Wing colors of the four species of Chrysozephyrus butterflies were analyzed by a spectrophotometer. As the dorsal wing surface of males showed a strong reflectance when the specimen was tilted, measurements were made by the tilting method. The dorsal wing surface of males which appears green to the human eye reflected UV (315–350 nm) as well as green light (530–550 nm). The reflectance rate of UV to visible green light varied among species with a higher rate for C. hisamatsusanus and C. ataxus, and a lower rate for C. smaragdinus and C. brillantinus. The peak wavelength and the peak height did not shift when the specimen was exposed to direct sunlight at least for 16 hr. Artificial removal of scales by scratching the wing surface decreased reflectance. Blue marks on the forewings of C. brillantinus, C. hisamatsusanus and C. ataxus females reflected UV to visible light of short wavelength, and orange marks on the dorsal surface of the forewing and the ventral surface of the hindwing of C. samaragdinus females showed a higher reflectance at longer wavelengths.
The colorful wing patterns of butterflies are thought to have some biological significance. In 1909, Darwin proposed that the color pattern of the dorsal wing surface has an intraspecific signaling role, including attraction of females by males in sexually dimorphic species, and that of the ventral wing surface displayed in a perched posture has inter-specific effects as protection from predators. Since then, some experiments have been carried out to address the functions of wing color patterns; a clue to mate recognition for Arginnis paphia (Magnus, 1958), Pieris rapae (Obara and Hidaka, 1968; Obara, 1970) and P. protodice (Rutowski, 1981), Colias eurytheme (Silberglied and Taylor, 1978) and Papilio xuthus (Hidaka and Yamashita, 1975), and anti-predatory effect as aposematic coloration for Heliconius erato (Benson, 1972) and as mimicry for some North American species (Brower, 1958a, b, c). However, the number of species examined is limited, and male-specific brilliant coloration has rarely been subjected to experimental studies (for exception, Lederhouse and Scriber 1996).
For understanding of the biological significance of butterfly wing colors, it is prerequisite to learn the range of colors or wavelengths reflected from the wing surface, along with the perceptible range of wavelengths in potential receivers such as conspecific individuals or predators. Concerning the latter problem, the perceived range is known to somewhat differ between humans and insects or other animals (Silberglied, 1979), and thus, analyses based on physical measurements of wavelengths reflected from butterfly wings are needed (Crane, 1954). In this study, we analyzed the wing colors of four species of Chrysozephyrus butterflies which show conspicuous sexually dimorphic coloration on their wings, with males showing brilliant green wings in contrast to females with mostly dark brown wings. Further, the effects of direct sunlight and dislodging of scales on potential change of color, and the effect on reflectance of angles in which light is reflected were examined.
MATERIALS AND METHODS
Measurement of wavelength
Four species of genus Chrysozephyrus distributed in Japan were examined; C. smaragdinus, C. brillantinus, C. hisamatsusanus and C. ataxus. The dorsal wing surface of males of these species appears metallic green to the human eye, and that of females basically dark brown with an orange and/or a blue mark only on the forewing. Ventral wing surfaces are similar between sexes with gray to brown, except for C. ataxus which shows a complex pattern in females, as shown in Fig. 1.
For measurement of wavelength of the light reflected from the wing surface, the central part of the wing was cut out in a 10×15 mm piece and settled on the stage of a spectrophotometer (Shimadzu, UV-240). The measured range was from 200 to 700 nm. For measurement of specific colored parts of the wing, those parts were used (CB21, CH21 and CA21 in Fig. 7B; CA21b and CA21w in Fig. 7E), by putting the corresponding areas on the left and right wings together (CS21o and CS21c in Fig. 7B).
In a preliminary observation, the dorsal wing surface of males was found to look more brilliant when viewed from the front or inside at a lower angle (nearly parallel to the wing surface), and thus, the specimen was set on the spectrophotometer at a tilt angle of 45° for the dorsal wing surface of males as mentioned next.
Tilt angle examination
A small rubber stage (10×10 mm at the base) whose top was cut at a certain angle (0°=no cut, 15°, 30°, 45 and 60°) was inserted between the fitting plate and the specimen. Thus, on the spectrophotometer, the incident light was applied to the wing piece from the direction shown with line L in Fig. 2, and the reflected light that returned in the same course was measured. The direction of tilt was at right angle to the frontal edge for the forewing and toward the base for the hindwing, which was determined by a naked-eye observation as to perceive the strongest reflection. For the examination of tilt angle (α in Fig. 2), the left fore- and hindwings of three individuals of C. smaragdinus (CS1, CS2 and CS3) were used.
Since butterflies are thought to loose their scales during active flight in the field which may cause wing color change, the effect of scale loss on reflectance was examined.
A piece of the wing specimen was softly swept off with a writing brush, and the number of cover scales on the surface was counted on photos taken under a microscope with illumination from oblique above; as the brilliant male coloration is derived from the green cover scales, only these scales (except for dark brown basal scales) were counted. For counting, five sampling sites were selected around the center of the wing; four corners and the center of a 3×3 mm square. An average value of these five counts was used as the number of scales for the wing. The specimen was then subjected to a spectrophotometric measurement.
This examination was made by using the right hindwings of three individuals of C. smaragdinus (CS1, CS2 and CS3).
Effect of direct sunlight
Since old specimens of butterflies show a faded wing color, probably due to a result of exposure to light, wing specimens were exposed to direct sunlight and their reflectance was measured at various intervals. For exposure, specimens were put on a cylindrical drum which was rotated once a day and whose axis was adjusted for the specimens to continuously receive the sunlight vertically. The light intensity was measured at the same time; the sensor of the photometer was also attached to the drum. Exposure was made between 9 a.m. and 3 p.m. on fine days from November 2000 to January 2001. The light intensity was 8.14±1.68×104 lux (mean±SD). The left fore- and hindwings of three individuals of C. smaragdinus (CS1, CS2 and CS3) were used.
Ten male and one female specimen were used for C. smaragdinus and C. brillantinus and five male and one female specimen for C. hisamatsusanus and C. ataxus. All of them were derived from field collections, except for the C. hisamatsusanus female which was difficult to capture in nature and thus, reared from eggs. The specimens used in the present experiments are summarized in Appendix.
For comparison of wavelengths and reflectance rates between two species (Table 1), a Mann-Whitney U-test was carried out with the use of Statview J-4.5 software (Abacus Concepts). For the cluster analysis for Fig. 8, Ward's method on JMP ver. 3 software (SAS Institute) was adopted.
The peak wavelength in UV and green regions (nm) and the ratio of the peak height of UV to that of green region. The left hindwings were used. Mean±SD.
1. The dorsal wing surface of the C. smaragdinus male
Reflectance on the dorsal surface of the four wings of a C. smaragdinus male (CS2) is shown in Fig. 3. The reflectance was measured tilted at an angle of 45°, except for LF0 which was measured without tilting. The four wings showed a similar pattern of reflectance with two large peaks at about 315 nm (UV region) and 525 nm (green region), with a minor difference between the fore- and hindwings; the forewings tended to reflect slightly shorter wavelengths than the hind-wings both in the UV and green regions, and the tendency was confirmed for the other two individuals (CS1 and CS3). The heights of the two peaks in the UV and green regions were not largely different; thus, a C. smaragdinus male reflected ultraviolet light nearly as strongly as visible green light.
Effect of tilt angle
When the tilt angle of the specimen was increased, the reflectance also increased (Fig. 4; also compare LF and LF0 in Fig. 3). At an angle of 45°, it was about three times as strong as that without tilting. On the other hand, the peak wavelength did not largely shift with the change in tilt angle. A high reflectance by tilting of the wing in such a way as shown in Fig. 2 suggests that green and UV lights are reflected strongly forward and inward.
Effect of scale loss
Effect of direct sunlight
The peak wavelength and the peak height did not change when the specimens were subjected to exposure to direct sunlight at least for 16 hr (Fig. 6).
2. Dorsal wing surface of males for the four species of Chrysozephyrus
The dorsal wing surface of males appeared similarly brilliant green in all four species (Fig. 1). Analyses by spectrophotometry revealed that these species also reflected UV light at different intensities among the species (Fig. 7A).
The height of the peak should be affected by the number of scales lost during flight activity in nature, as expected from the scale-loss experiment (ref. Fig. 5). Thus, the ratio of the peak height of UV region to that of the green region was compared among species. It was higher in C. ataxus and C. hisamatsusanus, and lower in C. smaragdinus and C. brillantinus (Table 1). A cluster analysis of color characteristics with use of average values of the peak wavelengths and the reflectance ratios revealed that C. smaragdinus and C. brillantinus are nearest (distance=0.6), followed by C. ataxus and C. hisamatsusanus (0.9), and the two groups differred by a distance of 2.8 (Fig. 8).
3. Dorsal wing surface of females for the four species
Four types are known for Chrysozephyrus females (Kawazoé and Wakabayashi, 1976); type A possesses an orange mark, type B a blue mark, type AB an orange and a blue mark, and type O has no mark on the dorsal surface of the forewing.
A large orange mark of C. smaragdinus showed a higher reflectance at a longer wavelength (CS21o in Fig. 7B). Blue marks of C. brillantinus, C. hisamatsusanus and C. ataxus reflected shorter visible lights and longer UV lights with a peak around 400 nm; 415, 420 and 395 nm, respectively (Fig. 7B) The dark hindwing showed no peak from 300 to 700 nm wavelengths for all the species examined (Fig. 7C).
4. Ventral wing surface of males and females
The ventral wing surface of these butterflies, except for C. ataxus, is basically gray to brown with white stripe and an orange mark at the posterior corner of the hindwing (Figs. 1, 7D and 7E). In C. ataxus, the male is white, showing a higher reflectance from 300 to 700 nm (CA 1 in Fig. 7D). The female shows a complex pattern composed of white (CA 21w in Fig. 7E) and brown (CA 21b), the former being similar to the ventral wing surface of the male.
The orange mark at the posterior corner of the hindwing of C. samaragdinus (CS21c in Fig. 7B) showed almost the same reflection pattern as the orange mark of the forewing (CS21o).
In the present study, the wing colors of Chrysozephyrus butterflies were analysed. The dorsal and ventral wing surfaces of females and the ventral wing surface of males were shown to exhibit color patterns expected from our optical perception, whereas the dorsal wing surface of males was found to strongly reflect ultraviolet light imperceptible to humans. So, we concentrate the discussion here on the reflection characteristic of male wings of C. smaragdinus and the UV-light reflection of Chrysozephyrus butterflies.
1. Reflection characteristic of male wings of C. smaragdinus
In some of the present experiments (Figs. 4, 5 and 6), some variation was seen in reflectance measurements, even when the same sample was measured repeatedly. Such variation seems to be attributable to a slight difference in the area sampled on the uneven surface of the wing with vanes, from one test to another, by the spectrophotometer that sampled a small area of 5×8 mm.
In spite of such variation, the dorsal wing surface of males reflected stronger light when it was measured at a larger angle (Fig. 4) from the perpendicular axis (see Fig. 2). Thus, the reflection by Chrysozephyrus butterflies is directional. This is the case for the UV and visible light peaks. Directional reflection is known for males of Gonepteryx rhamni (Nekurtenko, 1965), Phoebis rurina (Eisner et al., 1969), Eurema lisa (Ghiradella et al., 1972) and Colias eurytheme (Silberglied and Taylor, 1973). In Eurema lisa, Ghiradella et al. (1972) obtained the strongest reflection when the horizontally held wing was measured, under top illumination, at 40° inward (toward the body axis) from the perpendicular. Rutowski (1977) confirmed UV light reflection when he opened a pair of wings of a butterfly illuminated and observed from the top by more than 120° from each other. The latter method is similar to ours in that the incident and observed lights passed almost the same course. Thus, the angle that yields a strong reflection differs between Eurema and Chrysozephyrus; the former species reflected a strong UV-light at smaller angles from the perpendicular and the latter species at larger angles. The difference may be related to arrangement of scales on the wing (Shinkawa, pers. com.).
Examination of the dorsal wing surface of C. smaragdinus males revealed that the wavelength of light reflected from the forewing was slightly shorter than that from the hindwing both in UV and green light ranges (Fig. 3). This difference may be caused by the method of measurements; the tilt angle direction of the incident (and also the measured) light was adjusted to different courses relative to the base of the wing between the fore- and hindwings (Fig. 2). A shift in wavelength according to observation angles was shown in a sulfur butterfly (Ghiradella et al., 1972). For Chrysozephyrus butterflies, close examination by employment of incident light from various tilt angle directions is needed for each of fore- and hindwings.
Wing colors of C. smaragdinus males were unaffected by exposure to direct sunlight at least for 16 hr (Fig. 6). This duration corresponds approximately to 3 days of natural activity of this species, as they are known to be active 5 to 6 hr a day (Imafuku, unpublished). Butterflies being less likely always to receive direct sunlight perpendicularly on their wings in nature, as expected at time other than around noon, the exposure time applied in the present experiment seems to correspond to more days in nature. Even though such a long term exposure, the color of the wing never changed. On the other hand, colors of some other butterflies are said to fade in time, even during their life time (Crane, 1954). The stable nature of the wing color of C. smaragdinus males may be related to a color producing mechanism in the cover scales; by “interference” as expected from directional reflectance of this species. It is interesting to compare color stability to sunlight between butterflies with interference coloration and those with pigmentary coloration.
2. Ultraviolet Reflection in Chrysozephyrus butterflies
In the present experiments, the dorsal wing surface of males was found to reflect UV light as well as green light for all of the Chrysozephyrus species examined. The UV light reflection is known for many other butterflies belonging to Nympharidae, Acraeidae, Pieridae, Morphidae, Blassolidae, Papilionidae and Lycaenidae (Lutz, 1933, Makino et al., 1952, Crane, 1954, Mazokhin-Porshniakov, 1957, Obara and Hidaka 1968, Eisner et al., 1969, Silberglied and Taylor, 1973, Yonekubo and Saito, 1973, Rutowski, 1981). Reflection patterns of UV differs considerably among species. UV is reflected from the whole wing surface or a part of it. In the latter case, wings that look evenly monochrome to the human eye seems to appear in an utterly different pattern to their bearers' eyes, as expected in Gonepteryx rhamni (Mazokhin-Porshniakov, 1957), Anteos clorinde and Phoebis rurina (Eisner et al., 1969). In many species, the reflection pattern of UV differs between sexes; females reflect UV whereas males do not for Pieris rapae (Makino et al., 1952, Obara and Hidaka, 1968), P. protodice (Rutowski, 1981) and Belenois zochalia (Silberglied, 1979), and the situation is utterly opposite for Phoebis sennae (Crane, 1954), Gonepteryx rhamni (Mazokhin-Porshniakov, 1957), Eurema lisa (Ghiradella et al, 1972), Colias eurytheme and C. chrysotheme (Silberglied and Taylor, 1973). In either case, butterflies could discriminate sex by UV-light reflection, not possible by the human eye. In the case of Chrysozephyrus butterflies, sexual discrimination seems to be easier, because reflection patterns differ between sexes in the UV as well as the visible light range. A further interesting point in Chrysozephyrus butterflies is that male wings in this group look green, to us, the same color as that of leaves on which they habitually alight, but will appear, for butterflies, in a different color because of UV-light reflection, thus outstanding out from the background. Term “private channel” (Silberglied, 1984) for UV light, therefore, seems to be quite adequate for this situation, though recent investigations have been revealing UV perception in some birds (Bennet and Cuthill, 1994).
UV reflection by butterflies is known to be produced in two ways; absence of UV absorbing pigments, and “interference” based on thin-layers. In Pieris rapae, a pigment, leucopterine, is shown to be one of absorbents of UV, especially rich in male wings that reflect least UV light (Makino et al., 1952). Contrary to this, UV reflection in Chrysozephyrus and some other butterflies is based on interference. In Chrysozephyrus butterflies, both of UV and visible lights are thought to be produced by interference, because both are directional. Interestingly, the wavelength of the visible light (530–550 nm) is approximately twice of that of the shorter UV peak (260–270 nm) (Figs. 3 and 7A), and thus, it is probable that the visible light is a secondary reflection by a structure that produces the shorter UV light. Contrary to Chrysozephyrus butterflies, some Coliadinae butterflies are known to produce visible yellow light by pigments whereas UV light by interference as seen in Eurema (Ghiradella et al., 1972) and Colias (Silberglied and Taylor, 1973).
Perception of UV light is known for insects belonging to Hymenoptera (Kühn, 1927) and Coleoptera (Weiss, 1943). According to observations by Weiss, the strongest response was induced by light of 365 nm and the second peak by blue-blue green light of 492 nm. For butterflies, UV light perception is also proved by behavioral examinations (Obara, 1970, Rutowski, 1977, 1981, Silberglied and Taylor, 1978) and by physiological analyses (Bernard and Remington, 1991, Arikawa et al. 1987). In these studies, near UV, or ultraviolet A covering 315–400 nm (Silbergliede, 1979), is shown to be effective. In the case of Chrysozephyrus butter-flies, two peaks were detected in the UV range; one around 250 nm and the other around 320 nm. Only the 320 nm peak or its longer side shoulder could be perceived by these butterflies.
In the present study, the reflection pattern was found to be slightly different among species. Difference in UV reflection is known among relative species or subspecies; Colias (Silberglied and Taylor, 1973), Phoebis (Silberglied, 1979) and Pieris (Roland, 1978, Obara and Majerus, 2000). In the case of Chrysozephyrus butterflies, the difference is small. C. hisamatsusanus and C. ataxus reflected UV light more strongly than green light, whereas the pattern was reversed in C. smaragdinus and C. brillantinus (Table 1). Based on cluster analysis, the former two species and the latter two species are respectively categorized in different groups (Fig. 8). Though this does not always suggest a systematic relationship, it is interesting to know that the two groups differ from each other to some extent in ecological aspects. Species in the former group live in laural forests in the southern parts of Japan with the northern limit at Niigata and Kanagawa Prefecture, whereas those in the latter group live in deciduous broad-leaved forests over most of Japan including Hokkaido and Kyushu (Fukuda et al., 1984). Color patterns in butterflies should be considered from ecological viewpoints such as conspecific signaling and/or predator avoidance (Darwin, 1909), on the basis of systematic constraint inferred from morphological (Shirôzu and Yamamoto, 1956) and DNA analyses. The latter study is in progress (Saigusa and Odagiri, 2000).
We thank Hiroyuki Okazaki in Tokyo and Shin Goto of the Nature Center, Tanabe City, for providing specimens difficult to obtain, and Manabu Kajita of our laboratory for information of the UV-perception in birds.