How Glasswing Butterflies See Through Their Own Wings: The Strange Optics of Biological Transparency

Transparency is rare in animals, and for good physical reasons. Most biological tissue contains structures that scatter light: cell walls, lipid bilayers, protein crystals, mitochondria. Each interface between materials with different refractive indices reflects a fraction of the light passing through it, and biological tissue has thousands of such interfaces per millimeter. Even tissue with no pigment scatters light strongly enough to appear white (the way milk does, despite containing no opaque molecules) because there are too many interfaces for the light to pass through without reflection.

In water, this problem is solvable: aquatic animals have tissue refractive indices close to that of seawater (about 1.34 to 1.36), so the differential at the outer surface is small and many internal structures match the medium closely enough to disappear. The deep ocean is full of transparent animals: jellyfish, ctenophores, salps, larval fish, glass octopuses, almost all of pelagic copepod diversity. Transparency in water is so common it is the default for an entire trophic layer.

In air, transparency is much harder because the refractive index of air is essentially 1.0 and biological tissue is about 1.5. Every surface produces a strong reflection, and the cumulative effect makes air-living tissue inevitably visible. There are only a handful of terrestrial animals with substantially transparent body parts: glass frogs (whose ventral skin is partly transparent showing internal organs), some shrimp larvae, the wings of certain butterflies and dragonflies. The glasswing butterfly Greta oto is the most studied case, and the optics turned out to be more sophisticated than half a century of casual observation suggested.

The 2015 nanostructure discovery

Until the mid-2010s, the assumption among entomologists was that glasswing wings were transparent because the wing membrane was thin and the chitinous scales had degenerated to a sparse covering. Doris Gomez and colleagues at CNRS in Montpellier did the first thorough optical measurements in the early 2010s and found that the wing transmittance was about 90%, with reflectance below 2% across the visible spectrum. This was already surprising: a flat chitin surface in air should reflect about 5-7% from each face, for total reflection of 10-13%. The wing was somehow suppressing reflection well below what a smooth surface would produce.

The 2015 Siddique-Gomes-Lobo paper in Nature Communications used scanning electron microscopy to look at the transparent regions of Greta oto wings at high resolution and found a randomly-distributed array of nanostructures: pillar-like protrusions about 100-300 nanometers tall and 100-200 nanometers wide, packed at a density of about 3 per square micrometer. The pillars were not regular like a crystal lattice but instead arranged with significant disorder in both position and size.

This is the optical mechanism: when light passes through a graded refractive-index layer (one whose effective index varies smoothly from air to chitin rather than jumping abruptly at a surface), the reflection at the interface is suppressed. The nanostructures create an effective graded-index layer because at the wing surface itself the structures are tall and sparse (low effective index, mostly air), and deeper in they merge into solid chitin (high index). Light moving from one to the other does not encounter a single sharp interface but instead a gradual transition that produces no significant reflection.

This is essentially the same mechanism used in moth-eye anti-reflective coatings, which were developed for optical applications in the 1970s based on observations of nocturnal moth eyes. The moth-eye pattern uses regular sub-wavelength nanostructures arranged on a hexagonal lattice. The glasswing pattern uses irregular sub-wavelength nanostructures arranged randomly. Both work, and the regular pattern produces slightly lower reflection at any single wavelength while the irregular pattern produces lower average reflection across the broad visible spectrum.

Why disorder is the better strategy

The randomness turned out to be functionally important. A regular array of nanostructures has constructive-and-destructive interference patterns that vary with wavelength and angle, producing iridescence at oblique viewing angles. Regular moth-eye nanopatterns appear slightly colored when viewed from the side. A random arrangement averages over these interference effects and produces uniform low reflection at all angles and all wavelengths. For an animal whose entire survival strategy depends on being invisible from any direction, the angle-independence of disorder is the right optimization.

The biological challenge of producing irregular nanostructures is also lower than producing regular ones. Each scale-precursor cell on the wing produces its nanostructures by a self-organizing process during pupation, and the irregularity is just the natural consequence of biological variability between cells. A regular nanopattern would require coordination between cells that biology generally does not bother with when the function does not require it. The two-billion-year history of optimization in butterflies arrived at the random pattern partly because it is the better optical solution and partly because it is the easier developmental program.

The hydrophobic side benefit

The nanostructured surface has a side benefit that is not optical: it is strongly hydrophobic. Water droplets bead up and roll off rather than wetting the wing. This is the same effect that makes lotus leaves self-cleaning (the "lotus effect"), and it is mechanically inevitable: sub-wavelength roughness traps a layer of air at the surface and prevents water from achieving good contact with the chitin. For an animal flying through rain forest canopy with frequent rain, this is non-trivial: a wing covered in water droplets is heavy, light-scattering, and infection-prone.

The same nanostructures provide both transparency and self-cleaning. This is a common pattern in biological design where one structural choice solves multiple problems, and it makes biomimetic translation harder because the engineering target needs to capture both properties simultaneously to fully imitate the biology.

Biomimetic applications

The glasswing pattern has been a target for biomimetic anti-reflective coatings for displays, solar panels, and optical instruments. The advantages over the moth-eye pattern are the broader spectral range and the angle-independence. The challenge is that producing a controlled random nanostructure at scale is harder than producing a controlled regular one. Photolithographic techniques are good at making periodic patterns and poor at making controlled disorder.

Recent fabrication approaches have used colloidal self-assembly (allowing nanoparticles to settle into a random arrangement) and reactive ion etching of polymer substrates with intentionally varied etch rates. The Bing-Liu lab at Caltech demonstrated a glasswing-inspired solar cell coating in 2017 with reflection below 0.5% across the visible spectrum and across angles up to 60 degrees, compared to about 4% for the same cell with conventional anti-reflective coating. Production-scale fabrication remains expensive and the technique has not yet displaced moth-eye coatings in commercial applications, but the laboratory results are encouraging enough that the pattern continues to be researched.

The wider context

Biological anti-reflection is not unique to glasswing butterflies. The transparent wings of certain damselflies and lacewings have similar but less-studied nanostructures. The cornea of nocturnal insects (the moth-eye case) and the lens of some deep-sea fish use related strategies. The leg scales of certain Asian beetles produce structural color with intentional iridescence using the same kinds of nanostructures arranged in different geometries.

The unifying observation is that biology has been doing nano-optical engineering for hundreds of millions of years, generally arriving at solutions that match human engineering only after parallel investigation that began in the 1960s and is still ongoing. The glasswing butterfly evolved its current wing transparency about 30 million years ago, judging by molecular phylogenetics of the Ithomiine butterflies; the moth-eye nanostructures are at least 100 million years old based on fossil evidence; the deep-sea fish lens optimizations may be even older. The lab work of the 2010s rediscovered a fraction of what selection has already explored.

The deeper observation

The notion that biology and engineering converge on the same solutions because the physics constrains both is comforting and not quite right. The closer truth is that biology has explored the solution space far more thoroughly than human engineering has, often arriving at non-obvious solutions that human engineers would not have invented from first principles. The glasswing butterfly's random-nanostructure anti-reflection is a case where the obvious engineering answer (a regular array) is the slightly worse solution and biology found the better one by following its developmental constraints rather than its optimization gradient. Looking carefully at biological structures continues to be one of the most productive ways for human engineering to learn about its own assumptions.