The Paradise whiptail (Pentapodus paradiseus) has distinct reflective stripes on its head and body. The reflective stripes contain a dense layer of physiologically active iridophores, which act as multilayer reflectors. The wavelengths reflected by these stripes can change from blue to red in 0.25 s. Transmission electron microscopy revealed that the iridophore cells contain plates that are, on average, 51.4 nm thick. This thickness produces a stack, which acts as an ideal quarter-wavelength multilayer reflector (equal optical thickness of plates and spaces) in the blue, but not the red, region of the spectrum. When skin preparations were placed into hyposmotic physiological saline, the peak wavelength of the reflected light shifted towards the longer (red) end of the visible spectrum. Hyperosmotic saline reversed this effect and shifted the peak wavelength towards shorter (blue/UV) wavelengths. Norepinephrine (100 μmol l-1) shifted the peak wavelength towards the longer end of the spectrum, while adenosine (100μ mol l-1) reversed the effects of norepinephrine. The results from this study show that the wavelength changes are elicited by a change of∼ 70 nm in the distance between adjacent plates in the iridophore cells.
- fish reflector
- paradise whiptail
- multilayer reflector
- rapid colour change
- colour vision
The paradise whiptail, Pentapodus paradiseus (body length up to 30 cm; Fig. 1A), is a commonly found member of the bream family, inhabiting coastal waters of Queensland, Australia. It has a characteristic thread-like filament on the upper lobe of the caudal fin, which gives the fish its name. Paradise whiptails are mostly found over sandy bottoms near coral reefs and are commonly used as bait by fishermen, as they are often caught by line in large numbers. We could find no accounts in the literature regarding the social behaviour of these fish; however, we have observed them in loosely packed schools or groups.
Paradise whiptails have very distinct reflective stripes on their head and body (Fig. 1A). When we first observed these fish in tanks, we noticed that the colour of these stripes changes from blue to red within less than one second. These changes are unusually fast for fish iridophores, which have been reported to take several seconds or minutes (Kasukawa et al., 1987; Lythgoe and Shand, 1982).
Iridophores, both `active' and `passive', are common in many fish, and their likely functions in camouflage and communication (e.g. schooling and mating) have been discussed by several authors (see e.g. Denton, 1970; Denton and Rowe, 1994; Fujii et al., 1989; Herring, 1994). The blue damselfish (Chrysiptera cyanea), for example, normally displays a characteristic blue colouration, which is produced mainly by iridophores in the skin (Kasukawa et al., 1987). During stressful conditions, the fish changes its hue rapidly to ultraviolet (UV), seen as black by humans. These changes have been associated with simultaneous changes in the distance between adjoining reflecting plates that make up these iridophores (Kasukawa et al., 1986, 1987; Oshima and Fujii, 1987). Active iridophores have also been reported in the blue-green damselfish (Chromis viridis; Fujii et al., 1989), the common surgeonfish (Paracanthurus hepatus; Goda and Fujii, 1998), the neon tetra (Paracheirodon innesi; Lythgoe and Shand, 1982; Nagaishi and Oshima, 1989), the dark sleeper (Odontobutis obscura; Fujii et al., 1991) and the domino damsel (Dascillus trimaculatus; Goda and Fujii, 2001).
Many fish iridophores have been shown to be multilayer reflectors. Light reflected from a reflector of this kind is almost always coloured. Multilayer reflectors are characterised by the fact that they contain thin plates of a higher refractive index than the spaces separating them. In fish, it is assumed that the plates are made of guanine, which has a refractive index (n) of 1.83, while the spaces are believed to be made of cytoplasm, which has n=1.33 (Land, 1972). In an ideal multilayer reflector, the plates and spaces both have an optical thickness (actual thickness multiplied by refractive index) of a quarter of the wavelength reflected by the stack at normal incidence. `Ideal' here means that such a stack has the highest reflectivity in comparison with `non-ideal' reflectors, for which the plates and spaces differ in optical thickness. It thus becomes obvious that by varying the thickness of the plates and/or spaces, the wavelengths reflected from the stack can be changed. Another characteristic feature of these types of reflectors is that, as the angle of the incident light becomes more oblique, the reflected light shifts towards the shorter (blue/UV) end of the spectrum (Huxley, 1968; Land, 1972).
It seems quite clear from the evidence presented by Oshima and Fujii (1987) and Kasukawa et al. (1987) that the reflective changes in the blue damselfish are brought about by a change in the distance between the plates that are contained within the reflective cells of the skin. Furthermore, Oshima and Fujii (1987) showed that microtubular structures connect adjacent plates and that their assembly and disassembly may provide the mechanism for the reflective changes in the blue damselfish. Kasukawa et al. (1987) showed that the reflective changes are controlled by the sympathetic nervous system. Whilst noradrenaline (norepinephrine) causes the reflected wavelengths to change to the longer end of the spectrum, adenosine causes the reverse effect.
In the present study, we report reflective changes in the paradise whiptail and show that this unusually fast colour change is mediated by a change in the spacing between the plates of multilayer reflectors.
Materials and methods
Paradise whiptail (Pentapodus paradiseus Günther 1858) were linefished off the Moreton Bay Research Station on Stradbroke Island and transferred to holding tanks at the laboratory (UQ AEC number Phys/Ph/289/02/VTHRC). Fish were observed and video-recorded using a Sony digital camcorder (DCR-TRV238E). Images were grabbed at 25 frames s-1.
Spectral reflectivity measurements
For measurements of spectral reflectivity changes in a living paradise whiptail, one fish was placed in a small glass tank. Spectral reflectance measurements (300-800 nm) were obtained using a fibre optic spectrometer (S-2000; Ocean Optics Inc., Dunedin, FL, USA) and a pulsed xenon light source (220-750 nm; PX-2; Ocean Optics). The flash rate (up to 220 Hz) was controlled by a laptop running the S-2000 and was synchronised with the detector. A bifurcated fibre optic cable (1 mm diameter) provided illumination with 50% of its fibres, while the other 50% acted as detectors of the reflected light. A `Spectralon' white tablet was used as a 99% reflection standard. All measurements were made in a dark room to prevent the influence of stray light. The fibre optic cable was held by a small stage, enabling the angle of incidence to be determined. Measurements were made at approximately normal incidence to the skin surface of the reflective stripes on the head.
The same methods were used on skin preparation to investigate the effects of osmotic changes and drugs on the wavelengths of the reflected light. For measurements at oblique angles of incidence, two fibre optic cables were used (illumination, 200 μm; detector, 100 μm). The fibres were held by a small stage, and the angle of incidence was measured using a protractor.
Transmission electron microscopy
One paradise whiptail was killed by an overdose of clove oil (UQ AEC number Phys/Ph/289/02/VTHRC; Munday and Wilson, 1997) and fixed in 3.5% paraformaldehyde and 0.25% glutaraldehyde in 0.1 mol l-1 phosphate buffer. The reflective stripes were in a blue-reflective phase (resting phase) after fixation. Small pieces of skin containing the reflective areas were cut out and post-fixed in 2% osmium tetraoxide in 0.1 mol l-1 phosphate buffer. Specimens were washed and dehydrated in a graded series of acetone and embedded in Spurr's resin (Spurr, 1969). Sections were cut on a Reichert Jung Ultracut microtome and viewed on a JEOL JEM 1010 transmission electron microscope.
Perfusion of drugs and osmotic effects
Three paradise whiptails were killed by an overdose of clove oil. The tissue on the head and caudal peduncle containing the reflective stripes was cut out using a scalpel blade and placed in a petri dish. The tissue was equilibrated in a physiological saline, made according to Kasukawa et al. (1987): 125.3 mmol l-1 NaCl, 2.7 mmol l-1 KCl, 1.8 mmol l-1 CaCl2, 1.8 mmol l-1 MgCl2, 5.6 mmol l-1 d-glucose, 5.0 mmol l-1 Tris-HCl buffer, pH 7.2 (osmolality: 272 mOsm kg-1). The osmotic effects on the reflective properties were tested by making saline solutions of different osmotic strengths. A hyposmotic solution was made by diluting the saline in ultrapure water to concentrations varying from 10% to 75% (with osmolalities varying from 28 mOsm kg-1 to 194 mOsm kg-1, respectively). A hyperosmotic solution was made by adding 200 mol l-1 sucrose to the saline (460 mOsm kg-1). Noradrenaline (norepinephrine) and adenosine (Sigma; 100 mmol l-1) were diluted in the saline before experiments. The volume of the dish (approximately 10 ml) was then replaced with the experimental solution, and the reflective changes were recorded with the spectrometer (S-2000, Ocean Optics) described earlier.
Fig. 1A shows a photograph of a paradise whiptail, displaying a blue colouration of its reflective stripes. There are three reflective stripes on the head, of which the bottom stripe extends to the gill cover. Furthermore, there is one stripe running along each side of the body, as well as one stripe on each side of the dorsal fin. There appears to be no appreciable difference between males and females in the distribution of the stripes and the properties of their colour change.
Cycles of colour change
Fig. 1B shows a typical cycle of colour changes observed in the nose stripes of a paradise whiptail. Each reflective cycle can be divided into four phases: (1) `resting phase', which is dull blue (Fig. 1B, panel a), (2) `green flash phase' (Fig. 1B, panel b), (3) `red phase' (Fig. 1B, panel c) and (4) `recovery phase', during which the peak wavelength shifts from red through yellow and green back to blue (resting phase; Fig. 1B, panels d-f). The cycle shown in Fig. 1B can occur continuously for several minutes, although more commonly an interval from seconds to minutes may separate successive cycles. The change from the resting phase to the red phase (Fig. 1B, panels a-c) is accomplished within approximately 0.25 s. The reflective stripes may then remain in the red phase (Fig. 1B, panel c) for several seconds (typically 3 s). The recovery phase from red through green to blue (Fig. 1B,panels c-f) may last a few seconds (typically around 2 s), after which another cycle may occur. The initiation of this reflective cycle as well as the duration of any of its phases is entirely under the control of the animal. The colour change appears to occur nearly instantaneously across all reflective stripes of the body.
The body colouration also changes during a reflective cycle, becoming redder as the reflective stripes change from blue to red. This is, however, very weak compared with the change of the reflective stripes. The body colouration is probably caused by iridophores and pigment-containing chromatophores.
Spectral changes in living fish
The spectral measurements of the reflective changes shown by living fish are shown in Fig. 2A. In this figure, spectral curves are shown for all wavelengths reflected by these stripes during one cycle. It can be seen that the reflected light undergoes a change from blue (465 nm) through the green and yellow parts of the spectrum to red (650 nm).
Fig. 2B shows the ratio of reflectance half-width to peak wavelength [Δλ/λmax, whereΔ λ is the bandwidth at 50% reflectivity andλ max is the peak wavelength (Land, 1972)]. The reflectance half-widths increase with wavelength (see Fig. 2A). The regression line in Fig. 2B shows that, as a fraction of the peak wavelength, they increase slightly. This increase was significantly different from zero (ANOVA F1,63=8.83, P=0.0042). This is in contrast to what we expected to find. The predicted reflectance half-widths are also shown in Fig. 2B (blue line; calculated from Land, 1972). The possible significance of this is outlined in the Discussion.
Changes in reflected wavelength with angle of incidence
Fig. 2C shows the effect of varying the angle of the incident light source on the peak wavelength of the reflected light. Here, we show measurements of a group of iridophores reflecting yellow-green light at normal incidence. It can be seen that as the angle of the incident light becomes more oblique, the wavelength of the light reflected from the same group of iridophores moves towards the shorter (blue/UV) end of the spectrum. This behaviour is typical of multilayer interference reflectors.
Transmission electron microscopy (TEM)
In Fig. 3, we show electron micrographs of the skin from the reflective stripes on the head. Sections were cut perpendicular to the skin surface. Fig. 3A is a low-power micrograph that shows the arrangement of the iridophores and chromatophores in the skin of these fish. It can be seen that the plates contained within an iridophore are arranged circularly around the pigment-containing chromatophores. Note that the plates have broken away during the process of sectioning and that the white areas are those that normally contain the plates. Some of the areas that contained the plates have expanded considerably during examination with the electron microscope. Only specimens preserved as well as those shown in Fig. 3B,C,showing iridophores with intact plates, were used to measure plate and space thicknesses.
The plates have a mean thickness of 51.4±3.5 nm (mean ± s.e.m.; N=24), while the spaces varied from 26.3 nm to163.6 nm (mean=83.5±11.64 nm; N=9) (see Fig. 3B,C). The number of plates making up a stack within an iridophore varied between individual cells. On average, we counted 9.7 plates (N=31).
The results of the physiological experiments are shown in Fig. 4. Fig. 4A shows that reducing the osmotic pressure of the external solution by diluting the saline resulted in a wavelength shift to the longer (red) end of the spectrum. This wavelength shift could be reversed by returning the preparation to 100% saline (Fig. 4B). Increasing the osmotic pressure by adding 200 mmol l-1 sucrose to the saline shifted the reflected wavelengths to the shorter end of the spectrum by a further 50 nm (Fig. 4B). These results are consistent with the swelling of the spaces between the plates in hyposmotic solutions and shrinkage in hyperosmotic solutions.
The effects of norepinephrine and adenosine are shown in Fig. 4C,D. From Fig. 4C, it can be seen that topical applications of 100 μmol l-1 norepinephrine resulted in a shift in the peak wavelength to the longer (red) end of the spectrum, whilst applications of adenosine (100 μmol l-1) reversed the reflected peak to the shorter (blue) end of the spectrum (Fig. 4D). It took approximately 20 min before the wavelength shifts (blue to red, as well as red to blue) could be observed. Washing in saline following norepinephrine application resulted in the same spectral shift as during adenosine applications. However, it took approximately twice as long for the blue reflection to return.
In this paper, we report reflective colour changes in a tropical fish, the paradise whiptail Pentapodus paradiseus. This fish has very characteristic stripes on its head and body and the speed with which it changes the wavelengths reflected from these stripes is remarkable.
The evidence presented here is in strong support of the theory that the reflectors that make up these stripes act as multilayer reflectors. With increasing angle of incidence, the reflected wavelengths shifted towards the shorter (blue/UV) end of the spectrum, a feature typical of multilayer reflectors. As has been shown by Huxley (1968) and Land (1972), the light reflected from reflectors of this kind is invariably coloured if the thickness of the plates and the spaces separating them are comparable with the wavelength of light. In the ideal configuration, for which highest reflectivity is achieved for the minimum number of plates in the stack, each space and plate has an optical thickness of a quarter of the wavelength reflected by the stack at normal incidence.
Structural inferences from optical measurements
The change in peak reflectance during a colour change cycle, from 465 nm to 650 nm, in conjunction with the mean plate thickness of 51.4 nm, can tell us the extent of the change in the spacing of the plates (we assume that only the spacing changes, not the plate thickness). For a given peak wavelength, the sum of the optical thicknesses (actual thickness × refractive index) of each plate and space must equal half a wavelength, i.e. 232.5 nm for the 465 nm maximum and 325 nm for the 650 nm maximum. If we assume that the plates are made of guanine (n=1.83), as is common in fish, then their optical thickness is 51.4×1.83=94.1 nm. This means that the optical thickness of the spaces must increase from 138.4 nm (i.e. 232.5-94.1 nm) to 230.9 nm (i.e. 325-94.1 nm). Assuming the spaces are made of cytoplasm (n≈1.33), the change in the actual thickness of the spaces is from 103.3 nm to 172.3 nm, an increase of 67%.
Another calculation that can be made is how closely the stack approaches the `ideal' situation where n1t1=n2t2=λ/4 (where n is the refractive index and t is the actual thickness; n1t1 refers to the plates and n2t2 to the spaces). An ideal reflector has the highest reflectance for a given number of plates and also the widest reflected spectral waveband. For an ideal multilayer n1t1/(n1t1+n2t2)=0.5, and lower values indicate a departure from this condition. When the whiptail stack reflects blue (465 nm), n1t1/(n1t1+n2t2)=94.1/232.5, i.e. 0.405, which is close to ideal. When the stack reflects red light, however, n1t1/(n1t1+n2t2)=94.1/325, i.e. 0.290, which is non-ideal. For a guanine-cytoplasm multilayer with 10 plates, these figures predict that the half-width of the reflected waveband should be 0.24λmax and 0.21λmax for blue light and red light, respectively (fig. 7b from Land, 1972), whereas half-widths measured from Fig. 2A are both narrower than these values: 0.185λmax and 0.194λmax, respectively (see Fig. 2B). Possible explanations are that the measured curves are truncated on the short-wavelength side by the red screening pigment of the chromatophores or that the plates are not actually made of guanine but rather a substance of lower refractive index. The fracturing of the plates during preparation for electron microscopy suggests that they are crystalline, so the screening explanation is the more likely.
The shift in λmax towards the blue with increasing angle of incidence is consistent with the behaviour of multilayers. However, the shift (Fig. 2C) is less than expected. If λmax (0° incidence) is 515 nm, then for a 10-plate guanine-cytoplasm multilayer one would expect a shift to 420 nm (40°) and 360 nm (55°) (fig. 6 from Land, 1972). The measured values were 450 nm and 415 nm, respectively. The most likely explanation here is that the overall layout of the stacks in the iridophores is not flat but curved, so that even at high angles the incident beam meets some stacks at close to normal incidence. Another possible explanation is that the refractive index difference between the plates and the spaces is actually higher than guanine and cytoplasm would provide (i.e. the opposite of the change needed to fit the half-width data above). The curvature explanation is quite consistent with the known anatomy.
Osmotic and pharmacological changes
The data presented here suggest that the wavelength changes are elicited by a change in the distance between adjacent iridophore plates. This is supported by the finding that altering the osmolarity of the external solution resulted in a change in the reflected wavelength. Hyposmotic saline shifted the reflected wavelengths towards the longer (red) end of the spectrum, presumably caused by water diffusing into the cells and pushing the iridophore plates further apart. Removing intracellular water by returning the preparations to 100% saline, or even adding hyperosmotic saline, shifted the reflected wavelengths towards the shorter (blue/UV) end of the spectrum.
Some simple physiological experiments were conducted to investigate mechanisms of control of the reflective changes. We found that the reflective changes in the paradise whiptail are under the control of the sympathetic nervous system, which is in agreement with the findings of Kasukawa et al. (1986, 1987), Fujii et al. (1989) and Nagaishi and Oshima (1989) on other fish species. Applications of norepinephrine shifted the reflections of blue-reflecting iridophores towards the longer (red) end of the spectrum, whilst washing in saline reversed this effect, shifting the reflections back towards the blue parts. The shift from red to blue was speeded up by adenosine. In our study, we have not, however, investigated the controlling mechanisms in any more detail. Oshima and Fuji (1987) showed that in the blue damselfish the mechanism of reflective changes is probably based on a microtubular system.
In comparison with the studies of Kasukawa et al. (1986), we used very high concentrations of norepinephrine. The reflective stripes of the paradise whiptail were difficult to dissect intact and, as a consequence, we had to cut thick slices of muscle and cartilage containing the reflective stripes. Therefore, we had to increase the concentrations of norepinephrine in order to ensure that the drug diffused into the tissue.
When looking at the wavelengths of the different phases of a colour change cycle it becomes obvious that the green phases ('green flash' and green during `recovery phase') may be the most visible reflections of the entire cycle. Light in the sea is absorbed preferentially with depth, the blue-green spectral region (around 500 nm) being transmitted best, while the red, deep blue and UV parts are absorbed in the first 5-10 m of the water column (Jerlov, 1976; Tyler and Smith, 1970). The wavelengths of light best transmitted in coastal waters of Moreton Bay, where paradise whiptails occur, are those of the green parts of the spectrum (N.J.M., unpublished). There are a number of accounts in the literature presenting evidence for a strong correlation between spectral absorption of cone visual pigments of fish and the spectral composition of their environments (Levine and MacNicol, 1979). Many coastal fish, for example, have visual pigments absorbing in the green and yellow regions of the spectrum, compared with many off-shore species that have spectral sensitivities in the green and blue parts (Lythgoe et al., 1994). The visual pigments of paradise whiptails have, to our knowledge, not been studied to date; however, as coastal fish, their spectral sensitivities may be expected to be best in the longer-wavelength regions of the spectrum. Therefore, the contrast between the red-brown body colour and the green reflections may appear stronger to them than the contrast between the body colour and the other reflected wavelengths, which may make the green reflections much easier to detect.
It is remarkable to observe the rapidity with which the paradise whiptail can change the colour of its reflective stripes. This colour change is much faster than that reported for other fish (see e.g. Kasukawa et al., 1987; Lythgoe and Shand, 1982); however, it appears to be controlled by the same mechanisms (see e.g. Kasukawa et al., 1986). To date, we can only speculate about the functions of these colour changes. We observed that the paradise whiptail changes the reflections from blue to red, especially when excited by external stimuli, and so it appears likely that colour change in this fish plays some role in communication.
We would like to thank the Staff at the Moreton Bay Research Station (The University of Queensland) for accommodation and facilities for this work, and Maurizio Bigazzi for showing us where to catch paradise whiptail. L.M.M. is funded by a Post-Doctoral Fellowship from the Royal Society. M.F.L. would like to thank the Royal Society for a grant for travel to Australia. U.E.S. and N.J.M. are funded by the ARC.
- © The Company of Biologists Limited 2003