Content uploaded by Rafael Maia
Author content
All content in this area was uploaded by Rafael Maia
Content may be subject to copyright.
Iridescent structural colour production in
male blue-black grassquit feather barbules:
the role of keratin and melanin
Rafael Maia
1,2,
*, Joa
˜o Victor O. Caetano
3
,So
ˆnia N. Ba
´o
3
and Regina H. Macedo
1
1
Laborato
´rio de Comportamento Animal, Departamento de Zoologia, and
2
Programa de
Po
´s-Graduac¸a
˜o em Ecologia, Universidade de Brası
´lia, Brası
´lia 70910-900, Brazil
3
Laborato
´rio de Microscopia Eletro
ˆnica, Departamento de Biologia Celular,
Universidade de Brası
´lia, Brası
´lia 70919-970, Brazil
Iridescent coloration plays an important role in the visual communication system of many
animal taxa. It is known that iridescent structural colours result from layers of materials
with different refractive indexes, which in feathers usually are keratin, melanin and air.
However, the role of these materials in the production of structural iridescent coloration is
still poorly documented. Despite the great interspecific variation in the organization of such
structures in bird plumage, melanin layers are usually considered too opaque, suggesting its
main role is to delineate the outermost keratin layer and absorb incoherently scattered stray
light. We combined spectrometry, electron microscopy and thin-film optical modelling
to describe the UV-reflecting iridescent colour of feather barbules of male blue-black
grassquits (Volatinia jacarina), characterized by a keratin layer overlying a single melanin
layer. Our models indicate that both the keratin and the melanin layers are essential for
production of the observed colour, influencing the coherent scattering of light. The melanin
layer in some barbules may be thin enough to allow interaction with the underlying keratin;
however, individuals usually have, on an average, the minimum number of granules that
optimizes absorbance by this layer. Also, we show that altering optical properties of the
materials resulted in better-fitting models relative to the empirically measured spectra.
These results add to previous findings concerning the influence of melanin in single-layer
iridescence, and stress the importance of considering natural variation when characterizing
such photonic structures.
Keywords: barbule; feather; sexual selection; photonic crystal; thin film;
visual communication
1. INTRODUCTION
Visual signals play a major role in avian communi-
cation, inferred from the keen colour discrimination
(Cuthill 2006;Ha˚stad & O
¨deen 2008) and the diversity
and ubiquity of ornaments and colorations found in this
taxon (Andersson 1994). Coloration properties of
secondary sexual traits may provide reliable infor-
mation about an individual’s genetic quality ( Figuerola
et al. 1999), aggressiveness (Alonzo-Alvarez et al. 2004)
or reproductive strategy (Badyaev & Hill 2002), which
may be used by individuals during agonistic or mate
choice interactions (Searcy & Nowiki 2005).
Traditionally, mechanisms for production of feather
coloration have been subdivided into two categories.
Pigmentary colours result from the chemical proper-
ties of pigments and their concentration in feathers,
which allow the differential absorption and reflectance
of light of different wavelengths. The chief pigments in
this category are carotenoids, which produce red,
orange and yellow coloration, and melanins, respon-
sible for black, brown and rufous colours (McGraw
2006a,b). Although these pigments may combine to
produce several other hues, they are not responsible for
all known avian colorations—for example, no known
pigments produce blue coloration (Bagnara et al.2007),
and only rarely do they produce green coloration
(McGraw 2006c).
Structural colours are an alternative path to colour
production and rely on the organization and differential
refraction properties of the nanostructures that compose
feathers, producing an astonishing array of green, blue,
violet and ultraviolet tones (Kinoshita et al.2008).
J. R. Soc. Interface (2009) 6, S203–S211
doi:10.1098/rsif.2008.0460.focus
Published online 13 January 2009
One contribution of 13 to a Theme Supplement ‘Iridescence: more
than meets the eye’.
Electronic supplementary material is available at http://dx.doi.org/
10.1098/rsif.2008.0460.focus or via http://journals.royalsociety.org.
*Author and address for correspondence: Laborato
´rio de Comporta-
mento Animal, Departamento de Zoologia, Universidade de Brası
´lia,
Brası
´lia 70910-900, Brazil (rafa.maia@gmail.com).
Received 26 October 2008
Accepted 5 December 2008 S203 This journal is q2009 The Royal Society
The most common materials in these structures
are keratin, melanin (in the form of granules, called
melanosomes) and air (Prum 2006).
The ultraviolet component of the spectrum is invisible
to the human eye. However, several studies, ranging
from behavioural and electrophysiological data to opsin
coding of UV-absorbing cone genes, have indicated
that most birds have tetrachromatic vision and can see
light in the ultraviolet spectrum (Cuthill et al.2000;
Cuthill 2006). It remains controversial whether light in
this spectrum represents a special communication
channel for birds or whether it is just another chromatic
channel (Banks 2001;Haussmann et al.2003). In either
case, the universality of colour production in this range
(Eaton & Lanyon 2003), and therefore the role of
feather nanostructures in the production of this colour
component, should not be underestimated.
Structural coloration is usually produced by coher-
ent scattering of light, in which refractive particles are
organized in size, shape and distribution in a manner
that promotes a non-random interaction of different
refracting wavelengths. Certain wavelengths may be in
phase after interacting with the structure (interfering
constructively and being reinforced), while others will
be out of phase (and will thus interact destructively),
resulting in the observed colour (Kinoshita & Yoshioka
2005a). These structures may be organized in one, two
or even three dimensions (Prum & Torres 2003).
In feather barbules, unidimensional (i.e. thin film)
and bidimensional structures have been described
(e.g. Greenwalt et al. 1960;Prum et al. 1998;Andersson
1999;Zi et al.2003), and recent techniques have provided
new insights indicating that three-dimensional photonic
structures may also be found in the medullar cortex
of feather barbs (Shawkey et al. 2009). Structures
organized in one and two dimensions have the additional
feature of being iridescent, i.e. changing colour properties
(especially hue) as a function of the angle of light incidence
(Osorio & Ham 2002). This characteristic derives from the
coherent refraction of light through laminar structures, in
which angle affects the optical distance (a function of
linear distance and refractive index) that light will
encounter, therefore altering the phase relations in each
refracted wavelength (Kinoshita & Yoshioka 2005b).
Barbule nanostructures may be found in various
unidimensional organizations to produce iridescence,
resulting in great interspecific variability of structures
and colours. For example, a thick, absorbing melanin
layer underlying a keratin cortex results in violet-blue
coloration, as found in the velvet satin bowerbird
(Ptilonorhynchus violaceus minor;Doucet et al. 2006)
and some blackbirds and grackles (Shawkey et al.
2006). However, iridescent structural colours can be
obtained without the influence of a melanin layer in
coherent scattering, as appears to be the case of the
rock dove (Columba livia), where melanin granules are
considerably larger and randomly dispersed ( Yoshioka
et al. 2007). In this species, the green-violet iridescence
can be modelled as the product of a thin layer of keratin
in an air substrate ( Yin et al. 2006;Yoshioka et al.
2007). However, even in such cases, melanin can play
an important role by acting as a poor mirror, resulting
in higher peak reflectance and overall brightness
(Yin et al. 2006), by absorbing incoherently scattered
light, and by structurally defining the width of the
keratin layer during the developmental process
(as suggested by Doucet et al. 2006). Nonetheless,
both air and melanin may interact with keratin in
multilayer stacks to produce very bright iridescent
coloration of multiple hues, typically observed in
hummingbirds (Greenwalt et al. 1960).
Thin-film optical modelling is a very useful tool to
study animal iridescent coloration, allowing predictions
concerning the behaviour of light when interacting with
colour-producing structures, based on nanomorpho-
logical and optical properties; such predictions may
then be compared with the spectra obtained from the
biological structures (Shawkey et al. 2006;Yoshioka
et al. 2007). Furthermore, thin-film optical modelling
allows the manipulation of optical properties to test for
their adequacy in characterizing these structures
(Brink & van der Berg 2004;Doucet et al. 2006).
Despite these advantages, two major methodological
obstacles have hindered biologists pursuing an integra-
tive approach: the lack of a simple operational tool to
apply the necessary calculations for the models; and an
objective way of comparing the predicted and obtained
spectra, which is usually done visually and does not
allow a fine-scale fit to the model.
The blue-black grassquit (Volatinia jacarina)isa
small neotropical emberizid in which morphological
and behavioural characteristics are interrelated in a
complex communication process during social
interactions. During the breeding season, males moult
to a blue-black iridescent structural coloration, while
females maintain a cryptic plumage. Males defend
small, clustered territories by repeatedly performing
vertical leaping displays, during which their conspic-
uous plumage is exhibited to conspecifics in a variety of
angles of incident light ( Webber 1985). Though the role
of plumage coloration in social interactions remains
unexplored, coloration characteristics of this species
have been shown to be condition dependent ( Doucet
2002), and both the amount of nuptial plumage and
display characteristics are related to parasite load
(Costa & Macedo 2005;Aguilar et al. 2008).
It is very likely that visual communication in this
species, especially relative to plumage, is important in
social interactions, and may play a role in the high rate
of extra-pair copulation that has been documented
(Carvalho et al.2006). Hence, the blue-black grassquit
provides a potentially excellent model system to study
both proximate and evolutionary mechanisms that
promote the production and maintenance of iridescent
coloration. In this study, we identify the mechanisms of
structural colour production of male blue-black grass-
quits. To accomplish our objective, we combined
spectrometric measurements, transmission electron
microscopy (TEM) to characterize the nanostructures
of feather barbules and thin-film modelling. We further
investigate the role of optical properties of these
structures, identifying the effects of altering these
parameters upon coloration. For this purpose, we
developed a routine based on open access software, a
tool that can be widely applied to studies of the
mechanisms of iridescent structural colour production.
S204 Iridescence of blue-black grassquit R. Maia et al.
J. R. Soc. Interface (2009)
2. MATERIAL AND METHODS
2.1. Sampling
We captured and banded birds in the Fazenda A
´gua
Limpa (property of Universidade de Brası
´lia), Brazil
(158560S, 478560W) during the breeding seasons of
November–January 2005–2006 and 2006–2007. We
plucked three feathers from various body parts of
males, but for the present analyses we used only dorsal
feathers. If individuals were moulting, we only collected
feathers clearly identifiable as nuptial plumage.
All feathers were taped to a black card and individually
identified, wrapped in foil paper and kept in the
laboratory under dry and stable temperature conditions.
2.2. Colour measurement
For colour measurements, feathers were taped in an
overlaid manner to a black velvet substrate to simulate
their arrangement on the bird’s body. Feather reflec-
tance was measured with an Ocean Optics USB4000
spectrometer attached to a PX-2 pulsed xenon light
source (range 250–750 nm, Ocean Optics, Dunedin,
FL). All measurements were taken with R400-7
UV-VIS optic fibre reflection probes (400 mm diam-
eter), using unpolarized light and relative to a WS-1-SS
white standard (Ocean Optics).
To characterize iridescence, reflectance measure-
ments were taken from two measurement geometries:
using a bifurcated optic probe held perpendicular to the
feather surface (hereafter ‘normal geometry’, since both
light source and observer are parallel to the feather
surface normal), and with probe ends at separate angles
of illumination and measurement, specularly positioned
at 458from the normal (hereafter, ‘458’). These
measurement geometries were selected because a pilot
study indicated that they provide reliable information
on overall iridescent behaviour of the coloration,
offering the most saturated and repeatable measure-
ments. Optic fibres were held using a block sheath to
exclude ambient light and to maintain the optic probes
6 mm from the feather surface. We used SPECTRASUITE
software (Ocean Optics) to record and measure 50
sequential spectra from the feathers with an integration
time of 20 ms, and this procedure was repeated five
times for each sample. Between repetitions, to guaran-
tee that different parts of the feather surface were
sampled, we lifted the sheath from the feather surface
before positioning it for the new measurement. All
measurements were interpolated to a step width of 1 nm,
and calculations were performed based on the average
spectra of these five repetitions, from 320 to 700 nm.
2.3. Transmission electron microscopy
We prepared three whole barbs from dorsal feathers of
each male for TEM (protocol modified from Andersson
1999). Barbs were immersed for 30 min in a 0.25 M
NaOH solution and for 2 hours in formic acid : ethanol
(2 : 3 v/v). After bathing in distilled water, they were
dehydrated in acetone in increasing concentrations
(15 min in 30, 50, 70 and 90 per cent, and three 10-min
immersions in pure acetone). Inclusion was carried out
in increasing solutions of Spurr resin and acetone (3 : 1,
2 : 1, 1 : 1, 1 : 2 and 1 : 3, overnight in odd steps and for
6 hours in even steps, finalizing in pure Spurr for
6 hours). Barbules were then placed in Spurr blocks and
allowed to polymerize at 608C. Preliminary analyses
revealed that the keratin matrix and the melanosomes
were sufficiently electron dense, and thus no contrast-
ing was conducted to avoid depositing residues in
the material.
Polymerized blocks including barbules were cut
using a diamond knife on a Leica Reichert Supernova
ultramicrotome (Leica Microsystems, Austria). Sec-
tions were placed in 150 mesh grids and observed in a
JEOL JEM-1011 TEM (80 kV; JEOL, Japan). We
obtained micrographs at 10 000!magnification from
three barbules from the distal portion of different barbs
for each individual. Images were obtained with a digital
camera and digitally magnified for a final 25 000!
magnification. From these images, we obtained three
nanostructural measurements at six equidistant points
for each barbule: keratin cortex thickness; outer
melanin layer thickness; and number of granules in
the outer melanin layer. Melanin granule diameter was
estimated by dividing melanin layer thickness by the
number of granules in the section measured. We used
the mean of the six measurements of the three barbules
to characterize each individual. All measurements were
taken using IMAGEJ v. 1.38x (Rasband 1997–2007,
available at http://rsb.info.nih.gov/ij/index.html).
2.4. Models of colour production
We used thin-film optical modelling to characterize and
identify the influence of structures in colour production.
Our main objective was to identify the best-fitting
model to the observed spectra, considering the import-
ance of keratin and melanin layer properties obtained
from TEM. We used the transfer matrix method (see
Jellison 1993) to create four models, as described below.
Since we used unpolarized light in all measurements,
the s and p components of polarization were averaged
(Srinivasarao 1999).
To identify the influence of each component
structure in the production of blue-black grassquit
iridescent coloration, we considered four models that
define all possible two- and three-beam combinations of
interfaces produced by a set of two overlapping layers
(figure 1; for more details, see Doucet et al. 2006;
model 1
interface 1
interface 2
interface 3
model 2 model 3 model 4
air
keratin
keratin
melanin
Figure 1. Schematic of all thin-film optical models considered
in this study. Arrows represent incident and reflected beams
of light.
Iridescence of blue-black grassquit R. Maia et al. S205
J. R. Soc. Interface (2009)
Shawkey et al. 2006). Model 1 takes into account all
three interfaces and the thickness of both layers. Model 2
takes into account the two outermost interfaces and the
keratin cortex thickness. Model 3 takes into account
the two outermost interfaces and the thickness of the
melanin layer, thus considering that keratin thickness
does not influence the optical distance of the layer
(i.e. considered transparent). Finally, model 4 takes into
account only the two innermost interfaces and the
thickness of the second layer, once again considering
the keratin layer transparent. Although model 1 is the
only model that does not exclude any structures deemed
relevant, combining these models allows conjectures
about the relative importance of each layer to colour
production, since we can assess the effects of removing
individual components on the produced spectra.
In all the models, the melanin layer was considered to
be immediately below and parallel to the keratin layer,
ultimately defining its width. For simplicity, all models
were generated considering smooth layers. We initially
considered published estimates of real refractive indexes
(n) and extinction coefficients (k)ofair(nZ1.00,
kZ0.00), keratin (nZ1.56, kZ0.03) and eumelanin
(nZ2.00, kZ0.6) in all calculations (Land 1972;
Brink & van der Berg 2004). Since spectra obtained
from these models are very disparate, we visually
compared model fit to the observed data (Brink & van
der Berg 2004;Doucet et al.2006;Shawkey et al.2006).
All models were calculated using a script developed
for R software (R Development Core Team 2007;
available at http://cran.r-project.org). This user-
friendly script is a valuable tool for the integrative
study of animal structural coloration, as it allows the
calculation of any combination of layers and refractive
indexes, as well as the four predefined models
considered in this study. The programming code is
available in the electronic supplementary material or
upon author request.
We considered the observed spectrum as the average of
the measured spectra of all individuals, and all models
used the average measurements from nanostructures of
these individuals. Since barbule tilting relative to the
feather axis may produce differences in measured bright-
ness (Osorio & Ham 2002), all comparisons were made
with spectra normalized to have integrals of 1.
We selected the best general model through visual
inspection. However, since we had a considerably larger
sample size compared to similar studies, we calculated
the 95% confidence intervals (CIs) for the estimates of
hue mean, for both normal and 458geometries, and also
for the estimate of hue shift with geometry owing to
iridescence (DHueZHue
90
KHue
45
). We used the
obtained intervals to estimategoodness offit from models
with subtle differences owing to changes in optical
properties of the component structures. Although a
similar approach has been used to assess the model fit to
observed reflection in the Madagascan sunset moth
(Yoshioka et al.2008), the use of CIs allows for some
uncertainty of the calculated values owing to both
natural variation and measurement accuracy, therefore
providing a more conservative and biologically relevant
estimate of model adherence.
3. RESULTS
3.1. Barbule microstructure
We obtained and measured 75 micrographs from
barbules of 25 adult male blue-black grassquits.
Although melanin granules were also found scattered
within the matrix, several were distributed and
organized in a seemingly discrete layer, beneath a
thin and uniform keratin layer (figure 2).
Our measurements revealed that the keratin cortex
is thinner than the outer melanin layer, with the latter
comprising one to five melanin granules in thickness
(table 1). Although this melanin layer was usually
thicker than one granule, we never observed keratin–
melanin layer stacks (i.e. sequences of discrete keratin
and melanin layers). Furthermore, the distance
between the melanin granules in this outer layer was
consistently lower than 300 nm, therefore not allowing
individual interaction with light in all avian visible
wavelengths (300–700 nm; Cuthill et al. 2000) and
conferring an organization typical of single-layer thin-
film optical systems.
3.2. Spectrometric measurements
The colour spectra obtained from the measurements of
the same 25 feathers that were analysed nanostructu-
rally were characterized by a single peak in the range of
avian visible wavelengths (figure 3). Changes in the
2µm
Figure 2. Transmission electron micrograph of a representa-
tive barbule from a dorsal feather of the nuptial plumage of
male blue-black grassquits (magnification 10 000!). Black
ovals are melanin granules and dark grey areas are keratin.
Table 1. Nanostructural measurements of male blue-black
grassquit feather barbules, obtained from TEM. (Each
sample corresponds to the average of six measurements
from three barbule micrographs of different feathers of the
same individual.)
variable meanGs.e.m. (nZ25)
keratin cortex thickness (nm) 126.57G2.08
melanin layer thickness (nm) 421.76G12.29
melanin granules in layer 2.45G0.06
melanin granule diameter (nm) 177.69G5.17
S206 Iridescence of blue-black grassquit R. Maia et al.
J. R. Soc. Interface (2009)
measurement geometry shifted the wavelength of
peak reflectance, such that an increase in the angle
of incidence produced a decrease in hue. There was no
overlap in the CI estimates of mean hue at both
measured geometries (table 2).
3.3. Thin-film optical modelling
Two of the models considering the outer melanin layer
(models 1 and 2 in figure 4a,b) resulted in spectra
similar to the ones measured empirically. The shape of
the curve and the hue of the modelled spectra had a
good fit relative to the spectra measured at both
measurement geometries,andbothmodelledand
measured curves had similar behaviour with the
changing angle of light incidence, with hues shifting
to shorter wavelength and more saturated spectra
(i.e. greater spectral purity) owing to the increase in
the angle of incidence relative to the normal. Therefore,
the disposition of melanin granules in an organized
fashion seems indispensable for the production of the
observed colour by coherently scattering light.
Model 2 does not consider the width of the melanin
layer or the fact that light reaches the interface between
the melanin layer and the keratin matrix beneath it.
Hence, the similar behaviour of models 1 and 2
(resulting in nearly identical curves in figure 4a,b)
reveals that the melanin layer has isolating properties.
Models 3 and 4, which do not consider the keratin
cortex width, resulted in very different spectra from the
ones measured from blue-black grassquit feathers,
therefore highlighting the importance of the keratin
cortex to the production of the observed colour.
Owing to the considerable inter-individual variation
in the number of granules composing the melanin layer
(figure 5a), which consequently affects the layer width,
we ran alternative versions of model 1, with varying
widths of the melanin layer based on granule diameter
and number of granules, to identify the effect of this
variation upon the resulting spectra. We considered
that the difference in the number of granules resulted
from differences in the deposition of granules from the
keratin medulla to the melanin layer, therefore affecting
melanin layer width but not keratin layer width. Based
on the average granule diameter, only two melanin
granules sufficed to isolate the keratin cortex from the
core of the barbule, since model 1 converges to model 2
when a two-granule-thick melanin layer is used
(figure 5b). However, this also indicates that, considering
an average keratin cortex thickness, a one-granule-thick
melanin layer can be penetrated by light, which
interacts with the second melanin–keratin interface
and produces a colour spectrum with hue shifted to
longer wavelengths relative to a similar barbule with a
two-granule-thick melanin layer.
Despite these procedures, we still found some small-
scale differences between the observed and modelled
spectra that could not be explained by variation in the
wavelen
g
th (nm)
400 500 600 700
reflectance (arb. units)
0
0.001
0.002
0.003
0.004
0.005
Figure 3. Reflectance spectra of feathers from the nuptial
plumage of male blue-black grassquits measured at normal
(solid line) and 458(dashed line) measurement geometries
(dotted linesZ95% CI; nZ25).
Table 2. Average values and estimated confidence limits for
the hue of reflectance spectra obtained from feathers of male
blue-black grassquits at two different measurement
geometries, and for the difference in hue between those two
measurements (DHue).
measurement
geometry (nZ25) mean (nm)
95% CI
lower
limit
(nm)
upper
limit
(nm)
normal 464.2 455.8 473.3
458413.4 403.4 423.3
DHue 51.2 44.2 58.1
wavelen
g
th (nm)
400 500 600 700
reflectance (arb. units)
0
0.001
0.002
0.003
0.004
0.005
0
0.001
0.002
0.003
0.004
0.005 (a)
(b)
Figure 4. Comparison of the reflectance spectra of feathers
from the nuptial plumage of male blue-black grassquits (black
solid line, mean; black dashed lines, 95% CI ) and the
predicted spectra from the four models (blue, model 1; red,
model 2; green, model 3; grey, model 4) based on average
measurements of barbule nanostructures (nZ25) at
(a) normal and (b)458measurement geometries.
Iridescence of blue-black grassquit R. Maia et al. S207
J. R. Soc. Interface (2009)
variables considered. A comparison of the initially
modelled spectra and the empirically obtained esti-
mates indicates that hue values do not fall within the
95% CI limits (table 3). Therefore, we investigated
whether the changes in keratin and melanin opacity
would affect the goodness of fit of the modelled spectra
relative to the measured one by running the thin-film
models with a broad range of refractive index and
extinction coefficient values.
Considerable changes to the extinction coefficient of
both keratin and melanin did not improve the
predictive ability of the thin-film models (figure 6a,b,
dashed lines). However, relatively small changes to the
refractive index of both materials resulted in models
with a better fit. Variation in the keratin refractive
index estimated that values falling between 1.58 and
1.63 generated models that fall within the calculated CI
(figure 6a, solid line). A melanin refractive index
between 1.75 and 1.90 also improved the model fit
(figure 6b, solid line). Combining both variations in the
refractive index and the extinction coefficient further
suggests that the changes in the refractive index are
more relevant for obtaining a good model fit (table 3).
Also, variation in hue values owing to changes in the
angle of incident light remained fairly stable regardless
of the changes in the melanin opacity, suggesting that
the iridescent properties of the selected model are
robust despite subtle variations in optical properties of
the materials.
4. DISCUSSION
Male blue-black grassquit feather barbules present a
single keratin layer over a layer comprising melanin
granules. This simple arrangement is sufficient to
produce iridescent coloration, as confirmed by thin-
film optical modelling. Shifts in hue and shape of the
reflected spectrum derived from changing angles of
light incidence were well explained by the modelled
differences in the optical path that light encounters at
each angle. This result reinforces the conclusion that
these barbules are photonic structures that interact with
light as predicted by the thin-film models of refraction.
Thin-film optical modelling also revealed that light
can only penetrate the melanin layer and interact with
the keratin core when the melanin layer is sufficiently
thin (fewer than two melanin granules, on average).
Although this configuration can be found in some male
grassquit feathers, most individuals have melanin
layers of two to three granules, with little variation
across average values. This suggests that plumage
development in this species optimizes colour properties
through the production and placement of sufficient
granules to maximize absorbance, rarely generating
more melanin granules than those strictly necessary for
this colour production process. Therefore, the con-
structive interference of light in the single keratin layer,
isolated from the core of the barbule by a sufficiently
thick melanin layer, optimizes the reflection of short
wavelengths (UV–blue) that produce the characteristic
coloration of the nuptial plumage of male blue-black
grassquits, without requiring any additional pigments
or structures.
Several studies have described structures similar to
those found in the blue-black grassquit, also by
applying thin-film optical models. For instance, male
Table 3. Predicted values of hue and hue shift with
measurement geometry (DHue) obtained from selected
models considering different values of melanin opacity.
(Values in italics indicate hues that fall within the 95% CI
shown in table 2.)
model optical properties angle of light incidence
refractive
index (n)
extinction
coefficient
(k)
normal
(nm) 458(nm)
DHue
(nm)
2.0 0.6 447 396 51
0.4 437 384 53
0.8 453 402 51
1.8 0.6 466 414 52
0.4 453 406 47
0.8 470 417 53
2.2 0.6 432 385 47
0.4 420 377 43
0.8 440 391 49
melanin
g
ranules in outer la
y
er
1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4
frequency
0
1
2
3
4
5
6
7
wavelen
g
th (nm)
400 500 600 700
reflectance (arb. units)
0
0.001
0.002
0.003
0.004
0.005
(a)(b)
Figure 5. (a) Frequency distribution of the average number of melanin granules in each individual’s melanin layer and
(b) predicted spectra for a barbule with an average-width keratin layer measured for male blue-black grassquits, with total
isolation by the melanin layer (solid line, model 2) and considering a melanin layer of thickness of one (dashed line) or two
(dot-dashed line) average-sized melanin granules.
S208 Iridescence of blue-black grassquit R. Maia et al.
J. R. Soc. Interface (2009)
satin bowerbirds (Doucet et al. 2006), as well as
some grackles and blackbirds (Shawkey et al. 2006),
also produce iridescent colours with peak reflectance in
short wavelengths owing to a keratin cortex over a
single melanin layer. However, in other species such as
the hadeda ibis (Bostrychia hagedash), the melanin
platelets are hollow and could provide an extra inter-
face between melanin and air, but the melanin was
found to be highly opaque so that the resulting colour is
produced by the keratin cortex alone (Brink & van der
Berg 2004). The keratin layer in the barbules of this ibis
is over four times thicker than that of male blue-black
grassquits, resulting in a very different structural colour
with up to four peaks in the avian visible spectra.
The role commonly attributed to melanin in the
production of both iridescent and non-iridescent struc-
tural colours is that of absorbing incoherently scattered
light waves, therefore accentuating colour properties,
especially saturation (Prum 2006;Shawkey & Hill
2006). The relationship found between the colour and
melanin granule density in male satin bowerbirds
(Doucet et al.2006) is an example of this property in
iridescent feathers. This is similar to the case of an
amelanotic Steller’s jay (Cyanocitta stelleri ), which
exhibits a non-iridescent blue colour. In this example,
the absence of melanin in the quasi-ordered structure of
feather barbs does not allow the absorbance of
these incoherently scattered wavelengths, resulting in
white-coloured feathers (Shawkey & Hill 2006).
The melanin layer is usually considered to have very
little influence in coherent scattering and the resulting
production of iridescent colours, absorbing most of the
light that reaches it and therefore serving mostly to
define the thickness of the keratin cortex (Brink &
van der Berg 2004;Doucet et al. 2006). However, as
shown for some cowbird species, if the melanin layer is
sufficiently thin, light can interact with it and reach
underlying layers (Shawkey et al. 2006). Similarly, in
the blue-black grassquit, the melanin layer is relatively
thin owing to the stacking of very few melanin granules.
Indeed, variation in the number of melanin granules
(i.e. layer thickness) found for males of this species may
result in an altogether different organization of layers
and interfaces interacting with light. Our data suggest
that if the melanin layer is on average thinner than
two melanin granules, light can reach the underlying
melanin–keratin interface. The result is a change in the
coherently reinforced wavelengths and, consequently,
in the observed coloration. Thus, in accordance with
the findings in species that display feathers with similar
structures, the role of melanin in the nanostructural
organization of blue-black grassquit feather barbs
includes not only absorbance of incoherently reflected
light, but also an active role in thin-film structural
organization. The fact that the melanin substrate in
iridescent feathers may absorb the most incident light
does not mean that its relevance to colour production
resides solely in this property. Its interface with the
keratin cortex may also contribute to differential
reinforcement of wavelengths when the melanin layer
is sufficiently thin.
Our results also show that even considering the most
adequate model selected by visual inspection, an
improvement in model fit to empirical data can be
achieved when optical properties of the materials are
considered and changed. Ultimately, an ideal approach
would be to objectively measure optical properties of
keratin and melanin, using these values to generate
predictive thin-film models. This also highlights the
need for an objective methodology for comparing the
predicted and empirically obtained colour spectra.
The simple methodology applied in this study points
to the potential traps of simply comparing colour
curves visually. Also, we found that although peak
reflectance changed, the overall distance between peak
reflectance measured at both geometries (DHue) varied
little with changes in opacity.
Altering the values of both materials’ refractive
index improved model fit by altering the optical path
length ratio between the two layers (Kinoshita &
Yoshioka 2005b). Thus, since melanin has a higher
refractive index than keratin, it requires larger changes
to achieve the necessary ratio to improve the model fit.
Keratin is usually considered to have a refractive index
of between 1.5 and 1.55, with Brink & van der Berg
(2004) estimating values of 1.56–1.58 for the hadeda
ibis. This renders our estimates even higher than most
studies on insect cuticles and bird feathers. Melanin
optical properties, on the other hand, have received
considerably less attention, and little has changed since
Land’s (1972) suggestion of a refractive index of
approximately 2.0. Brink & van der Berg (2004) tested
380 400 420 440 460 480 500 520
keratin k
0
0.02
0.04
0.06
0.08
0.10
keratin n
1.40
1.45
1.50
1.55
1.60
1.65
1.70
hue (nm)
400 450 500 550 600
melanin k
0.4
0.6
0.8
1.0
melanin n
1.0
1.5
2.0
2.5
3.0
(a)
(b)
Figure 6. Predicted values of hue under normal incidence by
thin-film optical models considering different values for the
refractive index (n; solid lines) and the extinction coefficient
(k; dashed lines) of (a) keratin and (b) melanin. Vertical
dashed lines indicate the mean value of hue and its 95% CI
(grey area), as estimated from the reflectance spectra obtained
at normal incidence from feathers of male blue-black grassquits.
Iridescence of blue-black grassquit R. Maia et al. S209
J. R. Soc. Interface (2009)
different values for the extinction coefficient in the
hadeda ibis and suggested a minimal value of 0.6, in
which the melanin layer width lies around 70 nm.
We consider that taking a different value for the
refractive index of melanin can be biologically more
meaningful, since the optical properties available in the
literature actually refer to eumelanin, but both
eumelanin and phaeomelanin occur in all bird species
studied to date (McGraw 2006b). Since phaeomelanin
displays 10–30 per cent less absorbance in all wave-
lengths (Krishnaswamy & Baranoski 2004), a mixture
of both would result in a change to the refractive index
in the direction we have estimated, with little change to
the extinction coefficient, for the difference in absor-
bance between the melanin types is somewhat consist-
ent across wavelengths. Further studies should also
consider that the differences in the amount of each type
of melanin might even account for interspecific vari-
ation in estimated refractive indexes.
We thank Matthew D. Shawkey, Shinya Yoshioka, Ju
´nio
C. R. Cruz and Liliane A. Maia for insightful discussions
about thin-film modelling, Eduardo Leoni for clarifications
while developing the R script, Marina Ancia
˜es, Valdir
Pessoa and two anonymous reviewers for the extremely
valuable comments on the earlier versions of this manu-
script, and Roge
´rio Lionzo from Mopa Studio for assistance
in creating the schematic drawings. We also thank Arizona
State University and all organizers of the ‘Iridescence: more
than meets the eye’ conference for the opportunity to
participate in the meeting and in this special issue. This
work was funded by the Animal Behavior Society Develop-
ing Nations Research Grant, National Geographic Society,
Fundac¸a
˜o de Apoio a Pesquisa, CAPES/CNPq and Uni-
versidade de Brası
´lia.
REFERENCES
Aguilar, T. M., Maia, R., Santos, E. S. A. & Macedo, R. H.
2008 Parasite levels in blue-black grassquits correlate with
male displays but not female mate choice. Behav. Ecol. 19,
292–301. (doi:10.1093/beheco/arm130)
Alonzo-Alvarez, C., Doutrelant, C. & Sorci, G. 2004
Ultraviolet reflectance affects male–male interactions in
the blue tit (Parus caeruleus ultramarinus). Behav. Ecol.
15, 805–809. (doi:10.1093/beheco/arh083)
Andersson, M. 1994 Sexual selection. Princeton, NJ: Prince-
ton University Press.
Andersson, S. 1999 Morphology of UV reflectance in a
whistling-thrush: implications for the study of color
signaling in birds. J. Avian Biol. 30, 193–204. (doi:10.
2307/3677129)
Badyaev, A. V. & Hill, G. E. 2002 Paternal care as a
conditional strategy: distinct reproductive tactics associ-
ated with elaboration of plumage ornamentation in the
house finch. Behav. Ecol. 13, 591–597. (doi:10.1093/
beheco/13.5.591)
Bagnara, J. T., Fernandez, P. J. & Fuji, R. 2007 On the blue
coloration of vertebrates. Pigm. Cell Res. 20, 14–26.
(doi:10.1111/j.1600-0749.2006.00360.x)
Banks, A. N. 2001 For your eyes only? The role of UV in mate
choice. Trends Ecol. Evol. 16, 473–474. (doi:10.1016/
S0169-5347(01)02245-5)
Brink, D. J. & van der Berg, N. G. 2004 Structural colors from
the feather of the bird Bostrychia hagedash.J. Phys. D:
Appl. Phys. 37, 813–818. (doi:10.1088/0022-3727/37/5/025)
Carvalho, C. B., Macedo, R. H. & Graves, J. 2006 Breeding
strategies of a socially monogamous Neotropical passerine:
extra-pair fertilizations, behavior and morphology. Con-
dor 108, 579–590. (doi:10.1650/0010-5422(2006)108[579:
BSOASM]2.0.CO;2)
Costa, F. J. V. & Macedo, R. H. 2005 Coccidian oocyst
parasitism in the blue-black grassquit: influence on
secondary sex ornaments and body condition. Anim.
Behav. 70, 1401–1409. (doi:10.1016/j.anbehav.2005.03.024)
Cuthill,I.C.2006Colorperception.InBird coloration (eds G. E.
Hill & K. J. McGraw). Mechanisms and measurements, vol. I,
pp. 3–40. Cambridge, MA: Harvard University Press.
Cuthill, I. C., Partridge, J. C., Bennett, A. T. D., Church,
S. C., Hart, N. S. & Hunt, S. 2000 Ultraviolet vision in
birds. Adv. Stud. Behav. 29, 159–214. (doi:10.1016/S0065-
3454(08)60105-9)
Doucet, S. M. 2002 Structural plumage coloration, male body
size, and condition in the blue-black grassquit. Condor
104, 30–38. (doi:10.1650/0010-5422(2002)104[0030:SPC
MBS]2.0.CO;2)
Doucet, S. M., Shawkey, M. D., Hill, G. E. & Montgomerie, R.
2006 Iridescent plumage in satin bowerbirds: structure,
mechanisms and nanostructural predictors of individual
variation in color. J. Exp. Biol. 209, 380–390. (doi:10.1242/
jeb.01988)
Eaton, M. D. & Lanyon, S. M. 2003 The ubiquity of avian
ultraviolet plumage reflectance. Proc. R. Soc. B 270,
1721–1726. (doi:10.1098/rspb.2003.2431)
Figuerola, J., Munoz, E., Gutierrez, R. & Ferrer, D. 1999
Blood parasites, leucocytes and plumage brightness in the
cirl bunting, Emberiza cirlus.Funct. Ecol. 13, 594–601.
(doi:10.1046/j.1365-2435.1999.00354.x)
Greenwalt, C. H., Brandt, W. & Friel, D. D. 1960 Iridescent
colors of hummingbird feathers. J. Opt. Soc. Am. 50,
1005–1013. (doi:10.1364/JOSA.50.001005)
Ha˚stad, O. & O
¨deen, A. 2008 Different ranking of avian colors
predicted by modeling of retinal function in humans and
birds. Am. Nat. 171, 831–838. (doi:10.1086/587529)
Haussmann, F., Arnold, K. E., Marshall, N. J. & Owens,
I. P. F. 2003 Ultraviolet signals in bird are special. Proc.
R. Soc. B 270, 61–67. (doi:10.1098/rspb.2002.2200)
Jellison Jr, G. E. 1993 Data analysis for spectroscopic
ellipsometry. Thin Solid Films 234, 416–422. (doi:10.
1016/0040-6090(93)90298-4)
Kinoshita, S. & Yoshioka, S. 2005aFundamental optical
processes underlying the structural colors. In Structural
colors in biological systems: principles and applications
(eds S. Kinoshita & S. Yoshioka), pp. 3–26. Osaka, Japan:
Osaka University Press.
Kinoshita, S. & Yoshioka, S. 2005bStructural colors in
nature: a role of regularity and irregularity in the
structure. ChemPhysChem 6, 1443–1459. (doi:10.1002/
cphc.200500007)
Kinoshita, S., Yoshioka, S. & Miyazaki, J. 2008 Physics of
structural colors. Rep. Prog. Phys. 71, 076401(1-30).
(doi:10.1088/0034-4885/71/7/076401)
Krishnaswamy, A. & Baranoski, G. V. G. 2004 A bio-
physically-based spectral model of light interaction with
human skin. Comp. Graph. Forum 23, 331–340. (doi:10.
1111/j.1467-8659.2004.00764.x)
Land, M. F. 1972 The physics and biology of animal reflectors.
Prog. Biophys. Mol. Biol. 24, 77–106. (doi:10.1016/0079-
6107(72)90004-1)
McGraw, K. J. 2006aMechanics of carotenoid-based
coloration. In Bird coloration (eds G. E. Hill & K. J.
McGraw). Mechanisms and measurements, vol. I,
pp. 177–242. Cambridge, MA: Harvard University Press.
S210 Iridescence of blue-black grassquit R. Maia et al.
J. R. Soc. Interface (2009)
McGraw, K. J. 2006bMechanics of melanin-based coloration.
In Bird coloration (eds G. E. Hill & K. J. McGraw).
Mechanisms and measurements, vol. I, pp. 243–294.
Cambridge, MA: Harvard University Press.
McGraw, K. J. 2006cMechanics of uncommon colors: pterins,
porphyrins and psittacofulvins. In Bird coloration (eds
G. E. Hill & K. J. McGraw). Mechanisms and measure-
ments, vol. I, pp. 354–398. Cambridge, MA: Harvard
University Press.
Osorio, D. & Ham, A. D. 2002 Spectral reflectance and
directional properties of structural coloration in bird
plumage. J. Exp. Biol. 205, 2017–2027.
Prum, R. O. 2006 Anatomy, physics, and evolution of avian
structural colors. In Bird coloration (eds G. E. Hill & K. J.
McGraw). Mechanisms and measurements, vol. I,
pp. 295–355. Cambridge, MA: Harvard University Press.
Prum, R. O. & Torres, R. H. 2003 A Fourier tool for the
analysis of coherent light scattering by bio-optical
nanostructures. Integr. Comp. Biol. 43, 591–602. (doi:10.
1093/icb/43.4.591)
Prum, R. O., Torres, R., Williamson, S. & Dyck, J. 1998
Constructive inference of light by blue feather barbs.
Nature 396, 28–29. (doi:10.1038/23838)
Rasband, W. S. 1997–2007 IMAGE J. Bethesda, MD: US National
Institutes of Health. See http://rsb.info.nih.gov/ij/.
R Development Core Team 2007 R: a language for data
analysis and graphics. Vienna, Austria: R Foundation for
Statistical Computing. See http://cran.r-project.org/.
Searcy, W. & Nowiki, S. 2005 The evolution of animal
communication. Princeton, NJ: Princeton University Press.
Shawkey, M. D. & Hill, G. E. 2006 Significance of a basal
melanin layer to production of non-iridescent structural
plumage color: evidence from an amelanotic Steller’s jay
(Cyanocitta stelleri ). J. Exp. Biol. 209, 1245–1250. (doi:10.
1242/jeb.02115)
Shawkey, M. D., Hauber, M. E., Estep, L. K. & Hill, G. E.
2006 Evolutionary transitions and structural mechanisms
of avian plumage coloration in grackles and allies
(Icteridae). J. R. Soc. Interface 3, 777–783. (doi:10.1098/
rsif.2006.0131)
Shawkey, M. D., Saranathan, V., Pa
´lsdo
´ttir, H., Crum, J.,
Ellisman, M. H., Auer, M. & Prum, R. O. 2009 Electron
tomography, three-dimensional Fourier analysis and
colour prediction of a three-dimensional amorphous bio-
photonic nanostructure. J. R. Soc. Interface 6, S213–S220.
(doi:10.1098/rsif.2008.0374.focus)
Srinivasarao, M. 1999 Nano-optics in the biological world:
beetles, butterflies, birds and moths. Chem. Rev. 99,
1935–1961. (doi:10.1021/cr970080y)
Webber, T. 1985 Songs, displays and other behavior at a
courtship gathering of blue-black grassquit. Condor 87,
543–546. (doi:10.2307/1367957)
Yin, H., Shi, L., Li, Y., Quin, Y., Dong, B., Meyer, S., Liu, X.,
Zhao, L. & Zi, J. 2006 Iridescence in the neck feathers of
domestic pigeons. Phys. Rev. E 72, 051916(1-6). (doi:10.
1103/PhysRevE.74.051916)
Yoshioka, S., Nakamura, E. & Konoshita, S. 2007 Origin of
two-color iridescence in rock dove’s feather. J. Phys. Soc.
Jpn 76, 013801(1-4). (doi:10.1143/JPSJ.76.013801)
Yoshioka, S., Nakano, T., Nozue, Y. & Kinoshita, S. 2008
Coloration using higher order optical interference in the
wing pattern of the Madagascan sunset moth. J. R. Soc.
Interface 5, 457–464. (doi:10.1098/rsif.2007.1268)
Zi, J., Yu, X., Li, Y., Hu, X., Xu, C., Wang, X., Liu, X. & Fu,
R. 2003 Coloration strategies in peacock feathers. Proc.
Natl Acad. Sci. USA 100, 12 576–12 578. (doi:10.1073/
pnas.2133313100)
Iridescence of blue-black grassquit R. Maia et al. S211
J. R. Soc. Interface (2009)
A preview of this full-text is provided by The Royal Society.
Content available from Journal of the Royal Society Interface
This content is subject to copyright.
