Red ﬁsh, blue ﬁsh: trade-offs between
pigmentation and immunity in Betta splendens
Ethan D. Clotfelter,
*Daniel R. Ardia,
*and Kevin J. McGraw
Department of Biology, Amherst College, Amherst, MA 01002, USA,
Department of Biology,
Franklin and Marshall College, Lancaster, PA 17604, USA, and
School of Life Sciences, Arizona State
University, Tempe, AZ 85287, USA
Carotenoid pigments are responsible for many examples of sexually attractive red, orange, and yellow coloration in animals and
play an important role in antioxidant and immune defenses. Because vertebrates cannot synthesize carotenoids, limited dietary
availability may impose a trade-off between maintaining ornamental coloration and health. We used an experimental approach to
test the carotenoid trade-off hypothesis in the ﬁghting ﬁsh Betta splendens, by examining whether carotenoid allocation strategies
differ among conspeciﬁcs that exhibit a gradient of body coloration from blue to red. We found that male redness is underlain
by carotenoids and that females preferred to associate with red males over blue males, suggesting a sexually-selected advantage to
being red. Moreover, we found strong experimental support for the carotenoid trade-off hypothesis, as individuals that varied in
color did not appear to allocate carotenoids equally to both immune response and coloration. Redder ﬁsh given supplemental
carotenoids increased in both immune response (to a phytohemagglutination challenge) and redness compared with controls.
In contrast, bluer ﬁsh given supplemental carotenoids did not become more red but instead beneﬁted immunologically more so
than either control or redder supplemented ﬁsh. These results enhance our understanding of the evolution and plasticity of
carotenoid mobilization and utilization pathways in animals. Key words: carotenoids, coloration, immune response, pteridines,
sexual selection. [Behav Ecol 18:1139–1145 (2007)]
Pigment-based colors are common visual signals in the an-
imal kingdom (Needham 1974). Carotenoid pigments are
widely used to produce red, orange, and yellow coloration,
especially in ﬁshes, lizards, and birds (Evans and Norris
1996; Macedonia et al. 2000; Hill and McGraw 2006a). In
many instances, these carotenoid-based colors are sexually
attractive to prospective mates (Kodric-Brown 1993; Blount
et al. 2003; Maan et al. 2006). Carotenoid pigments also serve
a variety of physiological roles (Vershinin 1999), with one
major function as an immunostimulant and antioxidant
(Bendich 1989; Chew 1993; McGraw and Ardia 2003). Because
carotenoid pigments are derived from dietary sources and
cannot be synthesized de novo, their availability is under nu-
tritional control in a variety of taxa (Grether et al. 1999;
Alonso-Alvarez et al. 2004; Hill and McGraw 2006b). A conse-
quence of this scarcity is that carotenoid allocation is condi-
tion dependent (Hill and Montgomerie 1994; von Schantz
et al. 1999), which leads to a presumed trade-off between alloca-
tion to functions such as coloration and immunity (Lozano 1994;
von Schantz et al. 1999; Faivre et al. 2003; Alonso-Alvarez et al.
2004; Peters et al. 2004). Individuals with the brightest colors
are presumed to be those that have sufﬁcient carotenoids for
meeting both immunological and coloration functions and
hence are the healthiest and most desirable mates.
Several types of studies have attempted to elucidate such
a carotenoid trade-off. At a very basic level, the fact that di-
etary supplementation with carotenoids enhances both immu-
nity and coloration (Blount et al. 2003; McGraw and Ardia
2003; Alonso-Alvarez et al. 2004) suggests that carotenoid-
limited animals must dedicate carotenoids more to one or
another function or suffer both somatically and sexually. Sec-
ond, experimental manipulations of health status in animals
that deposit carotenoids in bare parts (e.g., beaks, legs, and
ﬂesh) have shown that immunocompromised animals fade in
color (Faivre et al. 2003; Peters et al. 2004), suggesting that
carotenoids are retrieved from colorful tissues to ﬁght patho-
genic or parasitic challenges. These studies have not tested
the alternative that carotenoid deposition or metabolism was
instead impaired by the immune challenge and thus whether
trade-offs cause these color changes remains unknown. Third,
in perhaps the best test of carotenoid trade-offs to date (Fitze
et al. 2007), it was recently found that when 2 types of caro-
tenoids (xanthophylls and carotenes) were provided to nes-
tling great tits (Parus major), the carotenoids used in plumage
coloration (xanthophylls) were not the carotenoids that inﬂu-
enced immunocompetence (carotenes). Through all this, how-
ever, we still await a rigorous, experimental test of the
carotenoid trade-off hypothesis in adult animals that display
sexually attractive, pigment-based coloration.
In contrast to this prior work, an ideal system for testing
carotenoid trade-offs would be in species that show distinct
color morphs that vary in their carotenoid dependency or
their ability to mobilize carotenoids (Sinervo and Lively
1996; Craig and Foote 2001; Craig et al. 2005; Pryke and
Grifﬁth 2006) and thus may employ different carotenoid allo-
cation strategies. In such a system, a key prediction for the
carotenoid trade-off hypothesis would be that individuals
that lack or have reduced carotenoid coloration should allo-
cate relatively more dietary carotenoids to their immune
system. Moreover, animals with extensive carotenoid coloration
should suffer decreased immunocompetence compared with
less carotenoid-colored animals given the same level of carot-
enoid uptake, due to increased allocation to coloration.
Therefore, the goal of this study was to test whether geneti-
cally based intraspeciﬁc differences in body coloration affect
relative allocation of carotenoids to coloration versus immune
system. We did so by examining the effect of dietary carotenoid
*These authors contributed equally to this work.
Address correspondence to E.D. Clotfelter. E-mail: edclotfelter@
Received 12 February 2007; revised 30 August 2007; accepted 5
Advance Access publication 10 October 2007
The Author 2007. Published by Oxford University Press on behalf of
the International Society for Behavioral Ecology. All rights reserved.
For permissions, please e-mail: firstname.lastname@example.org
supplementation on a range of color phenotypes in the
Siamese ﬁghting ﬁsh (Betta splendens). Artiﬁcial selection on
B. splendens has produced a range of color morphs, particu-
larly reds and blues, which provides a powerful tool for exam-
ining intraspeciﬁc carotenoid allocation strategies.
Our initial objectives were to determine the pigment basis
for redness in B. splendens and the extent to which red body
color in males is attractive to females. We used traditional
biochemical methods to measure tissue concentrations of
carotenoids and pteridines (drosopterins), another class of
pigments that can contribute to red and orange coloration
in ﬁshes (Dupont 1958; Henze et al. 1977). Pteridines can be
synthesized de novo (Hurst 1980); thus, it is important to con-
sider the possibility that ﬁsh compensate for carotenoid scar-
city by using pteridines as skin colorants (Grether et al. 2001).
We predicted that both carotenoids and drosopterins would
contribute to red coloration. With respect to female mate
choice, we predicted that female B. splendens would favor
red males over blue males in a dichotomous choice test. Little
is known about female preferences for male redness in either
wild-type or domestic stocks of this species, but female prefer-
ences for carotenoid-dependent coloration in males have been
found in many other ﬁshes (Kodric-Brown 1993; Candolin
1999; Maan et al. 2006).
Our main objective, however, was to test the trade-off hy-
pothesis by comparing carotenoid allocation strategies as
a function of a ﬁsh’s initial body coloration. We supplemented
dietary carotenoids to individuals over a range of body colors
from blue to red (measured with UV–Vis spectrophotometry)
in order to determine how the need to devote pigments to
skin color affected the ability to allocate carotenoids to color-
ation at the expense of the immune system. First, we tested
whether carotenoid supplementation increased redness and
enhanced the inﬂammatory response to phytohemagglutinin
(PHA). Similar work has been done on guppies (Poecilia retic-
ulata) (Grether et al. 2004) and salmonids such as rainbow
trout (Oncorhynchus mykiss) and sockeye salmon (Oncorhynchus
nerka) (Amar et al. 2000; Craig and Foote 2001; Amar et al.
2004). Based on these studies and those cited above, we pre-
dicted that B. splendens given supplemental carotenoids would
become redder as well as elevate their immune response.
After establishing that carotenoids boost both immune re-
sponse and coloration (see Results), we tested a central pre-
diction of the trade-off hypothesis: that initial body coloration
affects allocation of supplemented carotenoids. Because ﬁsh
could allocate additional carotenoids to either immune activ-
ity or coloration, we predicted that redder individuals would
use supplemented carotenoids to augment both coloration
and immunity, whereas less- or nonred individuals would use
carotenoids to improve immunity and not coloration and thus
show greater increases in immune response and smaller
changes in coloration compared with redder ﬁsh.
MATERIALS AND METHODS
Fish were housed in individual, visually isolated 1-l beakers
ﬁlled with municipal tap water that had been subjected to
reverse osmosis and reconstituted to a conductivity of 100–
150 lS. Water was changed at a rate of 25% every other day.
Fish were maintained at 27 C and a 12:12 h light:dark cycle.
Female preferences for male color
To establish whether female B. splendens have a mating pref-
erence for red males, we conducted a dichotomous mate
choice test in the laboratory. We obtained sexually mature,
male B. splendens from a commercial supplier. We measured
standard length (SL) 60.01 mm with digital calipers. Five
uniformly red males and 5 uniformly blue males were selected
and matched for size (SL red ¼37.61 60.56 mm, SL blue ¼
37.68 60.52 mm; t
¼0.10, P¼0.92). They were also
qualitatively matched for temperament, as measured by ago-
nistic responsiveness to their mirror image (Clotfelter et al.
2006). Female mate choice was assessed by placing focal fe-
males in a 15 315 330-cm tank; test males were housed in
two 15 315 315-cm tanks placed perpendicularly to the
female choice tank such that they could not see each other.
Females acclimated to the choice tank for 5 min, after which
time they were allowed to view males for a 5-min prechoice
period to ensure that they visited both males before data col-
lection began. We then measured the time spent by females
(in seconds) in each third of the tank during a 5-min choice
period. The positions of red and blue males were alternated
between the left and right tanks to eliminate the effects of
potential side biases by females.
To the human eye, predominant body coloration ranged from
blue to purple to red. Fish were paired with respect to these
categories of body coloration (see below for spectral analyses
of skin color), and then one member of each pair was
randomly assigned to the experimental (carotenoid supple-
mented) and control treatment groups. The carotenoid-
supplemented diet (in ﬂake form) contained the following
(percent by mass): spray-dried white ﬁsh meal (20%), wheat
ﬂour (20.2%), vegetable oil (0.9%), vitamins including vita-
min A in palmitate form (0.5%), water (58.1%), and b-carotene
(0.3%); a similar diet has been used for guppies (Grether et al.
2004). The control diet was identical with the exception that
a similar quantity of water was substituted for b-carotene.
Both diets were prepared by Ocean Star International, Inc.
(Snowville, UT) Sixty-one male B. splendens (28 controls and
33 carotenoid supplemented) were used in our diet experi-
ment. Fish were fed a ration equivalent to 5% of their body mass
twice daily for 8 weeks. Post hoc comparisons showed that ex-
perimental groups did not differ in either initial coloration or
body size (see Results).
Before and after the 8-week diet treatment, we measured the
reﬂectance of each ﬁsh using an Ocean Optics USB2000 spec-
trometer connected to a PX-2 pulsed xenon light. Fish were
removed from their home beakers and immobilized against
a moist sponge. We measured reﬂectance from a 1-cm diam-
eter region of the left side of the caudal peduncle for 10 s. We
assessed reﬂectance at 5-nm intervals over the wavelength
range of 300–700 nm using a 400-lm reﬂection probe (Ocean
Optics R400-7) held at a 45angle 5 mm from the sample
(Lahti 2006). Integration time was set at 100 ms, and reﬂec-
tance was averaged over 100 scans; boxcar smoothing was set
to 5. We standardized measurements with a diffuse tile made
of polytetraﬂuoroethylene that reﬂects .98% of light over all
sampled wavelengths (Ocean Optics WS-1).
We summarized reﬂectance data using principal compo-
nents (PCs) analysis (Jolliffe 1986), thought to be the most
appropriate means of reducing spectrophotometric data for
analysis (Cuthill et al. 1999). We reduced color data to 3 PCs,
which explained 96.2% of the variance in the sample, thus
creating 3 independent measures of color. PC1 (hereafter
‘‘brightness’’) explained 71.8% of the variance in the sample,
loaded negatively across the entire (300–700 nm) range of
wavelengths, and corresponded to differences in brightness
(Endler 1990). Individuals with high values of PC1 were those
with low brightness (i.e., low reﬂectance). PC2 (hereafter
1140 Behavioral Ecology
‘‘redness’’) explained 17.2% of the variation and loaded neg-
atively between 320 and 520 nm and highly positively between
600 and 700 nm, thus making PC2 an assessment of red versus
blue coloration. Higher values of PC2 were redder individuals.
Plots of PC loadings versus wavelength are shown in Figure 1.
We validated PC2 as an index of redness by regressing the
ﬁnal PC2 value for each ﬁsh on the wavelength at peak re-
ﬂectance from our reﬂectance curves, which yielded a signiﬁ-
cantly positive relationship (R
0.012). The reﬂectance curves we obtained from red ﬁsh were
qualitatively similar to those for the red ventral coloration of
threespine sticklebacks (Gasterosteus aculeatus) (Rush et al.
2003; Rick et al. 2004).
We assessed generalized cell-mediated immunity using expo-
sure to PHA, measured as an inﬂammatory response (Martin
et al. 2006). We previously validated this technique for use in
B. splendens (Ardia and Clotfelter 2006) by comparing the in-
ﬂammatory response of ﬁsh injected with PHA with those
injected with saline. Each individual was anesthetized in tri-
caine methanesulfonate and placed on a wet sponge under
a 6.33dissecting microscope. On the right side of the caudal
peduncle, 3–5 scales were removed to mark the injection site
for consistent measurements. Prior to injection, the thickness
of the caudal peduncle at the location of scale removal was
measured with a digital micrometer (60.001 mm accuracy)
3 times (F
¼34.3, P,0.0001, repeatability ¼0.94). After
measurements, each individual was injected at the location of
scale removal with 4 lg of PHA (L-8751, Sigma–Aldrich, St.
Louis, MO) in 2 ll of phosphate-buffered saline. After 24 h,
each ﬁsh was anesthetized again and the thickness of the
tissue at the location of injection and scale removal was re-
measured (repeatability F
¼22.1, P,0.0001, repeatabil-
ity ¼0.84). The response of each individual was recorded as
the difference (in millimeters) between postinjection thick-
ness and preinjection thickness (Smits et al. 1999).
At the end of the carotenoid supplementation period, we
euthanized ﬁsh and immediately removed a 0.5 30.5-cm sec-
tion of dermis and epidermis from the caudal peduncle; tissue
samples were stored at 80 C until analysis. Thawed tissue
was then ground in 2 ml methyl tertiary-butyl ether (MTBE)
for 2 min in a mixer mill (McGraw et al. 2003). The jar was
then rinsed with 1 ml MTBE to remove any residual pigment
and combined with the 2 ml extract in a 9 ml screw-cap glass
tube. We then added 2 ml of 1% NH
OH to the tube, vor-
texed it for 1 min, and then centrifuged it for 5 min at 3000
rpm. This method partitioned the carotenoids into the top
(MTBE) layer and the pteridines into the bottom (NH
layer. We used absorbance spectrophotometry separately on
the 2 fractions to determine carotenoid and pteridine concen-
trations based on standard calculations (McGraw et al. 2002).
Carotenoids in the MTBE fraction absorbed light maximally at
447 nm and presumably were yellow xanthophylls; pteridines
in the ammonium hydroxide fraction absorbed maximally at
490 nm and were presumably drosopterins (Grether et al.
2001). Thus, in our calculations, we used 2550 as the extinction
coefﬁcent for xanthophyll carotenoids (Bauernfeind 1981)
and 10 000 as the extinction coefﬁcient for drosopterins
(Wilson and Jacobsen 1977).
All variables met the assumptions of parametric statistics. We
used SAS 9.1.3 to conduct statistical analyses. We tested for the
effect of color on mate choice using a mixed model analysis of
variance (ANOVA) and compared initial allocation of ﬁsh to
treatments by color using an unpaired t-test. The effect of
supplementation on immune response, coloration, and tissue
pigment concentrations was tested using analysis of covari-
ance, with initial ﬁsh coloration and SL included as covariates.
Thus, we report least square means that account for the ef-
fects of these covariates. We included interaction terms in the
models that corresponded to our a priori predictions. Tests
were 2 tailed, and differences were considered signiﬁcant at
P,0.05. Means are shown with 6standard error (SE).
Pigment basis of coloration
To understand the basic relationship between pigments and
skin color in this species, we analyzed data for control ﬁsh
only. Tissue pigment concentrations were highly signiﬁcant
predictors of ﬁsh brightness (PC1) (F
0.0001). Drosopterins were positively correlated with PC1
¼15.8, b¼0.67, P¼0.001), indicating that high levels
of drosopterins were found in ﬁsh with low brightness. Caro-
tenoids had no relationship to ﬁnal brightness (F
b¼0.18, P¼0.29). Skin pigment concentrations were also
signiﬁcant predictors of ﬁsh redness (F
which varied with a continuous distribution deﬁned by PC2.
200 300 400 500 600 700 800
Principal Component 1 (“brightness”)
200 300 400 500 600 700 800
Principal Component 2 (“redness”)
Loadings of 2 PCs from a PCs
analysis of light reﬂected from
the caudal peduncle of male
Betta splendens, plotted against
light wavelength (nanometers).
(A) PC1 is a measure of skin
brightness, in which greater
values indicate lower bright-
ness (i.e., reﬂectance) and (B)
PC2 is a measure of redness.
Clotfelter et al. •Carotenoid trade-offs in Betta splendens 1141
Redder ﬁsh had more carotenoids (F
P¼0.03) but fewer drosopterins (F
Supplementation of B. splendens with dietary carotenoids
had weak effects on the concentrations of skin carotenoids
(controls ¼5.16 60.45 lg/g, N¼27; supplemented ¼5.96 6
0.33 lg/g, N¼32; t
¼1.46, P¼0.15) and drosopterins
(controls ¼3.95 60.49 mg/g, N¼26; supplemented ¼5.15 6
0.52 mg/g, N¼33; t
¼1.65, P¼0.10), tending to in-
crease both carotenoid and pteridine levels. However, dietary
carotenoid supplementation increased skin carotenoid con-
centrations in ﬁsh that were red at the start of the experiment
(initial PC2 score) (effect of supplementation: F
0.008; initial PC2 score: F
¼4.4, P¼0.04; initial PC2 3
¼6.8, P¼0.01). These ﬁsh also had
more drosopterins in their skin, but their drosopterin concen-
trations increased only marginally due to carotenoid supple-
mentation (effect of supplementation: F
initial PC2 score: F
¼4.3, P¼0.04); there was also no
interaction between initial PC2 score and the effect of supple-
¼0.08, P¼0.78). When we tested only ﬁsh
with initial positive redness (PC2) scores, we found a signiﬁ-
cant effect of carotenoid supplementation on skin carotenoid
¼2.7, P¼0.04; controls ¼5.17 10.41,
N¼13; supplemented ¼6.79 10.37, N¼18) but no differ-
ence in skin drosopterin concentrations (t
controls ¼4.75 10.7, N¼13; supplemented ¼5.34 10.6,
Female preference for red coloration in males
Female B. splendens (N¼23) spent signiﬁcantly more time in
the side of the tank nearest the red male (149.09 612.12 s) as
with the blue male (96.48 612.21 s) (Figure 2; mixed model
¼9.36, P¼0.004). Female preference was un-
affected by which pair of male stimulus ﬁsh we used (removal
of term led to no change in 2 log likelihood ratio). Note that
this was the only experiment in which ﬁsh were categorized
dichotomously (red vs. blue) rather than continuously (PC2).
Carotenoid supplementation boosts red coloration
Post hoc comparisons showed that we allocated ﬁsh to treat-
ment groups irrespective of their initial brightness (control ¼
0.90 61.34, N¼28; supplemented ¼0.27 61.53, N¼33;
¼0.31, P¼0.76), redness (control ¼0.59 60.78,
N¼28; supplemented ¼0.19 60.67, N¼33; t
P¼0.45), or SL (control ¼37.56 61.45 mm; supplemented ¼
39.06 60.34 mm; t
We were able to measure initial (before carotenoid supple-
mentation) and ﬁnal brightness and redness in 60 (27 con-
trols and 33 supplemented) of the 61 ﬁsh (erroneous ﬁnal
color measurement on one ﬁsh resulted in a signiﬁcant out-
lier; Cook’s distance ¼0.76). The carotenoid trade-off hypoth-
esis assumes that the delivery of supplemental carotenoids will
enhance coloration, and consistent with this, we found that
ﬁsh supplemented with carotenoids became signiﬁcantly
more red (PC2) (least square mean 6SE change in redness;
controls ¼2.13 60.51, N¼27; supplemented ¼0.33 6
0.46, N¼33; F
¼43.76, P,0.0001). Carotenoid supple-
mentation did not, however, induce a signiﬁcant change in
skin brightness (PC1) (control ¼0.43 615.04, N¼27;
supplemented ¼1.58 615.24, N¼33; t
Carotenoid supplementation boosts immune response
A second key assumption of the trade-off hypothesis is that
carotenoid supplementation boosts immune activity. Also con-
sistent with this, we found that carotenoid supplementation
signiﬁcantly increased the immune response of male B. splendens,
as measured by the swelling of the caudal peduncle in re-
sponse to PHA injection (Figure 3; mean postinjection swell-
ing in mm 6SE: control ¼0.087 60.01 mm, N¼28;
supplemented ¼0.17 60.02 mm, N¼33; carotenoid supple-
mentation group: F
¼12.68, P¼0.001). We obtained sim-
ilar results when we used the ratio of preinjection swelling to
postinjection swelling as our dependent variable (data not
Initial coloration affects trade-offs between immune
response and coloration
In support of a key prediction for the carotenoid trade-off
hypothesis—namely, that animals with less carotenoid-dependent
coloration should devote comparatively more carotenoids to
an immune response than should animals with more such
coloration—we found that initial body coloration affected al-
location strategies of carotenoids between coloration and im-
munity. Fish with initially low PC2 values (more blue) showed
a larger increase in immune activity (PHA response) than did
redder ﬁsh when supplemented with carotenoids (Figure 4;
Blue male Neutral zone Red male
Female Betta splendens (N¼23) spent more time (F
0.004) associating with red males than they did with blue males.
Immune response to PHA (mm)
Fish supplemented with dietary carotenoids (N¼33) were able to
mount a signiﬁcantly greater immune response to the PHA injection
than were the ﬁsh on the control diet (N¼28; F
1142 Behavioral Ecology
overall model: F
¼7.83, P,0.0001; initial redness: F
6.78, P¼0.01; initial redness 3supplementation: F
P,0.01; supplementation: F
¼15.5, P¼0.0002). Carot-
enoid supplementation increased redness (least square mean 6
SE change in redness; controls ¼2.13 60.51, N¼27; sup-
plemented ¼0.33 60.46, N¼33; F
but only in ﬁsh that were initially more red (Figure 5; initial
¼116.6, P,0.0001; initial redness 3supple-
¼85.3, P,0.0001). Supplementation de-
creased ﬁsh brightness, as indicated by higher PC1 values
(least square mean 6SE change in brightness; controls ¼
1.12 60.82, N¼27; supplemented ¼2.58 60.73, N¼
33), such that reﬂectance in less-bright ﬁsh decreased even
further in response to supplementation than in ﬁsh with high-
er initial brightness (initial brightness: F
0.0001; initial brightness 3supplementation: F
P¼0.001). Neither immune responses nor SLs were signiﬁ-
cantly related to changes in redness or brightness when they
were included as covariates (F1.2, P0.17).
Organisms are predicted to make trade-offs when resources
are scarce and serve multiple functions. Carotenoids are
thought to be an example of a scarce resource; they must be
acquired through the diet, and they provide beneﬁts to both
health and sexually selected coloration. We report experimen-
tal evidence demonstrating carotenoid trade-offs in B. splendens;
male B. splendens displayed different carotenoid allocation
strategies based on their initial coloration. Unlike other spe-
cies in which this trade-off has been examined, where the
ability to maintain carotenoid-based coloration is condition
dependent and results in a range of red and less-red pheno-
types, male B. splendens have genetically determined color
morphs. Redder individuals (positive PC2 values) provided
with supplemental carotenoids showed an increased inﬂam-
matory response to PHA and greater redness, whereas bluer
individuals (negative PC2 values) showed no change in color-
ation and instead mounted an even greater immune response.
In other words, because bluer ﬁsh were faced with an inher-
ently more relaxed carotenoid trade-off for health versus col-
oration (fewer carotenoids devoted to color), they apparently
diverted their accumulated pool of carotenoids more to one
function (immune response) than to the other (color).
Role of carotenoids in coloration, mate choice,
We ﬁrst found support for 3 key assumptions of the caroten-
oid trade-off hypothesis in B. splendens: that carotenoid pig-
ments 1) are used for coloration, 2) are a predictor of female
preference, and 3) boost both immunity and coloration when
in abundance. We found that coloration in B. splendens is un-
derlain by both carotenoids and pteridines, as is true for many
other red or orange color patches in ﬁsh (Henze et al. 1977)
and lizards (Macedonia et al. 2000). Carotenoid-supplemented
ﬁsh became redder in color, and naturally redder ﬁsh had
higher carotenoid concentrations in their skin as a result of
the supplementation. Furthermore, carotenoid supplementa-
tion decreased ﬁsh brightness, providing further evidence
that dietary carotenoids were allocated to skin coloration.
There was also evidence that redder ﬁsh had greater drosop-
terin concentrations in their skin. Grether et al. (2001) re-
ported that skin pteridine concentrations in Trinidadian
guppies covaried positively with natural carotenoid availability
due to population genetic differences in drosopterin content.
Grether et al. (2005) also found that experimental manipula-
tions of dietary carotenoids can marginally and inversely im-
pact skin drosopterin concentrations. Clearly, more work is
needed to better understand the complementarities and com-
petitions between these 2 classes of integumentary colorants.
Second, we report a role of red coloration in sexual selec-
tion in this species. Female B. splendens display a preference
for associating with red males. Such female preferences for
carotenoid-dependent signals in males have been reported in
other ﬁshes (Kodric-Brown 1993; Candolin 1999; Maan et al.
-10 -5 0 5 10
Immune response to PHA (mm)
Fish that were initially bluer (negative PC2 values; see text for
explanation) and received the carotenoid-supplemented diet
showed a larger boost in immune activity than did redder ﬁsh
(positive PC2 values) or than did ﬁsh on the control diet (overall
¼7.83, P,0.0001; initial redness: F
0.01; initial redness 3supplementation: F
-10 -5 0 5 10
Change in Redness
Carotenoid supplementation increased redness (PC2; see text for
explanation), and the change in redness varied depending on initial
redness (initial redness: F
¼116.6, P,0.0001; initial redness 3
Clotfelter et al. •Carotenoid trade-offs in Betta splendens 1143
2006) but have not been previously reported for B. splendens
or any member of its perciform family (Osphronemidae).
Finally, we demonstrated that dietary supplementation with
carotenoids signiﬁcantly increased the ability of male B. splen-
dens to mount an inﬂammatory response to PHA injection.
Our results provide additional evidence for the immunoen-
hancing role of carotenoids in vertebrates and that environ-
mental scarcity of carotenoids may lead to the evolution of
allocation strategies. Moreover, our use of the generalized
swelling response to PHA as a metric of immune response
(Ardia and Clotfelter 2006) complements the humoral meas-
ures (Amar et al. 2004) or allografting outcomes (Grether
et al. 2004) used in other ﬁsh studies.
Experimental support for the trade-off hypothesis: the
effect of initial coloration
In our key test of the carotenoid trade-off hypothesis, we pro-
vided evidence that individuals vary in their carotenoid allo-
cation strategy depending on their degree of carotenoid-
based skin coloration. We found that redder ﬁsh appeared
to allocate their supplemental carotenoids to both immune
response and color, as they increased in redness over the sup-
plementation period and increased their inﬂammatory re-
sponse to PHA relative to control ﬁsh. In contrast, bluer ﬁsh
(individuals with negative PC2 scores) given supplemental
carotenoids did not change color but instead mounted
a greater inﬂammatory response to the PHA challenge than
observed in either control or redder supplemented ﬁsh.
This study is the ﬁrst to demonstrate that, within members
of the same sex, individuals whose coloration is less caroten-
oid based have a qualitatively different carotenoid allocation
strategy than do redder conspeciﬁcs. Grether et al. (2004)
have previously shown that carotenoid enhancement of im-
munity in guppies is sex speciﬁc because males—and not
females—have carotenoid-based coloration. Most studies that
have attempted to shed light on carotenoid trade-offs in ani-
mals have examined species whose yellow-to-red coloration is
purely carotenoid based and is environmentally (condition)
dependent (Blount et al. 2003; McGraw and Ardia 2003;
Alonso-Alvarez et al. 2004; Peters et al. 2004). In such systems,
carotenoid trade-offs have been more difﬁcult to evaluate be-
cause there is no obvious group of animals that is constrained
in carotenoid allocation (i.e., even drab animals can divert
supplemental pigments to both immunity and coloration).
While we realize that this genetic color polymorphism has
been derived through artiﬁcial selection, we believe that these
results provide a model for examining intra- and interspeciﬁc
differences in carotenoid allocation strategies, as artiﬁcially
selected ﬁsh represent extremes of a naturally occurring color
gradient. Wild B. splendens possess both blues and reds to
varying degrees, and many of the domestic ﬁsh we used in
this study fell along a similar continuum. Furthermore, we
showed that even the bluest ﬁsh had carotenoids in their
dermis, demonstrating their ability to develop carotenoid-
based pigmentation. Thus, selection for uniformly blue and
red coloration in domestic strains led to associated changes in
carotenoid allocation strategies, leading to changes in both
coloration and carotenoid usage. Applying this approach to
examining differences among populations or closely related
species that differ in the extent of carotenoid-based coloration
may help elucidate the rate and extent of change in the costs
and beneﬁts of allocating carotenoids to coloration versus
Overall, our results indicate that body coloration in poly-
chromatic species can have a strong effect on carotenoid
allocation strategies. Research investigating the role of evolu-
tionary trade-offs between sexual ornaments and immunity
should examine underlying differences in carotenoid alloca-
tion strategies that may be caused by differences in coloration.
Dean of Faculty’s ofﬁce at Amherst College (E.D.C.); School
of Life Sciences and College of Liberal Arts and Sciences at
Arizona State University (K.J.M.).
We thank Greg Grether for advice on ﬁsh diets and Mark Lamon and
Nanette Bunker at Ocean Star International, Inc. for producing our
experimental and control diets. The Jeff Podos laboratory, particularly
David Lahti, at the University of Massachusetts kindly loaned us their
spectrometer and their expertise. Maureen Manning provided out-
standing logistical support through many stages of this project. Addi-
tional thanks to Alexandria Brown, Katie Moravec, and Neron
Thomas for laboratory and animal care assistance. This research was
conducted with the approval of the Institutional Animal Care and
Use Committee of Amherst College. Anne Houde and 2 anonymous
referees provided helpful comments on earlier versions of the
Alonso-Alvarez C, Bertrand S, Devevey G, Gaillard M, Prost J, Faivre B,
Sorci G. 2004. An experimental test of the dose-dependent effect of
carotenoids and immune activation on sexual signals and antioxi-
dant activity. Am Nat. 164:651–659.
Amar EC, Kiron V, Satoh S, Okamoto N, Watanabe T. 2000. Effects of
dietary beta-carotene on the immune response of rainbow trout
Oncorhynchus mykiss. Fish Sci. 66:1068–1075.
Amar EC, Kiron V, Satoh S, Watanabe T. 2004. Enhancement of innate
immunity in rainbow trout (Oncorhynchus mykiss Walbaum) associ-
ated with dietary intake of carotenoids from natural products. Fish
Shellﬁsh Immunol. 16:527–537.
Ardia DR, Clotfelter ED. 2006. The novel application of an immuno-
logical technique reveals the immunosuppressive effect of phytoes-
trogens in Betta splendens. J Fish Biol. 68:144–149.
Bauernfeind. 1981. Carotenoids and colorants and vitamin A precur-
sors. New York: Academic Press.
Bendich A. 1989. Carotenoids and the immune response. J Nutr.
Blount JD, Metcalfe NB, Birkhead TR, Surai PF. 2003. Carotenoid
modulation of immune function and sexual attractiveness in zebra
ﬁnches. Science. 300:125–127.
Candolin U. 1999. Male–male competition facilitates female choice in
sticklebacks. Proc R Soc Lond B. 266:785–789.
Chew BP. 1993. Role of carotenoids in the immune response. J Dairy
Clotfelter ED, Curren LJ, Murphy CE. 2006. Mate choice and spawn-
ing success in the ﬁghting ﬁsh Betta splendens: the importance of
body size, display behavior and nest size. Ethology. 112:1170–1178.
Craig JK, Foote CJ. 2001. Countergradient variation and secondary
sexual color: phenotypic convergence promotes genetic divergence
in carotenoid use between sympatric anadromous and nonanadro-
mous morphs of sockeye salmon (Oncorhynchus nerka). Evolution.
Craig JK, Foote CJ, Wood CC. 2005. Countergradient variation in
carotenoid use between sympatric morphs of sockeye salmon (On-
corhynchus nerka) exposes nonanadromous hybrids in the wild by
their mismatched spawning colour. Biol J Linn Soc. 84:287–305.
Cuthill IC, Bennett ATD, Partridge JC, Maier EJ. 1999. Plumage re-
ﬂectance and the objective assessment of avian sexual dichroma-
tism. Am Nat. 153:183–200.
Dupont A. 1958. Pteridines in the scales of ﬁshes. Naturwissenschaf-
Endler JA. 1990. On the measurement and classiﬁcation of color in
studies of animal color patterns. Biol J Linn Soc. 41:315–352.
Evans MR, Norris K. 1996. The importance of carotenoids in signaling
during aggressive interactions between male ﬁremouth cichlids (Ci-
chlasoma meeki). Behav Ecol. 7:1–6.
Faivre B, Gre´goire A, Pre´ault M, Ce
¨zilly F, Sorci G. 2003. Immune
activation mirrored in a secondary sexual trait. Science. 300:103.
1144 Behavioral Ecology
Fitze PS, Tschirren B, Gasparini J, Richner H. 2007. Carotenoid-based
plumage colors and immune function: is there a trade-off for rare
carotenoids? Am Nat. 169:S137–S144.
Grether GF, Hudon J, Endler JA. 2001. Carotenoid scarcity, synthetic
pteridine pigments and the evolution of sexual coloration in gup-
pies (Poecilia reticulata). Proc R Soc Lond B. 268:1245–1253.
Grether GF, Hudon J, Millie DF. 1999. Carotenoid limitation of sexual
coloration along an environmental gradient in guppies. Proc R Soc
Lond B. 266:1317–1322.
Grether GF, Kasahara S, Kolluru GR, Cooper EL. 2004. Sex-speciﬁc
effects of carotenoid intake on the immunological response to al-
lografts in guppies (Poecilia reticulata). Proc R Soc Lond B. 271:45–49.
Grether GF, Kolluru GR, Rodd FH, de la Cerda J, Shimazaki K. 2005.
Carotenoid availability affects the development of a colour-based
mate preference and the sensory bias to which it is genetically
linked. Proc R Soc Lond B. 272:2181–2188.
Henze M, Rempeters G, Anders F. 1977. Pteridines in skin of xipho-
phorine ﬁsh (Poeciliidae). Comp Biochem Physiol B. 56:35–46.
Hill GE, McGraw KJ. 2006a. Bird coloration: function and evolution.
Cambridge (MA): Harvard University Press.
Hill GE, McGraw KJ. 2006b. Bird coloration: mechanisms and meas-
urements. Cambridge (MA): Harvard University Press.
Hill GE, Montgomerie R. 1994. Plumage colour signals nutritional
condition in the house ﬁnch. Proc R Soc London B. 258:47–52.
Hurst DT. 1980. An introduction to the chemistry and biochemistry of
pyrimidines, purines, and pteridines. New York: John Wiley.
Jolliffe IT. 1986. Principal components analysis. New York: Springer-
Kodric-Brown A. 1993. Female choice of multiple male criteria in
guppies: interacting effects of dominance, coloration and court-
ship. Behav Ecol Sociobiol. 32:415–420.
Lahti DC. 2006. Persistence of egg recognition in the absence of cuckoo
brood parasitism: pattern and mechanism. Evolution. 60:157–168.
Lozano GA. 1994. Carotenoids, parasites and sexual selection. Oikos.
Maan ME, van der Spoel M, Jimenez PQ, van Alphen JJM, Seehausen O.
2006. Fitness correlates of male coloration in a Lake Victoria cichlid
ﬁsh. Behav Ecol. 17:691–699.
Macedonia JM, James S, Wittle LW, Clark DL. 2000. Skin pigments and
coloration in the Jamaican radiation of Anolis lizards. J Herpetol.
Martin LB, Han P, Lewittes J, Kuhlman JR, Klasing KC, Wikelski M.
2006. Phytohemagglutinin-induced skin swelling in birds: histolog-
ical support for a classic immunoecological technique. Funct Ecol.
McGraw KJ, Ardia DR. 2003. Carotenoids, immunocompetence, and
the information content of sexual colors: an experimental test. Am
McGraw KJ, Hill GE, Parker RS. 2003. Carotenoid pigments in a mu-
tant cardinal: implications for the genetic and enzymatic control
mechanisms of carotenoid metabolism in birds. Condor. 105:
McGraw KJ, Hill GE, Stradi R, Parker RS. 2002. The effect of dietary
carotenoid access on sexual dichromatism and plumage pigment
composition in the American goldﬁnch. Comp Biochem Physiol B.
Needham AE. 1974. The importance of zoochromes. Berlin
Peters A, Delhey K, Denk AG, Kempenaers B. 2004. Trade-offs be-
tween immune investment and sexual signaling in male mallards.
Am Nat. 164:51–59.
Pryke SR, Grifﬁth SC. 2006. Red dominates black: agonistic signalling
among head morphs in the colour polymorphic Gouldian ﬁnch.
Proc R Soc Lond B. 273:949–957.
Rick IP, Modarressie R, Bakker TCM. 2004. Male three-spined stickle-
backs reﬂect in ultraviolet light. Behaviour. 141:1531–1541.
Rush VN, McKinnon JS, Abney MA, Sargent RC. 2003. Reﬂectance
spectra from free-swimming sticklebacks (Gasterosteus): social con-
text and eye-jaw contrast. Behaviour. 140:1003–1019.
Sinervo B, Lively CM. 1996. The rock-paper-scissors game and the
evolution of alternative male strategies. Nature. 380:240–243.
Smits JE, Bortolotti GR, Tella JL. 1999. Simplifying the phytohaemag-
glutinin skin-testing technique in studies of avian immunocompe-
tence. Funct Ecol. 13:567–572.
Vershinin A. 1999. Biological functions of carotenoids—diversity and
evolution. Biofactors. 10:99–104.
von Schantz T, Bensch S, Grahn M, Hasselquist D, Wittzell H. 1999.
Good genes, oxidative stress and condition-dependent signals. Proc
R Soc Lond B. 266:1–12.
Wilson TG, Jacobsen KB. 1977. Isolation and characterization of pter-
idines from heads of Drosophila melanogaster by a modiﬁed thin-layer
chromatography procedure. Biochem Genet. 15:307–319.
Clotfelter et al. •Carotenoid trade-offs in Betta splendens 1145