Open Access
How to translate text using browser tools
1 March 2002 A Quantitative Analysis of Geographic Color Variation in Two Geotrupes Dung Beetles
Takashi Watanabe, Taketo Tanigaki, Hirotaka Nishi, Atushi Ushimaru, Tsuyoshi Takeuchi
Author Affiliations +

We conducted a quantitative analysis of geographic color variation in two species of dung beetles: Geotrupes auratus and G. laevistriatus. The reflectance of the dorsal surfaces was measured from 300 to 700 nm using a spectrophotometer. The reflectance curves for both beetles were bimodal; there were two distinct peaks, namely, the α peak, between 400 and 700 nm, and the β peak at around 300 nm. A stepwise discriminant analysis indicated that geographic color variation in Geotrupes beetles was primarily characterized by a shift of the α peak. Using beetles from three locations, we compared the wavelength (nm) of the α peak (λmax(α)) and its reflectance intensity (R(α)) to investigate sex and population differences. Intraspecific geographic variation in coloration was effectively detected by discriminant analysis of spectral reflectance curves. Our results showed that G. auratus and G. laevistriatus had similar coloration within each sampling location. Our study also revealed hidden sex differences in R(α); R(α) of males were significantly higher than those of females in both species. Since the dorsal surface of the beetles shows remarkable color variation, and coloration can be assessed objectively using reflectance spectra, Geotrupes beetles may be good model organisms to investigate geographic color variation.


As with many other morphological traits, animals can show dramatic geographic variation in coloration. Many different selective pressures, including thermoregulation, intraspecific communication, and predator avoidance, could generate this variation in color (Endler, 1978). Although researchers have long been interested in geographic color variation as a possible consequence of adaptation to local environments, the quantification and classification of the coloration of animals has been dependent on human perception. However, it is now recognized that using human vision to evaluate coloration could lead to erroneous conclusions (Bennent et al., 1994). If receivers of color stimuli, such as predators or individuals of the opposite sex, influence the evolution of coloration, we need to understand receiver visual perception. Many studies have shown that color vision among many taxa can be considerably different from that of humans in both the range of visible wavelengths and the mode of color information processing (Menzel and Backhaus, 1991; Jacobs, 1992; Brandt and Vorobyev, 1997; Vorobyev et al., 1998). Recently, mathematical models based on the physiological and psychophysical knowledge of animal vision have been developed and used for quantitative analysis of ‘apparent (observer-perceived) color’ (Vorobyev et al., 1998; Cuthill et al., 1999; Vorobyev & Menzel, 1999; Sumner and Mollon, 2000; Vorobyev et al., 2001). The results of these studies strongly suggest that we should not reduce spectral information to human standards, since such a reduction could result in a loss of critical color information. Therefore, objective methods for assessing color patterns are required, and should encompass the entire wavelength range of the receiver (Cuthill et al., 1999).

Another difficulty in evaluating animal coloration is that it often involves complex patterns. We usually do not have exact knowledge of which color components or combinations of color are ecologically important. Consequently, many researchers are still obliged to describe and categorize the color patterns of animals subjectively. Hence, studies of the evolution of geographic color variation in animals could benefit from the spectral information of an organism that has a simple, ideally monochromatic, coloration.

For the reasons mentioned above, some coleopteran insects may be good model subjects. Harris (1991) described color variation in the sand-burrowing beetle Chaerodes trachyscelides by using indices calculated by the CIELab system (CIE, 1978). He revealed that variation in the color of Chaerodes beetles matched the color of the sand of their resident beach. However, he only compared variation in the lightness component of the color.

Here, we report the geographic color variation of two dung beetles: Geotrupes auratus Motsch and G. laevistriatus Motsch. These two species range throughout most of the Japanese islands, feeding on the dung of mammals. These Geotrupes beetles may be good models for studying geographic color variation, as they show remarkable geographic color variation and have nearly monochromatic elytra. Preliminary geographic color variation in these two beetles has been described in an early study (Mizuno, 1964), which identified five color types for G. auratus and two color types for G. laevistriatus. Judgment of coloration, however, was based on human visual perception. In this paper, we attempt to establish an objective method for quantifying the coloration of monochromatic dung beetles. For this purpose, we calculated several variables from the reflectance spectra of the dorsal surface (elytra) of the beetles, and then isolated the parameter that most effectively characterized the geographic variation in the coloration. To our knowledge, this is the first study to evaluate the color of beetles objectively and examine geographic differences statistically.


We chose three sampling sites (Arashiyama, Otowayama and Nara Park; Fig. 1) where distinct color types (red, green and indigo, respectively) of Geotrupes auratus have been observed (Mizuno, 1964). We collected male and female G. auratus and G. laevistriatus with baited traps using the dung of horses. We also collected the beetles from the dung of unknown mammals found at the study sites. The sampling was conducted from early May to mid August 2001. We measured total body length and thorax width in the laboratory.

Fig. 1

Map of central Japan showing the positions of the sites from which samples were collected.


We measured the reflectance spectra of the dorsal surface of beetles with a Shimadzu spectrophotometer UV-240, which measures wavelength with an accuracy of 0.3 nm. In our analyses, we used the reflectance data in the range from 300 to 700 nm, in 5-nm increments. We used a deuterium light source with the spectrophotometer, and barium sulfate (Merck) as a white standard. Since the coloration of the beetles appears to be of structural origin, it should change according to the viewing angle. In this study, the measurement angle of reflectance was perpendicular to the dorsal surface. In addition, there appears to be color variation among different parts of the body. Therefore, we only measured the coloration of the dorsal surface (elytra), which occupies the largest part of the body, although other parts, such as the underside and legs of the beetles also show geographic color variation (Watanabe, personal observation).

Before measuring all the samples, we conducted pilot studies and found that spectra are usually bi-modal within the measured wavelength range (see Fig. 2 for representative spectra), with a peak (α peak) within the range of human visual sensitivity (400–700 nm), and another peak (β peak) near 300 nm. Although most β peaks are not real peaks (they show the upper limit of an increasing phase toward the shortest wavelength measured), we treat them as such for convenience.

Fig. 2

Representative reflectance spectra of the dorsal surface of two species of dung beetle. Variables extracted from the spectral data (81 points from 300 to 700 nm in 5-nm increments) include α and β peak wavelength (λmax(α) and λmax(β), respectively), reflectance intensity (R(α) and R(β), respectively), mean intensity, and minimum intensity (MIN).


Six variables were obtained from the reflectance spectra: the wavelength corresponding to the α and β peaks (λmax(α) and λmax(β), respectively), the reflectance intensity of the α and β peaks (R(α) and R(β), respectively), as well as the mean and minimum reflectance (Fig. 2). The mean reflectance was calculated as the average reflectance of the entire spectrum measured in 5-nm increments (300–700 nm; n=81). The minimum reflectance (MIN) was the minimum value of the reflectance intensity. We used these parameters to examine sexual and geographic differences in coloration within each species. These variables were not normally distributed; consequently, we used a nonparametric statistical test for our analyses. In an attempt to discriminate between collection sites based on color variation, we used a canonical discriminant analysis, which included all six color parameters. To determine the order of effectiveness of these variables in discriminating between collection sites, we adopted a stepwise method for the discriminant analysis.


General Patterns

There was no significant difference in total body length (TL) and thorax width (TW) between sexes within a population (Table 1). When pooling males and females, there was no significant difference in TL and TW among the three populations of Geotrupes auratus (Kruskal-Wallis test; H=1.674, p=0.4330 for TL; H=2.529, p=0.2824 for TW). In contrast, TL and TW of G. laevistriatus were significantly different among the three populations (Kruskal-Wallis test; H=26.541, p<0.0001 for TL; H=24.974, p<0.0001 for TW). Nonparametric multiple comparison tests (Dunn, 1964) revealed that both TL and TW of G. laevistriatus collected at Otowayama were significantly larger than those from the other two sites (TL: Arashiyama vs. Otowayama, Q=3.04, p<0.01; Arashiyama vs. Nara Park, Q=1.60, n.s.; Otowayama vs. Nara Park, Q=5.02, p<0.001, TW: Arashiyama vs. Otowayama, Q=2.93, p<0.05; Arashiyama vs. Nara Park, Q=1.60, n.s.; Otowayama vs. Nara Park, Q=4.88, p<0.001).

Table 1

Sampling sites, total body length (mm) and thorax width (mm) of the Geotrupes beetles. There was no significant difference between sexes within a population. Sample numbers are in parentheses.


We found two different types of spectra that characterize the reflectance spectra of each species (Fig. 3; A–F: averaged data are shown in 10-nm increments). All the spectra measured were bimodal, with a reflectance peak between 400 and 700 nm (α peak), and another peak near 300 nm (β peak). The α peak of G. auratus corresponds to the maximum intensity of the spectra (Fig. 3; A–C), but that of G. laevistriatus is moderate, and lower than the β peak (Fig. 3; D–F). Since most β peaks are not real peaks, as mentioned in the methods, we focused on the wavelength (λmax(α)) and reflectance intensity of the α peak (R(α)) in the following analyses.

Fig. 3

Reflectance spectra of the dorsal surface of beetles. Figs. illustrate the mean and standard deviation in 10-nm increments for all the samples at each sampling site.


Sex differences

There were highly significant differences in the mean reflectance of spectra of G. auratus (Table 2); males showed overall higher reflectance than females. Males of G. laevistriatus also showed higher mean reflectance than females, except at the Otowayama site (Table 2). There were no significant differences in the minimum reflectance of spectra (MIN), except in G. auratus at the Otowayama site (Table 2).

Table 2

Sexual differences in elytrons reflectance spectra. Sample numbers are in parentheses.


There were no significant differences in λmax(α) between sexes within a population (Table 3). On the other hand, the reflectance intensities of males were significantly higher than those of females in each species (Table 3). We also examined whether R(α) was related to the body length of males and females. However, no significant correlations were observed (data not shown).

Table 3

The λmax(α) (nm) and reflectance intensity (%) of a peaks with distinct of sex. Sample numbers are in parentheses.


Population comparisons

Since there were no significant differences in λmax(α) between the sexes, we compared the λmax(α) of different populations using combined (male and female) data. Fig. 4 shows the distribution of λmax(α). Distributions of λmax(α) were significantly different among populations (Kruskal-Wallis test; H=140.310, p<0.0001 in G. auratus; H=59.966, p<0.0001 in G. laevistriatus). Multiple comparison tests revealed significant differences among the three populations of G. auratus: Arashiyama vs. Otowayama, Q=3.57, p<0.01; Arashiyama vs. Nara Park, Q=10.41, p<0.001; Otowayama vs. Nara Park, Q=8.51, p<0.001. The λmax(α) of G. laevistriatus also differed significantly among two of the populations: Arashiyama vs. Otowayama, Q=6.08, p<0.001; Arashiyama vs. Nara Park, Q=7.44, p<0.001; Otowayama vs. Nara Park, Q=2.00, p>0.10 (not significant). The λmax(α) of G. laevistriatus was more variable than that of G. auratus at all three sites (at Arashiyama, coefficient of variation (CV)=2.1 in G. auratus and 6.6 in G. laevistriatus; at Otowayama, CV=2.8 in G. auratus and 8.4 in G. laevistriatus; and at Nara Park, CV=3.2 in G. auratus and 12.4 in G. laevistriatus).

Fig. 4

Histogram of the distribution of λmax(α) measured on the dorsal surface of G. auratus and G. laevistriatus. Data are categorized using 25 nm bins.


The differences in the shapes of the spectra were summarized by canonical discriminant analysis (Fig. 5a, 5b; Table 4a, 4b). A stepwise discriminant analysis of variance in the entire dataset (six variables) reduced the number of discriminating variables to 4 for male G. auratus, 3 for female G. auratus, 2 for male G. laevistriatus, and 3 for female G. laevistriatus (significance associated with Wilks' λ, p<0.05; Table 4a). Analyses revealed that λmax(α) was the most efficient parameter for discriminating between sampling sites in both species (Table 4a). This analysis was able to discriminate between sampling sites for both male and female G. auratus clearly (Wilks' λ=0.011 for males; Wilks' λ=0.013 for females). Although λmax(β) had a large effect on CAN1 scores in the final model (Table 4b), a discriminant function analysis using only λmax(α) was more effective at discriminating between sampling sites (Wilks' λ= 0.082 for males; Wilks' λ=0.078 for females). In comparison, discrimination analyses between sampling sites for male and female G. laevistriatus showed relatively poorer performance (Wilks' λ=0.359 for males; Wilks' λ=0.251 for females). The canonical scores of samples from Otowayama and Nara Park largely overlapped, although samples from Arashiyama could be clearly discriminated from those from the other two sites (Fig. 5b). The CAN1 scores for G. laevistriatus were highly dependent on λmax(α) (Table 4b), and the frequency distributions of λmax(α) of G. laevistriatus from Otowayama and Nara Park actually shared a wide wavelength range (Fig. 4).

Fig. 5

The canonical discriminant scores CAN1 and CAN2 of the dorsal surface spectra of G. auratus (a) and G. laevistriatus (b). Discriminant analyses were conducted for each sex, since some variables showed sex differences.


Table 4

Parameters in the canonical discriminant analysis using stepwise method.

(a) Stepwise improvements of discrimination



Our study demonstrates that reflectance spectra are useful for describing the monochromatic color of the dorsal surface of two dung beetle species, Geotrupes auratus and G. laevistriatus. Furthermore, we show that geographic color variation in each species is well characterized by the shift of α peaks. Since structural colors in the Geotrupes beetles seem to be produced by a multilayer reflector, the α peak of reflectance may depend on the thickness of the layers in a stack (Parker et al., 1998). Therefore, it seems appropriate to use λmax(α) as an index for evaluating color variation in Geotrupes beetles. Our analyses of λmax(α) also revealed hidden (imperceptible by humans) sexual dichromatism in both beetles; the reflectance intensity of males was significantly higher than that of females at λmax(α). Through objective spectral measurements, we may find hidden sexual dichromatism in many other species.

Recently, Favila et al. (2000) clearly demonstrated that cuticular color in a Scarabaeinae beetle (Canthon cyanellus cyanellus LeConte) is genetically controlled. If the coloration of G. auratus and G. laevistriatus is also genetically controlled, we should pay more attention to the frequency distribution of λmax(α) of a population. Coefficients of variation in λmax show that the coloration of G. auratus is more uniform than that of G. laevistriatus in all three populations. Furthermore, the frequency distributions of λmax(α) in G. laevistriatus largely overlap between the different sites. Differences in the degree of variation within a population may be explained by the differing levels of gene flow between the two beetle species. Habitats of G. auratus to be restricted to localized areas where wild mammals are more abundant, while G. laevistriatus are observed more widely, even in isolated green spaces within cities (Mizuno, 1964). Therefore, gene flow among G. auratus populations seems to be more restricted than in G. laevistriatus. Differences in the strength of selective pressures may also explain the differences in the degree of geographic variation within each species if the coloration of the beetles is determined by natural or sexual selection; selective pressures may be greater on G. auratus than on G. laevistriatus in a specific area. To understand these patterns of geographic variation, we need to determine the selective forces acting on the coloration, the level of gene flow between populations, and the interaction between these two selective pressures.

Interestingly, our results show that G. auratus and G. laevistriatus share similar coloration in each habitat. Although the coloration of G. laevistriatus appears dull to human observers, spectral analyses reveal that the α peaks of G. laevistriatus shift in the same direction as those of G. auratus. Additional populations should be surveyed to understand the patterns in geographic color variation in these two dung beetles.

Geographic color variation in G. auratus and G. laevistriatus does not appear to be maintained by thermoregulation, since coloration does not vary with latitude (Mizuno, 1964). Alternatively, coloration might serve a function in intraspecific communication, since there are sex differences in the reflectance intensity of the α peak. However, an intraspecific signal function does not explain the similarity of the coloration between the two Geotrupes beetles at each site. If the dung beetles are unpalatable to predatory animals, and the coloration functions as a warning signal, the color similarities between the two beetle species might be explained by Müllerian mimicry, since they share the same resources (the dung of mammals) and are frequently observed together in the same habitat. Studies of the diurnal activity of the beetles revealed that G. auratus and G. laevistriatus activities peak in the daytime (Sasayama et al., 1984). Therefore, their coloration may act as warning signals for visual diurnal predators, such as birds. Pluot-Sigwalt (1982, 1984) has demonstrated that some African dung beetles in the tribe Gymnopleurini have exceptionally numerous pygidial glands that generally secrete repellent substances. The bright color of these beetles may be interpreted as warning colors (Camberfort, 1991). To our knowledge, however, there is no evidence demonstrating the unpalatability of the two Geotrupes dung beetles.

In conclusion, our quantitative analysis effectively evaluated the coloration of two dung beetles: G. auratus and G. laevistriatus. Using the objective index, λmax, we can compare the geographic color variation of the beetles quantitatively. We found coloration similarity between G. auratus and G. laevistriatus within sampling locations, and also hidden sex differences in the reflectance intensity of the beetles. The Geotrupes beetles may be an ideal model system for investigating causal factors of geographic variation in animal color patterns.


We thank Arata Kochi for helpful suggestion on earlier stage of our work, and Seiko Mishima, Miwa Tanigaki and Akiko Watanabe for assistance in the field. Prof. Michio Imafuku kindly allowed us to use the spectrophotometer. Thanks are also extended to members of the Laboratory of Animal Ecology, Kyoto University, for their assistance and advice.



A. T. D. Bennett, I. C. Cuthill, and K. J. Norris . 1994. Sexual selection and the mismeasure of color. American Naturalist 144:848–860. Google Scholar


R. Brabdt and M. Vorobyev . 1997. Metric analysis of threshold spectral sensitivity in the honeybee. Vision Research 37:425–439. Google Scholar


Commission Internationale d'Eclairage (CIE) 1978. Recommendations on Uniform Colour Spaces, Colour Difference Equations, Psychometric Colour Terms. Supplement No.2 to CIE Publication No. 15 (E-1.3.1 1971/TC-1.3). Bureau Central de la CIE. Paris. Google Scholar


I. C. Cuthill, A. T. D. Bennett, J. C. Partridge, and E. J. Maier . 1999. Plumage reflectance and the objective assessment of avian sexual dichromatism. American Naturalist 153:183–200. Google Scholar


O. J. Dunn 1964. Multiple contrasts using rank sums. Technometrics 6:241–252. Google Scholar


J. A. Endler 1978. A predator's view of animal colour patterns. Evol Biol 11:319–364. Google Scholar


J. A. Endler 1996. Interacting effects of lek placement, display behavior, ambient light, and color patterns in three neotropical forest-dwelling birds. American Naturalist 148:421–452. Google Scholar


M. E. Favila, G. Ruiz-Lizarraga, and J. Malasco . 2000. Inheritance of a red cuticular color mutation in the scarab beetle Canthon cyanellus cyanellus LeConte (Coleoptera: Scarabaeidae). Coleopterists Bulletin 54:541–545. Google Scholar


Y. Cambefort 1991. Dung Beetles in Tropical Savannas. In “Dung Beetle Ecology”. Ed by I. Hanski and Y. Cambefort . Princeton Univ. Press. Princeton. Google Scholar


A. C. Harris 1991. Geographic variation for colour in the sand-burrowing beetle Chaerodes trachyscelides White (Coleoptera: Tenebrionidae) on New Zealand beaches analysed using CIELAB L* values. Biol J Linn Soc London 44:93–104. Google Scholar


G. H. Jacobs 1992. Ultraviolet vision in vertebrates. Am Zool 32:544–554. Google Scholar


R. Menzel and W. Backhaus . 1991. Color vision ininsects. In “The perception of colour”. Ed by P. Gourasa MacMillan. London. Google Scholar


T. Mizuno 1964. Geographic variation for colour in Japanese dung beetles genus Geotrupes. Iden 18:24–27. in Japanese. Google Scholar


A. R. Parker, D. R. McKenzie, and M. C. J. Large . 1998. Multilayer reflectors in animals using green and gold beetles as contrasting examples. J Exp Biol 201:1307–1313. Google Scholar


D. Pluot-Sigwalt 1982. Diversité et dimorphisme sexuel de glandes tégumentaires abdominals chez les Coléoptères Scarabaeidae. C R Acad Sci Paris III 294:945–948. Google Scholar


D. Pluot-Sigwalt 1984. Les glandes tégumentaires des Coléoptères Scarabaeidae: Répartition des glandes sternales et pygidiales dans la famille. Bull Soc Entomol Fr 88:597–602. Google Scholar


K. Sasayama, K. Nakamura, M. Manda, and I. Kurohiji . 1984. Diurnal prevalence and its seasonal change of dung beetles: relationships to weather elements. J Jpn Soc Grassland Sci 29:362–367. in Japanese. Google Scholar


P. Sumner and J. D. Mollon . 2000. Chromaticity as a signal of ripeness in fruits taken by primates. J Exp Biol 203:1987–2000. Google Scholar


M. Vorobyev, D. Osorio, A. T. D. Bennett, N. J. Marshall, and I. C. Cuthill . 1998. Tetrachromacy, oildroplets and bird plumage colours. J Comp Physiol A 183:621–633. Google Scholar


M. Vorobyev and R. Menzel . 1999. Flower advertisement for insects: Bees, a case study. In “Adaptive Mechanisms in the Ecology of Vision” Ed by S. N. Archer, M. B. A. Djamgoz, J. C. Loew, J. C. Partridge, and S. Vallerga . Kluwer Academic Publishers. Google Scholar


M. Vorobyev, J. Marshal, D. Osorio, N. H. de Ibara, and R. Menzel . 2001. Colourful objects through animal eyes. Col Res & Appl 26:214–217. Google Scholar
Takashi Watanabe, Taketo Tanigaki, Hirotaka Nishi, Atushi Ushimaru, and Tsuyoshi Takeuchi "A Quantitative Analysis of Geographic Color Variation in Two Geotrupes Dung Beetles," Zoological Science 19(3), 351-358, (1 March 2002).
Received: 30 September 2001; Accepted: 1 November 2001; Published: 1 March 2002
color measurement
dung beetle
geographic color rariation
Reflectance spectra
Back to Top