Resolution and sensitivity of the eyes of the Asian honeybees Apis florea, Apis cerana and Apis dorsata

The European honeybee, Apis mellifera, has been the object of considerable scientific curiosity and has served as the classical model organism in diverse fields such as behaviour, navigation, sensory biology,ecology and pollination biology for almost a century(Frisch, 1914; Frisch, 1927). Apis mellifera has eight congeners, all of which are Asian, and about which much less is known. Of the Asian honeybees, the cavity-nesting Apis cerana and the less well known species Apis koshevnikovi, Apis nigrocincta and Apis nuluensis are closely related to A. mellifera whereas the giant honeybees (Apis dorsata, Apis dorsata binghami and Apis laboriosa) and the dwarf honeybees (Apis florea and Apis andreniformis) form separate groups(Arias and Sheppard, 2005; Raffiudin and Crozier, 2007). Although A. mellifera is found in Africa, the Urals and Central Asia,it is allopatric to the south and south-east Asian species, many of which are sympatric in several tropical areas of Asia. For instance, in the Western Ghats in India, where we studied them, the most common Apis species, A. cerana, A. florea and A. dorsata co-occur.

All species of Apis are eusocial, building large nests with thousands of worker bees that collect nectar and pollen. Worker bees collect these resources from a large variety of flower species and can fly several kilometres between the hive and good nectar or pollen sources. From studies on A. mellifera, it is known that navigation and orientation as well as the detection and discrimination of suitable floral resources depend to a great deal on vision (Lehrer,1998; Chittka and Spaethe,2007). Like the vast majority of bees, A. florea and A. cerana, as well as the European A. mellifera ligustica,are restricted to foraging during the day. However, both the African honeybee, A. mellifera adansonii, and the Asian giant honeybee, A. dorsata, are reported to occasionally forage at night(Fletcher, 1978; Dyer, 1985). In both these species, nocturnal foraging has only been observed on moonlit nights, when the moon is at least half full, when there is no cloud cover and when no vegetation obscures this light source. By contrast, A. cerana is restricted to foraging during the day although it is able to fly at much lower temperatures (down to 10.5°C) than A. dorsata(Oldroyd et al., 1992).

Crepuscular or nocturnal foraging activity – requiring flight in rather dim light – has been reported from several genera in four families of the Apoidea (for a review, see Kelber et al., 2006), ranging from the minute Perdita bequaertiana to the large Xylocopa tranquebarica (Burgett and Sukumalanand, 2000; Somanathan et al., 2008). These bees, like all bees, possess apposition compound eyes. The transition to a dim-light environment is an apparent disadvantage to animals with apposition eyes because these eyes tend to have a very low absolute sensitivity, due to the small diameter of their facet lenses(see Warrant et al., 2004; Kelber et al., 2006; Somanathan et al., 2008). However, several reasons may have made this transition to a dim-light environment favourable for at least some species (see Wcislo et al., 2004). The matinal bees Xenoglossa and Peponapis, for instance, use nectar- and pollen-rich resources such as Cucurbita before other and often larger competitors take flight(Ordway et al., 1987). Species that fly during dusk, such as Perdita (Xerophasma) and Lasioglossum (Sphecodogastra), use equally rich pollen sources. These include the flowers of Oenothera spp., which these bees visit just after they open in the evening, and before the arrival of their obligate pollinators, the large hawkmoths(Kelber et al., 2006).

The only species of bee known to fly all night is the Indian carpenter bee X. tranquebarica, which seems to avoid competition with several equally large and much more abundant sympatric congeners such as Xylocopa tenuiscapa (Somanathan and Borges,2001; Somanathan et al.,2008). The latter facultatively nocturnal species, however, has also been found to collect pollen from a nocturnally flowering tree, Heterophragma quadriloculare, and is able to fly in dim light to do so. However, unlike X. tranquebarica, X. tenuiscapa can only fly on full-moon nights. Accordingly, Somanathan et al. found that the eyes of this species lack some adaptations for nocturnal vision that are found in X. tranquebarica (Somanathan et al.,2009). We expected that A. dorsata and A. m. adansonii may be similar cases because they extend foraging activity into the night if at least a half-moon is present in the sky(Fletcher, 1978) and if favourable food sources can be exploited at such times.

While Kerfoot (Kerfoot,1967) has recognised that nocturnal bees have unusually large ocelli, the adaptations of the eyes and the visual system for a nocturnal lifestyle have only recently been the focus of research. Warrant et al.(Warrant et al., 2004) gave a detailed report of the visual activity of the halictid bee Megalopta genalis, and Somanathan et al.(Somanathan et al., 2009) have described the eyes of X. tranquebarica and compared them with the eyes of sympatric diurnal congeners. Common features of nocturnal bees –and other nocturnal hymenopterans – are their huge ocelli and relatively large eyes with reasonably large facets and, most importantly, their unusually wide rhabdoms (Kerfoot, 1967; Menzi, 1987; Greiner et al., 2004a; Warrant et al., 2004; Greiner, 2006; Kelber et al., 2006; Greiner et al., 2007; Somanathan et al., 2009) (for a review, see Warrant,2008).

To better understand the importance of visual sensitivity and resolution for the feeding ecology of the Asian species of Apis, we have undertaken a study of the eyes and ocelli of the three most common species of Asian honeybees: A. dorsata, A. cerana and A. florea, with the goal of comparing their visual systems with those of the well-studied European honeybee, A. mellifera. Our expectation was that A. dorsata would possess visual adaptations that increase the sensitivity of their eyes, enabling flights on moonlit nights whereas the strictly diurnal A. cerana and A. florea would possess less sensitive eyes.

We observed and obtained individuals of Apis dorsata Fabricius 1793, Apis cerana Fabricius 1793 and Apis florea Fabricius 1793, from areas adjoining the Bhimashankar Wildlife Sanctuary in the Pune district of Maharashtra, India (19°21′–19°11′N,73°31′–73°37′E). During dry season flowering from November 2007 to April 2008, we observed flowers of 71 plant species in this plant community for honeybee visitation. Over several days, the timing of honeybee visits was observed for a single day for each plant species from 06:00 h to 11:00 h and from 15:00 h to 21:00 h. Additionally, we also made anecdotal observations of flight and foraging outside these windows of time. From these data, we obtained the earliest and latest daily foraging times for the three Apis species. Ambient light intensities measured on several mornings and evenings in different areas of the study site with an International Light IL 1700 radiometer (Peabody, MS, USA) have been reported by us in an earlier paper [see fig. 1 in Somanathan et al.(Somanathan et al., 2008)]. Solar and lunar data were obtained from the NASA website(http://aa.usno.navy.mil/).

The eyes of the bees were studied using standard methods described in detail elsewhere (e.g. Somanathan et al.,2009). Here, we give a short description of these methods. Measurements of body size, eye size and median ocellus size were made for workers of the three Apis species that were captured while visiting flowers. We used intertegular width, the distance between the wing bases, as an estimate of body size (e.g. Spaethe and Chittka, 2003; Somanathan et al., 2008). The number of ommatidia was determined using corneal nail polish replicas, following the methods adopted by Praagh(Praagh et al., 1980). Ommatidial diameters were determined in the fronto–ventral region of the eye where they were found to be the largest in other bees(Ribi et al., 1989; Greiner et al., 2004a; Kelber et al., 2006). Ocellar and ommatidial diameters were measured using light microscopy and scanning electron microscope (SEM) photographs.

The following histological procedures were performed to measure rhabdom diameters and lengths for the three species of Apis. After anaesthetising and decapitating bees, whole eyes were dissected and placed in fixative (2% glutaraldehyde, 2% paraformaldehyde, 2% sucrose in 0.15 mol l^–1 sodium cacodylate buffer) for 12–24 h. After rinsing repeatedly in buffer, the eyes were fixed in 1% osmium tetraoxide for one hour and embedded in epoxy resin. Longitudinal sections 1 μm thick were placed on a slide, dried on a hot-plate, stained with toluidine blue and photographed under a microscope. Ultrathin transverse sections were stained with lead citrate and uranyl acetate and studied under a JEOL 1240 Transmission Electron Microscope (Tokyo, Japan).

The spatial resolution of eyes is largely limited by the acceptance angle(half width of the photoreceptor's receptive field, in radians) of each ommatidium (Warrant and McIntyre,1993). For apposition compound eyes, the acceptance angleΔρ can be approximated by the ratio of the rhabdom diameter d and the focal length of the ommatidium f:Δρ≈ d/f(Stavenga, 2003; Frederiksen and Warrant,2008).

Observations of activity are summarised in Fig. 1. Apis ceranausually foraged earliest of all three species, confirming observations on this species by Oldroyd et al. (Oldroyd et al.,1992) in Thailand. The onset and offset of the foraging periods did not have constant relations to sunset and sunrise times, and thus light intensity, indicating a dependence of flight activity on temperature. In the warmer months (March and April), all three species started foraging around or even before sunrise. The minimum temperatures at night during the flowering season can be as low as 2°C in this site (range 2–14°C)(Somanathan and Borges, 2001). During this study, we recorded early morning temperatures between 07:00 h and 08:30 h that ranged between 16°C and 23°C during the flowering period. Nocturnal foraging by A. dorsata was observed on three occasions,during February and April. All three observations were made during the first few nights after full moon, when more than 70% of the moon was visible. In February 2008, A. dorsata were observed visiting flowers of the supra-annual mass-flowering species Memecylon umbellatum(Melastomataceae). These flowers open between 13:00 h and 19:00 h and A. dorsata were observed visiting them on a bright night, one day after full moon. Again in April 2008, we observed A. dorsata visiting flowers of Bridelia retusa (Euphorbiaceae) and Randia dumetorum(Rubiaceae), during the hours after midnight. These observations confirm earlier reports (Dyer, 1985; Kirchner and Dreller, 1993; Rao et al., 2001) that the species forages on moonlit nights. Like the cavity-nesting species A. cerana and A. mellifera but unlike A. florea, A. dorsata adds an acoustic component to its nocturnal dances, wherein length of sound signal was also correlated with distance to the food source(Kirchner and Dreller, 1993; Dreller and Kirchner, 1994). Because A. dorsata communicate on open combs, the acoustic component is thought to be correlated with regular nocturnal activity.

As indicated by their names, the dwarf honeybee A. florea and the giant honeybee A. dorsata differ in body size from the other two,similarly-sized species A. mellifera and A. cerana(Table 1). Dorso–ventral eye length increases linearly with body size but the scaling factor is considerably smaller than 1 (0.68) whereas the tongue length, as another example, increases with a scaling factor higher than 1 (1.7).

The eyes of all species of Apis are very similar in appearance:they are oval-shaped and have large numbers of sensory hairs between the ommatidia (see Fig. 2). These hairs differ in length between species: A. florea has the shortest(50 μm) and A. cerana the longest (125 μm) hairs, although even these hairs are shorter than those of A. mellifera (275 μm)(Neese, 1965).

As expected from their different body and eye sizes, the three study species differ with respect to ommatidial diameter and number of ommatidia(Table 1; Fig. 2D,F,H), both of which are parameters that are known to vary with body size (see Kelber et al., 2006). Apis cerana has a similar number of ommatidia compared with the similarly-sized A. mellifera (in the range of 5000), but it has smaller eyes and smaller ommatidia, indicating a lower sensitivity to light than its European congener. The ommatidia of A. dorsata are 29.5μm in diameter and thus almost 10 μm wider than those of A. florea. The latter species also has fewer facets (<3500) than the former (>6300). A thicker cornea and longer crystalline cones in A. dorsata (Table 1) are possibly due to their larger eyes.

In most bees that have been studied so far, the highest spatial resolution has been found in the fronto–ventral eye region, where both interommatidial angles and the acceptance angles of each ommatidium are smallest (Warrant et al.,2004; Somanathan et al.,2009). In our study species, we estimated the acceptance angles in this eye region from the focal length and the rhabdom diameter and found similar results for A. cerana and A. florea, which have very small acceptance angles of 1.2 deg. and 1.1 deg.(Table 1). This is in contrast to the larger A. dorsata with acceptance angles of 1.8 deg., similar to the European honeybee (1.7 deg.). We assume that, as with other diurnal bees (Greiner et al., 2004a; Somanathan et al., 2009),acceptance angles follow interommatidial angles in the two diurnal species and thus can be used as indicators for spatial resolution. By contrast, although A. dorsata has a considerably larger number of facets than both smaller bees, its larger acceptance angle indicates a lower spatial resolution, which mostly results from the large distal rhabdom diameter(Fig. 3). With a diameter of 3.2 μm, its rhabdom is clearly larger than that of all other congeners: 1.5μm in A. florea, 1.8μm in A. cerana and 2.0μm in A. mellifera (Table 1). As a comparison, the halictid M. genalis, flying in the dim twilight of the Panamanian rainforest, has rhabdoms that are 8 μm wide (Warrant et al., 2004)whereas the obligately nocturnal X. tranquebarica has rhabdoms that are 6 μm wide (Somanathan et al.,2008). Both these species have correspondingly larger acceptance angles and lower spatial resolution(Warrant et al., 2004; Somanathan et al., 2008).

The sensitivity of apposition compound eyes depends on three parameters:the facet diameter, the acceptance angle of each ommatidium and the length of the rhabdom (see Eqn 2). The rhabdom length, together with the absorption coefficient of the rhabdom k (assumed to be similar in all species), determines the fraction of light captured by the rhabdom that is absorbed by the visual pigment. Although rhabdom length differs between the four species of Apis, it has a minor influence, with the percentage of white light absorbed by the rhabdom[kl/(2.3+kl), see Eqn 2] ranging from 36% in A. florea to 48% in A. mellifera (with 43% in A. cerana and 44% in A. dorsata). Taken together, it is mostly the larger facet diameters and acceptance angles that make the eyes of A. dorsata (S=0.21μm²sr) twice as light-sensitive as those of A. mellifera (S=0.11 μm²sr), three times as sensitive as those of A. cerana (S=0.07 μm²sr) and seven times as sensitive as those of A. florea (S=0.03μm²sr). However, even the eyes of A. dorsata are 10 times less sensitive than those of the dim light specialists M. genalis and X. tranquebarica (both with S=2.7μm²sr). In the bumblebee, Bombus terrestris,Kapustjansky et al. (Kapustjansky et al.,2007) found that large-sized workers and drones had larger facet diameters and ocelli and were able to fly at slightly lower experimental illumination levels compared with smaller-sized individuals.

The rather modest differences in optical sensitivity of the eyes to white light does not explain how A. dorsata and A. m. adansoniican fly and forage on moonlit nights, when light intensities are 100 times dimmer than the lowest intensities at which we observed A. cerana foraging and 1000 times lower than those at which A. florea was found foraging. Neither does the 10 times higher sensitivity of the truly nocturnal bees M. genalis and X. tranquebarica explain their ability to forage at even dimmer light intensities. The large rhabdom diameters (that compromise spatial resolution) and the anatomy of the lamina monopolar cells, the interneurones that receive signals from the photoreceptors, indicate that M. genalis most probably uses spatial pooling of the receptor signals to improve the signal-to-noise ratio and to allow reliable vision at these dim light levels(Warrant et al., 2004; Theobald et al., 2005; Warrant, 2008). Warrant et al.(Warrant et al., 1996), in a study on the spatial resolution and sensitivity of A. m. ligustica,proposed that even honeybees use spatial and temporal pooling of receptor signals to improve the signal-to-noise ratio, thereby enabling them to see reliably even in dim light. The dendritic trees of lamina monopolar cells of A. mellifera do not have equally wide branches as those of M. genalis but they do have wider branches than those in other diurnal insects (Greiner et al.,2004b). Even in bright light, the behaviourally measured resolution of A. m. ligustica is coarser than expected from the optics of the eyes, an adaptation that allows for extremely good colour discrimination and probably also higher sensitivity. By contrast, bees of the closely related genus Bombus have coarser colour discrimination but higher spatial acuity (Dyer et al.,2008). We assume that adaptations similar to those in A. m. ligustica are found in A. m. adansonii, and possibly also in A. dorsata, and may enable these species to forage on moonlit nights. It is not impossible that the eyes of A. m. adansonii have enlarged rhabdom diameters making their eyes just as sensitive as those of A. dorsata. Behavioural studies of the spatial and colour resolution of the facultatively nocturnal African and Asian honeybees should thus prove interesting.

Ocellar diameters in the two smaller Asian species are similar to those of the European race of A. mellifera: 0.24 mm in A. florea and 0.26 mm in A. cerana (Table 1; Fig. 2E,G). The ocelli are considerably larger in A. dorsata where they measure 0.38 mm in diameter (Table 1; Fig. 2I). The optical sensitivity of the ocellus is proportional to the square of the ocellar diameter (Warrant et al.,2006); thus, endowing A. dorsata with ocelli that are twice as sensitive as any of the other three species. Again, compared with the ocelli of M. genalis (0.49 mm) and X. tranquebarica (0.95 mm) (Warrant et al., 2006; Somanathan et al., 2009), even the ocelli of A. dorsata appear relatively small. The ocelli of all bees that have been studied earlier are under-focused and do not generate images of any useful resolution (Warrant et al., 2006; Somanathan et al., 2008). We did not expect an exception in the Asian species of Apis and we thus did not obtain focal lengths.

To conclude, our results indicate that of all species of Apisstudied so far, A. dorsata has the most sensitive visual organs: both the eyes and the ocelli give this species the best preconditions to fly and orient visually in dim light. Yet, both visual organs of this species are only twice as sensitive as those of the European race of A. melliferawhereas those of the African race A. m. adansonii have never been studied. Compared with bee species such as X. tranquebarica or M. genalis that are active only in dim light, A. dorsata has eyes of lower sensitivity that cannot sufficiently explain its ability to fly,orient, recognise rewarding flowers and find its way back to the hive when only a half-moon is present in the sky (and light levels are a million times lower than during the day). We need to therefore assume additional adaptations at the neuronal level, specifically temporal and spatial summation as proposed for the European honeybee (Warrant et al.,1996) and M. genalis(Warrant et al., 2004; Theobald et al., 2005; Warrant, 2008) to explain this behaviour. The truly diurnal bees, A. florea and A. cerana,have less sensitive eyes but the potential for a higher spatial resolution than both A. mellifera and A. dorsata.

LIST OF ABBREVIATIONS

 
• d

  distal rhabdom diameter

 
• D

  maximum ommatidial diameter

 
• f

  focal length of the facet lens

 
• k

  peak absorption coefficient of the visual pigment

 
• l

  rhabdom length

 
• S

  optical sensitivity to white light

 
• s_o

  distance between striped object and lens

 
• λi

  spatial wavelength of the image of the striped pattern

 
• λo

  spatial wavelength of the striped pattern

 
• Δρ

  acceptance angle