1 Introduction

Butterflies are one of the most colorful insect groups in nature, with the scales being extensively studied from varied aspects, including scale development [1,2,3], camouflage properties of scales [4, 5], significance of scales in courtship behaviour and reproduction [6], thermoregulation [7, 8], and optical responses [9, 10]. Coloration of butterfly scales is often very dependent on the presence of particular micro/nanostructures that act as diffraction gratings, multilayer reflectors, or three-dimensional photonic crystals during interactions with light [11,12,13,14], and closely related to the angle of observation [15]. Such structural coloration, as well as the role of chemical pigments [16,17,18] produces a rich diversity of colors and photonic effects in butterflies. For example, green coloration has been previously attributed to the presence of a lattice structure organized in irregular domains [19] and the coexistence of gyroid configurations in individual butterfly wing scales [20]. The mechanisms of some optical phenomena supply vital references for practical applications. Recently, butterfly wing scales were biomimetically fabricated [21,22,23,24,25,26] to explore potential and efficient applications in various materials, functioning as sensors such as optical [27, 28], thermal [29], gas/vapor [30,31,32,33], humidity and pH [34], energy [35], tissue engineering [36] and anti-counterfeiting [37].

In wing scales of Lycaenid butterflies, which have green [38, 39], blue [39] or violet wings [39, 40], two types of abovementioned microstructures were intensively studied for their gradually changing color scheme [41,42,43,44,45]. Violet butterflies have cover scales that are supplemented with a photonic crystal structure, and have been scrutinised by researchers as bioinspirational optical functional materials [46, 47]. The cover scales of these butterflies were shown to be the only source of producing the blue coloration of varied hues regardless of the presence of both the ridge-like ground scales and the so-called androconia scales [48]. The dynamic optics is vitally significant in exploring optical mechanisms, functional materials and device development. However, previous research on butterfly scales generally focused on the static optical properties of cover scales, with few studies referring to dynamic optical properties of only the cover scales, seldom covering the contribution of other scales, or their flexibility [49,50,51].

Herein, the dynamic coloration effect of the scale microstructures was studied in two species of Lycaenid, i.e., Plebejus argyrognomon with violet wings and Polyommatus erotides with blue wings, based on the characterization of the scale flexibility and the in-situ measurements of the reflectance spectra in order to explore (1) the angle- dependent color changeability and (2) the color-mediating efficiency of three kinds of flexible scales during wing rotation. To the best of our knowledge, this is the first report that the three layers of composite scales broaden the light signal, rather than the single layer of scales, particularly the androconia scales which are shown to strongly contribute to the overall wing coloration. Furthermore, the flexibility of butterfly scales allows for the exposed ratio of the three-color scales to change and causes the color-mediating effect. These findings are expected to (1) provide new insights into the in-depth understanding of biological discoloration, and (2) provide inspiration for the fabrication of biomimetic flexible optical materials, particularly those that aim to broaden signal ranges or display discoloration in different directions.

2 Materials and methods

2.1 Insects

Two species of Lycaenid butterflies were chosen, with 12 Plebejus argyrognomon (dorsal violet wings) (Fig. 1a) and 12 Polyommatus erotides (dorsal blue wings) (Fig. 1b) samples collected in Beijing, China, where no specific permits were required for the described field studies. Only the dorsal surfaces of the naturally dried wings were used in this study. It is noted that the right wings exhibit identical features as the left wing, so only the left wings were used.

Fig. 1
figure 1

Arrangement of wing surface scales of lycaenid butterflies. (a) Plebejus argyrognomon, (b) Polyommatus erotides. (a1-a4, b1-b4) Schematics, micrographs and images show the three layers of scales stacked on the wing with a handle stemmed into a single tubular follicle extruding out of the wing cuticle in differently title angles of the three layers of scales relative to wing cuticle. ●, CS, cover scale. ▲, GS, ground scale. ★, AS, androconia scale

Fig. 2
figure 2

Microstructures and averaged reflectance spectra on the wing scales of lycaenid butterflies. (a1-a8) Plebejus argyrognomon, (b1-b8) Polyommatus erotides. (a1) AS with perforated structure (a4). (b1) AS with ridge structure (b4). (a2, b2) GS with similar lattice structure (a5, b5). (a3, b3) CS with similar pepper-pot structure (a6, a7, b6, b7). (a8, b8) Averaged reflectance spectra of 20 single scales on the horizontal oriented wings of lycaenid butterflies. The red arrow shows that, at the wavelength greater than 487 nm, the reflection intensity of AS is stronger than that of CS in P. argyrognomon. (a1-a3, b1-b3) optical micrographs. (a4-a7, b4-b7) SEM images. CS, cover scale. GS, ground scale. AS, androconia scale

2.2 Optical Photography

Optical epi-illumination micrographs were obtained by tilting the left wings downward (wing tip/costal margin rotating down) and upward (wing tip/costal margin rotating up) from 0° to 60° relative to the longitudinal and transversal axes of the butterfly body (Fig. S1) using a large depth-of-field 3D digital microscope (Keyence VHX-1000 C, Japan) and processed using Keyence digital VHX software (version 1.3.0.7).

2.3 Scanning Electron Microscopy

The wing surface microstructures were characterized using a scanning electron microscope (Hitachi S-4800, Hitachi High-Technologies Corporation, Tokyo, Japan). The wings were coated with a thin layer (ca.10 nm thick) of gold using an ion sputtering instrument (BAL-TEC, SCD005). The structural parameters of the scales, including length (i.e., the length from the base to the end excluding the pedicel), width (i.e., the size of the middle part of the scale), and scale length exposed to air (Fig. S1), were measured using ImageJ software. The mean values and standard deviations were calculated based on 7–23 replicates.

2.4 Reflectance Spectra of Single Scales

The reflectance spectra of single scales were measured using the micro angle-resolved spectrum system (Ideaoptics, Shanghai, China) in the visible wavelength range from 390 to 760 nm. The wings were fixed with the wing base on the glass slide using double-sided adhesive tape. The focal light spot is 12 μm in diameter. The single cover, ground and androconia scales were big enough to obtain reflectance measurements. The parameters were set with the coupled software Morpho. The incident angle increased from − 20° to 30° with a step of 5° (Fig. S1). The reflectance spectra were measured on 20 individual scales and averaged when measuring. Reference reflectance was obtained from STD-WS, the average of standard white. The data were analysed using the software Origin 8.0.

2.5 Reflectance Spectra of Multiple Scales

The reflectance spectra of multiple scales were measured using the macro angle-resolved spectrum system (Ideaoptics, Shanghai, China) in the UV-Vis wavelength range of 230–760 nm. The light source was equipped with a deuterium/halogen lamp, the half angle of the light cone is 2.9°, and the focal light spot is 1 mm in diameter. The parameters were set with the coupled software Morpho. The incident angle increased from 0° to 60° with a step of 5°. The wings were fixed with the wing base on the glass slide using double-sided adhesive tape with the wing tip aligned in four directions, which was parallel or perpendicular to the increase direction of the incident angle (Fig. S1). Reference reflectance was obtained from STD-WS, the average of standard white. The data were analysed using the software Origin 8.0.

2.6 2D Fourier Analysis

The digital SEM micrographs of structurally colored cover scales were analyzed using a Python program. The scale of each image (nm/pixel) was calculated from the number of pixels in the scale bar of the micrograph. A 784 × 784·pixel2 portion of each array was selected from each image for analysis. We calculated the 2D Fourier power spectrum, or the distribution of the Fourier coefficients with the 2D Fast Fourier Transform algorithm. The max spatial frequency was calculated from the distance of two neighboring pixels according to the scale bar of the micrograph. Then, the 2D Fourier power spectra were expressed in spatial frequency (μm–1) by dividing the max spatial frequency values by the length of the matrix. The 2D Fourier power spectrum resolves the spatial variation in refractive index in the tissue into its periodic components in any direction from a given point.

3 Results and Discussion

3.1 Arrangement and Color of Wing Surface Scales

The arrangement of wing surface scales was characterized using optical photography and a scanning electron microscope. As Fig. 1 shows, three main types of scales, cover, ground and androconia, are stacked layer by layer on the wing surfaces with their tips oriented to the wing distal end in both butterfly species studied (Fig. 1a1, a2, b1, b2, S2). Between them, some elongated scales were observed, and their density decreased towards the wing edge (Fig. S3). At the base, the cover and ground scales with a stout pedicel, and androconia scales with slender pedicels, are slantly stemmed into the scale socket extruding out of the wing cuticle (Fig. 1a3, b3). There are differences in the inclined angle (Fig. 1a4, b4) and color of the three scales between the two species. In addition, the differences are also reflected in the length of the visible part of both the cover and ground scales, particularly the shape, microstructure, width and the corresponding area of the androconia scale (see the details in Table 1).

Table 1 Comparison of the structural parameters of three layers of scales on the wing dorsal surfaces of lycaenid butterflies (Plebejus argyrognomon and Polyommatus erotides). The scale length refers to the length from the base to the end, excluding the pedicel of the scale. The scale width was measured at the middle of the scale. CS, cover scale. GS, ground scale. AS, androconia scale

3.2 Microstructure of Wing Surface Scale and Average Reflectance Spectra

Whilst investigating the scale microstructure, the dorsal surfaces of androconia scales (Fig. 2a1) in P. argyrognomon showed a perforated structure with quasi-regularly arranged fibrous knots underneath, whilst the non-perforated areas revealed cross ribs between the longitudinal ridges (Fig. 2a4). In P. erotides, the base of the androconia scale (Fig. 2b1) was seen to be cracked between the longitudinal ridges (arrow in Fig. 1b3), with the distal half showing a ridged structure (Fig. 2b4). The ground scales of both species (Fig. 2a2, b2) were seen to be mainly composed of lattice structures (Fig. 2a5, b5). The cover scales (Fig. 2a3, b3) consisted of pepper-pot photonic structures in both Lycaenid butterflies studied, with the diameter of middle pores ca. 170 ± 50 nm in P. argyrognomon (Fig. 2a6, a7) and ca. 190 ± 30 nm in P. erotides (Fig. 2b6, b7).

The averaged reflectance spectra of the single scales on the horizontally oriented wings of both Lycaenid butterflies are shown in Fig. 2a8, b8. The reflectance peaks of cover scales of approx. 395 nm in P. argyrognomon (the purple line in Fig. 2a8) and approx. 445 nm in P. erotides (the blue line in Fig. 2b8) are ascribed to the pepper-pot structure, with pore diameters (170 nm and 190 nm, respectively) corresponding to the wing color of violet and blue, respectively. For P. argyrognomon, the androconia scales (Fig. 2a8) showed a similar reflection peak as cover scales, which should contribute to the wing color, with indented structures on the top of ridges and fibrous knots in the perforated areas. When the wavelength exceeded 487 nm (indicated by the arrow in Fig. 2a8), the reflectance of the androconia scales became stronger than the cover scales, with a consistently weaker reflectance of the ground scales along the visible spectrum (Fig. 2a8). In P. erotides, the reflectance of cover scales were shown to be higher than that of the ground scales, with the weakest reflectance spectrum appearing for the androconia scales. The ground scales with lattice structures, and the androconia scales with ridges contribute relatively little to the blue coloration (Fig. 2b8).

3.3 Reflectance of a Single Scale on the Wing Surface

The reflectance spectra of single cover, ground and androconia scales of both Lycaenid butterflies in the visible wavelength range of 390–760 nm was measured as the incident angle changed from − 20° to 30° (inset in Fig. 3). A peak at an approximate wavelength of 390 nm was noted on the single cover scale (highlighted by arrow in Fig. 3a1) and androconia scale Fig. 3a3) of P. argyrognomon. Similarly with P. erotides, a distinct peak appears at 445 nm on the single cover scales (highlighted by arrow in Fig. 3b1) and androconia scale (Fig. 3b3), with one poorly visible peak evident at a broader wavelength of 390–550 nm on the ground scale (Fig. 3b2). The reflectance of the cover, ground and androconia scales of P. argyrognomon (5–27%, 2–4% and 7–22%, respectively, in Fig. 3a4) and those of P. erotides (5–13%, 3–7% and 4–9% in Fig. 3b4) were found to increase as the incident angle changed from − 20° to 30°.

Fig. 3
figure 3

Reflectance spectra of single scales on wings of lycaenid butterflies in the visible wavelength range of 390–760 nm. (a1-a4) Plebejus argyrognomon, (b1-b4) Polyommatus erotides. (a1, b1) CS, cover scale, (a2, b2) GS, ground scale and (a3, b3) AS, androconia scale. (a4, b4) The reflection intensity of single CS, GS and AS at the wavelength of about 390 nm and 445 nm as the incident angle changing from − 20° to 30°. The inset in the upper middle shows the direction of wing tip aligned with that of the incident angle, which changed from − 20° to 30° with a step of 5°. Red arrows show the peaks at approximately 390 nm in (a1, a3) and 445 nm in (b1, b3)

The reflectance was generally related to the inclined angle of scales relative to wing cuticle. In P. erotides, at 445 nm the reflectance of cover scales was always the highest followed by androconia scales and finally the ground scales, with an incident angle changing from − 20° to 30° (Fig. 3b4). The inclined angles of the three scales (34.5 ± 1.6°, 21.5 ± 1.0°, 15.3 ± 1.5°) also presented a declining trend. Similarly for P. argyrognomon (Fig. 3a4), with the reflectance of the androconia scale with the largest inclined angle (32.5 ± 1.3°) being highest when the incident angle changed from − 20° to 25°. At an incident angle of 30°, the higher reflectance of the cover scales than the androconia scales may be associated with the inclined angles of cover scales (30.5 ± 1.0°), when the incident light is perpendicular to the cover scale surface and causes the reflectance to increase. The ground scales have the lowest reflectance and maintained an almost identical amplitude at 390 nm due to the minimal surface area exposed to air due to the coverage of the upper cover scales, in addition to the lowest inclination angle (21.3 ± 1.5°) in P. argyrognomon (Fig. 3a4).

3.4 Reflectance of Multilayer Scale Composites on Wing Surface

The reflectance spectra of multiple scales of both Lycaenid butterfly wings were measured with the incident angle gradually increased from 0° to 60° in four directions at the UV-Vis wavelength range of 230–760 nm (Fig. 4). Excluding the deviant spectra measured at incident angles of 45°, 50°, 55° and 60°, all spectra have three prominent reflection peaks (Fig. 4). Two narrow peaks uniformly appeared at the wavelengths of 250 nm and 315 nm regardless of the direction of the incident angle (Fig. 4, S4). The third peak is narrower in the range from 390 to 500 nm and with a fixed peak at the wavelength of about 390 nm in P. argyrognomon (Fig. 4a1-a4, S4a1-a4). Similarly, in P. erotides the third peak is broader in the range from 390 to 550 nm and has a blue shift from the wavelength 430 to 390 nm with the incident angle changing from 0° to 60° (Fig. 4b1, b2, b4, S4b1, b2, b4). Interestingly, when the wing tip was aligned perpendicular to the incident angle, a narrow peak emerged at about 445 nm in P. erotides (Fig. 4b3, S4b3). Another narrow peak emerged at approximately 580 nm when the incident angle increased to 55° and 60°, and was evident in both Lycaenid butterflies (Fig. 4a3, a4, b3, b4), which is possibly due to the composite structure originating from the rotation of incident light or reflected from the back of the blue scale to achieve specific color changes. The reflectance of the second peak at 315 nm in P. argyrognomon (the red dotted line in Fig. 4a5-a8) and the third peak at 390–550 nm in P. erotides (the blue triangle line in Fig. 4b5-b8) were the highest among the three reflection peaks of the same species, which may contribute more to the wing color.

Fig. 4
figure 4

Reflectance spectra of the wing scale of lycaenid butterflies in the UV-Vis wavelength range from 230–760 nm. (a1- a8) Plebejus argyrognomon, (b1-b8) Polyommatus erotides. The insets in (a5-a8, b5-b8) show the direction of the incident angle increased from 0° to 60° with a step of 5° in the four aligned directions parallel with (a1, a2, b1, b2) or perpendicular to (a3, a4, b3, b4) the wing tip. (a5-a8, b5-b8) show the corresponding reflection intensity at the wavelength of 250 nm, 315 nm, 390–500 nm, 390–550 nm when incident angle increased from 0° to 60° in (a1-a4, b1-b4). The numbers demonstrate the reflectance peaks at wavelengths of approximately 250 nm and 315 nm in (a1, b1). The red arrows show the fixed peak at about 390 nm in (a1), the reflectance peak appears an inconspicuous blue shift from 430 to 390 nm with the increase of incidence in (b1), and peaks at approximately 445 nm in (b3), and 580 nm in (a3, a4, b3, b4)

The reflectance of multiple scales at the three peaks is related to the incident angle. As Fig. 4 shows, in P. argyrognomon, the reflectance initially decreases, then remains almost constant and finally significantly increases with an increase in the incident angle from 0°-15°, 15–45°, and 45–60° in all four directions (see the slopes of the fitting lines in Table S1). In P. erotides, the spectral reflectance clearly decreased when the incident angle increased from 0° to 15° and then remained almost constant when the wing tip was aligned parallel to the incident angle (Fig. 4b5, b6). However, when the wing tip was aligned perpendicular to the incident angle (Fig. 4b7, b8), the reflectance values of the wing changed similarly to P. argyrognomon, as the incident angle increased from 0°-15°, 15–40° and 40–60° (see the slopes of the fitting lines in Table S1).

3.5 Effects of Ground and Androconia Scales on Coloration in the Horizontal Wing

When the incident light illuminated multiple scales of the horizontal wing (Fig. 4), two reflectance peaks appeared in both species studied along the visible range of wavelengths, whereas peaks at 250 nm and 315 nm in the UV range remained the same. P. erotides had one broader spectrum (390–550 nm) and showed a larger reflectance angle (0–60°) than P. argyrognomon (390–500 nm, 0–45°). The highest reflectance peak at 315 nm on P. argyrognomon, and the peak range of 390–550 nm on P. erotides contributed more to the wing color (Fig. 4). Comparatively, although similar reflectance spectra were revealed for single ground scales (Fig. 3a2), the spectral reflectance of single cover and androconia scales in P. argyrognomon were found to be much stronger, with both spectral peaks in the short wavelength range of less than 500 nm (Fig. 3a1, a3). The stronger reflectance spectra illustrate that androconia scales play an important role in the enhancement of reflected light (Fig. 2a8). Similarly, in P. erotides, at approximately 445 nm, a single reflectance peak is seen for the single cover scales (arrows in Fig. 3b1) and the androconia scales (arrows in Fig. 3b3), with minimal light being reflected by the androconia scales (Fig. 2b8). Additionally, the one poorly visible peak at the wavelength of 390–550 nm on the ground scale is consistent with the third reflection peak of the multiple scales. The significantly different optical characteristics in the two studied Lycaenid butterflies are the outcome of the combined action of the three kinds of scales.

As Fig. 4 shows, when the rotating incident light illuminates the wings of P. argyrognomon and P. erotides, whilst aligned in four different directions (except for the deviated spectra), the consistency of the reflectance peaks at UV wavelengths of 250 nm and 315 nm (Fig. S3) illustrates that the UV color results from scattered light, filtered by a pigment absorbing in the blue and green wavelength ranges for the nondirectional dependence. In the visible region, the third peak in the range from 390 nm to 500 nm (Fig. 4a1-a4) in P. argyrognomon should be produced by both the cover scales and androconia scales (Fig. 3a1-a3, 2a8). In this case, the ground scales shielded by the cover scales contribute less to the reflectance spectra than the other two kinds of scales. In P. erotides, the spectral band of reflectivity in the range of 390 nm to 550 nm (Fig. 4b1-b4) should be the result of the three layers of cover scales, ground scales and androconia scales, as a result of the exposure of the ground scales to air and the appearance of reflectance spectra on all three scales (Fig. 3b1-b3, 2b8). The broader spectral bands in P. erotides (in the range from 390 nm to 550 nm) than those of P. argyrognomon (in the range from 390 nm to 500 nm) may be a result of the aggregation of weak peaks at the wavelength range from 390 nm to 550 nm of both the ground scales (Fig. 3b2) and androconia scales (Fig. 3b3, 2b8). In addition, the transmitted light reflected by the ground scales of P. argyrognomon located under the cover scales can additionally increase the reflectance. Simultaneously, the top part of the ground scales exposed to the air lose some reflected light, and so the contribution of the ground scales in P. erotides to the reflectance is stronger than that of P. argyrognomon (Fig. 2a8, b8).

The P. argyrognomon wing can reflect light in a small region, only deviating from the wing surfaces by approximately 45°, which is why the reflected light signal cannot be received in four directions (Fig. 4a1-a4). In comparison, for P. erotides, the incident light can be reflected for a broader angle of 60° when parallel to the direction of the wing tip (Fig. 4b1, b2), but the light can be reflected at almost the same angle of 45° as P. argyrognomon (Fig. 4a3, a4) when perpendicular to the direction of the wing tip (Fig. 4b3, b4). This should be associated with the inclined angles of the three types of scales (Fig. 2a4, b4). The range of the reflected signals implies that the P. erotides butterfly could be observed at a greater range of angles by other animals. In addition, for P. erotides specifically, when the incident light rotates perpendicular to the direction of the wing tip (as the insets demonstrate in Fig. 4a3, a4), the sharp peak at approximately 445 nm (red arrows in Fig. 4b3) may be ascribed to the combination of cover and androconia scales for reflectance peaks at the same wavelength as shown in Fig. 3b1, b3.

3.6 Flexibility of Scales and Their Role in Adjusting Wing Coloration

It was observed that the overlapped layers of scales on the butterfly wings were flexible (Fig. 5), with the tips of both the cover scales and androconia scales swaying under a light breeze (Movies S1, S2). When the wings were depressed downwards, only the tip of cover scales were visible (Fig. 5a1). When the wings were erected upwards, the cover and androconia scales in P. argyrognomon (Fig. 5a2) and P. erotides (Fig. 5b1) were seen to prominently bend. The ground scale was also shown to be flexible (Fig. 5b2). The backside of all scales was decorated with wavelike structures (Fig. 5a3, b3).

Fig. 5
figure 5

SEM images of the wing scale of lycaenid butterflies. (a1- a3) Plebejus argyrognomon, (b1-b3) Polyommatus erotides. (a1) When the wings were erected downwards, only the tip of CSs are visible. (a2) When the wings were erected upwards, the CSs prominently bend. (b1) When the wings were erected upwards, the CSs and ASs prominently bend. (b2) The GS is flexible as well. (a3, b3) The backsides of all scales are decorated with wavelike structures. CS, cover scale. GS, ground scale. AS, androconia scale

The optical epi-illumination micrographs further confirmed the flexibility of scales. When the wings tilted between angles of 0 to 60° downward (wing tip/costal margin rotating down) and upward (wing tip/costal margin rotating up) relative to the butterfly body’s longitudinal and transversal axes (in four different directions, as illustrated by the diagrams in Fig. 6a, b), all scales are seen to bend in the direction of tilting. Simultaneously, we observed a wing discoloration.

Fig. 6
figure 6

Optical micrographs of the left wings rotating in four directions relative to the butterfly body axis, as shown in the diagrams (a, b). (a1-a10) Plebejus argyrognomon, (b1-b10) Polyommatus erotides. The dash lines illustrate the original and rotated positions of the wings from 0° to 60°. The wing’s position at 0° is the horizontal plane, which is always perpendicular to the plumb line. (a1-a4, b1-b4) When the wing rotates downward (wing tip tilting down) relative to the butterfly longitudinal body axis, the wing color brightened. (a5-a8, b5-b8) When the wing rotates upward (wing tip tilting up) relative to the butterfly longitudinal body axis, the wing turned brown. (a9, b9) When the wing rotates downward (wing costal margin tilting down) relative to the butterfly transversal body axis, the wing color became slightly paler. (a10, b10) When the wing rotates upward (wing costal margin tilting up) relative to the butterfly transversal body axis, the wing color became paler (a10) or even whiter (b10). Angles 0°, 15°, 30°, 45° and 60° on the bottom left of each optical photo are the angles the wing rotation, angles 20°, 30°, 35° and 35° on the upper right of (a9, a10, b9, b10) are the titled angles of the scales relative to their original positions

As upper Fig. 6a, b shows, when the wing was tilted downward, i.e., rotating downward relative to the insect body’s longitudinal axis, the wing color became much brighter (Fig. 6a2, b2) than that of the horizontally placed wing (Fig. 6a1, b1). Particularly in P. erotides, the tilted wing was much bluer when rotated downward to an angle of 60° (Fig. 6b4). By magnifying the wing surfaces, the increase in color intensity was more pronounced (compare Fig. 6a2 with Fig. 6a4 and Fig. 6b2 with Fig. 6b4). In P. argyrognomon, the basal half of the androconia scale was covered by the above neighbouring cover scales when the wing was tilted to 45° (Fig. 6a3), and it was completely unseen when the wing was tilted to 60° (Fig. 6a4). In the latter case, the tips of cover scales completely overlapped with the androconia scales and became whitish, like the androconia scales (Fig. 6a4), indicating that the cover scales allow white light from the ventral side to pass through. In P. erotides, the ground scales and androconia scales in air were gradually shortened and became virtually invisible when the wing was tilted to an angle of 45° (Fig. 6b3). When the wing was tilted to 60°, the cover scales further overlapped and the color brightened (Fig. 6b4). The tilted elongated scales of P. erotides became dark brown (Fig. 6b4), while the androconia scales and elongated scales of P. argyrognomon became much brighter (Fig. 6a4).

When the wing was tilted upward relative to the longitudinal body axis (Fig. 6a5-a8, b5-b8), all scales correspondingly moved in the tilt direction. When the wing was tilted upward to 15°, the colors of the cover and androconia scales of P. argyrognomon became less bright than those in the wing horizontal position (compare Fig. 6a5 with Fig. 6a1). When the wing was tilted upward to 30°, the androconia scales stayed close to the slightly inclined neighbouring cover scales, becoming difficult to see (Fig. 6a6). The ground scales were clearly exposed when the wing was tilted to 45° (Fig. 6a7). All scales were positioned relative to the wing surface when the wing was tilted to 60° (Fig. 6a8). In this position, the wing showed a brownish color, which prominently differed from the wing color that appeared during downward rotation (Fig. 6a1-a4). In P. erotides, during upward wing rotation, all scales were more flexible and erect when the wing tilted to 45° (Fig. 6b7). At a tilt angle of 60°, the tip of the cover scale was bent downwards to the wing base (Fig. 6b8). The elongated scales of P. erotides were bright and their distal regions looked blue (Fig. 6b8), whereas in P. argyrognomon, the elongated scales were whitish (see Fig. 6a1-a4, a5-a8).

Similarly, when the wings were tilted either downward (wing costal margin tilting down) or upward (wing costal margin tilting up) relative to the butterfly transversal body axis (the lower diagrams of Fig. 6a, b), the scales had an identical tilting-induced deflection response. A deflection angle of approximately 20° was reached by the tilted scales about the horizontal axis, at a wing tilting angle of 30° (Fig. 6a9). When the wing was tilted to 60°, the angle of the scale inclination was much smaller in relation to the intact position, i.e., approximately 30° in P. argyrognomon (Fig. 6a10) and 35° in P. erotides (Fig. 6b10). Correspondingly, the wing color also changed for the moved scales (Fig. 6a9, a10, b9, b10), compared with the horizontally placed wings (Fig. 6a1, b1).

3.7 Rearrangement of Flexible Scales Mediating the Tilted Wing Coloration

When the P. argyrognomon wing was in the horizontal position, with the width and length of all three layers of scales exposed to air, the area ratio of cover scale to androconia (where ground scales are invisible) was roughly 4:1 ((58.62 × 49.77):(35.01 × 20.49)) (Fig. 6a1). The ratio for the three types of the P. erotides scales, i.e., cover, ground and androconia scales, was 4:3:1 ((59.74 × 24.26):(57.98 × 19.42):(20.07 × 18.95)) (Fig. 6b1). When the wing was tilted downward to 30° and 45°, the proportion of the two types of P. argyrognomon cover and androconia scales exposed to air was 5:1 and 6:1 (Fig. 6a2, a3), respectively, with the wing color becoming more violet for a larger area of the cover scale over the androconia scale when tilted to 60° (Fig. 6a4), which leads to the enhanced saturation of the wing coloration. In P. erotides, when the wing was tilted downward to 30°, the proportion of the cover and androconia scale was approximately 6:1 (Fig. 6b2). When the wing was tilted to 45°, only cover scales were visible, and when the wing tilted to 60°, the cover scales looked to overlap much more, which is why the wing color became much bluer (compare Fig. 6b3 with Fig. 6b4). Additionally, during wing rotation, the brown ground and dark androconia scales exposed to the air were tilted and unable to reflect light, thus, they become virtually invisible (Fig. 6b2, b3) and tuned the brightness of the wing color. In both Lycaenid butterflies, the greater the tilting angle (from 0° to 60°), the more significant the scale efficiency in mediating color reflectance.

When the wing was tilted upward, only the top section of cover and androconia scales were curved (Fig. 6a5-a8, b5-b8, 5a2, b1). With an increase in the tilting angle of the wing, unlike the cover and androconia scales, the area of ground scales exposed to air increased, possibly due to the inconspicuous bending (Fig. 5a2, b1) maintained by the rough broken structures of androconia scales, which restrained the ground scale tips (Fig. 1a3), with the cover scales blocked at the base of the flexible ground scales (Fig. 1b3, 5b2). Hence, the brown color of the ground scales became the main color as the tilting angle of the wing increased. Consequently, these moveable scales mediated wing coloration (Fig. 6a5-a8, b5-b8) because of the stronger reflectance, which predominantly originates from the large amount of ground scales exposed to air, and the offset of the weaker reflectance of the tilted or erect cover and androconia scales, which hardly affected coloration for their wavelike structures (Fig. 5a3, b3). In a downward tilted wing, cover scales were mainly exposed to the air, which is why the wing was violet or blue (Fig. 5a1).

Similarly, in the other two cases, when the wing was tilted forward or backward (lower Fig. 6a, b), wing rotation could efficiently mediate the color by changing the arrangement of scales and their ratios exposed to the air. However, the angle of the inclined scales relative to the intact position was much smaller in P. argyrognomon (approximately 30°) (Fig. 6a10) than in P. erotides (approximately 35°) (Fig. 6b10), possibly due to the scales of the latter being more flexible, as shown in Fig. 6b8, i.e., the scale tips bent down when the P. erotides wing rotated upward 60°, while the scales of P. argyrognomon were erect but not bent when the wing rotated to the same angle (Fig. 6b7).

3.8 Cover Scale Coloration Ascribed to the Coherence Scattering

The digital SEM micrographs of structurally violet (Fig. 7a1) and blue (Fig. 7b1) cover scales were analyzed using a Python program. As Fig. 7a2 shows, in P. argyrognomon the Fourier power spectrum exhibits a concentration of spatial frequencies within a circular region centered at the origin, suggesting the presence of an amorphous photonic structure—specifically, a structure characterized by short-range order. To determine the characteristic length scale, the Fourier power spectrum was averaged across all azimuth angles, enhancing the signal-to-noise ratio of the Fourier transform pattern depicted as the red line in Fig. 7a2. In the circular region, a distinct peak is observed, positioned at a spatial frequency of approximately 3.8 \( {\mu m}^{-1}\). The inverse of this value yields the characteristic length scale of the structure, which is approximately 263 nm. Structures seem disordered from their morphology, but they possess well-defined short-range order. Light scattering occurs due to the short-range order at the characteristic length scale of structures, leading to a reflectance peak at visible frequencies.

The Fourier power spectrum of the cover scales of P. erotides is shown in Fig. 7b2, which correspond to the digital SEM micrographs shown in Fig. 7b1. Again, a distinct peak is observed, positioned at a spatial frequency of approximately 2.76 \( {\mu m}^{-1}\). The inverse yields the characteristic length scale of approximately 362 nm, which is larger than the nanostructure of P. argyrognomon (Fig. 7a1, b1), which implies that the wavelength region of P. erotides’s optical response should be longer than that of P. argyrognomon (Fig. 7a2, b2).

3.9 Wing Optical Properties and the Biological Significance

The biological significance of butterfly wing coloration has an (1) intraspecific signalling role, (2) various interspecific effects [52] and (3) thermoregulation [7, 8]. In nature, butterflies can flap their wings to achieve specific discolorations, additionally, they can reduce their visibility to potential predators through a change of wing angle. Insects’ motion vision systems are achromatic and dependent on brightness rather than color contrast [53]. The two Lycaenid butterflies can readily achieve this goal with high efficiency by changing the position of their wings against a background. The color, pattern and small-scale reflective differences of the wings of Lycaenid butterflies are used for mate recognition [39, 54]. In the visible range, the three layers of the composite scales enhance the incident and reflected light from 20° (Fig. 3) to 55°, and even 60° (Fig. 4), maybe which is of great significance to the identification of signals between species. The UV reflection may be utilized as the spectral sensitivity of insect vision [55] and used as a recognition signal for mating purposes [56]. In Chrysozephyrus butterflies, two peaks at similar wavelengths (around 250 nm and 320 nm) were detected in the UV range. Only the 320 nm peak, or its longer side shoulder, could be perceived by these butterflies [52]. Both Lycaenid butterflies have wing scales with two reflectance peaks in the UV range of wavelengths (250 nm and 315 nm). Thus, these peaks do not represent a species-specific characteristic but might affect mate quality [57, 58].

As for the thermoregulation aspect, the photonic crystal-like behavior of the pepper-pot structure significantly reduces the penetration of light with wavelengths in the blue region of the spectrum into the body of the scales. This may shed some light on the adaptation that enhances the survival chances of the butterfly in a cold environment rich in blue and UV radiation, with the increase in the proportion of dark wing scales being of significance in thermoregulation [7]. The darker cuticle of the P. argyrognomon wing, compared to that of P. erotides (Fig. S5), may have a positive effect on the cooling rate for melanisation [59]. The elongated scales also exhibit tuneable coloration depending on the wing position, however, their significance in ecology or evolution remains unclear. These opening questions deserve to be further studied in the future.

4 Conclusions

In this paper, three types of layered scales in the wings of two species of Lycaenid butterflies were structurally and experimentally studied. The butterflies displayed violet and blue colors of varied intensity. It was demonstrated for the first time that the three overlapping layers of scales are flexible. When the butterfly wings tilted at angles from 0° to 60° in different directions, all scales correspondingly curved in the same direction as the wing. The relative motion of the scale surface resulted in corresponding changes in the wing anisotropic coloration or hue intensity. Reflectance measurements of the combined and individual scales and optical photographs illustrate that the interaction of the micro-structures of the three types of scales with incident light should be responsible for the discoloration of the scales. Such a biomechanical mechanism of discoloration can be potentially interesting for bioinspired flexible photonic materials.

Fig. 7
figure 7

2D Fourier power spectra of Scanning electron micrographs. (a1) Scanning electron microscope (SEM) image of the nanostructure and (a2) the corresponding Fourier power spectrum of Plebejus argyrognomon. (b1) SEM image of the nanostructure and (b2) the corresponding Fourier power spectrum of Polyommatus erotides. Color scale (from blue to red) indicates the relative magnitude of the squared Fourier components, which are dimensionless quantities. Direction from the origin indicates the direction of the 2D component waves in the image, and the distance from the origin indicates the spatial frequency (\( {\mu m}^{-1}\)) of each Fourier component. The Fourier power peaks (red pixels) demonstrate predominant short-range order at intermediate spatial frequencies. The distance from the origin is inversely proportional to the wavelength of the coherently scattered color