, Volume 92, Issue 10, pp 464–467

Spectral heterogeneity of honeybee ommatidia


  • Motohiro Wakakuwa
    • Graduate School of Integrated ScienceYokohama City University
  • Masumi Kurasawa
    • Graduate School of Integrated ScienceYokohama City University
  • Martin Giurfa
    • Centre de Recherches sur la Cognition AnimaleCNRS-Université Paul-Sabatier (UMR 5169)
    • Graduate School of Integrated ScienceYokohama City University
Short Communication

DOI: 10.1007/s00114-005-0018-5

Cite this article as:
Wakakuwa, M., Kurasawa, M., Giurfa, M. et al. Naturwissenschaften (2005) 92: 464. doi:10.1007/s00114-005-0018-5


The honeybee compound eye is equipped with ultraviolet, blue, and green receptors, which form the physiological basis of a trichromatic color vision system. We studied the distribution of the spectral receptors by localizing the three mRNAs encoding the opsins of the ultraviolet-, blue- and green-absorbing visual pigments. The expression patterns of the three opsin mRNAs demonstrated that three distinct types ommatidia exist, refuting the common assumption that the ommatidia composing the bee compound eye contain identical sets of spectral receptors. We found that type I ommatidia contain one ultraviolet and one blue receptor, type II ommatidia contain two ultraviolet receptors, and type III ommatidia have two blue receptors. All the three ommatidial types contain six green receptors. The ommatidia appear to be distributed rather randomly over the retina. The ratio of type I, II, and III ommatidia was about 44:46:10. Type III ommatidia appeared to be slightly more frequent (18%) in the anterior part of the ventral region of the eye. Retinal heterogeneity and ommatidial randomness, first clearly demonstrated in butterflies, seems to be a common design principle of the eyes of insects.


Since Karl von Frisch demonstrated color vision in the honeybee, Apis mellifera, almost a century ago (Frisch 1914), the visual system of this insect has been intensively studied at the behavioral, cellular, and molecular level. Intracellular recordings revealed that the honeybee eye is equipped with three classes of spectral receptors, peaking in the ultraviolet (UV, λmax=344 nm), blue (B, 436 nm), and green (G, 544 nm) wavelength range (Menzel and Backhaus 1991). At the molecular level, three distinct cDNAs encoding the visual pigment opsins were cloned: these opsins correspond to UV, B, and G absorbing visual pigments (Chang et al. 1996; Townson et al. 1998).

The honeybee eye is composed of ommatidia each containing nine photoreceptor cells, R1-9. The visual pigment molecules of the photoreceptors are localized in microvilli that form a photoreceptive rhabdomere. The rhabdomeres of R1-9 together construct a fused rhabdom along the central axis of the ommatidium (Gribakin 1975). The main photoreceptors R1-8 contribute the microvilli to the entire length of the rhabdom, but the basal R9 contributes microvilli only at the base of the ommatidium.

The localization of the spectral receptors in an ommatidium has been studied by electrophysiology coupled with dye injection (Menzel and Blakers 1976) and by analyzing light-induced structural changes in specific photoreceptors (Gribakin 1975). These studies have established a long-held view that the ommatidia in the main part of the eye contain an identical set of spectral receptors, with three UV (R1, 5 and 9), two B (R4 and 8) and four G (R2, 3, 6, and 7) receptors (Land and Nilsson 2002; Menzel and Backhaus 1991; Waterman 1981).

In previous years, however, evidence has accumulated showing that the eye of butterflies and moths is composed of randomly distributed spectrally heterogeneous ommatidia (Arikawa and Stavenga 1997; Briscoe et al. 2003; Qiu et al. 2002; Sauman et al. 2005; Stavenga 2002; White et al. 2003). Ommatidial heterogeneity has actually been suggested also in a wasp (Ribi 1978), a backswimmer (Schwind et al. 1984), and in flies (Hardie 1986). Spaethe and Briscoe recently showed that the bumblebee retina had three types of ommatidia as in lepidopterans (Spaethe and Briscoe 2005). However, the issue of ommatidial heterogeneity has not been addressed in the insect that constitutes the paradigm for color vision studies, the honeybee Apis mellifera. Here, we asked whether or not the long-held view that the ommatidia of the honeybee are identical is valid by performing histological in situ hybridization for detecting mRNAs encoding opsins corresponding to the UV, B, and G visual pigments.

Materials and methods


About 100 returning honeybee foragers, Apis mellifera, collected at the entrance of the hives kept either at the campus of Yokohama City University or the University of Tokyo, were used in the experiments.

In situ hybridization

We performed in situ hybridization in serial sections, each labeled with the probe that hybridizes to the mRNA encoding the opsin of the UV, B, or G visual pigment, respectively. The eyes were fixed in 4% paraformaldehyde in 0.1 M sodium phosphate buffer (pH 7.2) for 0.5–2 h at 25°C. After dehydration with an ethanol series, we embedded the eyes in Paraplast, which were sectioned at ca 8 μm thickness.

Probes for in situ hybridization contained almost the full length of the coding region (UV: 1274 bp, B: 1689 bp, G: 1096 bp) of the mRNAs encoding the UV, B, and G opsins. The corresponding cDNA region was first subcloned into pGEM-3zf(+) vector, and then digoxigenin (DIG)-labeled cRNA was generated using the DIG-RNA labeling kit. Specificity of the probes was confirmed by dot blot analyses. The labeling protocol has been reported elsewhere (Wakakuwa et al. 2004).

Ommatidial regionalization

We studied whether ommatidial types show regional distribution across the honeybee eye. We divided the main part of the compound eye, i.e. except for the dorsal rim area (DRA), in five regions: the anterior dorsal, posterior dorsal, frontal, anterior ventral, and posterior ventral regions (Fig. 1). We selected several sections containing clearly labeled ommatidia from each region collected from 2–4 individuals. We thus determined the frequency of three types of ommatidia (see below) in these regions. We excluded the ommatidia in the DRA here, because the labeling quality in those ommatidia was not very clear.
Fig. 1

Head and compound eye of Apis mellifera, with the nomenclature of eye regions: ad, anterior dorsal; pd, posterior dorsal; f, frontal; av, anterior ventral; pv, posterior ventral


Ommatidial heterogeneity

Figure 2a–c show a typical set of transverse sections after in situ hybridization. The labeling with the UV probe (Fig. 2a) yielded three types of ommatidia: type I with one labeled cell, type II with two labeled cells, and type III with no labeled cells. These three types could also be distinguished in the B probe-labeled sections, but in a compensatory manner (Fig. 2b). The G probe labeled six cells in all ommatidia (Fig. 2c).
Fig. 2

In situ hybridization of mRNAs encoding opsins of ultraviolet (a, d), blue (b, e), and green (c, f) absorbing visual pigments in three consecutive transverse (a–c) and longitudinal (d–f) sections. Dark blue signals indicate the localization of corresponding mRNA. Closed circles indicate a type I ommatidium where one photoreceptor was labeled with the UV probe (a) and another with the B probe (b). Dotted circles indicate a type II ommatidium that bear two labeled photoreceptors with the UV probe (a). Broken circles indicate a type III ommatidium that bear two labeled photoreceptors with the B probe (b). The G probe (c) labeled six photoreceptors in all ommatidia (white arrowheads). Matching of the ommatidia was only possible by the orientation of two short wavelength receptors (UV and B), one shown by a double-headed white arrow in (c). Black arrowheads in (d) and (e) indicate the cell body of presumptive UV (d) and B (e) receptors. Nuclei of the R9 basal photoreceptors should exist in the layer indicated by brackets in (f), where the G probe gave bulbous labeling, whereas the other two probes gave no labeling. BM, basement membrane. Scales = 5 μm (shown in (c), for (a–c)), 25 μm (shown in (d), for (d–f))

The longitudinal sections (Fig. 2d–f) were labeled after employing a slightly longer proteinase K treatment. Under this condition, labeling was often found only around the photoreceptor nuclei (Fig. 2d and e), although the exact reason for this is uncertain. Both UV and B probes failed to detect any signal in the region somewhat distal to the basement membrane, where the R9 nuclei should exist (shown by brackets in Fig. 2c). Instead, the G probe labeled the entire length of the retinal layer with some concentration of signals close to the basement membrane.


Having characterized the three types of ommatidia, we asked whether they are heterogeneously distributed across the eye. We analyzed sections labeled with the UV or B probe, which allowed us to identify the ommatidial type unambiguously (Fig. 2a and b). We estimated the number of ommatidia of each type in the five regions of the eye (Fig. 1). Type III appeared to occupy only about 10% of the total number of ommatidia. They seem to be more frequent in the anterior ventral region (Table 1): the region seems to be more densely populated by B receptors than other regions.
Table 1

Relative number and percentage of the three types of ommatidia in different regions of the compound eye. Data were collected from 2–4 individuals


Number of ommatidia (%)

Receptor type (%)

Type I

Type II

Type III



Dorsal, anterior

174 (38.2)

256 (56.3)

25 (5.5)



Dorsal, posterior

286 (56.3)

206 (40.6)

16 (3.1)




280 (45.1)

263 (42.4)

78 (12.6)



Ventral, anterior

157 (33.5)

226 (48.3)

85 (18.2)



Ventral, posterior

300 (46.6)

276 (42.8)

68 (10.6)



Total no.(mean%)

1197 (44.0)

1227 (46.1)

272 (10.0)




This work shows that the retina of the honeybee contains three types of ommatidia with different photoreceptor composition. The distribution of the three types appears random, at least locally. Although the in situ hybridization does not provide direct evidence about the cells' spectral sensitivity, we conclude that the cells labeled with the UV, B or G probe were UV, B or G receptors, respectively, because the spectral sensitivity in butterflies always matched the expression patterns of the opsin mRNAs (Arikawa et al. 2003).

Anatomical identification of the R1-8 in Fig. 2a–c is difficult because these cells twist except in the DRA (Wehner and Rossel 1985). Previous reports documented that R1 and R5 are both UV receptors, and R3 and R7 are G receptors. Other cells, R2, 4, 6, and 8 were considered to be B or G receptors, although their actual numbers and relative position were not determined (Menzel and Blakers 1976). Our results clearly indicate that each ommatidium contains one UV and one B receptor (type I), two UV receptors (type II), or two B receptors (type III). They are most likely R1 and R5. The other six are all G receptors.

Basal R9s were difficult to characterize. They have been assumed to be UV sensitive (Menzel and Snyder 1974), but our histology could not confirm this view. The region where the R9 nuclei should exist (shown by brackets in Fig. 2f) was labeled by the G probe but not with UV or B probe. Due to limited resolution, we cannot forcefully conclude that the signals in this region really correspond to the R9 nuclei, but our results at least do not support the view that the R9s are UV sensitive.

Spectral heterogeneity of ommatidia seems a design principle of compound eyes. The most conspicuous similarity is the specific combination of UV and B receptors in one ommatidial type, as in the honeybee type I ommatidia (Arikawa 2003). The fact that all ommatidia bear six (and not four, as previously thought) G receptors underlines the importance of this input channel in different visual performances. G receptor input is necessary in tasks that are relevant for a flying insect such as motion detection (Hausen and Egelhaaf 1989; Kaiser and Liske 1974) and target detection (Giurfa et al. 1997).

B receptors are presumably more frequent in the anterior ventral region of the honeybee eye. Similar regionalization was also found in Manduca (White et al. 2003). Both honeybees and Manduca use the ventral region of the eye to localize flowers on which they forage. Concentration of B receptors in the ventral region is probably related to a better ventral detection of targets providing contrast to B receptors (Giurfa et al. 1999). A possible nonuniform organization of the retina has been suggested on the basis of colored target discrimination experiments (Lehrer 1999; Menzel and Lieke 1983). These experiments showed that discrimination of dual-colored targets was highly dependent on the spatial distribution of colors in vertically presented targets, thus suggesting differential photoreceptor distribution across the compound eye. Here, we show that such hypothesis was correct and that the compound eye of the honeybee was more complex than previously thought.


We thank Dr. D.G. Stavenga for critical comments. Dr. T Kubo from the University of Tokyo, provided some honeybee samples. The work was supported by the Grants-in Aid for Scientific Research from the JSPS (Japan Society for the Promotion of Science) and the Grant for Promotion of Science from Yokohama City University to which Kentaro Arikawa belongs. Martin Giurfa was supported by the University Paul Sabatier (ATUPS fellowship), Yokohama City University and the CNRS (Center de la recherche scientific).

Copyright information

© Springer-Verlag 2005