Genetic switch in UV response of mimicry-related pale-yellow colors in Batesian mimic butterfly, Papilio polytes

See allHide authors and affiliations

Science Advances  08 Jan 2021:
Vol. 7, no. 2, eabd6475
DOI: 10.1126/sciadv.abd6475


In a Batesian mimic butterfly Papilio polytes, mimetic females resemble an unpalatable model, Pachliopta aristolochiae, but exhibit a different color pattern from nonmimetic females and males. In particular, the pale-yellow region on hind wings, which correspondingly sends important putative signals for mimicry and mate preference, is different in shape and chemical features between nonmimetic and mimetic morphs. Recently, we found that mimetic-type doublesex [dsx (H)] causes mimetic traits; however, the control of dimorphic pale-yellow colors remains unclear. Here, we revealed that dsx (H) switched the pale-yellow colors from UV-excited fluorescent type (nonmimetic) to UV-reflecting type (mimetic), by repressing the papiliochrome II synthesis genes and nanostructural changes in wing scales. Photoreceptor reactivities showed that some birds and butterflies could effectively recognize mimetic and nonmimetic pale-yellow colors, suggesting that a genetic switch in the UV response of pale-yellow colors may play essential roles in establishing the dimorphic female-limited Batesian mimicry.


Batesian mimicry is a classic example of adaptation where palatable species often resemble distasteful models to protect themselves from predators (1). Among many examples of Batesian mimicry in insects, Papilio polytes is known as a female-limited Batesian mimic butterfly (2, 3). Its wing color pattern is monomorphic in males but dimorphic (or polymorphic) in females. In the Okinawa islands of Japan, there are two forms of female wing patterns: the nonmimetic form (cyrus form), which is similar to that of the male, and the mimetic form (polytes form), which resembles an unpalatable sympatric model butterfly Pachliopta aristolochiae in appearance and flight behavior (2, 4). Although the mimetic females of P. polytes are protected from predators, their physiological life span is suggested to be shorter than that of nonmimetic females (5). In addition, it has been reported that male P. polytes prefer nonmimetic females to mimetic females (5); it is presumed that the wing pattern including pale-yellow regions lying in the center of the female hind wings is a visual cue (5, 6). However, there are other reports showing that males in P. polytes subspecies prefer mimetic wing patterns when choosing between inactive nonmimetic and mimetic females but prefer active females independent of female wing pattern when choosing between two females allowed to move (7, 8). Thus, further studies may be necessary to clarify the mate choice events in P. polytes.

Our recent study clarified that the chemical features of pigments included in the pale-yellow regions are different between mimetic and nonmimetic female P. polytes (9). The pigments in strip-shaped pale-yellow regions of hind wings in nonmimetic females and males (Fig. 1A) are composed mainly of papiliochrome II, which is widely observed in papilionid butterflies (10); they absorb ultraviolet (UV) light and emit strong fluorescence (Fig. 1B) peaked at 480 nm (9, 11, 12). Papiliochrome II is a compound consisting of kynurenine and N-β-alanyldopamine (NBAD); the aromatic amino group of kynurenine is bonded to the β carbon in the side chain of NBAD (13). However, in the mimetic females and the model P. aristolochiae, instead of papiliochrome II, the pale-yellow regions localized at the center of the hind wings have uncharacterized pigments that reflect UV light (Fig. 1C). Thus, the response to UV light is completely reversed between the pale-yellow pigments of mimetic (and model) and nonmimetic females (and males). Furthermore, the expression patterns of genes involved in kynurenine and NBAD synthesis are different between nonmimetic and mimetic females in the pale-yellow region during late pupal stages. The genes involved in both pathways are up-regulated in nonmimetic females, and those of the kynurenine pathway are severely repressed in mimetic females (9). These observations indicate that both the shape and synthesis of pigments in the pale-yellow region are switched between mimetic and nonmimetic morphs.

Fig. 1 UV response and ultrastructure of dorsal pale-yellow wing patterns in P. polytes and P. aristolochiae.

(A to D) Pale-yellow patterns photographed under (A) bright-field and (B and C) UV (375 nm) illumination, without a UV transmitting filter [(B); UV fluorescence] and with the filter [(C); UV reflectance]. Merged images of UV fluorescence and UV reflectance are shown in (D). Pale-yellow spots are numbered starting from the inside. (E) Scales from the pale-yellow pattern across morphs photographed under bright-field (top) and UV (375 nm) illumination (bottom; UV fluorescence). (F) Scanning electron micrographs (SEM) of scales [asterisks in (E)] from pale-yellow patterns. Scale bars, 1 cm (A to D), 500 μm (E), and 5 μm (F). Photo credit: Yûsuke KonDo and Tasuku Kitamura, The University of Tokyo.

UV reflectance and fluorescence are widely used as signals in animal communication (14). Some spiders use UV-reflecting webs to attract prey (15). Males of butterfly Pieris rapae recognize UV reflectance in female wings (16). In addition, UV reflectance is used for intraspecies communications and aposematic signals (17, 18). In contrast, fluorescence under UV light is an important signal in aquatic animals (19, 20). Although the functional role of fluorescence is unclear, some papilionid and Heliconius butterflies have fluorescence pigments in their wing scale structures (12, 21). In P. polytes, the pale-yellow region of nonmimetic morphs emits UV-excited fluorescence and potentially functions as mate choice signals for males. In contrast, similar to the model P. aristolochiae, the pale-yellow region of mimetic females reflects UV light and may be recognized by predator birds, which have visual pigments for UV. However, the structural and chemical differences in the pale-yellow regions between the two morphs and the molecular mechanisms that control these characteristics have not been well studied.

Female-limited Batesian mimicry of P. polytes is known to be controlled by a single autosomal and “supergene” locus H; the mimetic allele (genotype; HH or Hh) is dominant in females, while males are monomorphic regardless of the genotype (HH, Hh, or hh) (22). Recent studies revealed that the H locus constitutes a supergene that includes the doublesex (dsx) gene; an autosomal inversion (approximately 130 kb long) between mimetic (H-type) and nonmimetic (h-type) loci possibly reduces the recombination and maintains the dimorphic gene structure, such as mimetic type dsx [dsx (H)] and nonmimetic type dsx [dsx (h)] (2325). Furthermore, electroporation-mediated knockdown of dsx (H) in P. polytes mimetic female wings showed a mosaic change (or partial reversal effects) from memetic to nonmimetic coloration pattern, suggesting that dsx (H) induces mimetic wing patterns with centered pale-yellow regions and simultaneously represses nonmimetic patterns with strip-shaped pale-yellow regions (24). We recently reported that some genes, such as Wnt1, Wnt6, and abd-A, working under the control of dsx (H), are involved in the formation of pale-yellow and red spot regions (26); however, the mechanisms by which the chemical features and pigment formation in pale-yellow regions are switched between mimetic and nonmimetic female wings remain unclear.

In this study, dsx (H) was partially knocked down in mimetic female wings and each characteristic was compared between the small interfering RNA (siRNA) knockdown and siRNA-untreated wings to verify whether dsx (H) controls UV reflectance, fluorescence, and scale structures of pale-yellow pigments in the two types of females. In addition, to determine the mechanism by which papiliochrome II synthesis is controlled in the pale-yellow region, we performed functional analyses of genes in both kynurenine and NBAD synthesis pathways. Last, on the basis of the reflectance spectrum and sensitivities of photoreceptors of Papilio and birds as predators, we discuss how the four different pale-yellow regions are discriminated in mimetic and nonmimetic females and males in P. polytes and P. aristolochiae. These data help understand possible functional roles of the pale-yellow regions in Batesian mimicry and mate choice.


Optical features of the pale-yellow region between nonmimetic and mimetic females of P. polytes

To examine the optical features, such as UV reflectance and fluorescence, of pale-yellow regions among four different hind wings in mimetic, nonmimetic females, and males of P. polytes and P. aristolochiae, we first observed them and their scales under visible and UV light (Fig. 1). In the visible images for dorsal side, the shape and location of the pale-yellow region are different (Fig. 1A); the seven spots (1 to 7) that reside in seven different regions wedged between two radial veins are aligned like a band in males and nonmimetic females in P. polytes, whereas the five spots (1 to 5) that reside in five different regions are congregated in the center region in P. polytes mimetic females and P. aristolochiae. However, as it was difficult to discriminate the pale-yellow colors among the four different wings in bright field, we observed them under UV light (Fig. 1, B and C, and fig. S1A). In the mimetic females and P. aristolochiae, five centered pale-yellow spots showed strong UV reflectance without blue fluorescence (the UV reflectance image was modified with a filter and shown as red in Fig. 1C). In contrast, in the nonmimetic females and males, six to seven aligned pale-yellow spots (nos. 2 to 7 in males and nos. 3 to 7 in nonmimetic females) showed strong blue fluorescence without UV reflectance, while one to two inner spots showed weak UV reflectance. The merged image of Fig. 1 (B and C) demonstrates the counterchanged relationship between UV reflectance and fluorescence signals (Fig. 1D). It was reported that the pale-yellow region of nonmimetic females contains papiliochrome II, which is assumed to be responsible for the blue fluorescence under UV irradiation (9, 11). The above results indicate two types of optical properties of pale-yellow regions: One type is similar between nonmimetic females and males and the other is similar between mimetic females and the model butterfly.

Correlations between optical features and ultrastructure of scales in the pale-yellow regions

As the two types of pale-yellow regions seemed to be composed of different scales, we compared the scale structure of the pale-yellow spot at no. 4 among the four wings. Although we could not find the critical difference among each scale using bright field (Fig. 1E, top), we observed a strong blue fluorescence under UV light in the nonmimetic females and males but not in the mimetic females and the model butterfly [Figs. 1E (bottom) and fig. S1A]. We analyzed the ultrastructure of each scale using a scanning electron microscope (SEM) and observed longitudinal ridges and reticulated cross-ribs that usually exist in butterfly scales. As a result, we found a critical difference between fluorescent and nonfluorescent scales; between longitudinal ridges, large holes (black dots) appeared among the whitish area on nonmimetic scales, while many small holes without a whitish area were observed on the mimetic scales (Fig. 1F and fig. S1B) (27). Ridge spacing (ridge density) also differed between the nonmimetic and mimetic females, with the nonmimetic scale appearing to be narrower than that of the mimetic scale (Fig. 1F). On the basis of the observation of the vertical section of the fluorescent scales, we considered that the whitish area in the nonmimetic scales represented a “clogged” structure inside the scales and on the surface (high electron density; fig. S1C, left). On the other hand, the vertical section of the mimetic pale-yellow scales clearly showed the internal structure of the scales, especially the pillars or trabeculae that connect and presumably act as spacers between the upper and lower laminae (fig. S1C, right). These results suggest that not only the chemical features of pale-yellow pigments but also physical ultrastructures are switched between the two types of females.

Papiliochrome II is the primary pigment for emitting UV fluorescence in the nonmimetic pale-yellow scales

To examine whether papiliochrome II is responsible for the UV fluorescence in the pale-yellow region of the nonmimetic wing, we examined the UV response and ultrastructure of the pale-yellow region/scale of the nonmimetic, mimetic, and P. aristolochiae wings (the latter two wings were used as control) after extracting the pigments by dipping them in ethanol. In the mimetic and P. aristolochiae pale-yellow regions, the ethanol-treated wings showed no change in the UV response compared to the untreated wings. In ethanol-treated nonmimetic wings, however, the UV fluorescence was attenuated, and the UV reflection was enhanced in the pale-yellow region (Fig. 2A). When the pale-yellow scale surface was examined by SEM, we found that the ethanol-treated scale surface of the mimetic and P. aristolochiae wings did not change as compared to the untreated scale (Figs.1F and 2B). In the ethanol-treated nonmimetic wings, however, the substance covering the hole on the scale surface disappeared and the mimetic-like hollow holes were observed (Figs.1F and 2B). These results indicate that, as previously suggested (9), papiliochrome II is the primary substance for emitting the UV fluorescence in the pale-yellow spots of the nonmimetic wings.

Fig. 2 Changes in UV response of pale-yellow spots on hind wings after ethanol treatment.

(A) A hind wing dipped in 70% ethanol for 1 day (left) and an untreated hind wing (right). In mimetic female and P. aristolochiae, UV fluorescence and UV reflection were not changed. In the nonmimetic female, however, UV fluorescence was attenuated and UV reflection was enhanced in ethanol-treated wings. (B) Ultrastructure of ethanol-treated pale-yellow scale. The scales were collected from the fourth spot on the hind wing (see Fig. 1). No structural changes were observed in the mimetic female and P. aristolochiae. In the nonmimetic, however, the clogged structure seen in the intact ethanol-untreated scales (see and compare with Fig. 1F) disappeared almost completely, resulting in the mimetic-like ultrastructure. Scale bars, 5 μm. Photo credit: Shinichi Yoda, The University of Tokyo.

Different reflectance spectra for the pale-yellow region between nonmimetic and mimetic females of P. polytes

To verify the above notion, we measured the reflectance spectra of the pale-yellow spot at no. 4 in four wings (Fig. 3 and figs. S2 and S3). Reflectance spectra of spot no. 4 in nonmimetic females and males showed a similar curve that reached the bottom at a wavelength of approximately 380 nm, increased steeply from 400 nm, and then plateaued around 450 nm (Fig. 3, dotted green and black lines, respectively). In contrast, the spectra of spot no. 4 both in mimetic females and the model butterfly showed a curve that increased gradually from 300 to 600 nm (Fig. 3, red and yellow lines, respectively), which is different from the former two spectra. As Maya2000 was used for the measurements, we detected all wavelengths as reflectance when a light of a specific wavelength was irradiated (fig. S2); the fluorescence intensity (converted after UV irradiation) in the UV range (around 380 nm) was also included in the reflectance data in nonmimetic females and males. This indicates that the reflectance spectrum around 380 nm was detected more than the true value and that the reflectance spectrum in the blue fluorescent range (around 480 nm) was detected less than the true value. To obtain the appropriate reflectance spectra, we calculated and corrected the data in nonmimetic females and males, as described in the Supplementary Materials (fig. S3) and indicated as solid green and black lines in Fig. 3, respectively, which enhanced the difference in the spectrum between mimetic and nonmimetic females.

Fig. 3 Relative reflectance spectra of pale-yellow spots in P. polytes and P. aristolochiae.

Reflectance spectra correspond to males (black), nonmimetic females (green), mimetic females (red), si–dsx (H) mosaic pale-yellow patch of mimetic females (see the asterisk in Fig. 4B) (blue), and P. aristolochiae (yellow). As the optical spectrometer detected all wavelengths when a light with a specific wavelength was irradiated, even when the fluorescence intensity in the UV range (around 380 nm) was included in the reflectance data, we made a modification to raw spectra corresponding to male (broken black) and nonmimetic females (broken green) to take this into account and generate corrected spectra (black and green) (see text and Materials and Methods). K.D., knockdown.

Dsx (H) switches UV reflectance, fluorescence, and ultrastructure of pale-yellow scales

To verify whether dsx (H) regulates UV reflectance, fluorescence, and ultrastructure of scales in the pale-yellow region, we knocked down dsx (H) in mimetic females of P. polytes. dsx (H)–specific siRNA (si–Ppdsx-H) was injected just after pupation [Pupa day 0 (P0)]; electroporation mediated RNA interference (RNAi) was performed (24, 28), and phenotypic changes at the adult stage were observed (Fig. 4 and fig. S4). In the siRNA-treated side, we observed a pattern change from the mimetic to nonmimetic pale-yellow spots in bright-field observation [Fig. 4, A and B, and fig. S4, A and B (bright field)]. Although electroporation-mediated knockdown occurs in a mosaic fashion, we could not distinguish whether the scales in the siRNA-treated pale-yellow region were mimetic or nonmimetic type in the bright-field image. Under UV light, we could discriminate them easily, as nonmimetic scales should emit blue fluorescence while mimetic scales should reflect UV light [Fig. 4A and fig. S4A (UV fluorescence)]. Specifically, in the siRNA-treated pale-yellow spots, nonmimetic-like blue fluorescence was observed in some regions [Fig. 4C and fig. S4C (treated)] and UV reflection was weakened in those regions [Fig. 4D and fig. S4D (treated)]. It is remarkable that the merged image of Fig. 4 (C and D) and fig. S4 (C and D) on the treated side clearly showed an interchangeable relationship between UV fluorescence and reflectance [Fig. 4E and fig. S4E (treated)], indicating that the switch of optical features by dsx (H) occurred at the cellular (scale) level.

Fig. 4 UV response and ultrastructure of mosaic RNAi phenotype of dsx (H) in mimetic females.

(A) siRNA targeting dsx (H) was introduced into the left dorsal hind wing by electroporation. The detailed phenotype is represented in (B) and (C). (B to E) Pale-yellow patterns photographed under (B) bright-field and (C) UV (375 nm) illumination, without a UV transmitting filter [(C); UV fluorescence] and with the filter [(D); UV reflectance]. Merged images of UV fluorescence and UV reflectance are represented in (E). Similar to the previous results in (24), nonmimetic-like pale-yellow mosaic patches appeared (blue arrowheads). Pale-yellow spots are numbered starting from the inside. The asterisk in (B) indicates the ectopic pale-yellow patch sampled for the reflectance spectra measurement (Fig. 3). (F) Scales from pale-yellow pattern between the RNAi side (left) and the control side (right) photographed under bright-field (top) and UV (375 nm) illumination (bottom; UV fluorescence). (G) SEM of scales [asterisks in (F)]. Scale bars, 1 cm (A), 2 mm (B to E), and 25 μm (F). Photo credit: Yûsuke KonDo, Tasuku Kitamura, and Shinichi Yoda, The University of Tokyo.

We also measured the reflectance spectra for the si–Ppdsx-H treated area in mimetic females. The measured pale-yellow spot indicated by an asterisk in Fig. 4B appeared ectopically by the knockdown of dsx (H). Its spectrum showed that the curve decreased around 380 nm, which resembles that of nonmimetic females (Fig. 3, blue solid line), suggesting that the spot represents nonmimetic-type pale-yellow region emerging because of the repression of dsx (H). In contrast to the spectrum in the nonmimetic female (and male) pale-yellow region, the measured spectra showed lower reflectance in the 430- to 700-nm range. This may be due to the incomplete change by dsx (H) knockdown on the scales in this area, which caused the absorption of light in the 430 to 700 range. Furthermore, we also analyzed the ultrastructure of fluorescent scales that appeared ectopically by si–Ppdsx-H (Fig. 4F) and found that it showed clogged and “large hole” structures and higher ridge density similar to the pale-yellow scales in nonmimetic females and males (Fig. 4G and fig. S4F). The above results indicate that dsx (H) regulates not only chemical and optical features but also the physical ultrastructure of pale-yellow scales.

Dsx (H) represses the expression of genes involved in the kynurenine and NBAD synthesis pathways

As the main fluorescent pigment included in the pale-yellow region of nonmimetic females and males is papiliochrome II, it is presumed that dsx (H) suppresses the expression of some genes involved in the kynurenine and NBAD synthesis pathways (fig. S5A) in the hind wings of mimetic females. To examine the possibility of this regulation, we knocked down dsx (H) in mimetic female hind wings at P0, extracted RNA from si–Ppdsx-H treated and nontreated wings 10 days later [at P10; expression of genes in two pathways were induced (9)], and compared gene expression levels by quantitative reverse transcription polymerase chain reaction (RT-PCR). First, we confirmed that the knockdown of dsx (H) effectively decreased the expression of dsx (H) in the treated wings (Fig. 5). Next, we quantified the expression levels of the genes involved in the NBAD synthesis pathway—ebony, tyrosine hydroxylase (TH), and DOPA decarboxylase (DDC)—and the genes involved in the kynurenine synthesis, vermilion and kynurenine formamidase (kf) (fig. S5A). Compared to nontreated wings, expression of all genes involved in the NBAD pathway (ebony, TH, and DDC) and kynurenine synthesis (kf) increased critically (P < 0.05) and expression of vermilion showed a tendency to increase (P = 0.054) in the treated wing (Fig. 5). These results suggest that dsx (H) suppresses the expression of genes involved in both NBAD and kynurenine synthesis pathways directly or indirectly in mimetic female wings.

Fig. 5 Relative expression levels of the genes involved in NBAD and kynurenine pathway in the dsx (H) knockdown wing of mimetic females at P10.

We estimated the gene expression level by real-time PCR using RpL3 as an internal control. Error bars show the SD of 15 biological replicates. P values are given by paired one-sided Student’s t test or Welch’s t test. n.s., not significant.

Functional roles of NBAD pathway genes and laccase 2 in fluorescent pigment formation

To verify whether NBAD and kynurenine pathway genes are involved in the fluorescent scale formation in the pale-yellow regions of nonmimetic females or males, we knocked down several genes by electroporation-mediated RNAi in nonmimetic females (or males). When vermilion or kf in the kynurenine pathway was knocked down, no remarkable phenotypic changes were observed (fig. S6). This indicates the possibility that kynurenine is accumulated enough and does not need to be expressed to generate the fluorescent pigment papiliochrome II. When TH or DDC in the NBAD (melanin) pathway was knocked down, we observed that the pale-yellow region turned to a whitish color, which lost its blue fluorescence and instead showed a weak UV reflectance (fig. S7). Simultaneously, the black region also turned to a similar whitish color with UV reflectance. These results indicate that papiliochrome II or melanin pigment was not synthesized sufficiently because of the loss of their substrates.

Laccase 2 is known to polymerize various melanin-related substrates [3,4-dihydroxyphenylalanine (dopa), dopamine, N-acetyldopamine (NADA), and NBAD] into cross-linked pigments in the final steps (fig. S5A). When laccase 2 was knocked down in mimetic female wings, we observed that the pale-yellow, peripheral red spot, and black regions changed to a whitish color with UV reflectance (fig. S8), indicating the block of the cross-linking step in each color formation. Furthermore, in the pale-yellow region of nonmimetic females, knockdown of laccase 2 also caused a change in the UV response from blue fluorescence to reflectance (fig. S8), indicating the loss of papiliochrome II synthesis. This result was unexpected, as Laccase 2 has not been shown to be involved in NBAD and kynurenine syntheses pathways (fig. S5A). On the basis of these results, we speculate that Laccase 2 catalyzes the bonding reaction of NBAD and kynurenine to synthesize papiliochrome II.

We also found a critical phenotypic change by knockdown of ebony, which synthesizes NBAD from dopamine and β-alanine (Fig. 6 and figs. S5A and S9). In bright field, the color change of pale-yellow spots treated with si-ebony in nonmimetic females and males was not clearly observed [Fig. 6, A and B, and fig. S9, A and B (bright field)]. Under UV light, however, the bright blue fluorescence signals in the siRNA-treated side decreased in the knockdown region of pale-yellow spots [Fig. 6, A and C, and fig. S9, A and C (UV fluorescence)]. The UV reflectance was not clearly observed in the ebony knockdown region (Fig. 6D and fig. S9D), although the merged image showed weak red signals for UV reflectance in the region where the blue fluorescent signal was reduced (Fig. 6E and fig. S9E), indicating a change in the optical feature of nonmimetic pale-yellow pigment from UV fluorescence to UV reflectance. We also found that the ultrastructure of the scale, which lost its UV fluorescence by ebony knockdown, showed a mimetic female-type image with numerous small pores between the longitudinal ridges (Fig. 6, F and G, and fig. S9F). There was no clear difference in the ridge density (Fig. 6G and fig. S9F). These results indicate that ebony is involved in the synthesis of the fluorescent pigment papiliochrome II in the nonmimetic and male hind wings.

Fig. 6 UV response and ultrastructure of mosaic RNAi phenotype of ebony in nonmimetic forms.

(A) siRNA targeting ebony was introduced into the left dorsal hind wing through electroporation. The detailed phenotype is represented in (B) and (C). (B to E) Pale-yellow patterns photographed under (B) bright-field and (C) UV (375 nm) illumination, without a UV transmitting filter [(C); UV fluorescence] and with the filter [(D); UV reflectance]. Merged images of UV fluorescence and UV reflectance are represented in (E). Pale-yellow spots are numbered starting from the inside. (F) Scales from the pale-yellow pattern between the RNAi side (left) and the control side (right) photographed under bright-field (top) and UV (375 nm) illumination (bottom; UV fluorescence). (G) SEM of scales [asterisks in (F)]. Scale bars, 1 cm (A), 2 mm (B to E), and 25 μm (F). Photo credit: Yûsuke KonDo, Tasuku Kitamura, and Shinichi Yoda, The University of Tokyo.

Discrimination of pale-yellow colors in four different wings by human, bird, and butterfly photoreceptors

Here, we revealed pale-yellow scales in four different wings—nonmimetic females, males, and mimetic females in P. polytes and in P. aristolochiae—and showed two types of optical and physical features. The former two have UV fluorescent scales with clogged ultrastructure and higher ridge density, and the latter two have UV reflective scales with pored ultrastructure. This suggests that two types of pale-yellow colors are related to the interaction between conspecific sexual partners of butterfly and between prey and predators, such as birds, respectively. To address this question, we calculated the relative excitation of the different photoreceptors of human [short wavelength (S), middle wavelength (M), and long wavelength (L)], bird (blue tit, Parus caeruleus; UV, S, M, and L), and swallowtail butterfly (Asian swallowtail, Papilio xuthus; UV, S, M, and L) for four different pale-yellow pigments (Fig. 7). On the basis of the spectrum data of each pale-yellow region, we used the following equation for calculation (29)Qi=300700 I(λ)Ri(λ)where I (λ) is the spectral distribution of the reflectance of pale-yellow spot coloration, Qi is the quantum catch of receptor type i, and Ri (λ) is the sensitivity of receptor type i.

Fig. 7 Photoreceptor sensitivity in human, blue tit, and swallowtail butterfly.

Bar graphs of the relative excitation of the set of photoreceptors for pale-yellow coloration across P. polytes morphs and P. aristolochiae are shown. The reflectance spectra of Fig. 3 were convoluted with a CIE D65 standard illuminant spectrum and each of the photoreceptor spectral sensitivities. The obtained values were subsequently normalized to the highest excitation per morphs. Each bar represents the calculation for an individual photoreceptor. The broken lines and light blue bars indicate the relative excitations by considering the effect of UV absorbance and UV fluorescence, respectively. Human (Homo sapiens: top), blue tit (Cyanistes caeruleus: middle), and Asian swallowtail butterfly (P. xuthus: bottom) closely related to P. polytes.

Three photoreceptors in the human eye showed similar excitation patterns for all four pale-yellow colors in butterfly wings, suggesting the difficulty in discriminating these colors (Fig. 7, top). In birds (P. caeruleus) and butterfly (P. xuthus), although both have photoreceptors for UV (27, 29), the excitation value of UV was quite different between the two types of pale-yellow colors; the value was much lower for nonmimetic females and males than for mimetic females and the model butterfly (Fig. 7, middle and bottom). In addition, the patterns of the excitation values of S, M, and L were also divided into two different types, similar to the above results for the UV photoreceptor (Fig. 7, middle and bottom). In addition, when the value representing the UV absorption in the UV photoreceptor (dotted box) was subtracted and the value representing the blue fluorescence in the S photoreceptor (right blue box) was added, in nonmimetic females and males, the pattern of difference between the two types was enhanced (Fig. 7, middle and bottom). These data imply the possibility that Papilio butterflies and birds such as blue tit can discriminate the pale-yellow spots of mimetic females from that of nonmimetic females and that birds cannot distinguish the mimetic females from the model butterfly.


This study showed that the UV response of the pale-yellow region in the hind wings of P. polytes is completely different between nonmimetic females (and males) (termed as nonmimetic type) and mimetic females (termed as mimetic type) (or P. aristolochiae), although previous studies have also suggested this possibility (6, 30). The wavelength spectrum for the pale-yellow region of the nonmimetic type shown is consistent with previous spectrum data for papiliochrome II (31), indicating that chemical features such as the emission of blue fluorescent light by UV irradiation are caused mainly by papiliochrome II. In addition, electron micrographs of the scale structures in the pale-yellow region of the nonmimetic type showed clogged substances; papiliochrome II was filled in the frame structure composed of ridge and cross-rib, as shown in the hypothetical model in fig. S5B. In contrast, the pale-yellow region of the mimetic type showed UV reflectance and a scale structure with many holes by electron microscopy. Although we have attempted to extract and purify the pigments from the pale-yellow region of the mimetic type, it has not been successfully done so. For this reason, we considered melanin-related pigments to be highly cross-linked, as it could not be extracted into a solution. However, on the basis of the electron microscopy images, the pale-yellow scales of the mimetic type appeared to have few substances, suggesting that UV may be reflected by the frame structure or basal lamina on scales (fig. S5B). It has also been suggested that UV is reflected by the basal lamina in the yellow and red regions of Heliconius wings (32). Since it is known that lower ridge spacing (i.e., higher ridge density) in some Heliconius and Pieris butterflies tends to increase the reflectivity of UV-containing light (33, 34), the difference in ridge spacing between the nonmimetic and mimetic females may also play a role, to some extent, in enhancing the differences in visibility to UV irradiation. Previously, we found that expression of the genes involved in the kynurenine and melanin pathway during the late pupal stage (9 to 10.5 days) when the pale-yellow region was formed was induced in nonmimetic hh females but repressed in Hh mimetic females (Fig. 8A) (9). This observation supports the possibility that the pale-yellow region of the mimetic type is not filled with sufficient pigment substances (fig. S5B).

Fig. 8 Hypothetical model for genetic regulation of UV response and scale ultrastructure in the pale-yellow region across P. polytes morphs and P. aristolochiae.

(A) Gene expression patterns of dsx (H) and genes involved in the kynurenine/NBAD pathway across P. polytes morphs during the pupal stage. Dsx (H) is up-regulated during early pupal stage in mimetic females (Hh, pink line) but not in males with the same genotype (Hh, blue line). The expression pattern of nonmimetic females (hh) is not shown because of the lack of dsx (H). Expression patterns of the genes involved in the kynurenine/NBAD pathway (asterisk) during late pupal stage in nonmimetic (green) and mimetic females (pink) are from (9). (B) Genes involved in kynurenine/NBAD pathway are repressed by dsx (H) in the mimetic females (Hh), similar to the scales observed in P. aristolochiae, with numerous pores and higher UV reflectance than in the nonmimetic. In contrast to the mimetic females, genes involved in the kynurenine/NBAD pathway are probably up-regulated because of the lack of dsx (H) in the nonmimetic females or repression of the dsx expression in the males, resulting in the induction of kynurenine/NBAD pathway gene expression, leading to the formation of papiliochrome II associated with UV fluorescence under UV light. The scales of the nonmimetic have clogged ultrastructure with papiliochrome II.

The laccase 2 knockdown experiments also support the above model. Laccase 2 is known to cross-link substrates to synthesize polymerized pigments with different colors (fig. S5A) (35). It is assumed that the black region in Papilio wings is made of dopa melanin (or dopamine melanin) cross-linked with dopa (or dopamine) and absorbs UV light (fig. S5) (36). Thus, knockdown of the laccase 2 gene in the black region of the nonmimetic or mimetic wings changed the color from black to white and reflected UV (fig. S8). In addition, the knockdown of laccase 2 in the red spots of the mimetic wings also changed the color to white, which reflected UV (fig. S8). In both cases, the cross-linked color substances were considered not to accumulate on the scale because of the lack of laccase 2. Knockdown of TH and DDC also showed similar results as laccase 2 (fig. S7), which may be caused by the lack of melanin substrates. In addition, knockdown of ebony in the pale-yellow region of nonmimetic changed the UV responsiveness from fluorescence to reflectance, which may be caused by the reduced amount of NBAD necessary for papiliochrome II synthesis. These phenotypes and corresponding UV responses resemble those of the pale-yellow region of the mimetic type and thus support the abovementioned model in which UV is reflected against the frame structure (or basal lamina) of scales that are not filled with pigment substances (fig. S5B). In some male pierid butterflies, a UV-absorbing pigment called pterin is concentrated in pigment granules or beads that adorn the cross-ribs (37, 38). For instance, the male P. rapae crucivora has wing scales with a high density of beads with a low reflectance of the wings in the UV and a high reflectance in the visible wavelength range. The wing scales of the female P. r. crucivora have virtually no beads, however, and consequently, the wing reflectance in the UV is much higher than that of the male, but the reflectance at the visible wavelength is much lower (3941). The difference in UV response between the nonmimetic and mimetic appears to be similar to the case of the Pieris butterflies. Judged by images of electron micrographs, the scale structures of the pale-yellow region of P. aristolochiae wings are similar to those of the pale-yellow region of the mimetic type, and the inability to extract the constituent pigments (30) suggests that the substance may not be contained in the pale-yellow region of the model butterfly. In contrast, it is of interest that the knockdown of laccase 2 in the pale-yellow region of the nonmimetic type where papiliochrome II is accumulated showed a change in UV response from fluorescence to reflection (fig. S5B). This result indicates that the formation of papiliochrome II, which absorbs UV, was inhibited and supported by the experiment in which the pale-yellow pigment (papiliochrome II) of the nonmimetic was removed by ethanol (Fig. 2). Papiliochrome II synthesis is a type of phenol oxidation reaction (42); the enzyme mediated in this synthesis is unknown. This study, however, suggests that Laccase 2 mediates this reaction (fig. S5A). The lack of change in the pale-yellow region of the mimetic type by knockdown of laccase 2 is consistent with the above hypothesis that there is no pigment substance in the region (fig. S5B).

The most important finding in this study is that knockdown of mimetic dsx (H) in the pale-yellow region of the mimetic type caused a difference in not only the shape and pattern of coloration but also the optical properties such as UV response and electron microscopic level structure. The synthesis and accumulation modes of the pigments, including the observed physical structure, were switched from the mimetic type to the nonmimetic type (Fig. 4). This implies that the suppression of dsx (H) in the pale-yellow region of the mimetic type activates the synthesis of papiliochrome II, reproducing the chemical and physical structures of nonmimetic pale-yellow scales. As shown previously, dsx (H) is up-regulated during the early pupal stages in mimetic (Hh or HH) female wings, whereas it is hardly expressed in Hh male wings (24). It is clear that the dsx (H) gene does not exist in nonmimetic females (hh). Therefore, when dsx (H) is knocked down in mimetic female wings, dsx (H) expression decreases to the lowest level in nonmimetic female or male wings, which leads to the induction of papiliochrome II synthesis (Fig. 8B). As shown in Fig. 5, the suppression of dsx (H) resulted in an increased expression of five genes involved in the kynurenine and NBAD pathways, which are essential for papiliochrome II synthesis. This indicates that these genes are usually repressed by dsx (H) in mimetic female wings, which are comparable to the following reports. A recent study on Drosophila melanogaster reported that in females, a female-specific dsx isoform and Abdominal-B (Abd-B) activates bric-a brac (bab) expression, whereas in males, a male-specific dsx isoform represses bab, which allows for the expression of pigment synthesis genes, yellow and tan (43). In butterflies, it is known that melanin synthesis genes such as DDC and yellow regulate both the pigment and morphology of wing scales, while ebony affects coloration (44). As we have recently identified many transcription and regulatory factors as downstream target genes of dsx (H) (24), it is assumed that the pigment synthesis in the pale-yellow region is regulated through these factors. Wnt1 and Wnt6, which were induced by dsx (H), affected the coloration patterns; the knockdown of Wnt1 in the pale-yellow region of the mimetic type caused changes to the pale-yellow region of the nonmimetic type, and the knockdown of Wnt6 reduced the size of red spots (26). Furthermore, we recently found that the knockdown of the homeobox gene aristaless, another downstream gene of dsx (H), suppressed the expression of Wnt1 and Wnt6 and affected the formation of mimetic pale-yellow regions and red spots. These observations indicate that although further studies are necessary to clarify the whole gene network involved in the wing coloration and pattern formation of P. polytes, certain downstream genes of dsx (H) regulate the genes involved in the kynurenine and melanin synthesis pathway (Fig. 8B).

As diurnal birds can distinguish the intensity of UV reflection, UV-reflecting moths are often preyed by these birds (18). The pale-yellow region of the model butterfly P. aristolochiae strongly reflects UV, and together with red spots, they could potentially function as warning colors; both colors in the model butterfly and mimetic females in P. polytes could be useful signals for predator birds. In contrast, UV absorption and reflection are known as important factors in mate selection (45). The butterfly Polyommatus icarus ingests flavonoids from edible plants during its larval stage and deposits it as a UV-absorbing substance on wing scales in late pupal stage (46). The amounts of flavonoids are assumed to be related to the quality of the forage, which suggests that females prefer males retaining high UV-absorbing flavonoids as better mating candidates. In P. polytes, it has been suggested that males prefer nonmimetic females and the UV absorbing (and fluorescent) pale-yellow region in nonmimetic females (5, 6) and males is considered an important mating signal, although mimetic females use the UV-reflecting pale-yellow region as a form of camouflage. Thus, it is possible that the pale-yellow region of wings is important for both mate preference and prey-predation interaction. However, recent studies of the P. polytes subspecies have reported that males prefer mimetic females to nonmimetic females (8). The role of UV fluorescence in terrestrial animals is not known (47); although detailed mechanisms remain unknown, studies in Heliconius have shown that while 3-hydroxykynurenine (3-OHK) fluorescence is not involved in sexual selection or nonpredatory strategies, UV reflection and the 3-OHK yellow color are involved in sexual strategies (21). There have been reports on the maintenance mechanism of female polymorphisms in P. polytes. Recently, it has been suggested that bird predators may not be able to distinguish the hind wing colors among mimetic and nonmimetic females (7). Meanwhile, field experiments have reported mimetic females to be less predatory than nonmimetic females (48). Figure 7 shows that the pale-yellow region, which is difficult to discern in the human eye, may be discriminated by predator birds and Papilio butterflies. This indicates that P. polytes males and predator birds can recognize two types of females. Figure 7 also suggests that predator birds are unable to distinguish both spectra of poisonous butterflies and mimetic females and instead recognizes both of them as warning colors. These results suggest that the pale-yellow region may be deeply involved in predation and sexual strategies. However, this hypothesis is needed to be tested in some behavioral experiments with butterflies and with avian predators. Visual modeling in Heliconius has shown that tetrachromats may not be able to discriminate for presence or absence of 3-OHK and that two duplicated UV opsins, which confer sensitivity to 355 and 398 nm, may be needed to allow this contrast (49, 50). One of the key questions is whether the repertoire of opsins is different between the nonmimetic/mimetic and male/female in P. polytes, and if so, how it contributes to visual adaptation that associated with intraspecies recognition or mate preference remains to be demonstrated.



Adult P. polytes and P. aristolochiae females were purchased from Chokan-kabira at the Ishigaki Islands, Okinawa, Japan. P. polytes larvae were reared on Citrus unshiu (Rutaceae) leaves or on an artificial diet [5.6 ml of water, 20 μl of chloramphenicol, 0.8 g for first to second instar larvae or 1.2 g for third to fifth instar larvae of Insecta F-II (Nihon Nosan Kogyo), and 0.8 g for first to second instar larvae or 0.4 g for third to fifth instar larvae of Citrus natsudaidai (leaf powder) under long-day conditions (16-hour light/8-hour dark) at 25°C. Pupal samples were staged by the length of time after pupal ecdysis. P. aristolochiae larvae were reared on Aristolochia debilis leaves under long-day conditions (16-hour light/8-hour dark) at 25°C.

Quantitative RT-PCR

In a previous study (9), we examined the temporal expression pattern of pigmentation genes on pale-yellow spots of the nonmimetic female during late pupal stage (P9, P10, P10.5, and P11) and found that vermillion, kf, TH, and DDC were up-regulated at P10 (10 days after pupation). We therefore collected wing tissues at P10 when the most distinct changes in the expression of pigmentation genes were observed. The excised sample was collected from the third to fifth pale-yellow spot region in the hind wing. Total RNA was isolated using the TRIzol reagent (Sigma-Aldrich, St. Louis, MI, USA) and purified via phenol/chloroform extraction after treatment with 0.2 U of deoxyribonuclease I (TaKaRa, Kyoto, Japan) for 15 min at 37°C. Then, RT was conducted with random primers (N6) using a Verso cDNA synthesis kit (AB1453A; Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer’s instructions. The cDNA sample was purified by phenol/chloroform extraction and ethanol precipitation. The StepOne Real-Time PCR System (ABI, Waltham, MA, USA) was used to carry out real-time quantitative PCR. Analysis was carried out using StepOne software version 2.1, using the relative standard curve method. PCR was carried out using Power SYBR Green PCR Master Mix at 95°C for 15 s and 60°C for 1 min for 40 cycles. Table S1 summarizes the primers used for quantitative PCR.

Functional analysis by RNAi in hind wing using an in vivo electroporation method

The siRNA to be injected was designed using the siRNA design support software siDirect version 2.0 ( After obtaining the sequence of the open reading frame region of the target gene from PapilioBase, candidate sequences were searched for using siDirect and on the basis of the sequence information. To reduce the off-target effect, we used the BLAST search function of PapilioBase to investigate the specificity of the designed sequence to the target gene and selected one with high specificity from the candidate sequences. The synthesis of siRNA was contracted to FASMAC Corporation. Table S1 summarizes the details of the prepared siRNAs. Universal negative control siRNA (Nippon Gene) was used as a negative control. The siRNA (250 μM) was aliquoted into the glass needle and placed in a micromanipulator M 401 with a glass needle set under a stereoscopic microscope, and 1 μl was injected along the wing vein in the hind wing of the pupa. Following which, siRNA was introduced (five square pulses of 7.5 V, 280-ms width) by electrostimulation using an electroporator. Concomitantly, 1% phosphate-buffered saline (PBS) gel and a droplet of PBS were placed between the back wing and the electrode as a buffer solution. The detailed method follows that described in a previous paper (24). The photographs of all the individuals who performed the functional analysis are described collectively in the Supplementary Materials.


To observe UV reflectance and fluorescence patterns in the hind wing of P. polytes and P. aristolochiae, butterfly specimens were pinned to blackboards and photographs were captured under UV light (OptoCode Corporation, LED375N-100STND; peak wavelength, 375 nm). Specimens were irradiated with UV light from 30-cm distance 45° below (UV intensity, 3.0 to 3.5 mW/m2). To capture the UV fluorescence, we used Nikon D3100 with NIKKOR LENS AF-S DX Micro NIKKOR 85-mm f/3.5G ED VR lens or Nikon D70 with UV-Nikkor 105-mm f/4.5 lens. To capture the UV reflectance, we used Nikon D70 with UV-Nikkor 105-mm f/4.5 lens. A UV transmittance filter (SCHOTT UG11, Edmund Optics) was attached to the front of the lens using a Nikon AF-1 gelatin filter holder (Nikon). Since the SCHOTT UG11 transmits a small amount of infrared light, a near-infrared cut filter (DR655, Kenko Tokina) was attached to the camera to ensure that only UV light is transmitted as much as possible. The camera’s exposure mode was set to manual mode to accommodate the following camera settings. UV fluorescence was taken by 1/10-s exposure time and F6.3 aperture size; UV reflectance was taken by 1/10-s exposure and F4.5 aperture size. Photographs under bright field were taken by Nikon D70 with a UV-Nikkor 105 mm f/4.5 (without filters). Photographs of UV reflectance were extracted using a blue channel to truly depict UV reflectance, as the blue channel was closest to the UV region and colored in red by Adobe Photoshop software (Adobe Systems Co. Ltd., San Jose, CA, USA) to clarify UV reflectance pattern.

Scanning electron microscopy

To observe the wing scale ultrastructure, we obtained the surface imaging of scales from P. polytes male, nonmimetic female, mimetic female, and P. aristolochiae butterflies using SEM TM-1000 (Hitachi, Japan). To minimize variation due to natural aging, we used well-preserved specimens with little to no wing wear or scale loss. Samples for observation of cross-sectional structure were prepared by laying the wing with the region of interest facing down, wetting with ethanol, and freezing with liquid nitrogen. We then cross-sectioned the wing through the region of interest with a razor blade. Cross-sectional images of the nonmimetic and mimetic females were obtained from JCM-6000 (JEOL, Japan) after sputtering the surface with osmium using Os coater (Meiwafosis Neoc, Japan).

Reflectance spectrum of pale-yellow spots in P. polytes and P. aristolochiae

To investigate differences in color in the pale-yellow region among males, nonmimetic, mimetic females of P. polytes, and P. aristolochiae as a model species, we measured reflectance spectrum of the fourth pale-yellow spot (number “4” in Fig. 1) in the hind wing of these four species/morphs, which were fixed on a platform by adhesive tape, using a V-670 spectrometer (JASCO, Tokyo, Japan) that was equipped with an integrating sphere (diameter, 6 cm) and a focusing lens made of quartz glass. The focused spot had a rectangular shape of 1 mm in width and 3.5 mm in height. Reflectance was determined in a wavelength range of 300 to 700 nm with a white standard (6916-H422A, JASCO).

Fluorescence spectrum of pale-yellow spots on nonmimetic females and males in P. polytes

To examine the differences of coloration of pale-yellow spots between four species/morphs, we measured the fluorescence spectrum of the fourth pale-yellow spot in the hind wing, which was irradiated with UV light (365 nm) from 5-cm distance and 45° using spectrometer Maya2000 (Ocean Optics Inc., Largo, FL, USA) from 5-mm distance. Figure S10A shows the fluorescence spectra of nonmimetic and males at 365 nm.

Absorbance spectrum of pale-yellow spots in nonmimetic females and males in P. polytes

To calculate the contribution of the fluorescence of the pale-yellow pigments, we examined the UV wavelengths absorbed by the pigments in wings of nonmimetic females and males of P. polytes using the following procedure. Scales of the fourth pale-yellow spot from the inside of the hind wings (Fig. 1A) were collected from five individuals using cotton swabs and extracted by immersing them in 2 ml of 70% ethanol overnight at room temperature. Post-extracted scales and other foreign substances were precipitated and removed by centrifugation at room temperature at 5000 rpm for 10 min, and the supernatant was used as a sample. The extracted pigments were dried down completely by a centrifugal evaporator and then resuspended in 70 μl of 70% methanol. The absorbance spectrum was measured using UV-visible spectrophotometer Ultrospec 2100 pro (Amersham, UK) at 5-nm intervals (1 nm in the vicinity of the peak) in the UV region above 300 nm, which was contained in sunlight. Figure S10B shows the absorbance spectrum of nonmimetic and males at 365 nm.

Estimation of true reflectance and fluorescence in nonmimetic and male P. polytes

Reflectance spectra, which were measured using a V-670 spectrometer (JASCO), included reflectance and fluorescence in the UV (300 to 400 nm) region (broken lines in Fig. 3). Therefore, we estimated true reflectance in the UV (<400 nm) and fluorescence (430 to 530 nm) regions by using reflectance, fluorescence, and absorbance spectra, as described above. To estimate “true reflectance,” we calculated the ratio between UV reflectance and fluorescence by using the fluorescence spectrum. The fluorescence spectrum at 365 nm indicates two peaks: UV reflectance and UV fluorescence (figs. S3A and S10A). When λ (nm), which is the x-axis value of the graph, is converted to frequency ν (THz) by the following Eq. 1 and has measured both the area that was surrounded by the x axis (see Sr and Sf in fig. S2A), we were able to estimate the energy ratio between reflectance and fluorescence of 365 nm, as the area is proportional to its own energy. True reflectance of 365 nm was calculated using Eq. 2 as the ratio of energy (fig. S3A). We could also estimate the “true fluorescence” spectrum in the UV region from the absorbance spectrum (figs. S3B and S10B), as the excitation spectrum was identical to the absorption spectrum because the fluorescence intensity is proportional to the absorption (51). Therefore, the true fluorescence spectrum could be corresponded to constant multiples of absorbance spectra and have these values subtracted from the reflectance spectrum to calculate true reflectance (fig. S3C)ν=cλ(1)where c represents the speed of light (299,792,458 m/s)R(365)=r(365)×SrSr+Sf(2)where R and r represent the true reflectance and reflectance included fluorescence, respectively. Generally, fluorescence and absorbance spectra are enantiotopic with each other; the energy of fluorescence and absorbance are the same. Therefore, we adjusted the area of a constant multiple of absorbance spectrum and true fluorescence spectrum (see fig. S3B). We then added adjusted values to the reflectance spectrum to estimate true fluorescence. The spectrum of true reflectance is shown in Fig. 3 (black and green solid lines).

Perception model

For modeling receptor signals, we followed a formula proposed by Kelber and colleagues (29). Briefly, where the spectral distribution of a light stimulus, I(λ), is known, the quantum catch, Qi, of a receptor type i, with spectral sensitivity, Ri(λ), is given byQi=300700 I(λ)Ri(λ)(3)where integration is over the visible spectrum. The sensitivity Ri(λ) of the pyramidal cells used in the calculations were as follows: human (52), blue tit (36), and swallowtail butterfly P. xuthus (53). Since the color vision of closely related P. xuthus is much more evident than that of P. polytes, P. xuthus was used to calculate the relative excitation of the photoreceptors (Fig. 7). Since an ideal equal-energy white light source (E) is assumed as the light source, the value of the spectral graph based on its reflectance is proportional to the amount of energy E(λ); then, the spectrum in quantal units, I(λ), is given byI(λ)=λhcE(λ)(4)where h is Planck’s constant and c the velocity of light.

Statistical analysis

Statistical analysis was performed at least in three biological replicates. All data are presented as means ± SD. All statistical tests were two-sided unless indicated otherwise. Statistical differences were analyzed using Student’s t test or Welch’s t test (quantitative PCR), Holm’s multiple comparisons (relative fluorescent intensities of wing scales), and Tukey’s test (ultrastructural small hole ratios of scales); P < 0.05 was considered statistically significant.


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We thank T. Shimada and M. Kawamoto for providing citrus leaves. Funding: This work was supported by Grants in Aid for Scientific Research (KAKENHI) of the Japan Ministry of Education, Culture, Science, and Technology (MEXT) (20017007, 22128005, 15H05778, 18H04880, 20H04918, and 20H00474 to H.F.). Author contributions: K.Sak., S.Yod., I.S., T.Ki., and Y.K. provided organisms and performed RNAi experiments. Y.K., S.Yod., and R.O. took photographs. T.Ki. performed quantitative PCR. K.Sak., K.Sat., Y.K., and S.Yos. measured spectrum and performed simulations by using sensitivities of photoreceptors. H.F., S.Yod., T.Ki., Y.K., S.K., and T.Ko. wrote the manuscript. H.F. supervised this project. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Raw data related to Figs. 3 and 7 have been deposited in FigShare ( All other data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article