Arthropod Structure & Development 40 (2011) 509e520

Contents lists available at ScienceDirect

Arthropod Structure & Development

journal homepage: www.elsevier.com/locate/asd

The organization of honeybee ocelli: Regional specializations and rhabdom

arrangements
Willi Ribi a, c, Eric Warrant b, Jochen Zeil c, *
a

The Private University of Liechtenstein, Dorfstrasse 24, Triesen, FL-9495, Liechtenstein

Lund Vision Group, Department of Biology, University of Lund, Slvegatan 35, S-22362 Lund, Sweden
c
ARC Centre of Excellence in Vision Science, Division of Evolution, Ecology & Genetics, Research School of Biology, The Australian National University, Biology Place,
Building 46, Canberra ACT 0200, Australia
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 30 May 2011
Accepted 27 June 2011

We have re-investigated the organization of ocelli in honeybee workers and drones. Ocellar lenses are
divided into a dorsal and a ventral part by a cusp-shaped indentation. The retina is also divided, with
a ventral retina looking skywards and a dorsal retina looking at the horizon. The focal plane of lenses lies
behind the retina in lateral ocelli, but within the dorsal retina in the median ocellus of both workers and
drones. Ventral retinula cells are ca. 25 mm long with dense screening pigments. Dorsal retinula cells are
ca. 60 mm long with sparse pigmentation mainly restricted to their proximal parts. Pairs of retinula cells
form at, non-twisting rhabdom sheets with elongated, straight, rectangular cross-sections, on average
8.7 mm long and 1 mm wide. Honeybee ocellar rhabdoms have shorter and straighter cross-sections than
those recently described in the night-active bee Megalopta genalis. Across the retina, rhabdoms form
a fan-shaped pattern of orientations. In each ocellus, ventral and dorsal retinula cell axons project into
two separate neuropils, converging on few large neurons in the dorsal, and on many small neurons in the
ventral neuropil. The divided nature of the ocelli, together with the particular construction and
arrangement of rhabdoms, suggest that ocelli are not only involved in attitude control, but might also
provide skylight polarization compass information.
2011 Elsevier Ltd. All rights reserved.

Keywords:
Honeybees
Workers
Drones
Ocelli
Rhabdom organization
Polarization sensitivity

1. Introduction
In addition to the compound eyes, many insects possess three
single lens eyes, called ocelli (e.g. Wilson, 1975; Goodman, 1981;
Warrant et al., 2006; Berry et al., 2007a; reviewed by Mizunami,
1994). An ocellus typically consist of a lens, a vitreous body, an
iris, a corneageneous cell layer, a retina (including pigment cells)
and a neuropil in which photoreceptor axons make contact with
interneurons (Goodman, 1981).
Ocelli typically possess large elds of view and are usually
found to be heavily under-focused, although in some species the
optical focal plane lies in or near the retina (e.g. ies: Schuppe
and Hengstenberg, 1993; dragonies: Stange et al., 2002; Berry
et al., 2007a,c; locusts: Berry et al., 2007b; bees and wasps:
Warrant et al., 2006). Photoreceptors typically have spectral
sensitivities peaking in the UV and green (e.g. Wilson, 1975;
Geiser and Labhart, 1982; van Kleef et al., 2005; Berry et al.,

* Corresponding author. Tel.: 61 2 6125 5066; fax: 61 2 6125 3808.

E-mail address: jochen.zeil@anu.edu.au (J. Zeil).
1467-8039/$ e see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.asd.2011.06.004

2011) and interneurons possess large receptive elds, some of

them having the largest axon diameters in the insect brain (e.g.
Pan and Goodman, 1977; Goodman, 1981; Strausfeld and
Bassemir, 1985; Nssel and Hagberg, 1985; Mobbs, 1985; Berry
et al., 2006, 2007a).
Only in a few cases has it been possible to identify some of the
functional roles of ocelli (reviewed by Taylor and Krapp, 2007). In
dragonies, locusts and ies they are clearly important for the
control of head orientation around the roll and pitch axes (Stange,
1981; Taylor, 1981; Hengstenberg, 1993) and ocellar signals
contribute to the visual processing of stabilizing reexes (e.g.
Strausfeld and Bassemir, 1985; Parsons et al., 2006, 2010;
Simmons and de Ruyter van Steveninck, 2010). In the desert ant
Melophorus bagoti, the ocelli provide compass information
(Schwarz et al., 2011b) and in the desert ant Cataglyphis (Mote
and Wehner, 1980; Fent and Wehner, 1985), in Sarcophaga ies
(Wellington, 1953) and in bumble bees (Wellington, 1974) it has
been specically shown that the ocelli encode compass cues that
are derived from the pattern of polarized skylight. In hemipteran
bugs ocelli mediate negative phototactic behaviour (Lazzari et al.,
1998). In most other cases, the importance of ocelli can only be

510

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

indirectly assessed. Ocelli are predominantly found in ying

insects, although they are also well developed in some walking
insects (e.g. ants: Mote and Wehner, 1980; Fent and Wehner,
1985; Narendra et al., 2011; Schwarz et al., 2011a). Moreover,
compared with their day-active relatives, the lens diameters of
ocelli in night-active species of bees, wasps and ants are
massively increased (Kerfoot, 1967a; Kelber et al., 2006; Warrant
et al., 2006; Somanathan et al., 2008; Narendra et al., 2011).
Ocellar photoreceptors in night-active bees have high, yet variable absolute sensitivities and are comparatively slow, with
corner frequencies between 4 and 11 Hz (Berry et al., 2011),
which are much lower than the 12e25 Hz reported for photoreceptors in diurnal insects (Laughlin and Weckstrm, 1993). At the
neural level, the axons of ocellar interneurons project into the
optic lobes, the optic tubercles in the protocerebrum and to the
thoracic ganglia (reviewed Mizunami, 1994; for honeybees see
Pan and Goodman, 1977; Mobbs, 1985).
Despite much effort, the structure and possible function of the
ocelli of honeybees also remain elusive. The ocelli of the worker
bee have been previously described (Redikorzew, 1900; Gtze,
1928; Toh and Kuwabara, 1974; Kral, 1978; Goodman, 1981) and
the physiological properties of some interneurons are known
(Susuki et al., 1976; Guy et al., 1979; Milde, 1981, 1984a,b; Mobbs
et al., 1981). However, the morphological and optical details of
the worker bee ocelli, especially in comparison to the ocelli of
drones, have not been described in detail. We remedy this situation here by presenting an analysis of the external morphology,
the optics, the retina and the neuropil of the ocellar systems of
worker and drone honeybees. We show that all ocelli exhibit
dorso-ventral specializations with a bipartite lens, complementary
regionalization of the retina and separate dorsal and ventral
neuropils. We also provide anatomical evidence that ocellar
photoreceptors are likely to be sensitive to the plane of polarization of light.
2. Materials and methods
We used adult worker and drone bees (Apis mellifera L.), freshly
caught from hive entrances during January and February (Australian Summer) for optical investigations, and for scanning electron
microscope and light microscope analyses.

511

2.2. Scanning electron microscopy

The heads of both worker and drone bees were removed with
a scalpel and immediately xed in 2% paraformaldehyde and 2.5%
glutaraldehyde in phosphate buffer (pH 7.2e7.4) for 2 h. After
xation all hairs were removed, the heads were mounted on
double-sided sticky tape, and observed with a JEOL JSM 6400
scanning electron microscope at several different magnications.
2.3. Replica of the ocellar cornea
For the analysis of the outer appearance and arrangement of the
ocelli, and of the size and structure of their cornea, the ocelli were
painted with a thin layer of nail polish to produce corneal replicas.
After allowing the nail polish to dry for a few minutes the cornea
replicas were carefully removed and attened on a microscope
slide by cutting incisions with a micro-scalpel.
2.4. Analysis of rhabdom properties
Ocellar rhabdoms in bees have very elongated cross-sections
(e.g. Kral, 1978; Toh and Kuwabara, 1974; Berry et al., 2011) and
we were interested in determining how straight they are to assess
their potential polarization sensitivity. We imported digital
photographs of serial cross-sections through the ocellar retina into
a custom-made program (Jan M Hemmi and Robert Parker, The
Australian National University) written in Matlab (Mathworks,
Nattick, USA) that allowed us to determine the x/y coordinates of
ve equally spaced points along the long axis of all rhabdom crosssections in a section. We then determined the overall orientation of
rhabdoms as the connecting line between the two end points of
each elongated rhabdom cross-section. To determine how straight
rhabdoms are in the direction of the long axis of their cross-section,
we measured the orientation of the four line segments between
points 1 and 2, points 2 and 3, points 3 and 4, and points 4 and 5
(see inset Fig. 6a). We dene straightness as the difference between
segment orientation and the overall rhabdom orientation. Adding
the distances between the ve points also allowed us to determine
the length of rhabdom cross-sections. For comparison (see Fig. 8),
we conducted the same analysis with a section of the ocellar retina
of the night active bee Megalopta genalis, with kind permission
from Richard Berry (Berry et al., 2011).

2.1. Histology

2.5. Optics

For anatomical analysis, fully daylight adapted bees were

immobilized by cooling to 4  C. The head capsules were then
opened to allow xative to contact brain and compound eyes
directly. The brain was left in the head capsule to prevent
distortion and other artefacts and tissue was xed in a mixture
of 2% paraformaldehyde and 2.5% glutaraldehyde in phosphate
buffer (pH 7.2e7.4) for 2e3 h followed by 2% OsO4 in distilled
water for 1 h. After dehydration and embedding in Araldite
(FLUKA), semi-thin serial sections of 1 mm thickness were cut
with a diamond knife (DIATOME). Sections were stained with
toluidine blue, and analysed and photographed with a Leitz
photomicroscope.

The back focal distances and the focal lengths of ocellar lenses
were measured using a modication of Homanns (1924) hanging
drop method (see Warrant et al., 2006 for a more detailed
description of the method). A small piece of cuticle containing
either a lateral or a median ocellus was carefully dissected from
the head capsule of a freshly killed bee and placed in 0.9% sodium
chloride solution. All tissue was removed from the inner side of
the ocellar lens with a ne brush. The lens was then placed on
a droplet of physiological saline on a cover slip, in such a way that
its external surface remained in air. To prevent evaporation, we
constructed a chamber by mounting an O-ring on a microscope
slide. We then covered the upper side of the ring with a thin layer

Fig. 1. The ocelli of worker honeybees. (a) Frontal view. The median ocellus (arrow) is placed dorsally between the two compound eyes and looks upwards and forwards. The two
lateral ocelli are not visible, being completely covered by dense, long hairs. Scale bar 1 mm. (b) Side view. The restricted visual eld of the right lateral ocellus (arrow) is determined
by a special hair-free zone which gives the lateral ocelli clear view to the side and upwards. d: dorsal; p: posterior; Scale bar 1 mm. (c) Scanning electron micrograph frontal view of
worker bee ocelli with all hairs on the head capsule and the compound eyes having been removed. The normally hair-free zones at the ventral side of the ocelli can be recognized as
distinct patterns of smooth (at) cuticle, due to the lack of hair sockets (arrows). Scale bar 1 mm. (d) Dorsal view. The two lateral ocelli are clearly visible. Note the hair-free zones
providing unobstructed lateral and dorsal elds of view. Scale bar 1 mm. (e) Frontal view of a head replica. Hair-free zones (hfz) are clearly distinguishable from those regions of the
head that are normally covered by hairs (note hair sockets). ce: compound eye; hfz: hair-free zones mo: median ocellus; lo: lateral ocellus. Scale bar 300 mm.

512

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

of Vaseline. The cover slip was placed upside-down (with the lens
directed downwards) onto the O-ring. The back focal plane of the
lens was viewed through the cover slip and the saline with the
25 objective of a Zeiss microscope with the condenser removed.
A pattern of black-and-white stripes, made from translucent paper
and black tape pieces, was placed over the lamp in the microscope
foot at a distance s from the ocellar lens (s 113 mm). The spatial
wavelength lo of the stripe pattern was 8 mm. We then determined the position of best focus of the image of the stripe pattern
behind the ocellar lens. The spatial wavelength li of the image (in
mm), the spatial wavelength of the object lo and the object
distance s then allowed us to calculate the focal length f of the
ocellar lens:

f sli =lo
The nodal point N is situated at a distance of one focal length f in
front of the image and its position within the ocellar lens was
determined by measuring the back focal distance of the lens, that is,
the distance from the image to the back of the lens (as determined
by the calibrated focussing knob of the microscope) multiplied by
1.34 to correct for the refractive index of the saline droplet. The
distance of the nodal point from the proximal edge of the ocellar
lens is then simply the back focal distance subtracted from the focal
length.
2.6. Visual elds
The visual elds of ocelli were estimated by mounting worker
and drone bee heads with an insect pin on a goniometer. We
dened the vertical axis of the head as the line from the centre of
the median ocellus through the proboscis and the horizontal axis as
the line from the occipital opening to the centre of the basal scape
of the two antennae. Note, however, that the head posture of bees is
likely to differ depending on their ight speed and state of locomotion. We used the visibility of ocelli when heads were turned
through dened angles under a dissection microscope as a rough
measure of the maximal extent of visual elds.
3. Results
3.1. External morphology and visual elds
In the worker bee the three ocelli form an equilateral triangle
positioned on the apex of the head (Fig. 1c and d). In drones, the
three ocelli are slightly larger, sit closer together and are positioned
more frontally, squeezed in between the giant compound eyes
(Fig. 2). Ocellar lenses in both worker and drone bees have a slightly
ellipsoid shape (Figs. 1e and 2c), with major axes in the lateral ocelli
of the worker bee in the anterior-posterior direction and in the
ventral dorsal direction in the drone. The median ocellus is nearly
circularly symmetric in both sexes (Table 1).
The dense cover of hair on the worker and drone head make the
ocelli hard to recognize (Figs. 1a, b and 2a), although the head
cuticle just lateral of the lateral ocelli and just ventral of the median
ocellus is smooth and free of hairs (Figs. 1 and 2). Dense hair tufts
overlying the median ocellus on the worker bee head separate the
visual elds of all three ocelli in the horizontal direction and shade
the median ocellus from above (Fig. 1a and d). In the drone the
three ocelli are located at the very front of the head and are
arranged closely together (Fig. 2). They are also surrounded by
densely packed hairs that, however, leave an unobstructed view in
the frontal-ventral direction. In both worker and drone bees,
a special patch of dense, long hairs between the three ocelli limits
the visual eld in the dorsal direction.

The hairs around the ocelli act as a set of visors and shield the
median ocellus against direct sunlight from above. In the case of the
lateral ocelli, the surrounding hairs allow light to enter only from
the side and from above. The visibility of the ocellar lenses thus
allows us to roughly estimate their visual elds. The median ocellus
of the worker bee is invisible when the head is viewed directly from
the front, whereas it is fully visible when viewed from 20 above
the horizontal. At 30 above the horizontal the ocellus is covered
dorsally by hairs and when viewed at 40 it is no longer visible from
above. In elevation, the visual eld is centred at 25 above the
horizontal plane of the head and has a horizontal extent of
approximately 40 . The lateral ocelli only become visible at elevations larger than 50 and in azimuth they are fully visible 70
away from the midline. The centres of their visual elds therefore
lie approximately 70 to either side of the midline at an elevation
of 55 .
We investigated bees of different ages (16 d, 18 d, 23 d, 25 d) to
check whether the density of the hairs around the ocelli may be
subject to wear and tear, and thereby modify the visual eld.
However, we did not nd noticeable differences in hair cover in the
different ages of worker bees.
The ocelli of the drone look to the front and ventrally. Viewed
from straight ahead, a dense patch of hair completely obscures the
drone median ocellus but it is fully visible at an elevation of 15
below the horizon. In elevation, the visual eld of the drone median
ocellus is centred about 10 below the horizon with a horizontal
visual eld of approximately 25 either side of the midline. The
lateral ocelli become visible at about 10 azimuth either side of the
midline and their full extent is seen at 50 on either side of the
midline. The centres of their visual elds thus approximately view
the horizon over 45 on either side of the midline. Note, however,
that workers and possibly also drones hold their heads pitched
upwards in ight (e.g. Wehner, 1972; van Praagh et al., 1980).
3.2. Corneal lenses
The ocellar lenses are thickened areas of transparent cuticle
distal to the ocellar retinae and neuropils (Fig. 3). Frontal sections
through the lateral ocelli show a biconvex, thick lens with a cuspshaped (bipartite) outer surface and an asymmetrical inner
surface (Fig. 3aec). The shape of the corneal lens of the median
ocellus has the same cusp-shaped outer surface. Median and lateral
ocellar corneal replicas and SEM micrographs show a distinct
surface pattern in both sexes: a rough, uneven region with an
ellipsoid to roundish shape, covering two thirds of the lens surface
(Figs. 1e and 2c).
3.3. Optical properties of ocellar lenses
Focal length measurements were made from the lateral and
medial ocelli of two drones and three workers. Images formed
behind the corneal lenses were typically blurry and, compared with
the original striped object, of considerably reduced contrast. In all
cases, the measured focal lengths based on these images indicate
that lateral ocelli in both castes are heavily under-focused, that is,
the focal plane of the corneal lens lies behind the retina (Fig. 4, left
panels). The situation is different in the median ocelli where the
focal plane lies within the dorsal, but not the ventral retina (Fig. 4,
right panels). Due to the nature of the hanging drop method (in
which lens position on the drop is decided by gravity), the
measurements described below were unable to distinguish the
optical properties of different parts of the bipartite ocellar lens.
However, such differences may exist, as suggested by the division of
the ocellar retina into distinct dorsal and ventral regions (see
below).

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

513

Fig. 2. The ocelli of drone honeybees. (a) Frontal view. The three ocelli are arranged in a small triangle in a hair-free zone at the front of the head. The median ocellus (mo) is slightly
larger than the two lateral ocelli (lo), which are laterally constricted by the huge compound eyes (ce). The eld of view dorsally and ventrally is obscured by dense tuffs of hair. Scale
bar 1 mm. (b) Scanning electron micrograph of the dorsal view of the shaved drone head. The three ocelli sit on a raised part of the cuticle. The bulging compound eyes restrict the
ocellar eld of views in lateral directions. p: posterior. Scale bar 1 mm. (c) Frontal view of a drone replica. The ocelli of drones are slightly larger than those of the worker bee, sit
closer together on the front of the head, and are packed tightly between the giant compound eyes. Ocellar lenses have a rough surface and are surrounded by cuticle studded with
hair sockets (hs). d: dorsal. Scale bar 500 mm.

514

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

Table 1
Anatomical dimensions of lateral and median ocelli in honeybee workers and
drones.

Head size (mm)

Lateral ocelli minor diameter (mm)
Lateral ocelli major diameter (mm)
Median ocellus minor diameter (mm)
Median ocellus major diameter (mm)

Worker

Drone

3.76  0.07
293  6
315  4
264  5
295  21

4.39  0.13
305  3
396  4
364  6
372  3

Notes: for all measurements n 5; head size was measured as the frontal diameter
of the head, between the lateral edges of the compound eyes.

In workers, hanging drop measurements revealed that lateral

ocelli (n 3) have a focal length f of 367  49 mm and a posterior
nodal point N that is located 35  11 mm distal of the inner lens
surface (see Fig. 4). For the median ocellus (n 3), f 239  48 mm
and the posterior nodal point N is located 79  9 mm distal of the
inner lens surface. In drones (n 2), lateral and median ocelli have
focal lengths f of 308  42 mm and 220  5 mm, respectively, and
their posterior nodal points N are located, 85  11 mm and
102  19 mm respectively, distal of the inner lens surfaces.
Knowing the focal length f and the diameter D of the ocellar lens
we can now calculate the F-number F as F f/D. The F-number
predicts the light gathering capacity of an eye, with lower Fnumbers indicating a more sensitive eye (Warrant and McIntyre,
1993; Stavenga, 2003). Using ocellar lens diameters that are

averages of the major and minor diameters shown in Table 1, the

following F-numbers were calculated: 1.21 and 0.85 (worker lateral
and median ocelli respectively) and 0.88 and 0.60 (drone lateral
and median ocelli respectively). These F-numbers indicate that the
ocelli of drones are more sensitive than those of workers and that in
both castes the median ocellus is more sensitive than the lateral
ocelli.
3.4. The internal organization of ocelli
Proximal to the inner surface of the corneal lens lies an irregularly shaped vitreous body that appears darker than the lens in
toluidine blue stained sections (Fig. 3c and d). A single layer of
corneageneous cells e specialised epidermal cells that secrete the
corneal lens in development e lies between the vitreous body and
the retina. Densely packed small brownish pigment granules, as
seen in unstained sections, form a pigment sheath (iris) between
the lens and the receptor layer (retina) (Fig. 3c, see also Instausti
and Lazzari, 2000).
Vertical sections of all three ocelli, in worker and drone bees,
show a bipartite retina, with a dorsal part containing retinula cells
with long rhabdoms (50e65 mm) that receive light from horizontal
directions. The retinula cells of the ventral retina are about half that
length and receive light from the dorsal hemisphere (Fig. 3aed).
Rhabdoms in the ventral retina are located much closer to the lens
than rhabdoms of the larger dorsal retina. The two parts of the

Fig. 3. Histology of honeybee ocelli. (a) Semi-thin frontal section of a worker honeybee brain, showing the two lateral ocelli above the four calyces of the mushroom bodies. Note the
cusp in the outer surface of ocellar lenses (arrows). Section plane 42 mm from the front, stained with toluidine blue. ca: calyx of the median mushroom body; lo: lateral ocellus; mon:
neuropile of the median ocellus. Scale bar 100 mm. (b) Frontal section of a worker honeybee lateral ocellus. d: dorsal; dr: dorsal retina; i: iris; l: lens; v: ventral; vr: ventral retina.
Scale bar 100 mm. (c) Vertical section through the drone lateral ocellus. Note the cusp-shaped surface of the lens (arrow). Retinula cells in the dorsal retina (dr) are much longer than
in the ventral retina (vr) and their pigmentation is less dense compared with the ventral retina. vb: vitreous body; lcl: lens cell layer. Scale bar 50 mm. (d) Frontal section of a worker
lateral ocellus. kc: Kenyon cells of the mushroom body; rca: receptor cell axons; rcb: retinula cell bodies; sp: screening pigment. Scale bar 25 mm.

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

515

Fig. 4. Schematic diagram of ocellar optics in honeybee drones and workers as determined by the hanging drop technique (see Materials and methods and Results for details). The
focal plane is indicated by the dashed line relative to the nodal point (N). The lens is shown in orange, the vitreous body in yellow, the dorsal retina in dark green and the ventral
retina in light green. Note that in both drones and workers, the focal plane in lateral ocelli lies proximal to both the ventral and the dorsal retina, while it lies within the dorsal retina
in the median ocelli (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.).

ocellar retina also differ in the composition of screening pigments.

The shorter retinula cells in the ventral retina are densely packed
with red screening pigments along almost their entire length. In the
dorsal retina, only a few scattered pigments are dispersed over the
whole length of the retinula cells and at their proximal end around
the nuclei. The nuclei of receptor cells lie in the proximal third of
the retinula cells in both retinas and are oval in shape (major
diameter 7e11 mm). In both retinas, pairs of retinula cells form one
elongated rhabdom with cross-sections that are 1 mm wide with
a mean length of 8.7  2.4 mm (n 95) (Figs. 5a and 6d). Ocellar
rhabdoms thus form thin plates that are arranged parallel to the
direction of incident light.
In distal cross-sections of the retina of both worker and drone
bees, we counted up to 156 rhabdoms that form a fan-shaped
pattern of orientations (Fig. 5a and b), in which, however, horizontal orientations dominate (Fig. 6c, with horizontal 0 ; vertical 90 ). Because two paired retinula cells contribute microvilli
that are oriented perpendicular to the long axis of rhabdom crosssections, ocellar rhabdoms are likely to be highly sensitive to the
plane of polarization of light, provided these plates do not twist in
a distal-proximal direction and do not bend in lateral directions. We
did not nd evidence of twisting either in longitudinal sections or
in serial cross-sections through the dorsal and ventral retina. In
cross-section, however, these plates do not always appear to be
completely straight. We estimated the straightness of the rhabdoms along their cross-sectional long axis by determining the
difference between segment orientations and the mean orientation
of the rhabdom according to the upper inset in Fig. 6a (see Materials and methods for details). The mean absolute segment deviation from a straight line connecting the two end points of the

elongated cross-section is in most rhabdoms smaller than 20

(Fig. 6a). The lower inset in Fig. 6a shows differently modulated sine
waves sampled in the same way (numbers are means of absolute
segment deviation) to give an indication of what this measure
means in terms of straightness: most rhabdoms have a shallow
sine-wave shaped cross-section, which broadens, but would not
destroy the polarization sensitivity of retinula cells. Consequently,
the distribution of all signed segment deviations for 95 rhabdoms is
narrow and centred on zero (Fig. 6b).
Retinula cell axons from the dorsal and ventral retinas form two
distinctly separate nerve bundles as they pass to the rst synaptic
plexus where they make contact with second-order interneurons
(Fig. 7). Their diameters range between 1 and 3 mm. Axons from
retinula cells in the ventral retina form several distinct bundles that
connect to numerous small interneurons (axon diameters 2e4 mm),
while the receptor cell axons of the dorsal retina converge upon
a smaller number of large second-order neurons (see Pan and
Goodman, 1977).
4. Discussion
We have shown that the ocellar optics, the retina and the neuropils in honeybee workers and drones are divided in a dorsal and
a ventral part. In this respect, honeybee ocelli have a very similar
organization to that described in the lateral (Berry et al., 2007a), but
not the median ocellus in dragonies (Berry et al., 2007c) and to
that of the night-active bee M. genalis (Berry et al., 2011). This
division may simply reect the different ambient light intensities
available from the celestial and the terrestrial hemispheres, but
may also indicate that the two parts of the ocellar retina process

516

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

Fig. 5. The ocellar retina. (a) Cross-section of a retina of a worker lateral ocellus. Each
elongated rhabdom (arrows) is formed by two rhabdomeres of adjacent retinula cells
(circle). The rhabdoms are oriented in a fan-shaped pattern across the retina. Scale bar
20 mm. (b) How rhabdoms are arranged across the worker retina and their potential
polarization sensitivity. Red dots mark the ve points along each rhabdom that were
determined according to inset of Fig. 6a. Blue lines are perpendicular to the long axis of
rhabdom cross-sections and indicate the direction of maximal potential polarization
sensitivity of each rhabdom (for details see text) (For interpretation of the references to
colour in this gure legend, the reader is referred to the web version of this article.).

different kinds of information with the dorsal retina adapted for

resolving the horizon, and the ventral retina adapted for receiving
under-focused skylight (Stavenga et al., 1979; Berry et al., 2007a).
In addition we conrmed the nding by Toh and Kuwabara
(1974) that ocellar rhabdoms form elongated sheets with straight,
long and narrow distal cross-sections facing the incoming light.
Ocellar retinula cells thus must be very sensitive to the plane of
polarization of light. Indeed, the UV-sensitive, but not the BlueGreen sensitive ocellar photoreceptors analysed by Geiser and
Labhart (1982) have PS values around 4.7 (the ratio of maximum
to minimum response when stimulated via a rotating polarizer, c.f.
Greiner et al., 2007; Berry et al., 2011). In this context it is interesting to note that Velarde et al. (2005) found by in situ hybridization two opsins to be expressed in honeybee ocelli: a shortwavelength opsin AmUVop and a long-wavelength opsin
AmLop2. Their Fig. 3h appears to show a frontal section of the
median ocellus suggesting that AmUVops may only be expressed in
the ventral retina. This would clearly support our evidence for
functional differences between the dorsal and ventral retinae of
honeybee ocelli.
Apart from the Geiser and Labhart study, it is only in desert ants
that ocellar photoreceptors have been reported to be polarization
sensitive (Mote and Wehner, 1980) and where behavioural experiments have shown ocelli to be involved in providing compass
information from the pattern of polarized skylight (Fent and
Wehner, 1985; see also Schwarz et al., 2011b). The narrow platelike structure of ocellar rhabdoms has been documented in
honeybees and wasps (Kral, 1978) and in night-active bees

(Warrant et al., 2006). Warrant et al. (2006) noted the signicance

for polarization sensitivity of this rhabdom organization, as did
Schrer (1976) who interpreted the elongated cross-sections of
rhabdoms in the ventral retina of agelenid spiders as indicating
high polarization sensitivity. However, recordings from ocellar
photoreceptors in night-active bees provided no evidence for
polarization sensitivity (Berry et al., 2011).
We do not know the shape of rhabdoms in desert ants, but offer
the following explanation for why Berry et al. (2011) did not nd
ocellar photoreceptors in night-active halictine bees to be polarization sensitive. The polarization sensitivity of individual receptor
cells will depend on how straight the elongated cross-sections of
individual rhabdoms are. In nocturnal halictid bees (Megalopta)
these cross-sections are much more strongly curved (Fig. 8a,c) and
much longer (Fig. 8b) than in honeybees (Fig. 8b,d). Only a minority
of ocellar rhabdom cross-sections in Megalopta are short and
straight and are therefore likely to have been missed in electrophysiological recordings (Fig. 8c, thick white lines; from Berry et al.,
2011). The differences in the ocellar rhabdom organization in the
honeybee and in Megalopta are striking (e.g. Fig. 8a,b) and suggest
that polarization sensitivity is optimized in the former and minimized in the latter.
The potential polarization sensitivity of ocellar interneurons will
depend on how extensive and specic the convergence is of
receptor axons to ocellar interneurons. In all ocellar systems
investigated so far, receptor axons converge onto very few interneurons (e.g. locust: Patterson and Goodman, 1974; honeybee:
Heinzeller, 1976; Pan and Goodman, 1977; Milde, 1981, 1984a,b;
dragony: Patterson and Chappell, 1980; Berry et al., 2007c),
degrading the potentially high e-vector tuning of individual
receptors by pooling. However, the same is the case in the dorsal
rim of many insects, where the fan-like array of rhabdoms is
divided into 3 sub-elds in which the rhabdoms have very roughly
the same orientation and feed into POL neurons specic for each
sub-eld. Despite this convergence, these POL interneurons have
high polarization sensitivity (reviewed in Wehner and Labhart,
2006).
It is signicant in this context that we found distinct nerve tracts
associated with the ventral and the dorsal retina of all ocelli, with
ventral retina axons having smaller diameters and second-order
neurons being more numerous compared to those in the dorsal
neuropils. This suggests that in honeybees there is much less
convergence of photoreceptor axons to interneurons in the ventral
ocellar retina which looks skywards, compared to the dorsal retina
looking at the horizon. In honeybees, second-order neurons are
known to project into several different neuropils in the central
brain: some large neurons project into the posterior slope while
others project into the thoracic motor centres; thinner neurons
project into a number of other target neuropils (Pan and Goodman,
1977; Guy et al., 1979; Milde, 1984b; Mobbs, 1985), including the
central complex (Heinzeller, 1976), which has been shown to
receive input from the polarization sensitive dorsal rim of the
compound eyes in locusts and crickets (Homberg, 2004; Wehner
and Labhart, 2006).
Given the dorso-ventral specializations we describe here and
provided that ocellar photoreceptors are polarization sensitive,
what functional signicance would polarization sensitivity have?
We can see two inter-related reasons why polarization sensitivity
in ocellar photoreceptors and interneurons would be advantageous, considering the known and the potential functions of the
ocellar system.
In level, straight ight, the array of polarization sensitive
photoreceptors would be in a certain state of excitation depending
on the time of day and the direction of ight relative to the sun.
Receptors viewing the celestial hemisphere will receive highly

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

517

Fig. 6. The properties of ocellar rhabdoms. For each of 95 rhabdoms in the worker ocellar retina, the straightness of the cross-section was determined by measuring the deviation of
the orientation of four segments along the rhabdom from the line connecting the two end points as shown in the upper inset in (a). (a) Frequency distribution of the mean absolute
segment deviation per rhabdom. The lower inset shows expected means for different amounts of curvature of a hypothetical rhabdom. (b) Frequency distribution of all signed
segment deviations from the orientation of the line connecting the two end points of a rhabdom. (c) Overall rhabdom orientations relative to horizontal (0 ) across the retina. (d)
Frequency distribution of rhabdom length.

Fig. 7. Tangential section of a worker lateral ocellus. Each part of the bipartite retina
gives rise to a separate axon bundle (dorsal and ventral plexus). kc: Kenyon cells of the
mushroom body; a: anterior. Otherwise conventions as before. Scale bar 25 mm.

polarized light and depending on their orientation some will

absorb photons at a very high rate, while others will not. Photoreceptors in the dorsal retina will absorb rather few photons
overall, because light reected from the terrestrial hemisphere is
comparatively weak and in most cases un-polarized. A small rotation around the roll or pitch axis will change this pattern of excitation especially regarding the relative excitation of ventral and
dorsal photoreceptors. Looking through an array of polarisers thus
leads to an increase in the contrast between terrestrial and celestial
hemispheres and through this will provide a more robust signal for
attitude control, compared to pure luminance detectors. A second
reason why polarization sensitive ocellar photoreceptors would
offer advantages is that they could in addition to signalling changes
in roll and pitch also provide compass information by signalling
changes in the animals orientation around the yaw axis relative to
the pattern of polarized skylight. Indeed, three studies to date have
implicated the ocellar system in polarized light detection and its
contribution to the polarization compass: in the desert ant Cataglyphis (Fent and Wehner, 1985; see also Schwarz et al., 2011b), the

518

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

Fig. 8. A comparison of rhabdom properties between honeybees and the night-active bee Megalopta (histological material of Megalopta was kindly provided by Richard Berry and
Eric Warrant (see Berry et al., 2011)). (a) A comparison of the straightness of rhabdom cross-sections as measured by the mean absolute deviation of rhabdom segments per
rhabdom from the line connecting the two end points of rhabdom cross-sections. Inset shows the distribution of segment deviations (mean of all absolute segment deviations:
honeybee, 9.3  7.6 ; Megalopta, 19.7  15.1 ). Note that honeybee rhabdom cross-sections are much straighter than those in Megalopta. Rhabdoms in Megalopta were divided into 8
segments of equal length. (b) Frequency distributions of rhabdom length in honeybees and Megalopta. Honeybee rhabdoms are much shorter (8.7  2.4 mm, n 95) than those of
Megalopta (14.3  3.5 mm, n 84). (c) A horizontal cross-section through the retina of the median ocellus of Megalopta (semi-thin cross-section kindly provided by Richard Berry;
from Berry et al., 2011). Rhabdoms are emphasized by black lines. A few short and straight rhabdoms are labelled white (see text for explanation). Scale bar 20 mm. (d) A horizontal
cross-section through the retina of the median ocellus of a honeybee worker. Conventions as in (c).

y Sarcophaga (Wellington, 1953; see also von Philipsborn and

Labhart, 1990) and in the bumble bee Bombus terricola occidentalis (Wellington, 1954). Apart from this, the provision of polarization
compass information has so far been attributed to the specialized
dorsal rim area of the compound eyes in insects (reviewed by
Wehner and Labhart, 2006).
In most cases that have been investigated so far, the ocellar
optics produce an under-focused image on the retina (e.g. ies:
Schuppe and Hengstenberg, 1993; dragonies: Berry et al., 2007c;
locusts: Berry et al., 2007c) and our results show this also to be the
case in the lateral ocelli in honeybees. However, recent work on
dragonies and wasps has found that in some ocelli the image
plane actually does lie in the retina (Stange et al., 2002; Warrant
et al., 2006), suggesting the potential for higher contrast and
resolution. We now show this also to be the case for the dorsal
retina in the honeybee median ocellus. One of the demonstrated
roles of the ocellar system is its contribution to attitude control of
the head around roll and pitch axes by making use of the light
intensity differences between the celestial and the terrestrial
hemispheres (Hesse, 1908; Wilson, 1975; Stange, 1981; Stange and
Howard, 1979; Taylor, 1981). During ight, rotations around the

body axis (roll) are thought to be coded by changes in image

brightness experienced by the lateral ocelli, while changes in body
axis inclination (pitch) are instead coded by the median ocellus.
Experimental manipulations of the brightness experienced by the
median and lateral ocelli elicit head rotations in dragonies and
house ies (Stange, 1981; Hengstenberg, 1993; Schuppe and
Hengstenberg, 1993), corrective ight manoeuvres in honeybees
(Kastberger, 1990), but only minimal steering responses in the
blowy Calliphora, suggesting species differences in ocellar
involvement in ight control (Wehrhahn, 1984).
In this context, under-focused optics are benecial, because they
remove high spatial frequencies that may obscure the horizon line
(Wilson, 1975). On the other hand, horizon detection is particularly
improved in the dragony median ocellus, where the vertical and
horizontal lens radii are extremely different so that horizontal
contours are in focus and provide high contrast, while vertical
contours are defocused and provide low contrast (Stange et al.,
2002; van Kleef et al., 2005; Berry et al., 2007a,c). The spectral
and visual eld properties of ocellar interneurons associated with
the median ocellus in the dragony conrm that the retina is
specically adapted to resolve horizontally extended features of the

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

world across a narrow band of elevations (Stange et al., 2002; Berry

et al., 2007a,c). This is also reected in the directionally selective
responses of ocellar interneurons in the ventral cord of dragonies
(Zenkin and Pigarev, 1971). In honeybees, the median ocellus does
not show such an extreme astigmatism, but the observation that
the dorsal retina in the median ocellus is more in focus than the
ventral retina, again suggests a need for high contrast in those parts
of the retina that look at the horizon.
It has been noted for some time that ocelli provide information faster than the compound eyes. In bees, the latency of ocellar
and compound eye pathways was measured to be 6 to 9.5 ms in
descending multimodal neurons that receive inputs from both
the compound eyes and the ocelli (Guy et al., 1979). Recently,
Parsons et al. (2008, 2010) discovered that fast ocellar input also
reaches the motion sensitive neurons in the lobular plate in
blowies, where motion signals generated by specic rotational
movements of the head from the compound eyes and from the
ocelli are combined. Ocellar input arrives within 6 ms in these
neurons.
Lastly, we describe a number of external features of honeybee
ocelli that have not been noticed before, at least in bees, but are
likely to be functionally signicant. First, ocelli are associated with
distinct patches of hair-free cuticle and of hair-screens that affect
the visual eld of all three ocelli. Second, in both worker bees and
drones, the surfaces of the ocellar lenses are not smooth. Between
half and two thirds of each lens has a rough surface (Figs. 1e and 2c),
very similar to what can be seen in the ocelli of the nocturnal bee
Megalopta (Warrant et al., 2006). The functional signicance of this
lens surface modication is presently not clear, but we suggest that
it may serve to scatter and defocus incoming light, much like the
modied optics in the dorsal rim area of the compound eyes of
many insects (Schinz, 1975; Meyer and Labhart, 1981). Third, the
outer surface of ocellar lenses in honeybees is clearly bipartite,
divided by an incision that is oriented in the anterior-posterior
direction. As a consequence, one third of the lens (with one
surface radius) points dorsally, and two thirds (with another
surface radius) points either ventrally and laterally (as in the lateral
ocellus), or frontally (as in the median ocelli). In comparison, the
radius of the outer surface of the ocellar lenses of dragonies
appears to be uniform (Berry et al., 2007a,c). Fourth, the drone
ocellar system seems to be different from that of workers in that
drone ocelli are larger than in workers and do not sample the same
region of visual space. The three ocelli of the drone bee are positioned close together, sharing a fronto-dorsal visual eld, much like
dragonies, locusts and nocturnal sweat bees (Stange, 1981; Berry
et al., 2007b; Wilson, 1975; Warrant et al., 2006). In the worker
bee, in contrast, lateral ocelli are positioned at the apex of the head
and thus extend the ocellar visual eld laterally and dorsally. Why
is there this difference between drones and workers? In hymenopterans, ocellar size is known to be strongly correlated with light
intensity at peak activity: species active in dimmer light have larger
ocelli than those active in brighter light (e.g. Kerfoot, 1967a,b;
Warrant et al., 2006; Narendra et al., 2011). Drones, however, are
active in full daylight, as are worker bees. At this stage, we can only
speculate that the drones additional task of tracking small objects
during mating (van Praagh et al., 1980; Vallet and Coles, 1993; Gries
and Koeniger, 1993) may require a faster and more precise control
of head orientation around the three rotational axes, as has been
shown to be provided by ocellar input (Parsons et al., 2006, 2010).
The larger lens size of drone ocelli improves light sensitivity and
through this might provide a signicantly better signal to noise
ratio (e.g. Warrant and McIntyre 1993) for attitude control to
support the faster ight speeds and turning rates of drones
(Gmeinbauer and Crailsheim, 1993; Vallet and Coles, 1993; Gries
and Koeniger, 1993).

519

Acknowledgements
We thank Paul Helliwell and Ryszard Maleszka for the supply
of bees, the ANUs Centre for Advanced Microscopy for the use of
their facilities and Waltraud Pix for her help with analysing
rhabdom shapes and dimensions. We thank two reviewers for
their constructive comments. This work received nancial support
in the form of Visiting Fellowships from the Centre for Visual
Sciences to WR and from the ARC Centre of Excellence in Vision
Science to EW and WR. We acknowledge additional support from
the Private University of Liechtenstein, the Swedish Research
Council, the US Air Force of Scientic Research (AFOSR: Grant No.
FA8655-07-C-4011) and the ARC Centre of Excellence in Vision
Science.

References
Berry, R., Stange, G., Olberg, R., van Kleef, J., 2006. The mapping of visual space by
identied large second-order neurons in the dragony median ocellus. Journal
of Comparative Physiology A 192, 1105e1123.
Berry, R., van Kleef, J., Stange, G., 2007a. The mapping of visual space by dragony
lateral ocelli. Journal of Comparative Physiology A 193, 495e513.
Berry, R.P., Warrant, E.J., Stange, G., 2007b. Form vision in the insect dorsal ocelli: an
anatomical and optical analysis of the locust ocelli. Vision Research 47,
1382e1393.
Berry, R.P., Warrant, E.J., Stange, G., 2007c. Form vision in the insect dorsal ocelli: an
anatomical and optical analysis of the dragony median ocellus. Vision
Research 47, 1394e1409.
Berry, R.P., Wcislo, W.T., Warrant, E.J., 2011. Ocellar adaptations for dim light vision
in a nocturnal bee. Journal of Experimental Biology 214, 1283e1293.
Fent, K., Wehner, R., 1985. Ocelli e a celestial compass in the desert ant Cataglyphis.
Science 228, 192e194.
Geiser, F.X., Labhart, T., 1982. Electrophysiological investigations on the ocellar
retina of the honeybee (Apis mellifera). Verhandlungen der Deutschen Zoologischen Gesellschaft 75, 307.
Gmeinbauer, R., Crailsheim, K., 1993. Glucose utilization during ight of honeybee
(Apis mellifera) workers, drones and queens. Journal of Insect Physiology 39,
959e967.
Gtze, G., 1928. Untersuchungen an Hymenopteren ber das Vorkommen und die
Bedeutung der Stirnaugen. Zoologisches Jahrbuch Physiologie 44, 211e268.
Goodman, L.J., 1981. Organisation and physiology of the insect dorsal ocellar system.
In: Autrum, H. (Ed.), Handbook of Sensory Physiology, vol. VII 6B. Springer
Verlag, Berlin, pp. 201e286.
Greiner, B., Cronin, T.W., Ribi, W.A., Wcislo, W.T., Warrant, E.J., 2007. Anatomical and
physiological evidence for polarisation vision in the nocturnal bee Megalopta
genalis. Journal of Comparative Physiology A 193, 591e600.
Gries, M., Koeniger, N., 1993. Straight forward to the queen: pursuing honeybee
drones (Apis mellifera L.) adjust their body axis to the direction of the queen.
Journal of Comparative Physiology A 179, 539e544.
Guy, R.G., Goodman, L.J., Mobbs, P.G., 1979. Visual interneurons in the bee brain:
synaptic organisation and transmission by graded potentials. Journal of
Comparative Physiology 134, 253e264.
Heinzeller, T., 1976. Second-order ocellar neurons in the brain of the honeybee (Apis
mellifera). Cell and Tissue Research 171, 91e99.
Hengstenberg, R., 1993. Multisensory control in insect oculomotor systems. In:
Miles, F.A., Wallman, J. (Eds.), Visual Motion and its Role in the Stabilization of
Gaze. Elsevier, Amsterdam, pp. 285e298.
Hesse, R., 1908. Das Sehen der niederen Tiere. Fischer Verlag, Jena.
Homann, H., 1924. Zum Problem der Ocellenfunktion der Insekten. Zeitschrift fr
vergleichende Physiologie 1, 541e578.
Homberg, U., 2004. In search of the sky compass in the insect brain. Naturwissenschaften 91, 199e208.
Instausti, T.C., Lazzari, C.R., 2000. An ocellar pupil that does not change with light
intensity, but with the insect age in Triatoma infestans. Memrias do Instituto
Oswaldo Cruz 95, 743e746.
Kastberger, G., 1990. The ocelli control the ight course in honeybees. Physiological
Entomology 15, 337e346.
Kelber, A., Warrant, E.J., Pfaff, M., Walln, R., Theobald, J.C., Wcislo, W.T.,
Raguso, R.A., 2006. Light intensity limits foraging activity in nocturnal and
crepuscular bees. Behavioral Ecology 17, 63e72.
Kerfoot, W.B., 1967a. Correlation between ocellar size and the foraging activities of
bees (Hymenoptera; Apoidea). American Naturalist 101, 65e70.
Kerfoot, W.B., 1967b. The lunar periodicity of Sphecodogastra texana, a nocturnal bee
(Hymenoptera: Halictidae). Animal Behaviour 15, 479e486.
Kral, K., 1978. The orientation of the rhabdoms in the ocelli of Apis mellifera carnia
Pollm. and of Vespa vulgaris L. Zoologisches Jahrbuch Physiologie 82, 263e271.
Laughlin, S.B., Weckstrm, M., 1993. Fast and slow photoreceptors e a comparative
study of the functional diversity of coding and conductances in the Diptera.
Journal of Comparative Physiology A 172, 593e609.

520

W. Ribi et al. / Arthropod Structure & Development 40 (2011) 509e520

Lazzari, C.R., Reiseman, C.E., Insausti, T.C., 1998. The role of the ocelli in the
phototactic behaviour of the haematophagous bug Triatoma infestans. Journal of
Insect Physiology 44, 1159e1162.
Meyer, E.P., Labhart, T., 1981. Pore canals in the cornea of a functionally specialized
area of the honey bees compound eye. Cell and Tissue Research 216, 491e501.
Milde, J.J., 1981. Graded potentials and action potentials in the large ocellar interneurons of the bee. Journal of Comparative Physiology 143, 427e434.
Milde, J.J., 1984a. Ocellar interneurons in the honeybee: structure and signals of
L-neurons. Journal of Comparative Physiology A 154, 683e693.
Milde, J.J., 1984b. Ocellar interneurons in the honeybee: characteristics of spiking
L-neurons. Journal of Comparative Physiology A 155, 151e160.
Mizunami, M., 1994. Functional diversity of neural organization in insect ocellar
systems. Vision Research 35, 443e452.
Mobbs, P.G., 1985. Brain structure. In: Kerkut, G.A. (Ed.), Comprehensive Insect
Physiology, Biochemistry and Pharmacology, vol. 5. Pergamon Press, Oxford,
pp. 299e370.
Mobbs, P.G., Guy, R.G., Goodman, L.J., Chappell, R.L., 1981. Relative spectral sensitivity and reverse Purkinje shift in identied L-neurons of the ocellar retina.
Journal of Comparative Physiology A 144, 91e97.
Mote, M.I., Wehner, R., 1980. Functional characteristics of photoreceptors in the
compound eye and ocellus of the desert ant, Cataglyphis bicolor. Journal of
Comparative Physiology 137, 63e71.
Nssel, D.R., Hagberg, M., 1985. Ocellar interneurones in the blowy Calliphora
erythrocephala: morphology and central projections. Cell and Tissue Research
242, 417e426.
Narendra, A., Reid, S.F., Greiner, B., Peters, R.A., Hemmi, J.M., Ribi, W., Zeil, J., 2011.
Caste-specic visual adaptations to distinct daily activity schedules in Australian
Myrmecia ants. Proceedings of the Royal Society of London B 278, 1141e1149.
Pan, K.C., Goodman, L.J., 1977. Ocellar projections within the central nervous system
of the honey bee, Apis mellifera. Cell and Tissue Research 176, 505e577.
Parsons, M.M., Krapp, H.G., Laughlin, S.B., 2006. A motion-sensitive neurone
responds to signals from the two visual systems of the blowy, the compound
eyes and ocelli. Journal of Experimental Biology 209, 4464e4474.
Parsons, M.M., Krapp, H.G., Laughlin, S.B., 2010. Sensor fusion in identied visual
interneurons. Current Biology 20, 624e628.
Patterson, J.A., Chappell, R.L., 1980. Intracellular responses of procion lled cells and
whole nerve cobalt impregnation in the dragony median ocellus. Journal of
Comparative Physiology A 139, 25e39.
Patterson, J.A., Goodman, L.J., 1974. Intracellular responses of receptor cells and
second order cells in the ocelli of the desert locust Schistocerca gregaria. Journal
of Comparative Physiology 95, 237e250.
Redikorzew, W., 1900. Untersuchungen ber den Bau der Ocellen der Insekten.
Zeitschrift fr wissenschaftliche Zoologie 68, 581e624.
Schinz, R.H., 1975. Structural specialization in the dorsal retina of the bee, Apis
mellifera. Cell and Tissue Research 162, 23e34.
Schrer, W.-D., 1976. Polarisationsempndlichkeit rhabdomerialer Systeme in den
Hauptaugen der Trichterspinne Agelena gracilens (Arachnida: Araneae: Agelenidae). Entomologica Germanica 3, 88e92.
Schuppe, H., Hengstenberg, R., 1993. Optical properties of the ocelli of Calliphora
erythrocephala and their role in the dorsal light response. Journal of Comparative Physiology A 173, 143e149.
Schwarz, S., Narendra, A., Zeil, J., 2011a. The properties of the visual system in the
Australian desert ant Melophorus bagoti. Arthropod Structure and Development
40, 128e134.
Schwarz, S., Albert, L., Wystrach, A., Cheng, K., 2011b. Ocelli contribute to the
encoding of celestial compass information in the Australian desert ant Melophorus bagoti. Journal of Experimental Biology 214, 901e906.
Simmons, P.J., de Ruyter van Steveninck, R.R., 2010. Sparse but specic temporal
coding by spikes in an insect sensory-motor ocellar pathway. Journal of
Experimental Biology 213, 2629e2639.
Somanathan, H., Borges, R.M., Warrant, E.J., Kelber, A., 2008. Visual ecology of Indian
carpenter bees I: light intensities and ight activity. Journal of Comparative
Physiology A 194, 97e107.

Stange, G., 1981. The ocellar component of ight equilibrium control in dragonies.
Journal of Comparative Physiology 141, 335e347.
Stange, G., Howard, J., 1979. An ocellar dorsal light response in dragony. Journal of
Experimental Biology 83, 351e355.
Stange, G., Stowe, S., Chahl, J.S., Massaro, A., 2002. Anisotropic imaging in the
dragony median ocellus: a matched lter for horizon detection. Journal of
Comparative Physiology A 188, 455e467.
Stavenga, D.G., 2003. Angular and spectral sensitivity of y photoreceptors. II.
Dependence on facet lens F-number and rhabdomere type in Drosophila.
Journal of Comparative Physiology A 189, 189e202.
Stavenga, D.G., Bernard, G.D., Chappell, R.L., Wilson, M., 1979. Insect pupil mechanisms. III. On the pigment migration in dragony ocelli. Journal of Comparative
Physiology A 129, 199e205.
Strausfeld, N.J., Bassemir, U.K., 1985. Lobula plate and ocellar interneurons converge
onto a cluster of descending neurons leading to neck and leg motor neuropil in
Calliphora erythrocephala. Cell and Tissue Research 240, 617e640.
Susuki, H., Tateda, H., Kuwabara, M., 1976. Activities of antennal and ocellar interneurones in the protocerebrum of the honey-bee. Journal of Experimental
Biology 64, 405e418.
Taylor, C.P., 1981. Contribution of compound eyes and ocelli to steering of locusts in
ight. I. Behavioural analysis. Journal of Experimental Biology 93, 1e18.
Taylor, G.K., Krapp, H.G., 2007. Sensory systems and ight stability: what do insects
measure and why? Advances in Insect Physiology 34, 231e316.
Toh, Y., Kuwabara, M., 1974. The ne structure of the dorsal ocellus of the worker
honeybee. Journal of Morphology 143, 285e306.
Vallet, A.M., Coles, J.A., 1993. The perception of small objects by the drone
honeybee. Journal of Comparative Physiology A 172, 183e188.
van Kleef, J., James, A.C., Stange, G., 2005. A spatiotemporal white noise analysis of
photoreceptor responses to UV and green light in the dragony median ocellus.
Journal of General Physiology 126, 481e497.
van Praagh, J.P., Ribi, W.A., Wehrhahn, C., Wittmann, D., 1980. Drone bees xate the
queen with the dorsal frontal part of the their compound eyes. Journal of
Comparative Physiology 136, 263e266.
Velarde, R.A., Sauer, C.D., Walden, K.K., Fahrbach, S.E., Robertson, H.M., 2005.
Pteropsin: a vertebrate-like non-visual opsin expressed in the honey bee brain.
Insect Biochemistry and Molecular Biology 35, 1367e1377.
von Philipsborn, A., Labhart, T., 1990. A behavioural study of polarization vision
in the y, Musca domestica. Journal of Comparative Physiology A167,
737e743.
Warrant, E.J., McIntyre, P.D., 1993. Arthropod eye design and the physical limits to
spatial resolving power. Progress in Neurobiology 40, 413e461.
Warrant, E.J., Kelber, A., Walln, R., Wcislo, W.T., 2006. Ocellar optics in nocturnal
and diurnal bees and wasps. Arthropod Structure and Development 35,
293e305.
Wehner, R., 1972. Pattern modulation and pattern detection in the visual system
of hymenoptera. In: Wehner, R. (Ed.), Information Processing in the Visual
System of Arthropods. Springer Verlag, Berlin, Heidelberg, New York,
pp. 183e194.
Wehner, R., Labhart, T., 2006. Polarisation vision. In: Warrant, E.J., Nilsson, D.-E.
(Eds.), Invertebrate Vision. Cambridge University Press, Cambridge,
pp. 291e348.
Wehrhahn, C., 1984. Ocellar vision and orientation in ies. Proceedings of the Royal
Society of London B 222, 409e411.
Wellington, W.G., 1953. Motor responses evoked by the dorsal ocelli of Sarcophaga
aldrichi Parker, and the orientation of the y to plane polarized light. Nature
172, 1177e1179.
Wellington, W.G., 1974. Bumblebee ocelli and navigation at dusk. Science 183,
550e551.
Wilson, M., 1975. The functional organization of insect ocelli. Journal of Comparative Physiology 124, 317e331.
Zenkin, G.M., Pigarev, I.N., 1971. Optically determined activity in the cervical nerve
chain of the dragony. Biophysics 16, 309e317 (English translation from Biozika 16, 299e306).