2009 SOUTHEASTERN NATURALIST 8(4):723–732
Gender- and Size-based Variation in Wing Color in Large
Milkweed Bugs (Oncopeltus fasciatus) in Georgia
Andrew K. Davis*
Abstract - Milkweed bugs are aposematically colored, with orange and black on
their forewings, and the degree of both colors varies among individuals. Despite
the attention given to the warning nature of this color, there has been little research
directed at this variation. In this study, the subtle variation in wing colors
of one species of milkweed bug, Oncopeltus fasciatus (Large Milkweed Bug) was
measured to determine if the color variation was related to sex or body size. Fiftyeight
bugs were hand-collected at three sites in northeast Georgia, and their wings
scanned using a slide scanner and measured digitally using image analysis software.
Wings of females were larger than those of males in general, and the color
analyses showed statistically significant differences in wing hue between males
and females. Females also had darker black wing sections than males, which
could be evidence of a sex-related difference in immune function. Regardless of
sex, wings of larger bugs had deeper orange color and darker black, which may
increase the aposematic contrast. Finally, several differences in wing color were
found between sites, suggesting either site-level variation in host-plant quality or
relatedness among individuals within sites. This study is the first to quantify in
detail the wing colors of milkweed bugs and forms the basis for future research
into this little-studied aspect of this insect.
Introduction
The brightly contrasting orange and black aposematic coloration of milkweed
bugs has long fascinated researchers (e.g., Berenbaum and Miliczky
1984, Bowdish and Bultman 1993, Prudic et al. 2007). Their toxicity stems
from feeding on plants in the Asclepias (milkweed) genus, and because
they can be easily reared in captivity (Feir 1974), these bugs, especially
Oncopeltus fasciatus Dallas (Large Milkweed Bug), are ideal subjects for
studying questions related to warning coloration. Despite the attention given
to warning coloration in this insect, there has been little work focusing on
the color itself, which varies from yellow to orange to red in the Large Milkweed
Bug (A.K. Davis, pers. observ.). Therefore, basic questions relating to
this variation have yet to be addressed, such as whether there are gender- or
size-based differences in wing color. Only one study has examined colors
in milkweed bug wings, though this was not the main focus of the project
(Rodriguez-Clark 2004). Rodriguez-Clark (2004) was the first to assess the
color variation of milkweed bugs in a scientific fashion, by visually scoring
the dorsal color of bugs on a 5-unit scale as light-yellow, yellow-orange,
*D.B. Warnell School of Forestry and Natural Resources, the University of Georgia,
Athens, GA 30602; akdavis@uga.edu.
724 Southeastern Naturalist Vol. 8, No. 4
orange, red-orange, or red. While her study focused on the heritability of this
trait across generations, it is interesting to note that she found no significant
differences in wing color between adult males and females.
The measurement of color in insects and other small animals in research
has become easier and more objective in recent years with the
availability of digital cameras, scanners, and image analysis software
(Davis, in press; Davis and Grayson 2007; Davis et al. 2004, 2005, 2007).
This approach to quantifying color, whereby images of subjects are taken
under standardized lighting and software is used to assess the subjects’
color in the images, allows for extremely subtle differences in color
shades between individuals to be measured, which often are not discernable
to the naked eye. Further, such minor differences have been shown
to be important biologically in other insect species (Davis 2009, Davis
et al. 2007). With image analysis software, colors are measured in three
quantities: the hue (i.e., the difference between red, blue, green, etc.),
saturation (i.e., the degree or intensity of a given color, such as the difference
between pink and red), and the brightness of the color. In digital
images, all pixels in the image contain this information, and with image
analysis, the average pixel values for selected surface areas in the image
(i.e., such as a butterfly wing) are calculated. Conveniently, this breakdown
of colors into three parts allows each component to be separately
compared among individuals. For example, Davis et al. (2007) recently
found that the saturation alone of the orange color of Danaus plexippus
L. (Monarch Butterfly) wings was an important predictor of male mating
success, while the hue and brightness components were not important. In
addition, hue scores of migrant Monarchs appear to be different from that
of breeding and overwintering individuals (Davis 2009).
The degree of melanism (i.e., blackness) in insects is also an area where
image analysis can be utilized to objectively compare individuals, especially
in insects (Davis et al. 2005). Besides having three color components, all
pixels in digital images also have a “density” value, which is the brightness
value when the color information is removed and the image converted
to a greyscale form. Thus, the degree of “blackness” of a selected surface
(on a wing section, for example) can be scored as the mean density value
of all pixels in the selection. Similar systems have been used previously to
measure melanization levels in butterfl y wings (Davis et al. 2005, Ellers and
Boggs 2003) and beetle elytra (Thompson et al. 2002).
The current study is an examination of the natural variation in wing color
of Large Milkweed Bugs using an image analysis approach. Wild adults were
collected from three different locations in northeast Georgia, and both their
orange and black colors were assessed by scanning their wings with a fl atbed
scanner and using image analysis software. The possible variation in orange
color and melanism were then compared among sexes, as well as in relation
to body size.
2009 A.K. Davis 725
Methods
Collecting specimens
Adult Large Milkweed Bugs were collected by hand from Asclepias incarnata
L. (Swamp Milkweed) plants at three sites each separated by 5 km
around the city of Athens, Clarke County, GA during one week in August,
2006. At each site, there were between 5–15 plants and at least 15 individuals
per site were collected. All bugs were placed in plastic containers for
transport back to the lab, where they were killed by freezing. The sex of
each bug was identified following Rodriguez-Clark (2004), using the caudal
median point on the posterior margin of the 4th abdominal sternite, and the
triangular cleft pygidium.
Scanning wings
Before scanning, all bugs were held at room temperature for 20 minutes
to thaw. For each specimen, the left and right forewings were removed and
placed with the dorsal side down on an HP Scanjet 4670 see-thru vertical
scanner with a 35-mm slide scanner adapter. The slide scanner adapter was
ideal for scanning the wings because it 1) is designed to scan small items, and
2) illuminated the background, which standardized the amount of lighting
for every scan. The wings were thusly scanned at 1200 dpi, and the image
saved (Fig. 1). After all wings were scanned, a standard metric ruler was
scanned with the same settings to calibrate the image analysis software.
Figure 1. Diagram showing Large Milkweed Bug forewing characters measured.
Forewings from Oncopeltus faciatus (Large Milkweed Bug) (A) were removed from
the body and scanned face down (B), and their lengths, widths, and surface areas
measured with image analysis software (see methods). Next the orange and black
sections of the wings were digitally isolated (C), and color measure routines were
used to obtain average hue, saturation, and brightness values of orange sections and
average density values for black sections (D).
726 Southeastern Naturalist Vol. 8, No. 4
Measuring wing characteristics
The Fovea Pro image analysis software (Reindeer Graphics, Inc.) was
used to measure all forewings in this study, and image analysis methods
generally followed Davis et al. (2005) and Davis et al. (2007). Basic wingsize
traits were first measured for the left and right wings, including length
(mm), width (mm), and surface area (mm2). Then the left-right average of
each variable was calculated for each individual to use in analyses. For the
analyses of color traits (below), the wing area variable was used as the index
of body size.
Color features of the wings were measured as follows: all non-black
areas of the wings were selected first (Fig. 1C), and the Fovea Pro color
measure routine was initiated, which returned the average hue, saturation,
and brightness score for all pixels in the selection (typically over
10,000 pixels; Figure 1D). The same process was performed on the black
selections, but in this case the mean “density” values of all pixels were
used. Density is a computer-based, numerical value reflecting the degree
of “blackness” of a selection, and is useful as an index of “melanism” in
insects (Davis et al. 2005). In both cases, the orange and black values on
the left wings were measured first, then the right wings. Then the average
value of each was calculated to use in analyses, as was done with the size
measurements. The nature of the computer color-scoring was such that
hue values were scored on a scale from 0–360, while all other scores (saturation,
brightness, and density [melanism]) were on a scale from 0–255.
Note that in the melanism score, lower values represent darker, more intense
black colors (Davis et al. 2005).
Data analysis
Basic comparisons of wing size (wing length, width, area) between males
and females were made using Student’s t-tests. Analysis of covariance was
used to determine which variables related to wing color. Specifically, the
hue, saturation, brightness, and melanism were examined separately, and
in all cases, the independent variables included sex, site, and wing area as
a covariate. All two-way interactions were initially included in each model,
but were removed if found non-significant. All analyses were conducted using
Statistica 6.1 software (Statistica 2003).
Results
General results
A total of 58 adult Large Milkweed Bugs were collected and measured,
of which 26 (44.8%) were female and 32 (55.2%) were male (Table 1). Basic
comparisons of wing features between sexes revealed that female wings
were significantly larger than male wings in length, width, and total area
(Student’s t-test, P < 0.001 for all; Table 1). Moreover, this trend was consistent
across all three sites.
2009 A.K. Davis 727
Orange color results
In the analysis of orange hue scores, there was no support initially for
any of the two-way interaction terms in the ANCOVA model (P > 0.05).
Results from a simplified model with main effects only revealed no significant
effect of site (F2,53 = 1.56, P = 0.220), but a significant effect of
sex (F1,53 = 8.49, P = 0.005) and wing area (F1,53 = 4.56, P = 0.037). The
effect of sex was such that females tended to have higher hue scores (Fig.
2A), or in other words, were more yellow than males. The relationship
with wing size indicates that individuals with larger wings, regardless of
sex, tended to have lower hue scores, or were more orange, although this
Table 1. Summary of wing size measurements by sex and site. All sites were within 5 km of
Athens, GA and were plots of Asclepias incarnata (Swamp Milkweed). All individuals were
captured during one week in August, 2006. For all measurements, the average of the left and
right wings was used for each individual. Average wing values for each sex and site are shown,
with standard errors in parentheses. Asterisk denotes results of statistical comparisons (t-tests)
of traits of all females with all males.
Site Sex Wing length (mm) Wing width (mm) Wing area (mm2)
1 M 8.92 (0.23) 3.07 (0.06) 19.10 (0.82)
1 F 10.68 (0.66) 3.59 (0.24) 27.19 (3.30)
2 M 10.26 (0.18) 3.46 (0.06) 24.25 (0.71)
2 F 12.59 (0.20) 4.15 (0.07) 35.12 (1.08)
3 M 9.50 (0.39) 3.19 (0.11) 21.20 (1.50)
3 F 10.93 (0.25) 3.57 (0.07) 26.96 (1.06)
All sites M 9.60 (0.17) 3.26 (0.05) 21.68 (0.65)
All sites F 11.44* (0.26) 3.78* (0.09) 29.85* (1.26)
*P < 0.001.
Figure 2. Comparison of male and female color scores (hue, saturation, brightness,
and melanism [pixel density]). Shown are the mean values of all 32 males and 26
females, with standard error bars. Lower melanism scores represent darker shades
of black.
728 Southeastern Naturalist Vol. 8, No. 4
relationship was not strong in direct comparisons of the two variables (r =
-0.19, P = 0.163; Fig. 3A).
None of the two-way interactions were significant (P > 0.05) in the analysis
of orange saturation. In the model with main effects only, all three effects were
significant, although the differences in all cases were slight. Males tended to
have higher saturation scores than females (F1,53 = 7.94, P = 0.007; Fig. 2B),
meaning they were a deeper orange color, and there was a trend of higher scores
with increasing size of Large Milkweed Bugs (F1,53 = 7.66, P = 0.008; Fig. 3B),
so that bugs with larger wings tended to be deeper orange as well. There was
also differences in saturation between sites (F2,53 = 5.17, P = 0.009).
In the analysis of wing brightness, there were multiple significant interactions
and main effects, although their interpretation is difficult. The
interaction of sex*site (F2,48 = 4.10, P = 0.023) as well as visual inspection
of the categorized graph suggests a lack of consistency in the difference
between males and females for this trait. There was a similar lack of consistency
in the trends between brightness and wing area (i.e., in the wing
area*site interaction, F2,48 = 4.26, P = 0.020). The significant interaction between
sex and wing area (F1,48 = 4.23, P = 0.045) was such that while in both
sexes individuals with larger wings tended to have lower brightness scores,
the slope of the trend line was steeper in males than females (males: -1.02,
females: -0.27). Finally, there was a significant main effect of site (F2,48 =
4.50, P = 0.016) on brightness scores.
Figure 3. Relationships between Large Milkweed Bug wing size (average left and
right wing area) and all wing color variables in this study (hue, saturation, brightness,
and melanism [pixel density]). Lower melanism scores represent darker shades
of black.
2009 A.K. Davis 729
Melanism results
Interactions of site*sex and site*wing area were nonsignificant in the initial
model, but the interaction of sex*wing area was significant (F1,52 = 6.94,
P = 0.011). The main effect of sex approached significance (F1,52 = 3.80, P =
0.056; Fig. 2D), and there was a strong negative relationship with wing area
(F1,52 = 30.96, P < 0.001; Fig. 3D) such that individuals with larger wings
tended to have darker black wing sections. The interaction of sex*wing area
again showed that the slope of the negative relationship with wing area was
steeper for males than for females (males: -1.5, females: -0.5). Individuals
with larger wings had darker black wing sections, but the magnitude of this
relationship depended on the sex. Finally, there was a significant effect of site
on melanization scores (F2,52 = 3.59, P = 0.035).
Discussion
As this study was the first direct quantification of variation in wing color
of Large Milkweed Bugs, the questions addressed were inherently basic,
such as are wings of males and females differently colored, and is there a relationship
with body size? As for the first question, the data gathered indicate
that males and females of this species do differ in wing color—the wings of
males tend to be more orange, whereas wings of females are more yellow. At
the same time, females have darker black wing sections than males. These
results then represent the first confirmation of sexual dichromatism in this
species, which is in contrast to Rodriguez-Clark (2004), who found no sexbased
variation in wing color of this species (using manual color-scoring).
However, the functional significance of this sexual dimorphism is not clear
at this time. One possibility is that the greater melanism levels in females is
associated with differences between the sexes in immune function, which is
linked with cuticular melanism levels in insects (Wilson et al. 2001). While
it is not known if the sexes differ in immunity in this species, this question
may be a direction for future research; male and female immune parameters
such as hemocyte numbers (Feir 1964) could be compared, especially since
new methods of counting hemocytes in milkweed bugs have recently been
developed. These new methods also use image analysis (Davis 2007). This
idea becomes especially important when one considers that the sex-related
differences in wing melanism found here parallel those found in Monarch
Butterfl ies (where females are also darker than males), and in that species,
females have recently been shown to have higher concentrations of hemocytes
(Lindsey and Altizer 2009).
There were relationships between wing size (used here as a proxy of body
size) and certain color traits of milkweed bugs, but in some cases, the strength
of the relationships depended on the sex. In general, individuals with larger
wings tended to have deeper orange and darker black wing sections. These
relationships could be interpreted in light of the aposematic function of the
wing colors. Darker black and deeper orange colors may be associated with
large size, with the result that the aposematic contrast of these colors is more
730 Southeastern Naturalist Vol. 8, No. 4
apparent in the larger individuals (Prudic et al. 2007). On the other hand, the
ability to synthesize pigment may be tied with larval growth, in that the larvae
with optimal food resources may grow larger and produce more intense
pigmentation than those with poor resources. Evidence in support of this
idea comes from the similarities in wing color traits observed within sites
(discussed further below). Research in certain other insect species (Bembecinus
quinquespinosus Say [Digger Wasps]) also uncovered links between
body size and color, though the functional significance of the relationships
in that species remain unclear as well (O’Neill and Evans 1983, O’Neill et
al. 1989).
While not one of the main objectives of this study, there were several statistical
differences in wing colors found between the three sites where Large
Milkweed Bugs were collected. This result was surprising and may need to
be verified with additional data. However, it indicates that milkweed bugs
can have similar coloration within sites, which would support the idea that
there is some site-level variation in larval food resources. While all Large
Milkweed Bugs were collected from the same host plant species, Ascepias
incarnata, local variation could exist in the quality of these plants, which is
then refl ected in the degree of pigmentation on the Large Milkweed Bugs at
the sites. There is also the possibility that the Large Milkweed Bugs at each
site were related to one another. Indeed, it has been shown that wing colors
are heritable (Abbott 1968, Rodriguez-Clark 2004). In any case, this result
speaks to the need to ensure high genetic diversity or at least to account for
the collection site as a variable in future analyses.
This study represents an important first step into an area of research
ripe with questions that until now have not been easily addressed because
of technological limitations. With the color quantification methods now
available to researchers (similar to those outlined in this study), many questions
could now be readily addressed with Large Milkweed Bugs. These
questions could relate to those already addressed in other species, such as
linkages between external melanism and immune function (Wilson et al.
2001), melanism and population variation (Davis et al. 2005), or color and
mating success (Davis et al. 2007). Another idea would be to examine wing
color variations in relation to migratory propensity, an issue well-studied
within milkweed bugs (Dingle 1981, Dingle et al. 1980). This idea has been
examined already in Monarch Butterfl ies, and indeed there was evidence
found that wings of migrants are differently colored than those of breeding
individuals (Davis 2009).
On a related note, the approach used to quantify Large Milkweed Bug
wing colors in this study should be adaptable to the study of other insect species
as well. Most wings can be scanned fl at with a standard fl atbed scanner,
or with a slide scanner, and the scanner emits a standardized level of light
on the subject (but this must be specified beforehand by the user), so that the
variations among individuals in shades of black or hues of orange (or yellow,
blue, etc.) can be measured on-screen with the image analysis software
2009 A.K. Davis 731
(Davis, in press; Davis et al. 2005, 2007; Lindsey and Altizer 2009). For
insects with curved surfaces (beetle elytra, for example), digital photographs
can also work, providing that individuals are photographed under standardized
lighting (Davis et al. 2004, Todd and Davis 2007). Whatever the species
under study, the results of the current study will hopefully provide a framework
on which to build.
Acknowledgments
John Maerz provided help with the data for this project, and provided logistic
support. Sonia Altizer contributed useful comments about wing melanism in insects.
The manuscript was improved with comments from two anonymous reviewers.
Financial support during the writing of this manuscript was provided by the D.B.
Warnell School of Forestry and Natural Resources.
Literature Cited
Abbott, C.E. 1968. Inherited melanism in the Large Milkweed Bug, Oncopeltus fasciatus
(Heteroptera Lygaeidae). Annals of the Entomological Society of America
61:542.
Berenbaum, M.R., and E. Miliczky. 1984. Mantids and milkweed bugs: Efficacy of
aposematic coloration against invertebrate predators. American Midland Naturalist
111:64–68.
Bowdish, T.I., and T.L. Bultman. 1993. Visual cues used by mantids in learning
aversion to aposematically colored prey. American Midland Naturalist
129:215–222.
Davis, A.K. 2007. A simple, computer-assisted method for obtaining and counting
hemocytes in milkweed bugs. Entomologia Experimentalis et Applicata
123:203–206.
Davis, A.K. 2009. Wing color of Monarch Butterfl ies (Danaus plexippus) in eastern
North America across life stages: Migrants are “redder” than breeding and overwintering
stages. Psyche 2009. Available online at DOI 10.1155/2009/705780.
Davis, A.K., and K.L. Grayson. 2007. Improving natural history research with image
analysis: The relationship between skin color, sex, size, and stage in adult
Red-spotted Newts (Notophthalmus viridescens viridescens). Herpetological
Conservation and Biology 2:67–72.
Davis, A.K., B. Farrey, and S. Altizer. 2004. Quantifying Monarch Butterfl y larval
pigmentation using digital image analysis. Entomologia Experimentalis et Applicata
113:145–147.
Davis, A.K., B. Farrey, and S. Altizer. 2005. Variation in thermally induced melanism
in Monarch Butterfl ies (Lepidoptera: Nymphalidae) from three North American
populations. Journal of Thermal Biology 30:410–421.
Davis, A.K., N. Cope, A. Smith, and M.J. Solensky. 2007. Wing color predicts future
mating success in male Monarch Butterfl ies. Annals of the Entomological Society
of America 100:339–344.
Dingle, H. 1981. Geographic variation and behavioral fl exibility in milkweed bug
life histories. Pp. 57–73. In R.F. Denno and H. Dingle (Eds.). Insect life-history
patterns: Habitat and geographic variation. Springer-Verlag, New York, NY.
Dingle, H., N.R. Blakley, and E.R. Miller. 1980. Variation in body size and fl ight
performance in milkweed bugs (Oncopeltus). Evolution 34:371–385.
732 Southeastern Naturalist Vol. 8, No. 4
Ellers, J., and C.L. Boggs. 2003. The evolution of wing color: Male mate choice
opposes adaptive wing color divergence in Colias butterflies. Evolution
57:1100–1106.
Feir, D. 1964. Haemocyte counts on the Large Milkweed Bug, Oncopeltus fasciatus.
Nature 202:1136–1137.
Feir, D. 1974. Oncopeltus fasciatus: A research animal. Annual Review of Entomology
19:81–96.
Lindsey, E., and S. Altizer. 2009. Sex differences in immune defences and response
to parasitism in Monarch Butterfl ies. Evolutionary Ecology Online early: DOI:
10.1007/s10682-008-9258-0.
O’Neill, K.M., and H.E. Evans. 1983. Alternative male mating tactics in Bembecinus
quinquespinosus (Hymenoptera: Sphecidae): Correlations with size and color
variation. Behavioral Ecology and Sociobiology 14:39–46.
O’Neill, K.M., H.E. Evans, and R.P. O'Neill. 1989. Phenotypic correlates of mating
success in the sand wasp Bembecinus quinquespinosus (Hymenoptera: Sphecidae).
Canadian Journal of Zoology 67:2557–2568.
Prudic, K.L., A.K. Skemp, and D.R. Papaj. 2007. Aposematic coloration, luminance
contrast, and the benefits of conspicuousness. Behavioral Ecology 18:41–46.
Rodriguez-Clark, K.M. 2004. Effect of captivity on genetic variance for five traits in
the Large Milkweed Bug (Oncopeltus fasciatus). Heredity 93:51–61.
Statistica. 2003. Statistica version 6.1, Statsoft, Inc., Tulsa, OK.
Thompson, J.J.W., S.A.O. Armitage, and M.T. Siva-Jothy. 2002. Cuticular colour
change after imaginal eclosion is time-constrained: Blacker beetles darken faster.
Physiological Entomology 27:136–141.
Todd, B.D., and A.K. Davis. 2007. Sexual dichromatism in the Marbled Salamander,
Ambystoma opacum. Canadian Journal of Zoology 85:1008–1013.
Wilson, K., S.C. Cotter, A.F. Reeson, and J.K. Pell. 2001. Melanism and disease
resistance in insects. Ecology Letters 4:637–649.