on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

One contributhan meets th

*Author andEvolutionaryCT 06520, US†Present addSchool, Kans

Received 28 NAccepted 6 Ja

Development of colour-producing b-keratinnanostructures in avian feather barbs

Richard O. Prum1,2,*, Eric R. Dufresne3,4,5, Tim Quinn6,† and Karla Waters6

1Department of Ecology and Evolutionary Biology, 2Peabody Natural History Museum,3Department of Mechanical Engineering, 4Department of Chemical Engineering, and

5Department of Physics, Yale University, New Haven, CT 06511, USA6Department of Ecology and Evolutionary Biology, University of Kansas, Lawrence,

KS 66045, USA

The non-iridescent structural colours of avian feather barbs are produced by coherent lightscattering from amorphous (i.e. quasi-ordered) nanostructures of b-keratin and air in themedullary cells of feather barb rami. Known barb nanostructures belong to two distinctmorphological classes. ‘Channel’ nanostructures consist of b-keratin bars and air channels ofelongate, tortuous and twisting forms. ‘Spherical’ nanostructures consist of highly sphericalair cavities that are surrounded by thin b-keratin bars and sometimes interconnected by tinypassages. Using transmission electron microscopy, we observe that the colour-producingchannel-type nanostructures of medullary b-keratin in feathers of the blue-and-yellowmacaw(Ara ararauna, Psittacidae) develop by intracellular self-assembly; the process proceeds inthe absence of any biological prepattern created by the cell membrane, endoplasmicreticulum or cellular intermediate filaments. We examine the hypothesis that the shape andsize of these self-assembled, intracellular nanostructures are determined by phase separationof b-keratin protein from the cytoplasm of the cell. The shapes of a broad sample of colour-producing channel-type nanostructures from nine avian species are very similar to those self-assembled during the phase separation of an unstable mixture, a process called spinodaldecomposition (SD). In contrast, the shapes of a sample of spherical-type nanostructuresfrom feather barbs of six species show a poor match to SD. However, spherical nanostructuresshow a strong morphological similarity to morphologies produced by phase separation of ametastable mixture, called nucleation and growth. We propose that colour-producing,intracellular, spongy medullary b-keratin nanostructures develop their characteristic sizesand shapes by phase separation during protein polymerization.We discuss the possible role ofcapillary flow through drying of medullary cells in the development of the hollow morphologyof typical and spongy feather medullary cells.

Keywords: structural colour; self-assembly; b-keratin; coherent scattering;phase separation; spinodal decomposition

1. INTRODUCTION

Most organismal structural colours are produced bycoherent scattering, or interference, of ambient light bynanostructures with periodic spatial variation inrefractive index (Prum & Torres 2003). Here, coherentscattering means that the phase relationships amongthe scattered light waves determine the scattered lightspectrum (Prum & Torres 2003). Coherent scatteringincludes constructive interference, diffraction, Braggscattering and thin-film scattering. Coherent scattering

tion of 13 to a Theme Supplement ‘Iridescence: moree eye’.

address for correspondence: Department of Ecology andBiology, Yale University, PO Box 208015, New Haven,A (richard.prum@yale.edu).ress: University of Missouri at Kansas City Medicalas City, MO 64108, USA.

ovember 2008nuary 2009 S253

can be contrasted with incoherent scattering—such asRayleigh, Tyndall and Mie scattering—in which thephase relationships among the scattered waves arerandom (Prum & Torres 2003).

Although the great diversity of organismal nanostruc-tures shares this common physical basis in dielectricspatial periodicity (Prum & Torres 2003; Vukusic &Sambles 2003; Prum et al. 2004), the nanostructures ofdifferent organisms vary tremendously in the organiz-ation and composition of this periodicity (Parker 1999;Srinivasarao 1999; Vukusic & Sambles 2003).

How do these colour-producing nanostructuresdevelop? What physical and biological mechanisms doorganisms use to construct arrays of quasi-ordered andhighly periodic nanostructures? Surprisingly, thesefundamental questions have been little studied.Research on the development of colour-producingbiological nanostructures is critical for understanding

J. R. Soc. Interface (2009) 6, S253–S265

doi:10.1098/rsif.2008.0466.focus

Published online 23 February 2009

This journal is q 2009 The Royal Society

S254 Development of colour-producing b-keratin R. O. Prum et al.

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

condition dependence in structurally coloured com-munication signals (Prum 2006). Condition dependenceis predicted by some mechanisms of the evolution ofsignals used in social and intersexual communication(Andersson 1994).

A few studies indicate that colour-producingbiological nanostructures develop by either complexmechanisms of cellular physiology and growth or moresimple physical mechanisms of nanoscale self-assembly.A ground-breaking study by Ghiradella (1989) showedthat the bicontinuous, colour-producing, air–chitinnanostructures in the green wing scales of the butterflyMitoura gryneus develop from a prepattern formed bythe complex, reticulate growth of smooth endoplasmicreticulum and cellular membrane into the central bodyof each balloon-shaped, wing-scale cell. Followingestablishment of the membrane prepattern, extra-cellular chitin is deposited into the lumen of thesecomplex structures but outside the cell to create theoptical nanostructure. After cell death, the remainingcytoplasmic spaces between the chitin become filledwith air to create the refractive index difference thatproduces light scattering. Thus, internal nanostruc-tures in insect scales illustrate some of the subtledistinctions between nanostructural self-assembly andcellular physiological assembly. Although chitin is aprotein polymer formed by self-assembly of extra-cellular fibrils (Neville 1975, 1993), the nanoscalestructure of chitin inside butterfly and weevil scales iscreated by a biological prepattern of membranes, whichis probably determined by complex cellular physiologyand growth, rather than self-assembly.

Structurally coloured butterfly scales are extremelyvariable, and not all optical butterfly nanostructuresdevelop in the same way. Ghiradella (1974, 1989) hasalso shown that mechanical buckling of chitin isinvolved in the development of colour-producingsuperficial ridges of Colias butterfly wing scales.

Durrer & Villiger (1967) described the developmentof the hollow melanosomes that produce the structuralcolours in the feather barbules of a starling. The size ofthe central air cavity within each melanosome isdetermined by cellular mechanisms within melanocytesprior to transfer to the barbule keratinocyte.

In contrast to the highly regulated developmentalmechanisms in butterfly scales and hollow avianmelanosomes, Hemsley et al. (1994, 1996, 1998) havehypothesized that the exines of the cell wall spores andpollen grains of vascular plants consist of a self-assembled nanostructure of sporopollenin. Hemsleyet al. (1994, 1996, 1998) focus on Selaginella, in whichthe exines may form an iridescent, biological opal ofsporopollenin spheres in a hexagonal packing. In aseries of biomimetic experiments, Hemsley et al. (1994,1996, 1998) were able to simulate the development ofvery similar, complex nanostructures with the self-assembly of polymer colloids.

2. SPONGY MEDULLARY FEATHER KERATIN

The structural colours of avian feather barb rami areproduced by b-keratin nanostructures in the medullarycells of the barb. These quasi-ordered, or amorphous,

J. R. Soc. Interface (2009)

nanostructures function by coherently scattering lightfrom their air–b-keratin interfaces (Prum 2006; Prumet al. 1998, 1999). Colour-producing intracellularb-keratin nanostructures have independently evolvedmany times and are known from over 20 differentfamilies of birds (Prum 2006). These spongy air andb-keratin nanostructures come in two distinct classes(figure 1; Dyck 1976; Prum 2006). ‘Channel’ nano-structures consist of b-keratin bars and air channelswith elongate, tortuous and twisting forms(figure 1a,b). ‘Spherical’ nanostructures consist ofhighly uniform, spherical air cavities created by curvedb-keratin bars of various thicknesses and frequentlyinterconnected by tiny air passages (figure 1c,d ).

Here, we investigate the development of the colour-producing, spongy, channel-type nanostructure in theblue feather barb rami of the blue-and-yellow macaw(Ara ararauna, Psittacidae; figure 2) using transmissionelectron microscopy (TEM). Auber (1971/1972)observed the development of colour-producingmedullarybarb cells using light microscopy, but his method didnot allow him to describe the development of thecolour-producing nanostructure itself because it issmaller than the wavelengths of visible light.

Our observations support the conclusion that theseb-keratin nanostructures are self-assembled withinmedullary keratinocytes. To examine the mechanismof self-assembly further, we compared the shapes ofspongy medullary nanostructures from bird specieswith channel-type (nZ9) and spherical-type (nZ6)bird feathers with experimental data from a polymerblend undergoing phase separation by spinodal decom-position (SD; Takenaka & Hashimoto 1992).

3. SELF-ASSEMBLY BY PHASE SEPARATION

Self-assembly is a ubiquitous phenomenon wheremolecular interactions and thermal fluctuations guidethe formation of structures at the supermolecular,nano- and microscales (Jones 2002). We hypothesizethat the sphere and channel nanostructures in medul-lary cells self-assemble during the phase separation ofb-keratin from the remainder of the cytoplasm.

The stability of a mixture of two fluids depends onboth the interactions between individual molecules andthe concentrations of the components (Jones 2002).Entropy, embodied by random thermal fluctuations,typically favours the stability of a mixture. Repulsiveinteractions between dissimilar molecules favour theseparation of the two molecular species into twodistinct macroscopic phases. Vivid examples of thisphenomenon can be found in the kitchen. To thepleasure and frustration of many, ethanol and waterreadily form stable mixtures, while oil and water insalad dressing separate when they are not vigorouslymixed. The stability of a mixture can depend on therelative proportions of the fluids or other thermo-dynamic parameters, such as temperature.

The impact of the interplay between the entropy andmolecular interactions on the stability of a mixture isembodied in a phase diagram (figure 3; Jones 2002).The mixture is stable when the repulsive molecularinteractions between different molecules, measured by

(a) (c)

(b) (d)

(e) ( f )

120 nm

Figure 1. Transmission electron micrographs of colour-producing (a,b) channel-type and (c,d ) sphere-type nanostructures frommedullary cells of avian feather barb rami, and examples of phase separation by (e) spinodal decomposition (SD) and( f ) nucleation and growth. (a) Black-capped kingfisher Halcyon pileata (Alcenidae, KU48795); (b) Asian fairy-bluebird Irenapuella (Irenidae, KU101420); (c) blue-crowned manakin Lepidothrix coronata (Pipridae, KU87685); and (d ) red-leggedhoneycreeper Cyanerpes cyaneus (Thraupidae, KU88256). The dark-staining material is b-keratin and the unstained voidsare air. (e) Scanning electron microscopy image of nanoporous gold produced by SD of an Ag–Au alloy (from Erlebacher et al.2001). ( f ) Carbon dioxide bubbles produced by a nucleation and growth phase separation in beer. Magnifications: (a) 30 000!,(b) 30 000!, (c) 20 000!, and (d ) 25 000!. Scale bars: (a–d ) 500 nm.

Development of colour-producing b-keratin R. O. Prum et al. S255

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

the parameter c, are small compared with the randomenergy available from thermal fluctuations, measuredby kBT, where kB is Boltzmann’s constant and T is theabsolute temperature. Below a critical value of kBT/c,typically of the order of 1, the stability of a mixturedepends on its composition, the proportions of the twofluids. In figure 3, f indicates the volume fraction of onemolecule in the mixture. For example, fZ0 or 1 wouldbe a pure sample of molecule A or B, while fZ0.5 wouldbe a one-to-one mixture of A and B. While systemsconsisting almost entirely of one molecular species

J. R. Soc. Interface (2009)

remain stable, systems with intermediate compositionswill unmix or phase separate into two stable compo-sitions. Mixtures will be stable at all compositionsabove the phase-separation boundary (figure 3, solidline), but will phase separate below that line.

The mechanism by which phase separation proceedsdepends on the stability of the system to smallfluctuations in composition. When there is no acti-vation barrier to separation, a mixture is unstable andphase separation occurs through a relatively smoothand well-characterized process called SD. When an

(a)

(b) (c)c

a

sksk

Figure 2. (a) Adult blue-and-yellow macaw (A. ararauna) (photo by T. Laman, VIREO, with permission). (b) TEM image of asingle barb ramus showing the solid b-keratin cortex and the spongy medullary b-keratin. The rip in the section (a) is an artefact.(c) TEM image of the channel-type spongy medullary b-keratin. The solid line running through the image is the cell membraneboundary between two neighbouring medullary cells. Scale bars: (b) 2 mm and (c) 500 nm. a, sectioning artefact; c, cortex; sk,spongy medullary keratin.

S256 Development of colour-producing b-keratin R. O. Prum et al.

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

activation energy barrier stabilizes the system againstdemixing, a mixture is metastable and will not phaseseparate unless it is ‘seeded’ with an initial particlecomposed of the minority phase that is above a criticalsize (Jones 2002). This mechanism of phase separationis called nucleation and growth. Nucleation can occur byrandom thermal fluctuations or by interactions withedges or impurities (Jones 2002). In figure 3, the solidline, called the binodal line, marks the boundarybetween mixed and phase-separated systems. Justbelow the binodal, mixtures separate via nucleationand growth. Below the dashed spinodal line, systemsunmix by SD (figure 3).

Systems unmixing by SD display a characteristicsponge-like structure of interconnected channels of thetwo phases (figures 1e and 3). While the specificstructure in each region of a phase-separating sampleis different in every instance, the overall structure of SDis universal. The structures formed during nucleationand growth are more diverse. For simple fluids, theytypically consist of isolated spherical droplets of theminority phase that grow over time (figure 1 f ).The droplet size distribution depends strongly on thekinetics. If nucleation is fast and growth is slow,droplets are nearly identical or ‘monodisperse’, other-wise they can be of very different sizes or ‘polydisperse’.When the droplets are well separated, the overallstructure of the material is completely random, but, as

J. R. Soc. Interface (2009)

they get close to one another, the droplet positions canbecome correlated and even form periodic structures(Hemsley et al. 1994, 1996, 1998).

The morphologies produced by these two modes ofphase separation are strikingly similar to the channel andsphere classes of b-keratin nanostructures (figure 1e,f ).In this paper, we make a quantitative comparison ofavian nanostructures with experimental data on the SDof a polymer blend. Finding good correspondence, wehypothesize that the characteristic forms of thesebiological nanostructures develop by phase separation.

4. METHODS

4.1. Feather sampling and microscopy

We examined growing feather germs, or pin feathers,fromthebackand crownofanadultmaleblue-and-yellowmacaw (A. ararauna). Feather germs were 2–4 cm inlength at the time they were sampled. Each feather germwas fixed and stored in Karnovsky’s fixative (2.5% glu-taraldehydeand2.5%paraformaldehyde), andembeddedin plastic for transmission electron microscopy.

Pennaceous feather structure grows by a complexmechanism of helical growth within the tubular feathergerm (Lucas & Stettenheim 1972; Prum 1999; Prum &Williamson 2001). Feather growth involves threedifferent maturation gradients. The primary distal–proximal gradient involves maturation of feather cells

mixed

phaseseparated

N+G N+GSD

kBT

Figure 3. A schematic phase diagram showing the relationshipbetween the temperature (kBT/c), composition (f) and thestability of a mixture. kB is Boltzmann’s constant, T is theabsolute temperature and c is the strength of interactionsbetween dissimilar molecules. In feathers, f would representthe volume fraction of b-keratin. At high-temperatureconditions, a molecular mixture will be stable at anycomposition, but below the phase-separation boundary(solid line) the solution will phase separate. Mixtures belowthe spinodal line (dashed line) are unstable, and will phaseseparate by SD, which produces bicontinuous channels of eachmaterial (middle inset). Mixtures between the spinodal lineand the phase boundary are metastable, and will only phaseseparate through nucleation and growth (NCG), whichproduces spherical shapes (left and right insets).

Development of colour-producing b-keratin R. O. Prum et al. S257

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

as they emerge from the feather follicle out of the skin.Distal cells are older and more mature than proximalcells.Within a single horizontal section of a feather germ,a second ventral–dorsal gradient involves increasingmaturation of barb ridge cells as barb ridges growhelically around the tubular feather germ, towards thedorsal surface of the tubular feather germ, to fuse to therachis. Third, within a single barb ridge, cell maturationprogresses along a superficial–internal gradient; moresuperficial barbule plate cells mature before the moreinternal, or basal, cells of the barb ramus, which is theshaft of the barb.

Because we cannot follow individual cells as theymature, we documented barb ramus cell maturationand keratin self-assembly by tracking homologousmedullary cells within multiple barb ridges along thevarious maturation gradients. Thus, thin sections ofembedded feather germs were made at differentdistances from the base of the germ (between 1 and8.25 mm) and different radial positions around thetubular feather germ (between 608 and 308 from thedorsal midline). Medullary cells were completelymature at 10 mm from the base of the feather germ.

Thin sections were viewed with a JEOL EXIItransmission electron microscope, and images weredigitally captured using a Soft-Imaging Megaview IICCD camera (1024!1200 pixels).

The development of a series of homologous spongymedullary cells is described by following barb ridges attwo radial positions (608 and 1208 from the rachis) alongthe proximal–distal maturation gradient from the base

J. R. Soc. Interface (2009)

of the growing feather germ within the follicle and thecompletelymature barb cells. Thedistances inmillimetreswere measured from the base of the developingfeather germ, and are a proxy of cell age and develop-mental stage.

4.2. Comparative structural analysis

To test the SD hypothesis for self-assembly of spongymedullary b-keratin, we analysed transmission electronmicrographs (TEMs) of mature spongy medullarykeratin of 15 species from 12 different families ofbirds. The species were identified as nine spherical-typeand six channel-type nanostructures based on visualobservation of the nanostructures.

To quantify these structures, we estimate the staticstructure factor, S(k), from TEM images. The staticstructure factor can be directly measured with anappropriate scattering method, such as light scatteringor neutron scattering. It is essentially a diffractionpattern of a wave passing through the sample. Weestimated S(k) by calculating the azimuthal average ofthe Fourier power spectrum of TEM images in MATLAB.This method is identical to the ‘radial’ averages of theFourier power spectrum calculated by Prum & Torres(2003). A peak in S(k) corresponds to the predominantvariation in the image contrast at a wavelengthinversely proportional to the value of kmax, thewavevector or spatial frequency at the peak. Forcomparison across species, which have different charac-teristic length scales and different strengths of contrast,we normalized all values of k by kmax and all values ofS(k) by S(kmax). The normalized plots of S(k) forchannel and spherical classes of nanostructure werecompared with analogous light scattering data from aspinodally decomposing polymer mixture of polybuta-diene and polyisoprene (Takenaka & Hashimoto 1992).

5. RESULTS

5.1. Medullary cell and nanostructuredevelopment

At 1.5 mm above the base of the feather germ, distinctbarb ridges are well organized (figure 4a,b). Theperipheral barbule plate cells are well differentiatedand linearly organized into presumptive barbules, butthe medullary cells in the more basal (i.e. internal)positions within the barb ridge are small and poorlydifferentiated. In TEMs, the medullary cells are weaklyattached to one another, and separated by gaps(figure 4b). These gaps may be artefacts of sectioningfor TEM, but similar spaces were observed byMatulionis (1970) and Alibardi (2005) in chickembryos. No keratinization is occurring, but melano-somes are being transferred from melanocytes intomedullary keratinocytes. Melanocyte processes, andoccasionally entire melanocytes, migrate amidst themedullary cells within the barb ridge (figure 4a). Thismay be facilitated by the gaps between medullary cellsat this early stage.

At 3 mm, medullary cells exhibit greater organiz-ation and fewer gaps between cells (figure 4c).

cm

cm

uu

n

c

m

g

m

m

ml

m

bp

(a) (b)

(c) (d )

(e) ( f )

bp

Figure 4. Early stage of medullary cell development. (a) Separate barb ridges with differentiated barbule plates, weaklyorganized medullary cells and an occasional melanocyte. (b) Medullary cells are poorly differentiated, weakly interconnected andseparated by gaps. (c) Medullary cells elongate and more closely connected with fewer gaps. (d ) Medullary cells expand to theirmature size and boxy shape with enlarged nucleus. (e, f ) Cell membranes between medullary cells with unknown dark-stainingstructures. Distances from the base of the feather germ: (a,b) 1.5 mm, (c,d ) 4 mm, (e, f ) 5 mm. Scale bars: (a) 10 mm, (b) 3 mm,(c) 4 mm, (d ) 2 mm, (e) 1 mm and ( f ) 300 nm. bp, barbule plate; c, cortical cell; cm, cell membrane; g, intercellular gap;m, medullary cell; ml, melanocyte; n, nucleus; u, unknown structure.

S258 Development of colour-producing b-keratin R. O. Prum et al.

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

At 4–5 mm, the medullary cells have increased to theirmature size and realized their final shapes, andneighbour boundaries (figure 4d ). Neighbouring cellshave well-defined cell membranes bordering oneanother, and cells do not separate during TEM fixationand sectioning. Cell nuclei are large and less denselystaining. Many cells have unidentified darkly stainingstructures that are approximately 200 nmby100 nmthatoccur within 100 nm of the cell membrane (figure 4d–f ).

Starting at 4–5 mm, the medullary cells begin todevelop large ‘transparent’ or ‘clear’ (i.e. electron-lucent) regions, which expand in size to occupy themajority of the volume of the cell (figure 5a–d ). Ourobservations indicate that these clear regions arecytoplasmic ‘voids’ that lack most of the larger,electron-dense biological structures. These voids beginas one or a few small clear regions in the centre, whichunite as they expand into a single large void (figures 5and 6). As the voids expand, the nuclei, melanosomes

J. R. Soc. Interface (2009)

and other observable cellular structures becomerestricted to the extreme margins of the cell(figure 6a,b). These expanding voids do not appear tobe membrane bound or created by cytoskeletal fila-ments; rather, they are bounded by a line of darkmaterial that appears to be accumulated intracellularcontents that have aggregated at the expanding edge ofthe void (figure 5c,d ).

Starting at 5.5–6 mm, the medullary cells begin tokeratinize along their membrane boundaries withneighbouring cells (figure 5e, f ). This initial peripheralkeratinization creates interconnected linear keratinprocesses that are tens of nanometres to more than100 nm wide and up to 1 mm in length. These processesoccasionally branch, or bifurcate, at very small angles(less than 308) to create an anastamosing matt ofbranches. In cross section, these keratin processesappear as darkly staining circular bodies near the cellmembrane. At this stage, they closely resemble the

(a) (b)

(c) (d )

(e) ( f )

v

v

v

k

v

bb

b

v

mc

c

cm

v

n

Figure 5. Intermediate stage of medullary cell development. (a) Medullary cells expand to their full mature sizes with closelyadhering cell membranes. Clear, electron-lucent cytoplasmic voids begin to appear. (b) Medullary cell with expandingcytoplasmic voids that lack most electron-dense structures. (c,d ) Cytoplasmic voids showing little material within them. Themargins of the voids are sometimes marked by aggregated materials (black arrows), but in other places they clearly lack asurrounding membrane (white arrows). Some dark speckling in (d ) is a staining artefact. (e, f ) Peripheral b-keratin structuresthat radiate from the cell membranes into neighbouring cells. Distances from the base of the feather germ: (a–d ) 5 mm, (e) 6 mm,and ( f ) 7.5 mm. Scale bars: (a) 5 mm, (b) 2 mm, (c) 400 nm, (d ) 200 nm, and (e, f ) 1 mm. b, barbule; c, cortical cell; cm, cellmembrane; k, b-keratin; m, medullary cell; v, cytoplasmic voids.

Development of colour-producing b-keratin R. O. Prum et al. S259

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

initial keratinized structures of the non-spongy barbuleand ramus cortical cells (figure 6a,b). These initial linearkeratin processes do not extend into the centre of the cell;rather, they radiate at low angles that are nearly parallelto the cell membrane. In TEM sections of the end of amedullary cell, the linear processes showa tangled ‘crownof thorns’ form (figure 5e,f ). At their peak extent, theseperipheral b-keratin structures extend further from thecell membrane than the width of the b-keratin along thecell membrane in final mature cells (figures 2c and 5e,f ).This implies that these b-keratin structures are eithercompacted or remodelled during later development.

By 7 mm, every medullary cell appears as a single,large, entirely empty transparent void (white areas infigure 6a,b). Nuclei, melanosomes and other organellesare restricted to the extreme margins of the cells. Theperipheral keratin now forms a solid layer approxi-mately 750 nm thick. Although the medullary cells look

J. R. Soc. Interface (2009)

empty, subsequent development demonstrates thatthese featureless cells are indeed living.

After 7 mm, clouds of patchy grey (i.e. electron-dense) material begin to appear at the periphery of theclear voids within each cell (figure 6c,d ). This materialappears to have expanded from the periphery of thecells back into the cytoplasmic voids towards the centreof the cell. Keratinization of the spongy medullarymatrix then proceeds by broad regional precipitation ofa characteristic spongy pattern of keratin channels outof this amorphous granular cytoplasmic material.

Relatively few cells were observed in the inter-mediate stages of this process (figure 7), implying thatit proceeds rapidly, perhaps even within a few hours.Commonly, TEMs show neighbouring cells in substan-tially different stages of keratinization. Early stages ofspongy keratin polymerization are characterized by theappearance of a few keratin rods within a local region

(a) (b)

(c) (d)

v

v

gr

v

ml cmc

ml

k

k

c

bb

bv

Figure 6. Late intermediate stage of medullary cell development. (a) Barb ridge with all medullary cells occupied by a single,large cytoplasmic void. (b) Medullary cells with fully expanded central void (v). Melanosomes and nuclei are restricted to theextreme margins of each cell, and the peripheral b-keratin along the cell membrane has been compacted or remodelled. (c,d )A cytoplasmic void begins to shrink with the expansion of granular electron-dense material from the margins of the cell.Distances from the base of the feather germ: (a–d ) 7.5 mm. Scale bars: (a) 5 mm, (b) 3 mm, (c) 1 mm, and (d ) 600 nm. b, barbule;c, cortical cell; cm, cell membrane; gr, granular material; m, medullary cell; ml, melanosome; v, cytoplasmic void.

S260 Development of colour-producing b-keratin R. O. Prum et al.

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

(approx. 500 nm2) of granular cytoplasmic material.Other images later in the process show a more completenanostructure that is still vague and unresolved inspecific places (figure 7a–c). Cytoplasmic spaces amongdeveloping keratinized bars are occupied by granular,electron-dense materials including ribosomes andpossibly other organelles (figure 7a,b). Importantly,during spongy keratinogenesis, there are no indicationsof cellular intermediate filaments, microtubules, endo-plasmic reticulum, cell membrane, etc. to organize thespatial nanoscale patterning. Rather, the pattern arisesby self-assembly from within regions of granularcellular cytoplasmic materials.

As spongy keratinogenesis proceeds, the clear voidsare greatly reduced in size and increasingly restrictedtowards the centre and basal regions of each medullarycell (figure 7d ). Melanin granules become surroundedby and incorporated within the basal region of thespongy keratin (figure 7d, f ). Nuclei and other orga-nelles are progressively restricted to the basal regions ofeach cell towards the centres of the multicellular barbramus (figure 7d ).

When the medullary cells completely dehydrate anddie, the cytoplasmic spaces within the spongy nano-structure, the remaining cytoplasmic voids and thenuclear envelope become filled with air, creating thedielectric, quasi-ordered, photonic nanostructure that is

J. R. Soc. Interface (2009)

responsible for vivid structural colour production in themature plumage of the bird. Mature blue feather barbrami from blue-and-yellow macaw (A. ararauna) showexpansive regions of channel-type spongy medullarykeratin, melanin granules distributed in intermediatepositions and larger, basally distributed air-filled spaces(figure 2b,c). These large air-filled spaces were clearlyoccupied by cellular nuclei and the remaining cyto-plasmic voids of the cells when they died. Furtherevidence of remodelling or compression of the initial,peripheral b-keratin structures comes from the verynarrow (30–50 nm) continuous keratin volume markingthe position of the membrane junctions betweenneighbouring cells (figure 2c).

5.2. Test of the spinodal decompositionhypothesis

The SDmechanism of phase separation produces highlycharacteristic, scale-invariant morphologies, or shapes,of the two bicontinuous separating materials. Regard-less of size, the spinodal morphology will produce thesame spatial relationships between the regions of thetwo materials—here, b-keratin and cellular cytoplasm.To examine the hypothesis that the channel-typeb-keratin nanostructures in avian feather barb ramiself-assemble by SD, we compared the scale-independent

(a)

k

k

k

ml

ml

cc

r

sakc

k

v

n

c

c

kk

(b)

(c) (d )

(e) ( f )

k

ml

Figure 7. Development of spongy medullary feather b-keratin. (a,b) Emergence of curved bars of spongy b-keratin withingranular electron-dense materials including ribosomes. (c) Region of self-assembling b-keratin above an aggregation ofmelanosomes. (d ) Medullary cell with extensive spongy b-keratin peripheral to an array of melanosomes. Cytoplasmic voids aregreatly reduced and the nucleus is visible again. Keratin between the voids and the melanosomes is not completely assembled.(e, f ) Mature spongy b-keratin in final stages of assembly. Distances from the base of the feather germ: (a–f ) 7.5 mm. Scale bars:(a,e) 300 nm, (b) 200 nm, (c) 600 nm, (d ) 2 mm, and ( f ) 1 mm. b, barbule cell; c, cortical cell; k, spongy b-keratin; ml, melanosome;n, nucleus; r, ribosome; sak, self-assembling keratin; v, cytoplasmic void.

Development of colour-producing b-keratin R. O. Prum et al. S261

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

distribution of different spatial frequencies of variationin composition of spongy b-keratin nanostructureswith observations of light scattering by polymer blendsin SD (Takenaka & Hashimoto 1992).

First, we compared the shapes of the frequencydistributions of spatial frequencies in b-keratin–aircomposition (calculated from Fourier transforms ofTEMs) of mature channel-type nanostructures fromnine avian species with measurements of the spinodallydecomposing polymer mixture (figure 8a). Thenormalized spatial frequencies from the feather dataprovide a very accurate fit to the observed SDobservation at higher spatial frequencies (figure 8b).There is substantial variation at lower spatial frequen-cies, but these are probably produced by large-size-scaleartefacts in the TEM images, such as large-scalevariations in image darkness and finite sample effects.In contrast, spatial frequency profiles from spongymedullary keratin of sphere-type nanostructures from

J. R. Soc. Interface (2009)

six different birds species all show a second peak at higherspatial frequencies (between 1.6 and 2 k/kmax), which isabsent from the SD data, and often a strikingly narrowerpeak in spatial frequency.

These data support the conclusion that channel-type keratin nanostructures are self-assembled by aSD process. As expected from their qualitativemorphology, these data also indicate that sphere-typenanostructures deviate substantially in shape from theexpectations of the SD mechanism.

6. DISCUSSION

Nearly a century ago, Thompson (1917) responded tothe growing adaptationist trend in comparative,functional and evolutionary biology by arguing power-fully that physical and chemical forces have a funda-mental role in the development of organismal form.Since that time, Thompson’s ‘structuralist’ approach

0.5 1.0 1.5 2.0 2.5 3.00

0.2

0.4

0.6

0.8

1.0

1.2

k /kmax

0

0.2

0.4

0.6

0.8

1.0

1.2

(b)

(a)

S(k)

/S(k

max

)S(

k)/S

(km

ax)

Figure 8. Normalized spectra of the spatial frequencies ofvariation in composition of mature spongy medullary featherkeratin TEMs (coloured data points) and a polymer mix in SD(solid line; from Takenaka &Hashimoto 1992). (a) Nine speciesof channel-type nanostructures show strong congruence withSD at higher spatial frequencies above the peak. (b) Six speciesof sphere-type nanostructure show variation in the breadth ofthe peak and a distinct secondary peak at higher spatialfrequencies (approx. 1.6–2 k/kmax). (a) Channel-type speciessampled: Porphyrio martinica (Rallidae); Motmotus motmota(Motmotidae); Ifrita kowaldi (Cinclosomatidae); I. puella(Irenidae); Vireolanius pulchellus (Vireonidae); Cyanocoraxyncas; Cyanocorax beecheii (Corvidae); Muscicapa turquosa;and Sialia sialis (Turdidae). (b) Sphere-type speciessampled: Cotinga cayana (Cotingidae); Lepidothrix serena;L. suavissima; L. coronata (Pipridae); Tangara seledon; andC. cyaneus (Thraupidae).

S262 Development of colour-producing b-keratin R. O. Prum et al.

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

has often been eclipsed by the adaptationist perspectivein evolutionary and developmental biology (Gould &Lewontin 1979; Wake 1991; Amundson 2001, 2005).Thus, although no modern biologists would deny therole of physical and chemical forces in the developmentand evolution organismal form, research directly onthese processes is still rather rare. As proposed byHemsley and colleagues (Hemsley et al. 1994, 1996,1998; Hemsley & Griffiths 2000), this research is ananoscale exploration of Thompson’s hypothesis con-cerning abiotic physical forces in developmental biology.Here, we examine whether crucial details of the develop-ment of colour-producing b-keratin nanostructures inbird feathers may be described by two physical mechan-isms of soft-condensed-matter physics—specifically,capillary flow in drying suspensions, andphase separation.

6.1. Capillary flow in a drying cell

TEMs of spongy medullary keratin developmentindicate that a complex sequence of cellular physiologicalevents establishes unusual cellular conditions preceding

J. R. Soc. Interface (2009)

spongynanostructure self-assembly—thedevelopment oflarge, clear intracellular voids. The TEMs indicate thatthe materials within the cytoplasm are moving towardsthe periphery of the cell, producing an expanding centralregion that is devoid of larger biological structures orelectron-dense materials (figures 5 and 6). Using lightmicroscopy, Auber (1971/1972) observed the develop-ment of clear areas within spongy medullary cells.Matulionis (1970) described these structures in embryo-nic chick barb rami using TEM. Careful examinationof the edges of the voids indicates that they arenot membrane-bound vacuoles, as hypothesized byMatulionis (1970) and Alibardi (2002, 2005). Rather,the margin of each void appears to be defined by theaccumulation of cellular materials, similar to flotsamwashed up on a beach (figure 5c,d ). The cells are stillliving at this stage; ribosomes, the nucleus and melano-somes will all reappear at later stages of development(figure 7d ). Interestingly, mature medullary cells fromfeather barbs and rachis have a large hollow volume atthe centre, which apparently develops by thismechanism(see below; Lucas & Stettenheim 1972).

We propose that this cytoplasmic void is formed bycapillary flow as the cells begin to dry. This sameprocess produces ring stains in drying drops of coffee orred wine (Deegan et al. 1997; Tsapis et al. 2005).Evaporation from the surfaces of a drying drop of liquidwill produce capillary flow towards the ‘pinned’ edges ofthe drop, which cannot retract as the drop loses volumebecause of strong capillary attractions to the materialat the surface. Owing to the constraint on shrinking theedges of the droplet, fluid will travel towards the dryingedges to compensate for the evaporative loss, creating aflow or net evaporative flux (Deegan et al. 1997). Thiscapillary flow will transport dispersed materials(solutes) towards the pinned edges of the drop,resulting in a ring-shaped stain of deposited materialsaround the edges of a drying drop. Originally describedin drying droplets on a two-dimensional surface(Deegan et al. 1997), the phenomenon is known toform shells in drying three-dimensional droplets ofpolymer colloids (Tsapis et al. 2005).

We propose that drying in the medullary keratino-cytes creates capillary flow towards the margins of thecells, and forms a volume of cytoplasm at the centre ofthe cell that is devoid of larger molecules andorganelles. The two requirements of this process—drying and pinned edges—are both met in these cells.Drying is a fundamental part of the keratinocytematuration. All feather keratinocytes achieve theirmature functions when they are dead and dry.Peripheral keratinization at the cell boundaries thatprecedes void formation would simultaneously (i) cutoff the cytoplasm of each cell from diffusion of waterfrom the dermal pulp of the feather germ, and(ii) structurally constrain the cell from shrinking involume as it dries.

The capillary flow hypothesis raises the question ofwhy the flow of cellular materials to the margins ceasesand granular materials reappear and expand from themargins of the cells. The magnitude of capillary flowdepends on the rate of drying. We hypothesize thata slowing of the rate of drying reduces capillary flow and

Development of colour-producing b-keratin R. O. Prum et al. S263

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

provides opportunity for cytoplasmic materials todiffuse back into the cytoplasm, creating the conditionsthat immediately precede the self-assembly of thespongy medullary keratin.

6.2. Self-assembly

Our TEM observations of development of colour-producing, spongy medullary feather keratin indicatethat these nanostructures are self-assembled. Themorphologies of the spherical- and spongy-type nano-structures closely match the morphologies produced bytwo distinct processes of nanoscale self-assembly byphase separation—SD and nucleation and growth.There is no evidence within these feather cells of abiological prepatterning mechanism of intermediatefilaments, cell membrane, endoplasmic reticulum orother structural precursors to the nanostructure.Rather, the nanostructure emerges with its ownintrinsic and characteristic form. Similarly, Matulionis(1970) was unable to identify any intermediate filamentprecursors to feather keratin.

Keratinization in the medullary cells begins peri-pherally near the cell membrane as in other featherkeratinocytes of the barb cortex and barbules. Appa-rently, the expansion of the cytoplasmic voids disruptsthe continuation of this process, and establishes theconditions for subsequent spongy nanostructure self-assembly. It is clear that the development of traditionalsolid b-keratin and the spongy medullary b-keratinare very distinct processes. Interestingly, there may bea role for some structural, cellular prepattern in theinitiation of the peripheral keratin. The initial peri-pheral keratin bars radiate at low angles away fromsites along the cell membrane.

6.3. Phase-separation kinetics

The role of phase separation in b-keratin self-assemblywas suggested by the striking similarities of thestereotyped morphologies produced by the two funda-mental mechanisms of phase separation—SD andnucleation and growth (figure 1e, f )—and the physicalprocess of b-keratin self-assembly (Brush 1983).

Images of SD in well-characterized physical systemsshow a striking congruence with channel-type b-keratinnanostructures in bird feather barbs (figure 1a,b,e).Furthermore, the sphere-type spongy keratin nano-structures show strong resemblance to the sphericalmorphologies produced by nucleation and growth phaseseparation in metastable mixtures (figure 1c,d, f ).Although the two fundamental classes of spongykeratin nanostructure were recognized over 30 yearsago (Dyck 1976, 1978; Prum et al. 1999; Prum 2006),the phase-separation hypothesis provides the firstexplanation of (i) why there are two distinct classes ofnanostructure, and (ii) why they have the specific shapes.

Our analysis of the size-independent variation in theshapes of spongy keratin nanostructures documentsstrong congruence between the structure factor (or theform of spatial variation) of a mixture in SD and abroad sample of channel-type feather nanostructuresfrom nine different species (figure 8a). Substantial

J. R. Soc. Interface (2009)

variations among datasets at small spatial frequencieswere a consequence of large-scale deviations in imagecontrast and edge effects. By contrast, shapes of sphere-type nanostructures show distinct deviations from theshape of spinodal materials at higher spatial frequencies(figure 8b). The additional peak in S(k) for thesesystems is characteristic of dense collections of spheres,which can be produced by nucleation and growth.

Heterogeneity within each cell in those factors thatinfluence kinetic interactions of b-keratin and thecytoplasm (i.e. rate of drying, protein concentration,etc.) will create additional spatio-temporal patterning.For example, the wavefront of phase separation thatappears to move through the cell volume is notaccounted for by a simple description of SD.

SD produces a characteristic shape that expands insize and scale over time (Jones 2002). However, inexperiments on phase-separating colloids, this charac-teristic shape expansion breaks down when thestructures become large enough for gravity, or otherexternal forces, to perturb the purely intermolecularinteractions in the mixture (Aarts & Lekkerkerker2004). Consequently, in order for b-keratin nanostruc-tures to produce the appropriate species- and sex-specific structural colour when the feather cells aremature, the expansion of the spinodal spatial patternmust be halted and solidified (or ‘quenched’) at theappropriate size. Although phase separation provides apowerful explanation of the development of the shape ofthese b-keratin nanostructures, it raises a new questionabout the physical determination of nanostructure size,which is of critical importance to the ultimatecommunication function of these nanostructures inthe lives of birds. We expect that the coupling of phaseseparation and the irreversible polymerization ofb-keratin (Brush 1983) halts the phase-separationprocess and that competition between the rates ofphase separation and polymerization selects theultimate feature size of the nanostructures.

In conclusion, we propose that spongy medullarykeratin nanostructures develop by phase separation ofthe intracellular mixture of b-keratin and cytoplasm.The channel-type nanostructures develop by spinodalcomposition of unstable mixtures, and sphere-typenanostructures develop by nucleation and growth ofmetastable mixtures. We further propose that self-assembly of spongy medullary keratin can be repro-duced and studied in vitro. Our results extendHemsley’s hypothesis (Hemsley et al. 1994, 1996,1998; Hemsley & Griffiths 2000) of nanoscale self-assembly from plant spore development to vertebrateintegumentary appendages.

6.4. Evolution of spongy feather nanostructures

Our observations provide new insights into theevolution of colour-producing spongy medullary barbnanostructures. Medullary cells of the feather rami andrachis are identified by their large size, boxy shape andthe central, air-filled space (e.g. Lucas & Stettenheim1972). Air-filled medullary cells function to providemechanical support to the rami and rachis of thefeather, and are thus critical for the maintenance of

S264 Development of colour-producing b-keratin R. O. Prum et al.

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

a planar vane that is critical to avian flight and manyother feather functions. It is well understood that ahollow structure can provide more structural integrityfor the same volume of material (Vogel 2003). Hollowfeather medullary cells have convergently evolved withhollow plant sclerenchyma cells through selection fortheir mechanical properties (Niklas 1992).

Birds have apparently evolved to use the physicalmechanism of capillary flow during drying to create thehollow, mechanically advantageous morphology oftheir medullary feather cells. This novel developmentalmechanism also apparently created the opportunity forsubsequent establishment of appropriate conditionsfor phase separation of b-keratin and cytoplasm withinthe cell. Apparently, these conditions are not created insolid cortical and barbule cells in which keratinizationproceeds continuously from the periphery to fill the cellvolume (Matulionis 1970). Colour-producing spongykeratin has never evolved in any other type of feathercell (Prum 2006).

Thus, colour-producing spongy keratin evolvedthrough the co-option and specialization of mechanismsof development of hollow medullary cells that initiallyevolved to provide mechanical integrity to featherbarbs and rachis. Many lineages of birds haveindependently evolved to exploit phase-separationmechanisms within medullary cells to create opticalnanostructures out of b-keratin. The parallel evolutionof precisely similar nanostructures in so many differentlineages of birds indicates that the parameters necess-ary to produce dynamic phase separation duringkeratinogenesis are easily achieved and easily evolvedin medullary keratinocytes.

6.5. Self-assembly and condition dependence

Some mechanisms of sexual selection predict thatfemale preferences will evolve through natural selectionto prefer male traits that honestly communicate matequality or condition (Andersson 1994). Signal honestyis hypothesized to be reinforced by either dependenceon a limiting nutrient (e.g. dietary carotenoids) or asubstantial, physiologically costly investment.

In general, the self-assembly process does not requireeither limiting nutrients or extensive physiologicalinvestment, as hypothesized by condition-dependentsexual selection models. During development, spongymedullary cells establish the initial conditions for self-assembly of the colour-producing nanostructurethrough b-keratin synthesis and drying. By relying onpredictable physical forces to construct the nanostruc-ture, medullary cells eliminate many potential physio-logical mechanisms to mediate condition dependenceduring development. Furthermore, the additionalamounts of b-keratin forming the actual bars withinthe spongy matrix are a trivial additional physiologicalinvestment compared with the volume of keratin in theentire feather.

Undisturbed phase separation results in thecomplete separation of the two materials (e.g. oil andwater).Weare not surewhat causes the phase-separationprocess in spongy medullary cells to halt at thespecific size scale appropriate to produce the colour

J. R. Soc. Interface (2009)

required for that cell. But the self-assembly mechanismimplies that it may be the irreversible polymerization ofb-keratin that arrests the phase-separation process.Thus, the size scales of the spongy nanostructure, whichdetermine the colour produced, may be controlled by thekinetic properties of the specific b-keratin moleculesexpressed in latemedullary development. Little is knownabout the variation in molecular structure and kineticsof the b-keratins found in bird feathers and nothing isknown specifically about the b-keratins expressed inspongy medullary cells (Brush 1978, 1993).

Another way to assess mechanisms of variation instructural coloration is to examine variation within aplumage. For example, microscopic observation ofthese structurally coloured bird feathers indicatesthat the process is under strict control. For example,the primary flight feathers of many rollers (Coracias)change in structural colour from deep blue to turquoisein the middle of the vane. In the transition zonebetween colours, one can see that individual medullarycells faithfully produce either the deep blue or turquoisecolour. The structural colour transition across the wingis accomplished with a mosaic of cells with two distinctcolours rather than a series of cells that graduallytransition between the two colours.

Thus, it appears that the medullary cells physio-logically control the conditions necessary to initiateself-assembly, but that the deterministic physicalprocess of self-assembly leaves little opportunity forsubsequent variation in the outcome of development.Although additional features of these feathers—such asthe shape, thickness and surface of the barb cortex, thedistribution of melanosomes and the presence andconcentration of carotenoid pigments—may influencethe colour produced by the feather, the self-assembly ofthe nanostructure itself reduces the potential forcondition-dependent trait expression in structuralcoloration during feather development.

We thank Drs Tom Pollard and Andrew Miranker for theirhelpful comments, and the organizing committee of the 2008Iridescence Conference at Arizona State University for theinvitation to participate. Financial support for this researchwas provided by Yale University, and grants from theNational Science Foundation to R.O.P. (DBI-0078376) andE.R.D. (CAREER CBET-0547294). VIREO gave permissionto reproduce the photo in figure 2a by T. Laman.

REFERENCES

Aarts, D. G. A. L. & Lekkerkerker, H. N. W. 2004 Confocalscanning laser microscopy on fluid–fluid demixing colloid–polymer mixtures. J. Phys. Condens. Matter 16,S4231–S4242. (doi:10.1088/0953-8984/16/38/035)

Alibardi, L. 2002 Keratinization and lipogenesis in epidermalderivatives of the zebra finch, Taeniopygia guttatacastanotis (Aves, Ploecidae) during embryonic develop-ment. J. Morphol. 251, 294–308. (doi:10.1002/jmor.1090)

Alibardi, L. 2005 Cell structure of developing barbs andbarbules in downfeathers of the chick: central role of barbridge morphogenesis for the evolution of feathers.J. Submicrosc. Cytol. Pathol. 37, 19–41.

Amundson, R. 2001 Adaptation, development, and thequest for common ground. In Adaptationism and optimality

Development of colour-producing b-keratin R. O. Prum et al. S265

on 23 March 2009rsif.royalsocietypublishing.orgDownloaded from

(eds S. Hecht Orzack & E. Sober), pp. 303–334. New York,NY: Cambridge University Press.

Amundson, R. 2005 The changing role of the embryo inevolutionary thought: roots of evo-devo. Cambridge, UK:Cambridge University Press.

Andersson, M. 1994 Sexual selection. Princeton, NJ: PrincetonUniversity Press.

Auber, L. 1971/1972 Formation of ‘polyhedral’ cell cavities incloudy media of bird feathers. Proc. R. Soc. Edinb. 74,27–41.

Brush, A. H. 1978 Feather keratins. In Chemical zoology,vol. 10 (ed. A. H. Brush), pp. 117–139. New York, NY:Academic Press.

Brush, A. H. 1983 Self-assembly of avian 4-keratins.J. Protein Chem. 2, 63–75. (doi:10.1007/BF01025168)

Brush, A. H. 1993 The origin of feathers. In Avian biology,vol. 9 (ed. D. S. Farner, J. S. King & K. C. Parkes),pp. 121–162. London, UK: Academic Press.

Deegan, R. D., Bakajian, O., Dupont, T. F., Huber, G., Nagel,S. R. & Witten, T. A. 1997 Capillary flow as the cause ofring stains from dried liquid drops. Nature 389, 827–829.(doi:10.1038/39827)

Durrer, H. & Villiger, W. 1967 Bildung der Schillerstrukturbeim Glanzstar. Zeitschrift fur Zellforschung 81, 445–456.(doi:10.1007/BF00342767)

Dyck, J. 1976 Structural colours. Proc. Int. Ornithol. Congr.16, 426–437.

Dyck, J. 1978 Olive green feathers: reflection of light from therami and their structure. Anser 3(Suppl.), 57–75.

Erlebacher, J., Aziz, M. J., Karma, A., Dimitrov, N. &Sieradzki, K. 2001 Evolution of nanoporosity in dealloying.Nature 410, 450–453. (doi:10.1038/35068529)

Ghiradella, H. 1974 Development of ultraviolet-reflectingbutterfly scales: how to make an interference filter.J. Morphol. 142, 395–410. (doi:10.1002/jmor.1051420404)

Ghiradella, H. 1989 Structure and development of iridescentbutterfly scales: lattices and laminae. J. Morphol. 202,69–88. (doi:10.1002/jmor.1052020106)

Gould, S. J. & Lewontin, R. C. 1979 The spandrels of SanMarco and the Panglossian paradigm: a critique of theadaptationist programme. Proc. R. Soc. B 205, 581–598.(doi:10.1098/rspb.1979.0086)

Hemsley, A. R. & Griffiths, P. C. 2000 Architecture in themicrocosm: biocolloids, self-assembly and pattern forma-tion. Phil. Trans. R. Soc. A 358, 547–564. (doi:10.1098/rsta.2000.0545)

Hemsley, A. R., Collinson,M.E.,Kovach,W.L., Vincent, B.&Williams, T. 1994 Role of self-assembly in biologicalsystems: evidence from iridescent colloidal sporopolleninin Selaginella megaspore walls. Phil. Trans. R. Soc. B 345,163–173. (doi:10.1098/rstb.1994.0095)

Hemsley, A. R., Jenkins, P. D., Collinson, M. E. & Vincent, B.1996 Experimental modeling of exine self-assembly. Bot.J. Linn. Soc. 121, 177–187. (doi:10.1006/bojl.1996.0031)

Hemsley, A. R., Vincent, B., Collinson, M. E. & Griffiths, P.1998 Simulated self-assembly of spore exines. Ann. Bot.82, 105–109. (doi:10.1006/anbo.1998.0653)

Jones, R. A. L. 2002 Soft condensed matter. Oxford, UK:Oxford University Press.

Lucas, A. M. & Stettenheim, P. R. 1972 Avian anatomy—integument. Washington, DC: US Department of Agricul-ture Handbook.

J. R. Soc. Interface (2009)

Matulionis, D. H. 1970 Morphology of the developing downfeathers of chick embryos. Zeitschrift fur Anatomie undEntwicklungsgeschichte 132, 107–157. (doi:10.1007/BF00523275)

Neville, A. C. 1975 Biology of the arthropod cuticle.New York, NY: Springer.

Neville, A. C. 1993 Biology of fibrous composites. Cambridge,UK: Cambridge University Press.

Niklas, K. J. 1992 Plant biomechanics: an engineeringapproach to plant form and function. Chicago, IL: ChicagoUniversity Press.

Parker, A. R. 1999 Invertebrate structural colours. InFunctional morphology of the invertebrate skeleton(ed. E. Savazzi), pp. 65–90. London, UK: Wiley.

Prum, R. O. 1999 Development and evolutionary originof feathers. J. Exp. Zool. (Mol. Dev. Evol.) 285, 291–306.(doi:10.1002/(SICI)1097-010X(19991215)285:4!291::AID-JEZ1O3.0.CO;2-9)

Prum, R. O. 2006 Anatomy, physics, and evolution of avianstructural colors. In Bird coloration, vol. 1 (eds G. E. Hill& K. J. McGraw). Mechanisms and Measurements,pp. 295–353. Cambridge, MA: Harvard University Press.

Prum, R. O. & Torres, R. H. 2003 A Fourier tool for theanalysis of coherent light scattering by bio-opticalnanostructures. Integr. Comp. Biol. 43, 591–602. (doi:10.1093/icb/43.4.591)

Prum, R. O. & Williamson, S. 2001 A theory of the growthand evolution of feather shape. J. Exp. Zool. (Mol. Dev.Evol.) 291, 30–57. (doi:10.1002/jez.4)

Prum, R. O., Torres, R. H., Williamson, S. & Dyck, J. 1998Coherent light scattering by blue feather barbs. Nature396, 28–29. (doi:10.1038/23838)

Prum, R. O., Torres, R. H., Williamson, S. & Dyck, J. 1999Two-dimensional Fourier analysis of the spongy medullarykeratin of structurally coloured feather barbs. Proc. R.Soc. B 266, 13–22. (doi:10.1098/rspb.1999.0598)

Prum, R. O., Cole, J. & Torres, R. H. 2004 Blueintegumentary structural colours of dragonflies (Odonata)are not produced by incoherent Tyndall scattering. J. Exp.Biol. 207, 3999–4009. (doi:10.1242/jeb.01240)

Srinivasarao, M. 1999 Nano-optics in the biological world:beetles, butterflies, birds, and moths. Chem. Rev. 99,1935–1961. (doi:10.1021/cr970080y)

Takenaka, M. & Hashimoto, T. 1992 Scattering studies ofself-assembling processes of polymer blends in spinodaldecomposition. II. Temperature dependence. J. Chem.Phys. 96, 6177–6190. (doi:10.1063/1.462635)

Thompson, D. A. W. 1917 On growth and form. Cambridge,UK: Cambridge University Press.

Tsapis, N., Dufresne, E. R., Sinha, S. S., Riera, C. S.,Hutchinson, J. W., Mahadevan, L. & Weitz, D. A. 2005Onset of buckling in drying droplets of colloidal suspen-sions. Phys. Rev. Lett. 94, 018 302. (doi:10.1103/Phys-RevLett.94.018302)

Vogel, S. 2003 Comparative biomechanics: life’s physicalworld. Princeton, NJ: Princeton University Press.

Vukusic, P. & Sambles, J. R. 2003 Photonic structures inbiology. Nature 424, 852–855. (doi:10.1038/nature01941)

Wake, D. B. 1991 Homoplasy: the result of natural selection,or evidence of design limitations? Am. Nat. 138, 543–567.(doi:10.1086/285234)