Download Unique structure and optics of the lesser eyes of

Available online at www.sciencedirect.com
Vision Research 48 (2008) 1061–1073
www.elsevier.com/locate/visres
Unique structure and optics of the lesser eyes of the box jellyfish
Tripedalia cystophora
A. Garm a,*, F. Andersson b, Dan-E. Nilsson a
a
Department of Cell and Organism Biology, Lund University, Zoology building, Helgonavägen 3, 22362 Lund, Sweden
b
Center for Mathematical Sciences, Lund University, Sweden
Received 9 August 2007; received in revised form 10 December 2007
Abstract
The visual system of box jellyfish comprises a total of 24 eyes. These are of four types and each probably has a special function. To
investigate this hypothesis the morphology and optics of the lesser eyes, the pit and slit eyes, were examined. The pit eyes hold one cell
type only and are probably mere light meters. The slit eyes, comprising four cell types, are complex and highly asymmetric. They also
hold a lens-like structure, but its optical power is minute. Optical modeling suggests spatial resolution, but only in one plane. These
unique and intriguing traits support strong peripheral filtering.
Ó 2008 Elsevier Ltd. All rights reserved.
Keywords: Cubomedusa; Vitreous cells; Optical model; Refractive index; Visual field
1. Introduction
Cubozoa is a small exotic group of cnidarians found in
tropical and subtropical waters around the world. They are
known for their immense stinging powers but another truly
amazing feature of these animals is their visual equipment.
It has been known for well over 100 years that the medusa
stage of the cubozoans, box jellyfish, all possess four sensory structures, the rhopalia, each carrying a similar set
of six eyes (Berger, 1898, 1900; Claus, 1878; Hertwig &
Hertwig, 1878; Schewiakoff, 1889). These six eyes are of
four types: two relatively large medial eyes with spherical
lenses, a lateral pair of pigment pit eyes, and a lateral pair
of pigment slit eyes.
There has been much speculation on the usage of the eyes
in box jellyfish (Hamner, Jones, & Hamner, 1995; Kinsey,
1986; Pearse & Pearse, 1978). In a recent study we have
shown that medusae of the Caribbean species Tripedalia
cystophora and the Australian species Chiropsella bronzie
*
Corresponding author.
E-mail address: [email protected] (A. Garm).
0042-6989/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.visres.2008.01.019
perform a visually guided obstacle avoidance response
(Garm, O’Connor, Parkefelt, & Nilsson, 2007). Observations made by other authors working with other species of
box jellyfish support the presence of this behavior. (Hamner
et al., 1995; Hartwick, 1991; Matsumoto, 1995). Attraction
to light is another behavior which seems to be present in
most cubomedusae. T. cystophora has been shown to be
attracted to light in their natural habitat where they swim
in and out of light shafts in between the mangrove roots
(Buskey, 2003; Stewart, 1996). It has also been noted that
they are attracted to a point light source and can ‘‘remember” the direction (Romanes, 1876). Our electrophysiological studies have revealed the lens eyes to be slow and colorblind (Coates, Garm, Theobald, Thompson, & Nilsson,
2006; Garm, Coates, Seymour, Gad, & Nilsson, 2007).
Several studies have dealt with the morphology of box
jellyfish eyes, with the two lens eyes receiving by far the
most attention (Berger, 1898; Kozmic et al., 2004; LaskaMehnert, 1985; Laska & Hündgen, 1982; Matsumoto,
1995; Nilsson, Coates, Gislén, Skogh, & Garm, 2005; Pearse & Pearse, 1978; Piatigorsky, Horwitz, Kuwabara, &
Cutress, 1989; Piatigorsky & Kozmic, 2004; Yamasu &
Yoshida, 1976), since these eyes are built like typical cam-
1062
A. Garm et al. / Vision Research 48 (2008) 1061–1073
era-type eyes, and constitute an amazing analog to cephalopod and vertebrate eyes. Both the cubozoan lens eyes
have a more or less spherical lens and a well-developed
hemispherical retina. Their photoreceptors are mono-ciliated. The cilium is 20–40 lm long and dissolves into
numerous microvilli (Laska & Hündgen, 1982; Pearse &
Pearse, 1978). The lower lens eye of the Australian species
Chironex fleckeri has a few thousand photoreceptors and
the small lens eye a few hundred (Pearse & Pearse, 1978).
The morphology of the two other eye types, the slit and
pit eyes, has briefly been touched upon by a few authors
(Berger, 1898; Laska & Hündgen, 1982; Martin, 2002; Satterlie, 2002). They describe the shape of the pigment cups,
give some data on the photoreceptors, and refer to them as
simple eyes. Interestingly, Satterlie (2002) suggests from
light microscopy that there might be a ‘‘lens-like” structure
in the slit eye and therefore possible image formation.
Nothing is known about the functions of the smaller eyes,
but if the slit eye contains a lens it would indicate that this
eye is more complex than previously believed. Thorough
morphological examination based on detailed transmission
electron microscopy is needed to verify the presence and
nature of this structure. Detailed electron microscopical
data are also needed to appreciate the visual capacities of
the slit and pit eyes, and thereby to obtain a more complete
understanding of the visual system of box jellyfish.
It is the aim of the present work to examine the structure
of the pit and slit eyes of the box jellyfish T. cystophora,
and to use this information to analyze their optics with special attention on the possibility of spatial resolution. We
show that the pit eyes hold a single cell type only and are
probably mere light meters. The slit eyes have four cell
types, including two types of photoreceptors and vitreous
cells forming a lens-like structure. They may have spatial
resolution but along the transverse axis of the eye only.
alternative method was therefore developed to obtain information on
the refractive indices. Graded refractive indices have been directly measured from the lenses of the two large lens eyes (Nilsson et al., 2005). These
measurements are used to extrapolate the indices from the vitreous cells
under two assumptions. It is assumed that the vesicles found in the vitreous cells contain the same kind of crystallin as the lenses of the lens eye. It
is also assumed that the electron density in TEM-sections can be used as a
measure of the relative crystallin concentration. It could be added that
even if the first assumption should be wrong, it should not influence the
results much. The electron density is roughly proportional to the tissue
density (in this case the protein concentration), which in turn is proportional to the refractive index.
Under these assumptions, TEM-sections through the center of the lens
from three lower lens eyes were used to create a calibration curve between
electron density (relative pixel darkness) and refractive index (Fig. 1). The
good fit between electron density gradient and measured refractive index
gradient confirms the validity of the method. The calibration yielded the
following equation for the conversion: n = ((255 – D)/223 + 4.791)/3.88,
where n is the refractive index and D is the pixel value in 8-bit grayscale.
The relative electron density in the TEM-micrographs is strongly influenced by the staining procedure and camera settings. All the TEM-sections
were therefore carefully stained using the same protocol, and all the digital
pictures were normalized to the same contrast spectrum using an area of
the section without any tissue as the white reference, and the pigment
granules of the photoreceptors as the black reference. For the same reason, TEM-sections with diverging thickness were not used.
2.2. Optical modeling
The refractive indices obtained from the TEM-micrographs of the vitreous cells were used to model their refractive power. Due to the complex
shape of the vitreous cell population, and the arrangement of the crystallin
in vesicles with a size close to the wavelength of visible light (0.3–1.5 lm),
normal geometrical approximation of the optics does not necessarily
apply. Instead we used simulations of wave front propagation based on
the Helmholtz equation (see Appendix A).
Due to the computational complexity of making full 3D wave simulations, we have restricted our analyses to 2D simulations. The result from
these simulations proved sufficient to obtain a general understanding of
the optical properties of the vitreous cells. The 2D waveform propagation
was simulated directly on five selected TEM-sections (three transverse and
two longitudinal). The refractive indices of the vitreous vesicles were deter-
2. Materials and methods
Adult male and female T. cystophora were hand collected in the mangrove swamps at La Parguera, Puerto Rico. The rhopalia were removed
with a pair of scissors and fixed in 2% glutaraldehyde, 2.5% paraformaldehyde, and 3% sucrose in 0.15 M sodium cacodylate buffer. After 2–3 days
in the fixative the rhopalia were washed in buffer, dehydrated in a series of
ethanol and acetone, and embedded in Epon 812 resin. The material used
for TEM was post-fixed in 1% osmium tetraoxide in 0.15 M buffer for 1 h
at room temperature prior to dehydration. Fifty-nanometer sections were
cut with a diamond knife on a Leica ultra microtome. The TEM-sections
were mounted on single slot grids, contrasted with uranyl acetate (20 min
at room temperature), and lead citrate (3 min at 5 °C), and observed in a
JEOL 1240 microscope. The material for light microscopy was sectioned
in 1-lm sections on a glass knife and stained with methylene blue. All pictures were generated and manipulated digitally.
2.1. Determination of refractive index in the slit eye
The structure filling the slit eye is transparent but has little refractive
power (see Section 3). We will therefore not use the term lens, but rather
refer to these specialized cells as vitreous cells. Due to the small size
(90 25 20 lm), and soft nature of the group of vitreous cells, it was
not possible to remove them intact for interference light microscopy. An
Fig. 1. Calibration curve between refractive index and relative electron
density in TEM-sections of the lower lens eye. The good fit between
measured refractive index and electron density is used as an indirect
method to obtain data on the refractive index of the slit eye lens (see
Section 2 for details). Broken line and open circles show the measured
refractive indices through a radius on the lower lens (see Nilsson et al.,
2005, for details). Solid line and filled diamonds show the corresponding
relative electron density of the lens material following a radius. Error bars
indicate SEM, n = 3.
A. Garm et al. / Vision Research 48 (2008) 1061–1073
mined as described above. On each section the modeling was done with
wave fronts coming at nine different angles and with three different wavelengths (400, 500, and 600 nm). The refractive index of ‘‘normal” cytoplasm and photoreceptor microvilli was set to 1.33 and 1.37,
respectively (Nilsson, Land, & Howard, 1988).
Since modeling the wave front propagation found that the vitreous
cells have only weak optical power, the receptive fields were assessed using
geometrical optics and normal ray tracing, ignoring refraction in the vitreous cells. This was done for both the pit and the slit eyes in a custom made
program for MATLAB 7.0.1 (The Mathworks, Inc., Natick, MA).
3. Results
3.1. The arrangement of the eyes
The rhopalia containing the eyes are situated in small
groves, the rhopalial niches, midway between the pedalia
holding the tentacles. They are attached to the bell by a
stalk approximately half the length of the rhopalia. A full
sized rhopalium of T. cystophora is approximately
500 lm long. The two lens eyes are situated along the midline of the rhopalium with the upper lens eye pointing
upwards (Fig. 2). The lower lens eye sits centrally on the
rhopalium and points obliquely downwards (Fig. 2). The
slit eyes are situated laterally between the upper lens eye
and the lower lens eye with their openings pointing fronto-laterally (Fig. 2). The pit eyes sit laterally to the upper
lens eye, and they point upwards and slightly laterally
(Fig. 2). The slit and pit eyes are situated closely together
on the rhopalium, and internally they are separated/connected by a thin layer of neuropil only (not shown).
3.2. Morphology of the slit eye
The opening of the slit eye of T. cystophora is occupied
by a group of vitreous cells forming a very complex structure. Directly underneath these cells is a keel-shaped retina
holding two cell types: pigmented photoreceptors and nonpigmented photoreceptors. The slit eye also contains a
fourth cell type, which are non-sensory pigment cells. A
schematic drawing can be seen in Fig. 3. The eye is covered
by a mono-layered epithelium. The pigment cup of the slit
1063
eye of an adult T. cystophora is 90–110 lm long, has an
external maximum depth of 50–55 lm, and a maximum
width of 30–35 lm (Fig. 4A and B). The slit-shaped opening is 80–90 lm long.
The lower and medial part of the pigmented area is composed of about 200 pigmented photoreceptors (Figs. 3 and
4B). They are of the ciliated type but their outer segments
are very small, only 2–5 lm long and <5 lm3. The pigmented part lying just proximal to the outer segment is
5–10 lm long and filled with spherical to elliptic pigment
granules having diameters between 0.3 and 2 lm. These
photoreceptors have their nuclei arranged in one or two
layers with the innermost layer lying just proximal to the
pigment area (Figs. 3 and 4A, B). When present, the cells
with their nuclei in the 2nd layer have a thin 3- to 4-lmlong region with few organelles between the nuclei and
the pigment area. The nuclei are surrounded by numerous
mitochondria. The photoreceptors have an irregular
shaped ending a little proximal to the nuclei, and there
are no signs of them having any axons. The receptors seem
to be in synaptic contact with the rest of the nervous system
at their cell bodies, and this unusual system is under current investigation. In total the pigmented photoreceptors
are 20–25 lm long.
On the lateral side of the eye a group of 30–40 non-pigmented photoreceptors are situated (Fig. 5A and B). Their
outer segments retain the 9 2 + 2 cilia configuration for a
long distance (30–60 lm), and are sent in between the pigmented photoreceptors in the middle part of the eye. When
reaching the retina they also dissolve in irregular microvilli.
The outer segments of both receptor types are arranged
in layers when the eye is seen in transverse section (Figs. 3,
4D, and 6C–G). In the widest part the outer segments form
7–8 layers, each layer having a thickness varying between
0.5 and 1.5 lm (Fig. 4C and D). Together the layers form
a wedge-shaped retina in cross-section, 2–5 lm wide and
7-lm deep in the medial half of the eye (Fig. 4A and D).
In the lateral half it gets progressively smaller and stops
approximately 10 lm from the edge of the eye (Fig. 4A).
Non-sensory pigment cells form the upper and lateral
part of the pigment slit. Their pigmented region is about
Fig. 2. The visual system of the box jellyfish Tripedalia cystophora (A) is arranged on four identical sensory clubs, the rhopalia, having a similar set of six
eyes each. (B) and (C) show schematic drawings of the rhopalium of T. cystophora seen frontally (B) and laterally (C). The two lens eyes lie on the midline
with the upper lens eye pointing upwards and the lower lens eye pointing obliquely downwards. The pair of slit eyes is situated laterally between the lens
eyes. They point latero-frontally and slightly downwards. The pit eyes are found on either side of the upper lens eye and they point upwards and slightly
laterally.
1064
A. Garm et al. / Vision Research 48 (2008) 1061–1073
Fig. 3. Schematic drawing of a cross-section of the slit eye. The eye is
strongly asymmetric with the lower part of the pigment cup being very
short. This part is made of the pigmented photoreceptors. The upper part
of the pigment cup is made of non-sensory pigment cells. The area in front
of the outer segments (OS) is filled by a lens-like structure made of vitreous
cells. The slit eye also holds a fourth cell type, non-pigmented photoreceptors, which are not shown since they appear in another region of the
eye. The drawing is not to scale. CR, ciliary rootlet; Nu, nucleus; PG,
pigment granules; VV, vitreous vesicles.
10-lm long, 2–3 lm wide, and made of rather regular
spherical granules 0.5–1 lm in diameter (Fig. 4B). The
nuclei are arranged in several layers (Fig. 4B). Few mitochondria are found surrounding the nuclei, and the cells
end just proximal to the nuclei. There is a total of approximately 200 non-sensory pigment cells.
The vitreous cells fill the entire slit-shaped opening and
together they form a structure which is canoe-shaped in
longitudinal section (Figs. 4A and 6). In transverse section
it is highly asymmetric since it expands downwards (Figs.
3, 4B, and 6). There are about 40 vitreous cells, and they
are elongate with their long axis perpendicular to the long
axis of the eye (Figs. 4B and 5C). They are 15–25 lm long
and 5–7 lm wide, and most of their cytoplasm is filled with
closely packed vesicles (Figs. 3 and 5C, D), presumably
containing the same kind of crystallin as found in the lenses
of the lens eyes (Piatigorsky & Kozmic, 2004; Piatigorsky
et al., 1989). These vesicles are 0.3–1.5 lm in diameter. It
is interesting to note that the vesicles are arranged in
rounded clusters within the cells giving the entire light
gathering surface a lobed appearance (Figs. 3–6). The
refractive indices of the vesicles calculated from TEM-sections vary between 1.38 and 1.48. The distribution of
refractive index showed no clear pattern except for the
upper part of the vitreous cells being rather homogeneous
(Figs. 4A, B and 6), while the lower part seems to have a
more random refractive index distribution (Figs. 4B and
5D). Several of the vitreous cells have their nuclei situated
in between the vitreous vesicles (Figs. 4B and 5C).
3.3. Receptive fields of the slit eye
Modeling the wave front propagation in the slit eye
showed that the optical effects of the vitreous cells are negligible, and their overall effect is a rather weak scattering of
the incoming light. The maximum intensity changes at the
retina are in the range of 20% of the incoming intensity
(Fig. 7). All the examined sections, angles, and wavelengths
had similar effects. The small effect present comes almost
entirely from the lobed surface. We therefore proceeded
with standard geometrical optics and ray tracing without
the vitreous cells, to approximate the receptive fields of
selected photoreceptors and selected areas of the retina
(Figs. 8 and 9). The ray tracings were performed on 3D
reconstructions of the pigment screens made from TEM
(n = 5) and LM (n = 3) sections of the slit eyes.
The receptive fields were modeled for nine separate photoreceptors at different locations in the retina (Fig. 8). The
receptors were from three different positions along the long
axis of the retina (lateral, central, and medial, see Fig. 6G),
and from three different levels (distances from the vitreous
cells) in the retina at each position (distal, mid, and proximal, see Fig. 6C for arrangement). All the receptive fields
displayed the same asymmetry along the transverse axis,
with a steep sensitivity gradient in the upper part and relatively slow changes in the lower part (Figs. 8 and 9). Along
the longitudinal axis of the eye all of the nine modeled photoreceptors had very similar receptive fields with maximum
half-widths between 135° and 170°. Along the transverse
axis of the slit eye the receptive fields fell into three groups
according to their position. The proximal receptors had a
maximum transverse half-width of 18–22°, the mid-receptors 34–46°, and the distal receptors 68–76°. Interestingly,
they all had the same upper limit to their receptive fields
(Fig. 8). Due to the similarities found for receptors from
the same level in the retina, we also modeled the combined
receptive field for all the proximal, all the mid, and all the
distal receptors (Fig. 9A–C). The combined receptive fields
have half-widths very similar to the individual receptors
from the same level. The receptive field of the entire retina
was also modeled, and had a 50% curve very similar to
those of the combined mid-receptors, but with a more pronounced asymmetry (Fig. 9B and D).
3.4. The pit eyes
The pit eyes are the smallest eyes. The pigmented area is
a more or less spherical cup with an outer diameter of 25–
30 lm, and an inner diameter of about 12 lm (Fig. 10A).
There are no signs of any lens-like structure in the pit eyes.
Also, in their cellular components, they are less complex
than the slit eyes, since they are composed of pigmented
photoreceptors only. Morphologically, they are very similar to the pigmented photoreceptors found in the slit eye
retina, and there are about 300 of them in each pit eye.
The individual size and arrangement of the outer segments
could not be worked out from the TEM-micrographs
(Fig. 10A–C), but the total volume of the retina implies
that the outer segments are very small with a mean volume
of about 3 lm3.
Since no data were obtained on the arrangement of the
outer segments, receptive fields of individual receptors cannot be modeled. Instead, we treated the entire mass of
A. Garm et al. / Vision Research 48 (2008) 1061–1073
1065
Fig. 4. TEM-micrographs of the slit eye. In sections along the long axis the slit is filled with a boat-shaped lens-like structure (A, white outline) made of
vitreous cells (VC), which lie directly on top of a keel-shaped retina formed by the outer segments (OS) of the photoreceptors (PPR). In transverse section
the slit eye is highly asymmetric (B). The upper part is made by non-sensory pigment cells (NSPC) covering the lens. The lower and lateral part is made by
PPRs which do not cover the VCs (B). The crystallin containing vesicles are gathered in spheres giving the light collecting surfaces of the VCs a lobed
appearance (A and B, white outline). The PPRs are ciliary (C and D, arrowheads) and very small, the distal part of the inner segment is only about 1.5 lm
in diameter (C). The cilia dissolve in irregular membrane loops and together they form a keel-shaped retina (B and D). The dotted line in (A) indicates
approximate location of the section in (B) and the dotted line in (B) indicates approximate location of the section in (A). BB, basal body; M,
mitochondrion; Np, neuropil; Nu, nucleus; PG, pigment granules.
outer segments as one unit and performed ray tracing to
map the receptive field of the entire retina of the pit eye
(Fig. 10D and E). The result was a close to circular receptive field with a half-width of approximately 60°.
4. Discussion
The visual system of box jellyfish is highly complex, and
it comprises 24 eyes of four morphological types. An obvious assumption is that the eyes serve different functions,
and that each eye type is designed specifically to extract
the visual information needed for its particular function
(for a general discussion on special vs. general purpose eyes
see Land & Nilsson, 2006). Our previous work on the lens
eyes of T. cystophora supports this assumption, since it
demonstrates the presence of highly selective peripheral
information filtering (Coates et al., 2006; Garm et al.,
2007; Nilsson et al., 2005). Our present results are also well
in line with the assumption of special purpose eyes, and we
are now in a position to address what type of filtering takes
place in the lesser eyes.
Well-developed and seemingly advanced eyes have been
the subjects of detailed optical investigations, and there is a
wealth of such studies of eyes throughout the animal kingdom (e.g. Land & Nilsson, 2002). Simple pigment pit eyes
are typically much less studied in terms of their detailed
1066
A. Garm et al. / Vision Research 48 (2008) 1061–1073
Fig. 5. TEM-micrographs of the slit eye. About 40 non-pigmented photoreceptors (NPPR) are found in the lateral part of the slit eye (A). Their outer
segments remain normal cilia for a long distance (B) but dissolve in irregular microvilli when they reach the retina (not shown). From the electron density it
is indicated that the crystalline content of the vitreous vesicles varies with the location within the eye (A and C). In a longitudinal section of the upper part
of the vitreous cell population the vesicles are rather homogeneously stained (A, see also Fig. 4A and B) whereas in the lower part a more or less random
arrangement of the electron density is seen (C). In the lower part of the eye close to the full range of refractive indices can be seen within the same vitreous
cell (D, numbers indicate refractive indices). Ci, cilium; EC, epithelial cell; Np, neuropil; Nu, nucleus; PPR, pigmented photoreceptors; VC, vitreous cells.
optical function and performance. Here we present one of
the first detailed optical studies of such small eyes, resulting
in information important when assessing their biological
role. We also establish some methodological principles specific to investigations of seemingly simple eyes. Because of
their small dimensions, a wave optics approach initially
seemed necessary. We have demonstrated that simplifying
the wave-optics modeling to 2D provides a good overall
insight into the optical properties of an eye. It was found
that the vitreous cells of the slit eye have minimal optical
effects, and thus it is reasonable to use geometrical optics
for 3D models of the receptive fields. In general the errors
induced by using this method on small structures (e.g. not
taking diffraction and interference into account) have little
effect on the broad receptive fields found in eyes without
image forming optics. Thus, we conclude that geometrical
optics would normally be sufficient for optical modeling
of small and relatively simple eyes.
4.1. Function of the vitreous cells
The slit eyes of T. cystophora contain transparent cells
covering the entire opening through which light can reach
the retina. Similar structures have been reported from the
slit eyes of another box jellyfish, Carybdea rastonii (Satterlie, 2002). These cells are visible as glittering structures in
A. Garm et al. / Vision Research 48 (2008) 1061–1073
1067
Fig. 6. Schematic drawings of the slit eye. The vitreous cell population in the slit eyes has a surface made of oval lobes (A). The lower part is also lobed but
here the lobes are hemispheres (B). In cross-section the eye is highly asymmetric with the pigment screen covering only the upper and medial part of the
eyes (C–F, thick black line). The lower part of the vitreous cell population has a variable refractive index (C-F, dotted pattern) whereas the upper part has
a more homogeneous refractive index (C–F, gray area). In longitudinal section it is seen that the retina is also asymmetric and becomes progressively
smaller towards the lateral part of the eye (G). It stops about 10 lm from the edge of the vitreous cells. The dotted lines indicate areas of cross-sections in
C–F. The outer segments of the photoreceptors are modeled as boxes lying in layers perpendicular to the long axis of the eye (C–G).
live eyes, and we initially assumed that they might have
lens-like refracting properties. Our wave optics analysis
proved this assumption wrong. The refracting power of
the cell mass is minute, and it only causes a slight scattering
of the incoming light. We consequently refer to these cells
as vitreous cells. Most of their weak optical effect comes
from the lobed surface, but the internal distribution of
refractive indices has almost no effect. It should be remembered that we have only measured the refractive indices
indirectly. However, due to the very short distance between
the vitreous cells and the receptors, the refractive indices
would have to be much higher than our estimates in order
to have a significant effect on the directional propagation of
light.
The arrangement of the vitreous cell population with its
asymmetry, the lobed surface, and the small vitreous vesicles gathered in larger spheres, is peculiar and unique,
and the function of these cells is not obvious. A hint on
the function may come from the fact that all light reaching
the retina must first pass through the vitreous cells. With
refraction ruled out, we are left with polarizing properties
or spectral filtering. There is nothing in the way of ordered
ultrastructure in the vitreous cells that would indicate any
polarizing properties, and the retina is not arranged with
the aligned photoreceptor membranes typical in polarization vision (Labhart & Meyer, 2002; Nilsson & Warrant,
1999). It is more likely that the colorless vitreous cells serve
to remove UV light. There are strong indications that the
1068
A. Garm et al. / Vision Research 48 (2008) 1061–1073
Fig. 7. Models of wave front propagation through the vitreous cells of the slit eye. In both the cross-sections and longitudinal sections the model predicts
that more or less the entire optical effect originates from their lobed surface (A and D). The red color indicates light intensities higher than the original
wave and green indicate lower intensities. White arrow indicates the direction of the incoming light. The effects are more easily seen, when the structure is
removed from the picture (B and E). In the gray scale white indicates a doubling of the intensity and black indicates no light. The modification of the wave
front caused by the vitreous cells and the retina at the level of the broken red line is shown in (C) and (F). It is seen that in both cases the effect lies within
±20% of the incoming intensity.
visual system of box jellyfish uses opsin-based photopigments (Coates et al., 2006; Garm et al., 2007), and most
opsin-based photopigments have a secondary absorption
peak (b-peak) in the UV region (Govardovskii, Fyhrquist,
Reuter, Kuzmin, & Donner, 2000). If the receptor is meant
to pick up information in a narrow spectral band provided
by the primary absorption peak (a-peak), this UV absorption may need to be removed. Indeed, the spectral sensitivity of the lens eyes suggests that UV filters are present here
(Coates et al., 2006; Garm et al., 2007). Electrophysiologi-
cal recordings from the slit eyes are needed to test the
hypothesis that the vitreous cells are UV filters in the slit
eye.
4.2. Receptive fields and spatial resolution of the slit and pit
eyes
An essential question in assessing the role of slit and pit
eyes is whether there is any form of spatial resolution in
these eyes. Even though the pit and slit eyes lack image
A. Garm et al. / Vision Research 48 (2008) 1061–1073
1069
Fig. 8. Receptive fields of slit eye receptors. The receptive fields from nine individual receptors were modeled using ray tracing. The receptors represent
three areas of the retina (medial, central, and lateral) and three different levels in each area (proximal, mid, and distal, see Fig. 6C for arrangement). Dark
red indicates the directions from where the receptor receives the most light. Dark blue indicates directions from where no light reaches the receptor and the
broken black line indicates the 50% curve. Each square represents 22.5° 22.5°. It is seen that the receptive fields are similar within the same retinal level
irrespective of the area. In the horizontal plane all the half-widths of the receptive fields are broad, 135–170°. In the vertical plane the different retinal layers
vary. The distal receptors have broad receptive fields, 70°, whereas the proximal has more narrow receptive fields, 20°. Note that all the receptors have
approximately the same upper limit to their receptive fields and are asymmetric in the transverse plane.
Fig. 9. Receptive fields of large areas of the slit eye retina. Dark red indicates the directions from where the receptor receives the maximum light. Dark
blue indicates directions from where no light reaches the receptor and the broken black line indicates the 50% curve. Each square represents 22.5° 22.5°.
When performing summation of all the receptive fields in the same retinal layer similar results as for the individual receptors are seen (A–C, compare with
Fig. 8). When the whole retina is summed the 50% curve is very similar to that of the mid-row of receptors but the asymmetry is more pronounced (B and
D).
1070
A. Garm et al. / Vision Research 48 (2008) 1061–1073
Fig. 10. TEM-micrographs and modeling of the pit eyes. The pit eyes are close to spherical when cut through the midline (A). The pit is completely filled
with the outer segments of the photoreceptors, which are cilia (B, arrowhead) dissolving into a dense mass of microvilli (Mv) (B and C). All the
photoreceptors of the pit eyes are pigmented (PPR). The opening of the pit eye points upwards and slightly laterally (A and D). In the geometrical model
(D) the outer segments are pooled since no data are available on their orientation. Ray tracing on the entire retina returns a circular receptive field with a
half-width of about 60° (E). Dark red indicates the directions from where the receptor receives the most light. Dark blue indicates the directions from
where no light reaches the receptor and the broken black line indicates the 50% curve. Each square represents 22.5° 22.5°. BB, basal body; CR, ciliary
rootlet; OS, outer segments.
A. Garm et al. / Vision Research 48 (2008) 1061–1073
forming lenses, they may still have spatial resolution based
on differential shading by the pigment screen. We were
unable to determine the size and position of individual
outer segments in the pit eye, and consequently we cannot
say if there is a potential for spatial resolution in this eye.
Still, the morphology of the pit eye with an opening of
more than half the diameter of the retina would allow only
very crude spatial resolution, if any at all, and suggests a
function as a mere light meter.
The receptive fields from the receptors in the slit eye do
indicate a potential for spatial resolution but in an unconventional form. All of the modeled receptors have almost
identical half-widths along the longitudinal axis of the
eye, thus no spatial resolution can be present in this plane.
Along the transverse axis of the eye the receptors from different levels (depths) in the retina have receptive fields with
clearly different half-widths (Fig. 8). All the receptors have
receptive fields with about the same upward limit, but the
more distally they are located in the retina the further their
receptive fields reach downwards. This could provide spatial resolution along the transverse axis of the eye of about
10–15°. Another interesting aspect of the receptive fields of
the slit eye receptors is the strong asymmetry along the
transverse axis. In the upward edge the sensitivity gradient
is steep, but downwards the sensitivity drops slowly over a
large angle. It is noteworthy that similar asymmetries were
also found for the receptive fields of the receptors in the
lens eyes of T. cystophora (Nilsson et al., 2005). Taken
together, the arrangements of receptive fields in the slit
eye suggest that it is optimized to detect vertical movement,
or to make ‘‘centre-surround” comparisons exclusively
along the transverse axis of the eye.
4.3. Morphological complexity
We have seen that the optical models of the pit and slit
eyes suggest different functional complexity. This difference
is also reflected in the morphological complexity of the two
eye types. The pit eye, which probably has little or no spatial resolution, is formed exclusively by pigmented photoreceptors. This is similar to what has been found for other
cnidarian pigment cup eyes (Blumer, v Salvini-Plawen,
Kikinger, & Büchinger, 1995; Singla, 1974; Singla &
Weber, 1982a, 1982b; Toh, Yoshida, & Tateda, 1979;
Yamamoto & Yoshida, 1980). We found the slit eye to
have a much more complex morphology, even more complex than that suggested by previous work on these eyes
(Laska & Hündgen, 1982). It is composed of four cell types
including two different types of photoreceptors, which
potentially allow for extraction of specific spatial, temporal, or spectral information.
One intriguing aspect shared between the slit and pit
eyes is that they both have compact retinas formed by large
numbers of receptor cells (250 and 300 receptors, respectively). Consequently the receptive outer segments of each
receptor cell must be unusually small. Many small and closely packed photoreceptors normally indicate high spatial
1071
resolution but as we discussed above this cannot be the case
for the slit and pit eyes of T. cystophora. Even though no
direct counts have been made, similarly sized eyes in hydrozoans also seem to possess a relatively high number of
small photoreceptors (Singla, 1974; Singla & Weber,
1982a, 1982b), and in other parts of cnidarian nervous systems many small cells collectively form common information channels (Garm, Poussart, Parkefelt, & Nilsson,
2007; Grimmelikhuijzen & Westfall, 1995; Mackie, 2004).
If there is a specific tendency for parallel redundancy in
the organization of cnidarian neural and sensory systems,
it may very well have a developmental rather than direct
functional explanation.
The extremely small size of the outer segments means
that the photoreceptors will have very low sensitivity and
suffer relatively more from intrinsic noise and photon
noise. The obvious solution to these problems would be
to sum the signals from many receptors onto a smaller
number of neurons (Warrant, 2004). This would increase
the sensitivity and improve the signal-to-noise ratio. The
fact that many receptors share the same receptive field in
both examined eyes suggests that such pooling is taking
place in the nervous system.
4.4. Multiple eye types
With the new data presented here it becomes relevant to
speculate about the functional reason of having slit and pit
eyes in addition to the much larger lens eyes. Even though
the lens eyes are known to be significantly out of focus,
they provide better spatial information, and pick up much
more light, than the slit and pit eyes. It would seem that
any visual information is better received by the lens eyes.
An explanation for the need of several eyes is emerging
from recent studies of the lens eyes (Garm et al., 2007; Nilsson et al., 2005), where severe spatial and temporal filtering
occurs already at the level of optics and receptor cells. This
extremely peripheral filtering of information is likely to
represent efficient solutions to the tasks served by the lens
eyes, but it also means that much visual information is irretrievably lost at an early stage. If the slit and pit eyes have
receptors with different temporal properties, or different
opsins, they could serve visual tasks that the lens eyes cannot. Further work on visually guided behaviors in box jellyfish is likely to provide input to the question of the
different roles of the four eye types.
Acknowledgments
We greatly appreciate the skilled lab-work performed by
Rita Wallén and all the help received at the field station in
La Paguera. Comments on the manuscript offered by Prof.
Eric Warrant are much acknowledged. Grant No. 6212002-4873 from the Swedish Research Council (to D.-E.
N.) and Grant No. 2005-1-74 from the Carlsberg Foundation (to A.G.) are likewise acknowledged.
1072
A. Garm et al. / Vision Research 48 (2008) 1061–1073
Appendix A
Wave front propagation is simulated based on the
Helmholtz equation,
4p2
nðxÞuðxÞ ¼ 0; x 2 R2
ðA:1Þ
k2
where x denotes the position, n(x) the refractive index,
u(x) the total field, and k denotes a wave number. We consider solutions
DuðxÞ þ
u¼/þw
ðA:2Þ
to (A.1), where /(x) denotes the incoming electromagnetic
field, w(x) the scattered electromagnetic field and where /
satisfies the free space Helmholtz equation,
D/ðxÞ þ
4p2
n0 /ðxÞ ¼ 0;
k2
where n0 is the refractive index of surrounding water (1.33)
and where w solves,
4p2
nðxÞ
4p2
nðxÞ
DwðxÞ þ 2 n0
1 wðxÞ ¼ 2 n0
1 /ðxÞ;
n0
n0
k
k
subject to the Sommerfeld’s radiation condition:
pffiffi @w
2p pffiffiffiffiffi
i
lim r
n0 w ¼ 0; r ¼ jxj:
r!1
@r
k
For decomposition of the form (A.2), we will refer to u as
the total wave, to u as the incident wave, and to w as the
scattered wave. We then reformulate (1) as the Lippmann–Schwinger equation,
Z
Z
wðxÞ þ Gðy xÞnðyÞwðyÞ dy ¼ Gðy xÞnðyÞ/ðyÞ dy;
where
G¼
p2 i
2p pffiffiffiffiffi
n
jxj
;
H
0
0
k
k2
and where H0 denotes the zero order Hankel function of
the first kind. We solve the (non-approximated) stationary
wave propagation problem by solving the Lippmann–
Schwinger integral equation subject to Sommerfeld’s radiation condition. For more details about the numerical
implementation used, cf. Andersson & Holst (2005).
References
Andersson, F., & Holst, A. (2005). A fast, bandlimited solver for
scattering problems in inhomogeneous media. Journal of Fourier
Analysis and Applications, 11(4), 471–487.
Berger, E. W. (1898). The histological structure of the eyes of cubomedusae. Journal of Comparative Neurology, 8(3), 223–230.
Berger, E. W. (1900). The cubomedusae. Memoirs from the Biological
Laboratory of the Johns Hopkins University, 4, 1–83.
Blumer, M. J. F., Salvini-Plawen, L., Kikinger, R., & Büchinger, T.
(1995). Ocelli in a cnidaria polyp: the ultrastructure of the pigment
spots in Stylocoronella riedli (Scyphozoa, Stauromedusae). Zoomorphology, 115, 221–227.
Buskey, E. J. (2003). Behavioral adaptations of the cubozoan medusa
Tripedalia cystophora for feeding on copepod (Dioithona oculata)
swarms. Marine Biology, 142, 225–232.
Claus, C. (1878). Ueber Charybdea marsupialis. Arbeiten aus dem
zoologischen Institut Universität Wien, 1(2), 1–56.
Coates, M. M., Garm, A., Theobald, J. C., Thompson, S. H., & Nilsson,
D. E. (2006). The spectral sensitivity in the lens eyes of a box jellyfish,
Tripedalia cystophora. Journal of Experimental Biology, 209,
3758–3765.
Garm, A., Coates, M. M., Seymour, J., Gad, R., & Nilsson, D. E. (2007).
The lens eyes of the box jellyfish Tripedalia cystophora and Chiropsalmus sp. are slow and color-blind. Journal of Comparative Physiology A,
193, 547–557.
Garm, A., O’Connor, M., Parkefelt, L., & Nilsson, D. E. (2007). Visually
guided obstacle avoidance in the box jellyfish Tripedalia cystophora
and Chiropsella bronzie. Journal of Experimental Biology, 210,
3616–3623.
Garm, A., Poussart, Y., Parkefelt, L., & Nilsson, D. E. (2007). The ring
nerve of the box jellyfish Tripedalia cystophora. Cell and Tissue
Research, 329, 147–157.
Govardovskii, V. I., Fyhrquist, N., Reuter, T., Kuzmin, D. G., & Donner,
K. (2000). In search of the visual pigment template. Visual Neuroscience, 17, 509–528.
Grimmelikhuijzen, C. J. P., & Westfall, J. A. (1995). The nervous system
of cnidarians. In O. Breidbach & W. Kutsch (Eds.), The nervous
systems of invertebrates: An evolutionary approach (pp. 7–24). Basel:
Birkhäuser Verlag.
Hamner, W. M., Jones, M. S., & Hamner, P. P. (1995). Swimming,
feeding, circulation and vision in the Australian box jellyfish, Chironex
fleckeri (Cnidaria, Cubozoa). Marine and Freshwater Research, 46,
985–990.
Hartwick, R. F. (1991). Observations on the anatomy, behaviour,
reproduction and life cycle of the cubozoan Carybdea sivickisi.
Hydrobiologia, 216/217, 171–179.
Hertwig, O., & Hertwig, R. (1878). Das Nervensystem und die Sinnesorgane
der Medusen. Leipzig: Vogel.
Kinsey, B. (1986). Barnes on box jellyfish. Townsville: James Cook
University.
Kozmic, Z., Daube, M., Frei, E., Norman, B., Kos, L., Dishaw, J. J., et al.
(2004). Role of pax genes in eye evolution: A cnidarian PaxB uniting
Pax2 and Pax6 functions. Developmental Cell, 5, 773–785.
Labhart, T., & Meyer, E. P. (2002). Neural mechanisms in insect
navigation: Polarization compass and odometer. Current Opinion in
Neurobiology, 12, 707–714.
Land, M. F., & Nilsson, D. E. (2002). Animal eyes (Willmer, P., &
Norman, D. (Eds.)) (pp. 1–221). Oxford: Oxford University Press.
Land, M. F., & Nilsson, D. E. (2006). General-purpose and specialpurpose visual systems. In E. J. Warrant & D. E. Nilsson (Eds.),
Invertebrate vision (pp. 167–210). Cambridge: Cambridge University
Press.
Laska, G., & Hündgen, M. (1982). Morphologie und Ultrastruktur der
Lichtsinnesorgane von Tripedalia cystophora Conant (Cnidaria, Cubozoa). Zoologischer Jahrbuch der Anatomie, 108, 107–123.
Laska-Mehnert, G. (1985). Cytologishe Veränderungen während der
Metamorphose des Cubopolypen Tripedalia cystophora (Cubozoa,
Carybdeidae) in die Medusa. Helgoländer Wissenschaftliche Meeresuntersuchungen, 39, 129–164.
Mackie, G. O. (2004). Central neural circuitry in the jellyfish Aglantha: A
model ‘‘simple nervous system”. Neuro-Signals, 13, 5–19.
Martin, V. J. (2002). Photoreceptors of cnidarians. Canadian Journal of
Zoology, 80, 1703–1722.
Matsumoto, G. I. (1995). Observations on the anatomy and behaviour of
the cubozoan Carybdea rastonii Haacke. Marine and Freshwater
Behaviour and Physiology, 26, 139–148.
Nilsson, D. E., Coates, M. M., Gislén, l., Skogh, C., & Garm, A. (2005).
Advanced optics in a jellyfish eye. Nature, 435, 201–205.
Nilsson, D. E., Land, M. F., & Howard, J. (1988). Optics of the butterfly
eye. Journal of Comparative Physiology A, 162, 341–366.
A. Garm et al. / Vision Research 48 (2008) 1061–1073
Nilsson, D. E., & Warrant, E. J. (1999). Visual discrimination: Seeing the
third quality of light. Current Biology, 9, 535–537.
Pearse, J. S., & Pearse, V. B. (1978). Vision in cubomedusan jellyfish.
Science, 199, 458.
Piatigorsky, J., Horwitz, J., Kuwabara, T., & Cutress, C. E. (1989). The
cellular eye lens and crystallins of cubomedusan jellyfish. Journal of
Comparative Physiology A, 164, 577–587.
Piatigorsky, J., & Kozmic, Z. (2004). Cubozaon jellyfish: An Evo/Devo
model for eyes and other sensory systems. International Journal of
Developmental Biology, 48, 719–729.
Romanes, G. J. (1876). Preliminary observations on the locomotor system
of medusae. Philosophical Transactions of the Royal Society London,
66, 269–313.
Satterlie, R. A. (2002). Neural control of swimming in jellyfish: A
comparative story. Canadian Journal of Zoology, 80, 1654–1669.
Schewiakoff, W. (1889). Beiträge zur kenntnis des Acalephenauges.
Morphologisches Jahrbuch, 15(1), 21–60.
Singla, C. L. (1974). Ocelli of hydromedusae. Cell and Tissue Research,
149, 413–429.
1073
Singla, C. L., & Weber, C. (1982a). Fine structure of the ocelli of
Polyorchis penicillatus (Hydrozoa: Anthomedusae) and their connection with the nerve ring. Zoomorphology, 99, 117–129.
Singla, C. L., & Weber, C. (1982b). Fine structure of the ocellus of Sarsia
tubulosa (Hydrozoa, Anthomedusae). Zoomorphology, 100, 11–22.
Stewart, S. E. (1996). Field behavior of Tripedalia cystophora (class
Cubozoa). Marine and Freshwater Behaviour and Physiology, 27(2-3),
175–188.
Toh, Y., Yoshida, M., & Tateda, H. (1979). Fine structure of the ocellus of
the hydromedusan, Spirocodon saltatrix. I. Receptor cells. Journal of
Ultrastructure Research, 68, 341–352.
Warrant, E. J. (2004). Vision in the dimmest habitats on Earth. Journal of
Comparative Physiology A, 190, 765–789.
Yamamoto, M., & Yoshida, M. (1980). Fine structure of ocelli of an
anthomedusan, Nemiopsis dofleini, with special reference to synaptic
organization. Zoomorphology, 96, 169–181.
Yamasu, T., & Yoshida, M. (1976). Fine structure of complex ocelli of a
cubomedusan, Tamoya bursaria Haeckel. Cell and Tissue Research,
170, 325–339.