dyer and evolution of floral pigments have been extensively stud-
et al. 2012; Muchhala et al. 2014; Shrestha et al. 2016). The ied (e.g., Koes et al. 1994; Mol et al. 1998; Grotewold 2006;
colours of flowers are due to two basic optical principles: Rausher 2008; Hopkins and Rausher 2011; Zhao and Tao
(1) reflection and scattering of light by the floral structures, 2015; Sheehan et al. 2016), much remains unknown about
and (2) selective absorption in a specific wavelength range how light propagates in a flower and how backscattering
by floral pigments (van der Kooi et al. 2016, 2019). Scat- structures and pigments are tuned.
tering occurs because petals consist of media with different We recently studied the pigmentation and scattering prop-
refractive indices, such as cell walls, air cavities, water-filled erties of 39 species of flower, and found that flowers of the
vacuoles, and thus incident light is reflected and scattered at common poppy, Papaver rhoeas, are exceptional in various
their interfaces (Kay et al. 1981; Kevan and Backhaus 1998; ways. P. rhoeas has fairly large and exceedingly thin petals,
yet they are deeply coloured and are relatively strong scat-
terers (van der Kooi et al. 2016). These findings raise the
Electronic supplementary material The online version of this
article (https​://doi.org/10.1007/s0035​9-018-01313​-1) contains question as to the anatomical details of P. rhoeas flowers
supplementary material, which is available to authorized users. that cause the extreme optical characteristics, and whether
related species share similar properties.
* Casper J. van der Kooi Poppies are a group of genera in the subfamily Papa-
c.j.van.der.kooi@rug.nl
veroideae of the Papaveraceae, which is an early diverging
1
Groningen Institute for Evolutionary Life Sciences, eudicot family. They include species of Papaver, Meco-
University of Groningen, 9747 AG Groningen, nopsis, and Eschscholtzia, and comprise many iconic spe-
The Netherlands cies well-known for their showy flowers. Several of these
2
Computational Physics, Zernike Institute for Advanced species, such as P. rhoeas, P. somniferum (opium poppy),
Materials, University of Groningen, 9747 AG Groningen, Meconopsis grandis (Himalayan poppy) and Eschscholtzia
The Netherlands

13
Vol.:(0123456789)
364 Journal of Comparative Physiology A (2019) 205:363–372

californica (California poppy) are popular ornamental and to estimate the contribution of the absorbing pigments.
garden plants. For a few species, the characteristics of their Measurements were taken from flowers of at least three
flower colours have been studied in some detail. For exam- individuals per species.
ple, the chemistry and vacuolar pH have been studied for We also investigated the angle dependence of the reflec-
blue-flowered Meconopsis species (Yoshida et al. 2006). E. tance by illuminating the flat surface of a P. rhoeas petal
californica has been investigated because of its pigmenta- with a rotatable fibre and collecting the reflected light with
tion via the specific pigment eschscholtzxanthin (e.g., Strain another rotatable fibre, positioned at the mirror angle of
1938) and because of its ultrastructure (Wilts et al. 2018). the illumination. The latter fibre was fitted with a polar-
Here, we investigate the flower colours of P. rhoeas, P. izer, which allowed measurement of reflectance spectra as a
dubium (long-headed poppy), the closely related Meconop- function of angle of light incidence for both TE (transverse
sis cambrica (Welsh poppy), which has both yellow and electric) and TM (transverse magnetic) light.
orange colour morphs, and Argemone polyanthemos (crested
prickly poppy). Using photography, spectrophotometry, opti- Anatomy
cal modelling, and various anatomical techniques, we show
that a high pigment content together with scattering air holes The thickness of the petals was measured on pieces placed
cause the typical coloration of these thin flowers. We discuss in between two cover slips with a thickness gauge. We meas-
our results in context of the plant’s ecology and visual ecol- ured the thickness for each flower five times on a transect
ogy of their pollinators. from the proximal to distal part of a petal, for 3–5 indi-
viduals per species (Table 1). The pigment distribution of
the flowers was examined via transverse sections of flower
Materials and methods pieces. Flower pieces were embedded in 6% agarose solution
at a temperature of approximately 55 °C, i.e., close to the
Plant material and photography temperature of solidification. Transverse sections were cut
using a sharp razor blade and immediately examined with
All flower samples were obtained from road sides and the Zeiss Universal microscope. Satisfactory results could
meadows around the campus of the University of Gron- be obtained only for M. cambrica. Essentially, the same dis-
ingen, except for flowers of A. polyanthemos, which were tribution was observed for the other studied species, but the
grown from seeds purchased at De Bolster, Epe, The Neth- very thin Papaver petals precluded obtaining presentably
erlands. Flowers were photographed with a digital cam- clear pictures.
era (Nikon D70) equipped with an F Micro-Nikkor macro
objective (60 mm, f2.8, Nikon, Tokyo, Japan). Petal details Optical modelling
were photographed with an Olympus DP70 digital camera
mounted on an Olympus SZX16 stereomicroscope (Olym- We used the measured reflectance and transmittance spectra
pus, Tokyo, Japan), or with a Zeiss Universal microscope of the petals to calculate the petals’ overall absorption and
(Zeiss, Oberkochen, Germany) equipped with a DCM50 scattering parameters. A flower can be considered as a stack
camera (Mueller Optronic, Erfurt, Germany). of layers, where different layers have specific scattering and
pigmentation properties (van der Kooi et al. 2016). Light
Spectrophotometry microscopical observations as well as thickness measure-
ments on P. rhoeas and P. dubium flowers suggested that the
Reflectance and transmittance spectra of petals were meas- flowers are composed of only a few cell layers (Table 1). In
ured with an integrating sphere (for technical details and
measurement procedures, see Stavenga and van der Kooi Table 1  Thickness measurements (in µm)
2016). The sphere’s measurement area (its aperture) is
Species No. Min Max Mean SD
approximately 5 mm. In contrast to the commonly used individual
reflectance probe, an integrating sphere allows measuring plants
the absolute amount of backscattering as well as the modu-
lation of the spectrum (for further discussion on the sphere Papaver rhoeas 5 56 266 108 62
and probe, see Vukusic and Stavenga 2009). The pigment Papaver dubium 5 48 319 101 65
absorbance spectrum was measured with a microspectro- Meconopsis cambrica yellow 5 74 279 157 65
photometer (MSP, see Stavenga and van der Kooi 2016). Meconopsis cambrica orange 5 73 289 142 29
A piece of flower was immersed in water; the measurement Argemone polyanthemos 3 131 407 266 89
area was a square with side length ~ 10 µm. We subtracted Per petal, thicknesses were measured five times on a transect from
the long-wavelength absorbance, which is due to scattering, proximal to distal (see “Materials and methods”)

13
Journal of Comparative Physiology A (2019) 205:363–372 365

line with observations by Kay et al. (1981), our anatomical bordered by a thin white line (Fig. 1a). The reflectance and
investigations showed that the petals consist of three main transmittance values at very long wavelengths (> 900 nm)
layers, i.e., a pigmented upper and lower epidermis, with can be used to quantify the backscattering by the petal,
an unpigmented (mesophyll) layer in between. We hence because in that wavelength range absorption by pigments is
deployed the optical model that we developed for under- negligible (see also van der Kooi et al. 2016). For both the
standing the colours of the Chilean bellflower, Nolana para- distal and proximal areas, the transmittance is ~ 0.65 and
doxa, which combines the Kubelka–Munk theory for absorb- the reflectance ~ 0.35 (Fig. 1), meaning that the petals scat-
ing and scattering media with a layer-stack light-propagation ter approximately 35% of incident light. Even in the case of
model (Stavenga and van der Kooi 2016). Using measured the proximal (base) area, which is deeply black coloured,
transmittance spectra as well as adaxial and abaxial reflec- the reflectance and transmittance curves plateau at similar
tance spectra, we could calculate the absorbance parameter amplitude in the long wavelength range. Although the long-
K* = Kd and scattering parameter S* = Sd, where K and wavelength reflectance of ~ 0.35 may seem to be in conflict
S are the absorption and scattering coefficients and d the with the blackness of the base area, the reflectance in the
petal thickness. The modelling showed that asymmetric pet- visible wavelength range is small and the gradual increase
als consisting of one pigmented and one unpigmented layer in reflectance at wavelengths > 600 nm is too small to give
cause very different adaxial and abaxial colours. However, a colourful signal. The low transmittance and reflectance at
identical adaxial and abaxial reflectance spectra result when shorter wavelengths must be due to strongly light-absorbing
the petal is homogeneously pigmented or symmetrically pigments. The different slopes of the distal and proximal
organized into three layers, and the pigment is equally dis- spectra at wavelengths > 550 nm indicate different pigments.
tributed in the two peripheral layers. Using the calculated In the ultraviolet wavelength range, the transmittance and
absorption and scattering parameters in a calculation of reflectance of the distal area is distinctly higher than the cor-
the transmittance and reflectance spectra for a symmetri- responding value for the proximal part, which also indicates
cal, three-layer case yielded spectra virtually identical to the presence of a different pigment.
the experimentally measured spectra. Finally, the transmit- The long-headed poppy (P. dubium) displays similarly
tance of a homogeneously pigmented layer with absorption beautiful red flowers, although this species rarely has black
coefficient K and thickness d is calculated as: basal areas (Fig. 1d). The spectral slopes of the distal and
proximal spectra in the long wavelength range are somewhat
T = exp(−K ∗ ), (1) similar, but the spectral values in the ultraviolet measured at
so that the absorbance is the proximal area are clearly lower than those for the distal
D = −log10 (T) = gK ∗ = K � , (2) area, indicating different amounts of ultraviolet-absorbing
pigment (Fig. 1e,f). The Welsh poppy (M. cambrica) has
with g = log10(e) = 0.4343.
an orange (Fig. 2a) and a yellow morph (Fig. 2b), and their
transmittance and reflectance spectra are similar for all petal
Vision modelling areas (Fig. 2c, d). The valleys in the spectra indicate the
presence of a pigment absorbing mainly in the blue wave-
We investigated the visibility of the flowers with a pollina- length range. The crested prickly poppy (A. polyanthemos)
tor-subjective view for known poppy pollinators, i.e., honey has white flowers that are somewhat larger than for other
bees. We analysed the measured reflectance spectra under species. The valley in the spectra of A. polyanthemos indi-
D65 ambient light against a green leaf background as before cate a pigment absorbing exclusively in the ultraviolet wave-
(van der Kooi et al. 2016), with two well-established vision length range. For A. polyanthemos, there were no differences
models, i.e., the color hexagon model (Chittka 1992) and the in reflectance between different flower areas.
receptor noise-limited model (Vorobyev and Osorio 1998). Ultraviolet photography showed that there are no great
Both models yield values that correlate with the flower con- differences in coloration of the petals, but the flowers as
trast as perceived by bees. Green contrast was calculated as a whole feature a contrasting ultraviolet pattern. Flower
per Spaethe et al. (2001). patterns with ultraviolet-absorbing centres and ultraviolet-
reflecting peripheries are known as “bulls-eye patterns”, due
to the absence of ultraviolet reflection of the anthers and
Results stigma (Fig. S1).

Poppies vary in coloration and pigmentation Pigmentation and scattering is localised

The common poppy (P. rhoeas) features typically red The transmittance and reflectance spectra of flowers depend
flowers, with at the petal base often a distinctly black area on both the absorption of incident light by pigments and the

13
366 Journal of Comparative Physiology A (2019) 205:363–372

Fig. 1  Habitus pictures and

spectra of two exemplary
poppies. a–c P. rhoeas. d–f P.
dubium. Transmittance (T) and
adaxial (Rad) and abaxial (Rab)
reflectance spectra of distal
(b–e) and proximal (c–f) petal
areas. Scale bars (a, b) 1 cm

scattering by the floral structures. We observed the petals epi-illumination (Fig. 3d). The elongated epidermal cells
at high magnification with our Zeiss microscope (“Mate- seem to be compartmentalised (Fig. 3c, d), but the compart-
rials and methods”), applying epi-illumination as well as mentalisation is an optical illusion caused by the wavy cell
transmitted light (Fig. 3). With epi-illumination, the faint walls. The petal’s wavy cell walls are more clearly observed
linear surface reflections observed at the proximal part of a in oil immersion: epi-illumination reveals serpentine-wavy
P. rhoeas petal reveal very elongated, convex cells (Fig. 3c). cell boundaries that appear bright, whereas in transmitted
In line with reflectance spectra obtained for this black area light mostly the cell interior is brightest (Fig. 3e, f).
(Fig. 1a, c), the images showed that backscattering in the The flowers of M. cambrica feature a similar complemen-
visible wavelength range by the components inside the tary behaviour of reflection and transmission (Fig. 4a, b).
proximal cells is low (Fig. 3a). In transmitted light, the cells For yellow M. cambrica flowers, transverse sections further-
nevertheless appear purplish (Fig. 3b), corresponding to more revealed that the short-wavelength absorbing, yellow
the non-negligible transmittance in the violet and red wave- transmittant pigment is located in only the upper and lower
length range (Fig. 1c). epidermal cells, whereas the mesophyll layer in between is
The appearance of the distal petal areas is different. Epi- unpigmented (Fig. 4c). A similar epidermal deposition of
illumination also creates an assembly of faint, linear sur- pigment has been described for P. rhoeas (Kay et al. 1981)
face reflections, but the reflections from inside the petal are and M. grandis (Yoshida et al. 2006), and is typical for
bright-red (Fig. 3c). In transmitted light, the cells are also anthocyanins (Kay et al. 1981; Lee 2007).
red, but at locations complementary to those appearing with

13
Journal of Comparative Physiology A (2019) 205:363–372 367

Fig. 2  Flowers and spectral characteristics of the studied species. a, c reflectance spectra b yellow M. cambrica; d orange M. cambrica; f A.
Yellow and orange type Meconopsis cambrica. e Argemone polyan- polyanthemos. Scale bar a, c, 1 cm; e 2 cm
themos. b, d, f Transmittance (T) and adaxial (Rad) and abaxial (Rab)

The bright cell boundaries observed in epi-illumination Pigments absorb exclusively in the ultraviolet
versus the bright cell interior under transmitted light sug- or well into the red wavelength range
gest the presence of scattering structures around and/or in
between the (epidermal) cells. We investigated the pres- To obtain a closer view of the absorption properties of the
ence of air cavities further by cutting a petal of a common various poppy pigments, we calculated the absorbance of
poppy (P. rhoeas), about perpendicularly to the axes of the flowers shown in Fig. 1 (see “Materials and methods”).
the epidermal cells, and exposing the cut surface to water. The absorbance spectra indicate that the distal areas of P.
Initially, the images were turbid (not shown), but starting rhoeas and P. dubium petals contain the same pigment, with
from the cut surface the image clarified, and after a few a lower density in P. dubium (Fig. 6a). The pigment in the
minutes a clear image of the flower area resulted (Fig. 5). proximal area of P. rhoeas is different, however, as it has a
Figure S2 shows the same flower area at three different much higher absorbance tail in the red as well as an ultravio-
planes of focus: the upper epidermis, the interior, meso- let peak. The absorbance spectra of the orange and yellow
phyll layer, and the lower epidermis layer (Fig. S2a–c, M. cambrica are proportional, meaning that the two morphs
respectively). The turbid area (Figs. 5, S2a–c, top) repre- feature the same pigment, but the amount is much higher in
sents an area with strongly scattering air cavities, and thus the orange morph. Finally, the absorbance spectrum of A.
appears dark under transmission. The elongated, homoge- polyanthemos is almost fully restricted to the ultraviolet,
neous red coloured cells (Figs. 5, S2a–c, middle area) have which is typical for flowers that appear white to the human
become clearly visible, because the air cavities were filled eye (Chittka et al. 1994; Kevan et al. 1996; van der Kooi
with water by capillary action. The transected epidermal et al. 2016).
cells became transparent, due to leakage of the red colour- The absorbance spectra of Fig. 6 represent only an
ing pigment (Figs. 5 double-headed arrow, S2a–c bottom). approximation of the absorption properties of the various

13
368 Journal of Comparative Physiology A (2019) 205:363–372

Fig. 3  P. rhoeas petal areas observed close up (all top view). a, b nation. Bottom row, b, d, f images observed with transmitted light.
Proximal area. c–f Distal areas. a–d Petal pieces in air. e, f Petal piece Scale bars a–d 50 µm; e, f 10 µm
immersed in oil. Top row, a, c, e images observed with epi-illumi-

poppy pigments, because the different species of flower normal, there are clear differences in reflectance amplitude
will have different scattering properties, especially given between horizontally and linearly polarised light (Fig. 8a,
their markedly different thicknesses (Table 1). We previ- b). In the blue and green wavelength ranges, the polarisation
ously developed an optical model that considers a flower as response of the petal was very similar to that of a water sur-
a stack of absorbing and scattering media, which we success- face, with a Brewster angle at ~ 55° (Fig. 8c). For ultraviolet
fully applied to study the optics of the Chilean bellflower, wavelengths, this was less distinct, but nevertheless oblique
buttercups and other flowers (Stavenga and van der Kooi incident light causes significant polarisation.
2016; van der Kooi et al. 2016, 2017). We applied the same
model to P. rhoeas flowers and thus obtained the spectra of
the absorption and scattering parameters for the distal and Discussion
proximal petal parts (Fig. 7). The scattering parameters are
clearly wavelength dependent, which is related to the absorp- Our study on the optical properties of poppy flowers revealed
tion by the pigment. When we compare the experimental that they feature an interesting suite of optical and anatomi-
absorbance spectra with the modelled absorbance spectra, it cal traits. Papaver flowers are exceedingly thin (Table 1),
appears that scattering has only a minor effect (for calcula- but contain a high amount of pigment and are fairly strong
tion procedures, see “Materials and methods”). reflectors. Indeed, the (long-wavelength) reflectance value of
35% is high when considering their very minimal thickness
Obliquely incident light causes polarised reflections (van der Kooi et al. 2016).
How can poppy flowers, with their average thickness
The reflectance spectra shown in Figs. 1 and 2 were meas- of only three cell layers, be such efficient reflectors? The
ured for normal illumination, when surface reflection is gen- answer comes with the interior structure and shape of their
erally very small (van der Kooi et al. 2014, 2015). Under cell walls. Microscopical investigation of individual cells
large angles, however, surface reflection will increase, and showed that the serpentine-curved epidermal cell walls in
when reflected by a smooth surface it becomes polarised combination with surrounding structures strongly scatter
(Wehner 2001). Indeed, under large angles relative to the the light flux propagating in the petal interior, thus causing

13
Journal of Comparative Physiology A (2019) 205:363–372 369

Fig. 6  Absorbance of poppy petals. a Absorbance spectra of distal

and proximal areas of P. rhoeas and P. dubium petals. b Absorbance
spectra of petals of orange and yellow type M. cambrica and of A.
Fig. 4  Structure of yellow Meconopsis cambrica flowers. a Close-up polyanthemos petals
of epi-illuminated petal (top view). b The same area as in a trans-
illuminated (top view). c Cross-section of the petal. Scale bars a, b
20 µm; c 50 µm

Fig. 7  Absorbance and scattering in a P. rhoeas flower. The absorb-

ance D of the distal (dist) and proximal (prox) part of the petal is
Fig. 5  A cut P. rhoeas petal immersed in water with focus at the identical to that of Fig. 6a. The parameter K′ = 0.4343K*, where K*
upper epidermis level in transmission. The cut area, where the pig- is the absorption parameter that follows together with the scattering
ment is gone, is indicated by the double-headed arrow; in the top part parameter S* from the Kubelka–Munk layer-stack model analysis
the air holes are still intact, causing the blackness. Scale bar 50 µm

13
370 Journal of Comparative Physiology A (2019) 205:363–372

occur just below the epidermal cell layers in Papaver flow-

ers. The mean thickness of flowers of M. cambrica and A.
polyanthemos is significantly higher than the mean thick-
ness of P. rhoeas and P. dubium (Table 1), but the long-
wavelength reflectance values are similar (compare Rad in
Figs. 1, 2). An interesting open question remains as to how
widespread the highly reflective typical cell structure is. The
curvy cell walls occur in Papaver and Meconopsis, which
are closely related (Kadereit et al. 2011; Xie et al. 2014), but
we also observed similar cell shapes in the phylogenetically
more diverged A. polyanthemos and Chelidonium majus (not
shown). Hence, the typical cell shape—perhaps in thin petals
in conjunction with the strongly scattering air cavities—may
be widespread throughout Papaveraceae. Given the curvy
cell walls also occur in leaves of many angiosperms (Vőfély
et al. 2018), they are likely to occur widely in flowers too.
The studied poppy species feature a wide array of colours,
due to various floral pigments, with absorption maxima for
different species peaking over the whole visible wavelength
range (Fig. 6). Furthermore, the proximal and distal parts of
P. rhoeas and P. dubium flowers contain different pigments.
The red, orange and yellow flower colours are likely due
to anthocyanin pigments (Grotewold 2006; Ng and Smith
2016), though chemical investigations are required to deter-
mine the exact type of pigment. Our observations that the
pigments are water-soluble and occur only in the epidermal
cell layers (Figs. 4, 5) indeed suggest the presence of antho-
cyanin pigments. Flower colour polymorphisms, as seen in
M. cambrica, also most commonly occur with anthocyanin-
based colours (Narbona et al. 2018).
For flowers of P. rhoeas, there appears to be an ultraviolet
colour polymorphism linked to the visual system of their
Fig. 8  Angle-dependent reflectance of a P. rhoeas flower. a Reflec- pollinators. In Europe, flowers of P. rhoeas reflect ultravio-
tance spectra for TE-polarised light incident at angles 0, 10, …70°. let light (Daumer 1958; Lunau 1993; van der Kooi et al.
b Reflectance spectra for TE-polarised light. c Reflectance values at
2016; this study) and are pollinated by bees (McNaughton
wavelengths 340, 440 and 540 nm—which correspond to the peak
sensitivities of the honey bee’s photoreceptors (Peitsch et al. 1992)— and Harper 1960; Proctor et al. 1997), but in Israel, the flow-
as a function of angle of light incidence ers lack ultraviolet reflection and are pollinated by beetles
(Dafni et al. 1990). Given that bees exhibit little sensitivity
in the red part of the spectrum and high sensitivity in the
the bright appearance in epi-illumination and being dark in ultraviolet (e.g., Peitsch et al. 1992), and some important
transmitted light (Figs. 3, 4). In addition, the interior air flower-visiting Mediterranean beetle species are sensitive
cavities are very prominent scatterers, because of the great in the red part of the spectrum (e.g., Martínez-Harms et al.
difference in refractive index between the plant material and 2012), the lack of ultraviolet reflection in the eastern Medi-
air. Immersion of a cut petal shows that the air cavities are terranean seems to be a convergence for beetle pollination
gradually filled with water via capillary suction, thereby (Dafni et al. 1990). The chemical characteristics of the flo-
reducing the reflection (Fig. 5). Similar within-flower air ral pigments in different geographic regions will shed light
cavities have previously been demonstrated to exist in flow- on how ultraviolet-absorbing pigments were lost when P.
ers of Caltha palustris (Whatley 1984) and California poppy rhoeas expanded its geographic range to Europe.
(Wilts et al. 2018), and have been shown to enhance the In contrast to the natural variability in ultraviolet reflec-
brilliance of buttercup flowers (Vignolini et al. 2012; van tion, there is no significant variation in red coloration of the
der Kooi et al. 2017). two Papaver species, suggesting that the red coloration is
Intriguingly, scattering does not increase with petal thick- more developmentally constrained and/or it serves a relevant
ness, suggesting that the strong scattering structures only signalling function for bees also. Indeed, although bees are

13
Journal of Comparative Physiology A (2019) 205:363–372 371

Table 2  Vision modelling results and interior air spaces, combined with black markings and
Species Green contrast Hexagon units RNL units
anthers create a conspicuous flower.

P. rhoeas 0.04 0.23 7.3 Acknowledgements The authors thank Hein Leertouwer and Marten
P. dubium 0.01 0.23 6.8 Staal for practical support, two reviewers for comments on the manu-
script and Dr. Friedrich Barth for inviting us to contribute to this spe-
M. cambrica yellow 0.21 0.18 7.3
cial issue. CJvdK was financially supported by a VENI Grant (Grant
M. cambrica orange 0.16 0.30 11.6 number 016.Veni.181.025, provided by the Dutch NWO) and DGS was
A. polyanthemos 0.33 0.21 9.7 financially supported by AFOSR/EOARD (Grant FA9550-15-1-0068).

Stimuli were compared against an average green backdrop, with D65 OpenAccess This article is distributed under the terms of the Crea-
illuminant and spectral sensitivities for honey bees (see “Materials tive Commons Attribution 4.0 International License (http://creat​iveco​
and methods”) mmons​.org/licen​ses/by/4.0/), which permits unrestricted use, distribu-
tion, and reproduction in any medium, provided you give appropriate
credit to the original author(s) and the source, provide a link to the
Creative Commons license, and indicate if changes were made.
commonly thought to be insensitive to red colours, Chittka
and Waser (1997) have shown that this is not the case. The
spectral sensitivity of bees extends until at least 650 nm, and
red poppy flowers feature noticeable reflection below that
wavelength (Fig. 1). Interpretation of the colours with a bee References
subjective view showed that the red petals are strongly con-
Barth FG (1991) Insects and flowers. The biology of a partnership.
trasting to a green background (Table 2). On the other hand, Princeton University Press, Princeton
the green contrast, which is a (long-distance) visual signal Chittka L (1992) The colour hexagon: a chromaticity diagram based on
used by some insect pollinators including bees (Giurfa et al. photoreceptor excitations as a generalized representation of colour
opponency. J Comp Physiol A 170:533–543
1996; Spaethe et al. 2001; Hempel de Ibarra et al. 2014;
Chittka L, Menzel R (1992) The evolutionary adaptation of flower
van der Kooi et al. 2019), is very low for P. rhoeas and P. colours and the insect pollinators’ colour vision. J Comp Physiol
dubium (Table 2). Nonetheless, the fairly large floral display, A 171:171–181
the high colour contrast and the within-flower contrast in P. Chittka L, Waser NM (1997) Why red flowers are not invisible to bees.
Isr J Plant Sci. https​://doi.org/10.1080/07929​978.1997.10676​682
rhoeas and P. dubium may yield a visually strong signal. Chittka L, Shmida A, Troje N, Menzel R (1994) Ultraviolet as a com-
Furthermore, on a clear day and under oblique illumina- ponent of flower reflections, and the colour perception of Hyme-
tion, the surface reflection may enhance the (long-distance) noptera. Vis Res 34:1489–1508
visibility. Bees and other flower visitors may perceive the Dafni A, Bernhardt P, Shmida A et al (1990) Red bowl-shaped flowers:
convergence for beetle pollination in the Mediterranean region.
specular reflection as a flash, though it seems unlikely that Isr J Bot 39:81–92. https:​ //doi.org/10.1080/002121​ 3X.1990.10677​
any polarisation effect per se (Fig. 8) acts as a cue; for exam- 134
ple, because the part of the bee eye that generally faces the Daumer K (1958) Blumenfarben, wie sie die Bienen sehen. Z Vergl
flower is not polarisation-sensitive (for further discussion Physiol 41:49–110
Dyer AG, Boyd-Gerny S, McLoughlin S et al (2012) Parallel evolution
on polarised signalling by plants, see Wehner and Bernard of angiosperm colour signals: common evolutionary pressures
1993; van der Kooi et al. 2019). linked to hymenopteran vision. Proc R Soc B 279:3606–3615.
The differently coloured anthers, the dark red in the proxi- https​://doi.org/10.1098/rspb.2012.0827
mal part of the petal and—if present—the black markings Ellis AG, Brockington SF, de Jager ML et al (2014) Floral trait varia-
tion and integration as a function of sexual deception in Gorteria
furthermore create a within-flower colour contrast, mak- diffusa. Philos Trans R Soc Lond B Biol Sci 369:20130563. https​
ing the centre of the flower where the pollen are located to ://doi.org/10.1098/rstb.2013.0563
visually stand out (Figs. 1, 2, S1). In addition to generating Giurfa M, Vorobyev M, Kevan P, Menzel R (1996) Detection of col-
within-flower contrast, the black markings may increase oured stimuli by honeybees: minimum visual angles and receptor
specific contrasts. J Comp Physiol A 178:699–709
attractiveness to pollinators by means of sexual deception. Grotewold E (2006) The genetics and biochemistry of floral pigments.
Experiments with artificial P. rhoeas “flowers” and pollina- Annu Rev Plant Biol 57:761–780. https​://doi.org/10.1146/annur​
tors in natural Israeli habitats showed that visitation by male ev.plant​.57.03290​5.10524​8
beetles increases with black markings, presumably because Hempel de Ibarra N, Vorobyev M, Menzel R (2014) Mechanisms,
functions and ecology of colour vision in the honeybee. J Comp
males are looking for female beetles that often dwell in Physiol A 200:411–433
these flowers (Dafni et al. 1990). Similar results have been Hopkins R, Rausher MD (2011) Identification of two genes causing
reported for the fly-pollinated South African daisy, Gorteria reinforcement in the Texas wildflower Phlox drummondii. Nature
diffusa (Johnson and Midgley 1997; Ellis et al. 2014). 469:411–414
Johnson SD, Midgley JJ (1997) Fly pollination of Gorteria diffusa
In summary, our study shows that despite their minimal (Asteraceae), and a possible mimetic function for dark spots on the
thickness and floppy appearance, poppy flowers exhibit an capitulum. Am J Bot 84:429–436. https​://doi.org/10.2307/24460​
intriguing anatomy. High pigment content, wavy cell walls 18

13
372 Journal of Comparative Physiology A (2019) 205:363–372

Kadereit JW, Preston CD, Valtueña FJ (2011) Is Welsh Poppy, Meco- absorbing layer stack model. Planta 243:171–181. https​://doi.
nopsis cambrica (L.) Vig. (Papaveraceae), truly a Meconopsis? org/10.1007/s0042​5-015-2395-0
New J Bot 1:80–88. https​://doi.org/10.1179/20423​4811X​13194​ Strain HH (1938) Eschscholtzxanthin: a new xanthophyll from the
45300​2742 petals of the California poppy, Eschscholtzia californica. J Biol
Kay QON, Daoud HS, Stirton CH (1981) Pigment distribution, light Chem 123:425–437
reflection and cell structure in petals. Bot J Linn Soc 83:57–83. van der Kooi CJ, Wilts BD, Leertouwer HL et al (2014) Iridescent
https​://doi.org/10.1111/j.1095-8339.1981.tb001​29.x flowers? Contribution of surface structures to optical signaling.
Kevan PG, Backhaus WGK (1998) Color vision: ecology and evolu- New Phytol 203:667–673. https​://doi.org/10.1111/nph.12808​
tion in making the best of the photic environment. In: Backhaus van der Kooi CJ, Dyer AG, Stavenga DG (2015) Is floral iridescence a
WGK, Kliegl R, Werner JS (eds) Color vision: perspectives from biologically relevant cue in plant-pollinator signaling? New Phytol
different disciplines. De Gruyter, Berlin 205:18–20. https​://doi.org/10.1111/nph.13066​
Kevan P, Giurfa M, Chittka L (1996) Why are there so many and so van der Kooi CJ, Elzenga JTM, Staal M, Stavenga DG (2016) How to
few white flowers? Trends Plant Sci 1:252–527 colour a flower: on the optical principles of flower coloration. Proc
Koes RE, Quattrocchio F, Mol JNM (1994) The flavonoid biosynthetic R Soc B 283:20160429. https​://doi.org/10.1098/rspb.2016.0429
pathway in plants: function and evolution. BioEssays 16:123–132 van der Kooi CJ, Elzenga JTM, Dijksterhuis J, Stavenga DG (2017)
Lee DW (2007) Nature’s Palette. The science of plant color. University Functional optics of glossy buttercup flowers. J R Soc Interface
of Chicago Press, Chicago 17:20160933. https​://doi.org/10.1098/rsif.2016.0933
Lunau K (1993) Interspecific diversity and uniformity of flower colour van der Kooi CJ, Dyer AG, Kevan PG, Lunau K (2019) Functional
patterns as cues for learned discrimination and innate detection of significance of the optical properties of flowers for visual signal-
flowers. Experientia 49:1002–1010 ling. Ann Bot 123:263–276. https​://doi.org/10.1093/aob/mcy11​9
Martínez-Harms J, Vorobyev M, Schorn J et al (2012) Evidence of red Vignolini S, Thomas MM, Kolle M et al (2012) Directional scattering
sensitive photoreceptors in Pygopleurus israelitus (Glaphyridae: from the glossy flower of Ranunculus: how the buttercup lights up
Coleoptera) and its implications for beetle pollination in the south- your chin. J R Soc Interface 9:1295–1301. https:​ //doi.org/10.1098/
east Mediterranean. J Comp Physiol A 198:451–463. https​://doi. rsif.2011.0759
org/10.1007/s0035​9-012-0722-5 Vőfély RV, Gallagher J, Pisano GD et al (2018) Of puzzles and pave-
McNaughton IH, Harper JL (1960) The comparative biology of closely ments: a quantitative exploration of leaf epidermal cell shape.
related species living in the same area. I External breeding-barri- New Phytol 221:540–552. https​://doi.org/10.1111/nph.15461​
ers between Papaver species. New Phytol 59:15–26. https​://doi. Vorobyev M, Osorio D (1998) Receptor noise as a determinant of col-
org/10.1111/j.1469-8137.1960.tb061​97.x our thresholds. Proc R Soc Lond B 265:351–358
Mol J, Grotewold E, Koes R (1998) How genes paint flowers and seeds. Vukusic P, Stavenga DG (2009) Physical methods for investigating
Trends Plant Sci 3:212–217 structural colours in biological systems. J R Soc Interface 6:S133–
Muchhala N, Johnsen S, Smith SD (2014) Competition for hum- S148. https​://doi.org/10.1098/rsif.2008.0386.focus​
mingbird pollination shapes flower color variation in Andean Wehner R (2001) Polarization vision—a uniform sensory capac-
Solanaceae. Evolution 68:2275–2286. https​://doi.org/10.1111/ ity? J Exp Biol 204:2589–2596. https​://doi.org/10.1016/0042-
evo.12441​ 6989(77)90005​-0
Narbona E, Wang H, Ortiz PL et al (2018) Flower colour polymor- Wehner R, Bernard GD (1993) Photoreceptor twist: a solution to the
phism in the Mediterranean Basin: occurrence, maintenance and false-color problem. Proc Natl Acad Sci USA 90:4132–4135
implications for speciation. Plant Biol 20:8–20 Whatley JM (1984) The ultrastructure of plastids in the petals
Ng J, Smith SD (2016) How to make a red flower: the combinatorial of Caltha palustris L.. New Phytol 97:227–231. https​://doi.
effect of pigments. AoB Plants 8:plw013. https:​ //doi.org/10.1093/ org/10.1111/j.1469-8137.1984.tb041​26.x
aobpl​a/plw01​3 Wilts BD, Rudall PJ, Moyroud E et al (2018) Ultrastructure and optics
Peitsch D, Fietz A, Hertel H et al (1992) The spectral input systems of the prism-like petal epidermal cells of Eschscholzia califor-
of hymenopteran insects and their receptor-based color-vision. J nica (California poppy). New Phytol 219:1124–1133. https​://doi.
Comp Physiol A 170:23–40 org/10.1111/nph.15229​
Proctor M, Yeo P, Lack A (1997) The natural history of pollination. Xie H, Ash JE, Linde CC et al (2014) Himalayan-Tibetan plateau
Harper Collins, New York uplift drives divergence of polyploid poppies: Meconopsis vigu-
Raguso RA (2004) Flowers as sensory billboards: progress towards ier (Papaveraceae). PLoS One 9:e99177. https​://doi.org/10.1371/
an integrated understanding of floral advertisement. Curr Opin journ​al.pone.00991​77
Plant Biol 7:434–440 Yoshida K, Kitahara S, Ito D, Kondo T (2006) Ferric ions involved
Rausher MD (2008) Evolutionary transitions in floral color. Int J Plant in the flower color development of the Himalayan blue poppy,
Sci 169:7–21 Meconopsis grandis. Phytochemistry 67:992–998. https​://doi.
Sheehan H, Moser M, Klahre U et al (2016) MYB-FL controls gain org/10.1016/j.phyto​chem.2006.03.013
and loss of floral UV absorbance, a key trait affecting pollinator Zhao D, Tao J (2015) Recent advances on the development and regula-
preference and reproductive isolation. Nat Genet 48:159 tion of flower color in ornamental plants. Front Plant Sci 6:261.
Shrestha M, Lunau K, Dorin A et al (2016) Floral colours in a world https​://doi.org/10.3389/fpls.2015.00261​
without birds and bees: the plants of Macquarie Island. Plant Biol
18:842–850. https​://doi.org/10.1111/plb.12456​ Publisher’s Note Springer Nature remains neutral with regard to
Spaethe J, Tautz J, Chittka L (2001) Visual constraints in foraging jurisdictional claims in published maps and institutional affiliations.
bumblebees: flower size and color affect search time and flight
behavior. Proc Natl Acad Sci USA 98:3898–3903
Stavenga DG, van der Kooi CJ (2016) Coloration of the Chilean
Bellflower, Nolana paradoxa, interpreted with a scattering and

13
 © 2019. This work is published under
http://creativecommons.org/licenses/by/4.0/(the “License”). Notwithstanding
the ProQuest Terms and Conditions, you may use this content in accordance
with the terms of the License.