LUND UNIVERSITY

PO Box 117221 00 Lund+46 46-222 00 00

A simple visual system without neurons in jellyfish larvae

Nordstroumlm Karin Walleacuten Rita Seymour Jamie Nilsson Dan-E

Published inRoyal Society of London Proceedings B Biological Sciences

DOI101098rspb20032504

2003

Link to publication

Citation for published version (APA)Nordstroumlm K Walleacuten R Seymour J amp Nilsson D-E (2003) A simple visual system without neurons injellyfish larvae Royal Society of London Proceedings B Biological Sciences 270(1531) 2349-2354httpsdoiorg101098rspb20032504

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authorsandor other copyright owners and it is a condition of accessing publications that users recognise and abide by thelegal requirements associated with these rights

bull Users may download and print one copy of any publication from the public portal for the purpose of private studyor research bull You may not further distribute the material or use it for any profit-making activity or commercial gain bull You may freely distribute the URL identifying the publication in the public portalTake down policyIf you believe that this document breaches copyright please contact us providing details and we will removeaccess to the work immediately and investigate your claim

Received 25 April 2003Accepted 30 June 2003

Published online 18 September 2003

A simple visual system without neurons in jellyfishlarvaeKarin Nordstrom1 Rita Wallen1 Jamie Seymour2 and Dan Nilsson1

1Department of Cell and Organism Biology Zoology Lund University Helgonav 3 223 62 Lund Sweden2Department of Zoology and Tropical Ecology James Cook University McGregor Road Smithfield Cairns QLD 4878Australia

Earlier detailed studies of cnidarian planula larvae have revealed a simple nervous system but no eyes oridentifiable light sensing structures Here we describe the planula of a box jellyfish Tripedalia cystophoraand report that these larvae have an extremely simple organization with no nervous system at all Theironly advanced feature is the presence of 10ndash15 pigment-cup ocelli evenly spaced across the posterior halfof the larval ectoderm The ocelli are single cell structures containing a cup of screening pigment filledwith presumably photosensory microvilli These rhabdomeric photoreceptors have no neural connectionsto any other cells but each has a well-developed motor-cilium appearing to be the only means by whichlight can control the behaviour of the larva The ocelli are thus self-contained sensoryndashmotor entitiesmaking a nervous system superfluous

Keywords rhabdomeric photoreceptor planula pigment-cup ocelli Tripedalia cystophoraChiropsalmus sp Chironex fleckerii

1 INTRODUCTION

Cnidarians are simple radially symmetrical invertebratesmost classes of which have a metamorphic life cycleincluding a swimming larva a stationary polyp and a free-swimming medusa (Muller amp Leitz 2002) Cnidarian lar-vae are of the planula type The early development hasbeen described morphologically and molecularly in greatdetail (Okada 1927 Hayward et al 2001 2002) Afterfertilization of the egg and several cell divisions a flattenedblastula is formed resembling a shallow bowl The upperand lower ectoderm separate centrally and after gastru-lation a second tissue layer the endoderm (or gastroderm)appears After the formation of cilia the free-swimmingplanula is complete within 24ndash72 h for most species(Martin 2000 Groger amp Schmid 2001 Hayward et al2001) The larvae are simple with a surface epitheliumcontaining motor cilia and microvilli and occasional sen-sory neurons and nematocytes (Muller amp Leitz 2002)After some time spent swimming the planulae settle withthe anterior pole to the substrate to form polyps whichcan propagate asexually through fission or form medusaethrough strobilation metamorphosis or budding Themedusae can subsequently reproduce sexually to formnew larvae (Muller amp Leitz 2002) The medusae of somespecies release the gametes directly into the water whereasothers retain the embryos until the planulae are mature(Berrill 1949) Our present investigation of cubozoan lar-vae reveals a simpler organization than in most other cnid-arian larvae and the body plan is radially symmetrical

Two types of photoreceptor are found in animals therhabdomeric type where microvilli constitute the photore-ceptive part of the cell and the ciliary type where oneor more modified cilia form the photoreceptive structure(Salvini-Plawen amp Mayr 1977) Rhabdomeric photoreceptors

Author for correspondence (dan-enilssoncobluse)

Proc R Soc Lond B (2003) 270 2349ndash2354 2349 Oacute 2003 The Royal SocietyDOI 101098rspb20032504

are found in invertebrate groups such as arthropods andcephalopods whereas ciliary photoreceptors are typicallyfound in vertebrate eyes It has long been known thatsome cnidarian medusae have eyes of a surprisingly elab-orate type and the most complex examples are found incubozoans (see Pearse amp Pearse 1978 Laska amp Hundgen1982) Cnidarian photoreceptors are of the ciliary type(Bouillon amp Nielsen 1974 Singla 1974 Martin 2002) Ithas also been reported that cnidarian larvae respond tolight stimuli (Cargo 1979) but photosensory structureshave remained unknown We here report that the planulalarvae of cubomedusae have prominent photoreceptorsand that they break the rule by having rhabdomeric photo-receptors Our ultrastructural study further reveals thatthere is no nervous system to which the ocelli can passvisual information

2 MATERIAL AND METHODS

(a) AnimalsGravid females of Tripedalia cystophora were collected in man-

grove swamps near La Parguera Puerto Rico Larvae wereexcised in seawater and transferred to fixative For comparisonwe used cubozoan planulae of a second family namelyChiropsalmus sp and Chironex fleckerii collected at the northbeaches of Cairns Queensland Australia Unfortunately thestate of preservation (in formalin) was not good enough for adetailed ultrastructural study of larvae of the last two species

(b) HistologyFor transmission electron microscopy (TEM) larvae were

fixed in 25 glutaraldehyde and 25 paraformaldehyde incacodylate buffer for at least 2 h They were then rinsed in caco-dylate buffer stained in 1 OsO4 for 1 h dehydrated in an etha-nol series embedded in Epon resin and polymerized Sectionswere cut with a diamond knife and stained with lead citrate anduranyl acetate in an LKB ultrastainer The sections were exam-ined in a JEOL 1200 EX transmission electron microscope

2350 K Nordstrom and others A simple visual system in jelly sh larvae

(a)

(c)

(b)

20 m m

2 m m

20 m m

3 RESULTS

(a) The structure of the larvaTripedalia planula larvae (figure 1a) have a pear-like

shape with a maximum diameter of ca 80 m m and a totallength of 120 m m while the Chiropsalmus and Chironex

Proc R Soc Lond B (2003)

Figure 1 (a) Longitudinal cross-section through the larva ofTripedalia cystophora showing the general appearance of theanimal The anterior pole is broader (to the right in thepicture) and protruding ocelli (arrows) are found only in theposterior half of the larva The dermal cell layer encases thegastrodermal cell mass There is no mesogloea or basalmembrane between the two tissue layers (b) A close-up ofthe gastroderm showing a lightly stained interstitial-like cellwith densely packed ribosomes but few other organellesThe remaining gastrodermal cells are darker stained andhave a more complex morphology The electron-densegranules within the darker cells have a diameter of up to4 mm (c) This shows the appearance of the epidermal layerwith its columnar organization The cell nuclei are locatedclose to the gastroderm The epidermal cells aremonociliated and covered with surface microvilli There is ahigh abundance of lysosomes and ribosomes in thecytoplasm of these cells The apical borders of neighbouringcells are joined with desmosomes

larvae are rounded but of a similar size They have a single-layer ectoderm covering a more amorphous endodermgastroderm There are two morphologically distinct celltypes in the gastroderm (figure 1b) One type stains lightlyin TEM and resembles interstitial cells with ribosomesand a few other organelles (Martin amp Chia 1982) Themore common type is the yolk cell which stains darkerand contains electron-dense lsquoyolkrsquo granules These gran-ules are large (up to 4 m m in diameter) and alter the cellshape accordingly

The ectodermal epithelium is equally simple and con-tains only three distinguishable cell types These are themonociliated epithelial cells (figure 1c) a few nemato-cytes and 10ndash15 photoreceptor cells (described in detailbelow) The epithelium is organized as a columnar mono-layer with the cell nuclei close to the epithelial base Theectoderm is thicker in the anterior and posterior end ofthe larva Each cell is ca 2 m m wide and 12 m m tall andits surface is covered with microvilli The apical bordersof neighbouring cells are joined with desmosomes Theepithelial cells contain many mitochondria lysosomes andribosomes (free or as rough endoplasmic reticulum)which suggests a high metabolic rate Otherwise theyresemble typical epithelial cells seen in a wide array ofmetazoans There is no basal membrane or mesogloeabetween the two tissue layers

We did not find any neurons or neuron-like structuresin any part of the body The Tripedalia larvae were col-lected directly from female medusae and may not havebeen fully mature Despite this they swam freely after exci-sion and stayed healthy and active for several hours insmall containers of sea water Our comparison of maturelarvae of Chiropsalmus and Chironex also support the con-clusion that cubozoan larvae do not possess a nervous system

(b) Ocellar structureThere are ca 10ndash15 ocelli in each larva In Tripedalia

these are located in the posterior end compared with thedirection of movement and they are evenly spaced for-ming a regular hexagonal pattern Despite being locatedin the lsquowrongrsquo end the ocelli face forward and stick outas if struggling to see in the forward direction (figure 1a)In Chiropsalmus the ocelli are located around the middle

A simple visual system in jelly sh larvae K Nordstrom and others 2351

and in Chironex in the anterior end In neither do theyprotrude outside the epithelial surface as in Tripedalia

Each Tripedalia ocellus is a single photoreceptor cellsimilar to the other epithelial cells but with fewer ribo-somes (figure 2a) Each photoreceptor cell contains a largelysosome-like granular vesicle located close to the nucleusThe photoreceptor contains a cup of screening pigmentthe cavity of which is filled with microvilli projecting fromthe inside of the cup at all sides of the cavity Becausethese microvilli are located in the cup and represent theonly large membrane area in the cell we can convincinglyargue that they must be the sites of photoreception Thepigment cup forms a 3 m m thick screen of pigment gran-ules surrounded by plasma membranes The opening ofthe cup has a diameter of ca 2 m m and is open to theenvironment The pigment cup restricts the sensitivityangle such as to effectively form a directional light meteraiming laterallyforward The swimming larvae are rotat-ing at ca 2 revolutions s21 around their longitudinal axisAs a consequence the ocelli are performing a continuouscircular scanning of the forward direction

The photoreceptor has a single cilium protruding frominside the pigment cup but as this is practically identicalto the cilia of the other epithelial cells (figure 2bc) weinterpret these as being non-photoreceptive Photorecep-tive cilia are usually heavily modified (Salvini-Plawen ampMayr 1977) but those of the larval ocelli are indis-tinguishable from regular motor cilia complete with a dis-tinct 912 filament structure a basal body with a secondperpendicularly arranged centriole and a striated rootletJudging from the structure and anchoring the cilium mustbe assumed to be motile rather than sensory We did acareful search for microvilli protruding from the cilia onlyto find that all microvillar attachments on the numeroussections we analysed protrude directly from the membraneinside the pigment cup The photoreceptor must thus becategorized as rhabdomeric with a non-sensory motor-cilium

Despite careful studies of numerous specimens andelectron microscopy micrographs we were unable to findanything that could possibly be interpreted as axons Nordid we find any signs of synaptic or electrical (gapjunction) connections between the photoreceptor cells orany other cell in the larvae This is in agreement with thecomplete lack of a nervous system There is thus no struc-tural basis for believing that sensory information is in anyway communicated between cells in the larva

4 DISCUSSION

(a) An extremely simple body planWe have demonstrated the amazing simplicity of

cubozoan planula larvae being composed of only five celltypes (figure 3a) Such a simple larval organization isreminiscent more of that in Porifera than in other Cnida-ria Between the two tissue layers of anthozoan planulaea basal membrane or mesogloea have been described forAcropora millepora (eg Hayward et al 2001) or forPocillopora damicornis (Vandermeulen 1974) but we foundno sign of such structures in cubozoan larvae Instead thetissue layers are in direct contact with each other withoutsupporting or separating structures in between

Proc R Soc Lond B (2003)

(a)

V

N

MV

CR

CR

2 m m

2 m m

1 m m

(b)

(c)

Figure 2 In (a) a cross-section through a photoreceptor cell(ocellus) reveals the protruding pigment cup A largelysosome-like vacuole (V) is located close to the nucleus(N) which is similar to those of other epithelial cells Apartfrom microvilli the pigment cup contains a single cilium(arrow) The close-up in (b) shows the microvillar mass(MV) an abundance of mitochondria and the ciliary rootlet(CR) For comparison the very similar ciliary rootlet (CR)of a monociliated epidermal cell is shown in (c)

2352 K Nordstrom and others A simple visual system in jelly sh larvae

(a)

(b)

Figure 3 Schematic drawings of (a) the planula larva ofTripedalia cystophora and (b) a close-up of an ocellus In (a)the two tissue layers and a total of only five different celltypes reveal the extreme simplicity of the larval body Theepidermal cell types are apart from the normal monociliatedmotor cells the photoreceptors (ocelli) and nematocytesThe gastroderm is a homogeneous mass of two cell typeslsquoyolkrsquo cells and lightly staining cells The ocellus (b) has afully developed kinocilium photoreceptive microvilli and apigment cup all combined in the same cell Also indicatedin the ocellus are the striated ciliary rootlet the nucleus andmitochondria Ocellar morphology together with the absenceof a nervous system indicate that each ocellus operates as anindependent sensoryndashmotor unit

The cubozoan gastroderm is filled with electron-denselarge granules presumably containing proteins and lipidsThese may add to the larvarsquos buoyancy and aid in long-term survival or serve as nutrition during the metamor-phosis to a polyp In Acropora the larval gastroderm isextremely rich in lipids which assures a survival time ofup to months (Muller amp Leitz 2002) and similar-lookinggranules of Pocillopora larvae have been identified as lipid-containing (Vandermeulen 1974) The nutritious contentof the planula gastroderm has illustratively been calledyolk and when the larvae get older the granules have beenshown to get vacuolized as the contents are metabolized(Widersten 1968)

Acropora larvae have an oral pore opening into a gastriccavity (Hayward et al 2002) but we found no such struc-ture in the larvae of Tripedalia Chironex or ChiropsalmusScyphozoan larvae (Aurelia aurita and Cyanea capillata)also lack a gastric cavity (Widersten 1968) Interestinglyin the swimming Acropora planula the oral opening is pos-teriorly compared with the direction of movement

Proc R Soc Lond B (2003)

(Hayward et al 2002) The anteriorndashposterior polarity ofa planula is defined not only by the direction of swimmingbut also by the expression pattern of transcription factorsand the formation of nerve cells (Groger amp Schmid 2001)The development of the Podocoryne carnea larvae showssimilarities to Bilateria (Groger amp Schmid 2001) whereasthe Tripedalia larvae described here have a pure radialsymmetry also with respect to ocellar positions

In Tripedalia planulae we could only identify two gastro-dermal and three ectodermal cell types (figure 3a) Othercnidarian larvae are generally more complex in their cellcomposition (Widersten 1968 Muller amp Leitz 2002) Lar-vae of the scyphozoan Cassiopeia xamachana are the onlyexception they have only two ectodermal and two gastro-dermal cell types (Martin amp Chia 1982) Five ectodermalcell types have been described in hydra and anthozoanlarvae epitheliomuscular cells with a motor-cilium glandcells anteriorly located neurosensory cells ganglionicnerve cells and nematocysts (Vandermeulen 1974Muller amp Leitz 2002) but the Tripedalia larvae have onlytwo of these Because larvae of another cubozoanCarybdea is similarly reported to have a simple epidermis(Okada 1927) this simplicity can be assumed to be gen-eral for cubozoan larvae Pigmented spots on Carybdearastonii and C alata planulae have been reported earlier(Okada 1927 Arneson amp Cutress 1976) but have notbeen identified as photoreceptors As a result of thepresent investigation it is likely that the larval pigmentspots of Carybdea are indeed photoreceptor cells and thatthis condition is general for cubozoan larvae

Surprisingly we could find no nerve cells in the larvaealthough a simple nervous system is known from planulaeof other cnidarian groups In hydrozoan (Martin 2000)and anthozoan (Hayward et al 2001) larvae there is anerve net located between the gastroderm and the ecto-derm Two nerve cell types are found in the hydrozoanPennaria tiarella larval ectoderm sensory nerve cells witha broad base and a narrow neck projecting to the surfaceand ganglionic nerve cells located between the gastrodermand ectoderm (Martin 2000) Similarly in the Acroporalarva Arg-Phe-amide (RFamide)-positive neurons arelocated as a nerve net along the basal membrane withprojections towards the surface (Hayward et al 2001)Sensory RFamide-positive cells first appear in 24ndash30-hour-old hydrozoan Podocoryne larvae and later a differentsubset of tubulin-positive nerve cells appear from theanterior part (Groger amp Schmid 2001) The only reportedcases of cnidarian larvae without a nervous system are thepreviously mentioned Carybdea and Cassiopeia (Okada1927 Martin amp Chia 1982) where the degree of cellulardifferentiation resembles that of Tripedalia

(b) A uniquely simple visual systemThe photoreceptor of the Tripedalia planula has a pig-

ment cup contained within the cell (figure 3b) Larval eyesof similar optical simplicity have been described in the cili-ated larva of the polychaete Platynereis dumerilii but as inmost such examples its pigment cup is formed by a separ-ate pigment cell (Arendt et al 2002) Unicellular photo-receptors and presumed photoreceptors without axonsare indeed known to exist in the animal kingdom (Salvini-Plawen amp Mayr 1977 Blumer et al 1995) but these sim-ple traits are not combined in the same species nor do

A simple visual system in jelly sh larvae K Nordstrom and others 2353

they exist in animals without a nervous system Theunicellular photoreceptors of Tripedalia are unique in theirsimplicity yet they are highly specialized for photorecep-tion

Simple eye cups were probably the only existing eyetype before the Cambrian explosion (Nilsson 1996 Fer-nald 2000) and despite being a simple optical solution itcould clearly provide useful visual information to the ani-mal Simple omni-directional photosensitivity providesinformation on general luminance conditions which arefunctions of depth in the water and time of day Direc-tional photoreceptors such as those reported here can beused to guide the animal to brighter or dimmer parts ofthe environment By the motion of the animal itself it pro-vides the simplest form of spatial vision which althoughcoarse may be enough to seek out suitable habitats

The photoreceptor of Tripedalia larvae has bothmicrovilli and a cilium We believe the microvilli to bethe site of photoreception because the single cilium lacksmembrane extensions for housing large quantities of photo-pigment and is indistinguishable from conventionalmotor-cilia (figure 2) The evolutionary origin of photo-receptors has been suggested to be a general epithelial cellwith a cilium and many microvilli which subsequentlygave rise to all the possible photoreceptors seen today(Salvini-Plawen amp Mayr 1977) The photoreceptordescribed here (figure 3b) is close to such an origin It alsoshows some similarity to the simple photoreceptor of theadult hydromedusa Leuckartia octona claimed to be ciliary(Singla 1974)

Ciliary and rhabdomeric photoreceptors have beenassigned various degrees of evolutionary significance(Salvini-Plawen amp Mayr 1977 Nilsson 1994 Fernald2000) but one should recall that the two types can some-times coexist within the same species and in rare caseseven within the same eye (Eakin amp Brandenberger 1981)Both types appear to have evolved independently severaltimes (Salvini-Plawen amp Mayr 1977 Nilsson 1994) Thestandard 912 pattern of microtubules is usually heavilymodified in ciliary photoreceptors (Salvini-Plawen amp Mayr1977) a modification that receptors in Tripedalia larvaedo not share Typical cnidarian photoreceptors includingthose in the eyes of Tripedalia medusae have ciliary photo-receptors with microvilli projecting from the ciliary mem-brane It appears that cubozoan planulae are the firstexample of cnidarians with rhabdomeric receptors

Discrete photosensory structures of planulae have notbeen described before The negative phototaxis of the scy-phozoan Cyanea larva has been attributed to an anteriorlylocated apical tuft of sensory neurons (Svane amp Dolmer1995) However the apical tuft is generally believed to bemechanosensitive (Widersten 1968 Muller amp Leitz 2002)The cubozoan planulae investigated here have no apicaltuft

Photoreceptors with a normal motor-cilium and theabsence of neural or electrical (gap junction) contact toother epithelial cells are unusual features which mayexplain the functional principle of cubozoan larvae If theocelli are autonomous sensoryndashmotor units they may con-trol the swimming direction of the larva by acting as inde-pendent rudders The ocellar cilia need not propel thelarva as this is done by the normal epidermal cells Bysimply flexing and stretching the ocellar cilia in response

Proc R Soc Lond B (2003)

to light the ocelli may steer the larva towards particularluminance conditions This hypothesis is supported byseveral observations The fact that the ocelli do not con-tain more mitochondria than other epidermal cells sug-gests that propulsion is not a part of ocellar functionBecause photoreception is highly energy-consuming theocelli would need more mitochondria than epithelial cellsto support both systems Neither is there any indicationof a direct involvement of ocellar cilia in photoreceptionbecause the cell has many microvilli for this purpose Thisonly leaves a steering role for the ocellar cilia The anchor-ing of cilia with stout rootlets further points to an effectorfunction of the cilia The lack of a nervous system andthe even spacing and distribution of ocelli are strong indi-cators of ocelli operating as independent units with bothsensory and effector properties

It is particularly interesting to note that the ocelli arespread exclusively across the posterior half of the larvabut still pointing more or less in the anterior directionCubozoan planulae and indeed cnidarian planulae in gen-eral (Widersten 1968) rotate around their long axis dur-ing swimming In Tripedalia this rotation occurs with arate of ca 2 revolutions s21 The outwardndashforward point-ing ocelli will thus continuously scan the envelope of acone in the forward direction If the ocellar cilia stretchor relax in response to light they will consistently causemore drag during a part of the spinning cycle and conse-quently be able to steer the larva The only reason forhaving forward-pointing ocelli in a posterior locationwould be that this is a good place for the effector functionof ocelli in this species If this is correct the behaviour ofcubozoan larvae will depend crucially on the positioningof the ocelli Our comparative material from Chironexfleckerii and Chiropsalmus sp reveals different positions forthe ocelli and it also appears that the ocelli are more flushwith the epidermal surface in these species In yet anothercubozoan genus Carybdea Okada (1927) reports pigmentspots in a central band around the larva It is tempting tobelieve that ocellar placement and morphology is tunedfor finding different settling habitats in different species

The unparalleled simplicity makes cubozoan larvalocelli attractive model organisms for studies of geneticcontrol networks and how these can be modified to resultin different morphologies and behaviours With only twotissue layers five cell types and no nervous system (figure 3a)cubozoan larvae are among the most simply organized ani-mal life-forms Although conspicuous their ocelli (figure 3b)represent the simplest visual system yet described

The authors are grateful to Professor Em Rolf Elofsson andDr Eric Hallberg for comments and suggestions with inter-preting the ultrastructure Thanks to Melissa Coates for assist-ance with collecting larvae in Puerto Rico KN and DNreceived support from the Swedish Research Council grantnos 2432000 and 621-2002-4873

REFERENCES

Arendt D Tessmar K de Campos-Baptista M I Dorres-teijn A amp Wittbrodt J 2002 Development of pigment-cupeyes in the polychaete Platynereis dumerilii and evolutionaryconservation of larval eyes in Bilateria Development 1291143ndash1154

2350 K Nordstrom and others A simple visual system in jelly sh larvae

(a)

(c)

(b)

20 m m

2 m m

20 m m

3 RESULTS

(a) The structure of the larvaTripedalia planula larvae (figure 1a) have a pear-like

shape with a maximum diameter of ca 80 m m and a totallength of 120 m m while the Chiropsalmus and Chironex

Proc R Soc Lond B (2003)

Figure 1 (a) Longitudinal cross-section through the larva ofTripedalia cystophora showing the general appearance of theanimal The anterior pole is broader (to the right in thepicture) and protruding ocelli (arrows) are found only in theposterior half of the larva The dermal cell layer encases thegastrodermal cell mass There is no mesogloea or basalmembrane between the two tissue layers (b) A close-up ofthe gastroderm showing a lightly stained interstitial-like cellwith densely packed ribosomes but few other organellesThe remaining gastrodermal cells are darker stained andhave a more complex morphology The electron-densegranules within the darker cells have a diameter of up to4 mm (c) This shows the appearance of the epidermal layerwith its columnar organization The cell nuclei are locatedclose to the gastroderm The epidermal cells aremonociliated and covered with surface microvilli There is ahigh abundance of lysosomes and ribosomes in thecytoplasm of these cells The apical borders of neighbouringcells are joined with desmosomes

larvae are rounded but of a similar size They have a single-layer ectoderm covering a more amorphous endodermgastroderm There are two morphologically distinct celltypes in the gastroderm (figure 1b) One type stains lightlyin TEM and resembles interstitial cells with ribosomesand a few other organelles (Martin amp Chia 1982) Themore common type is the yolk cell which stains darkerand contains electron-dense lsquoyolkrsquo granules These gran-ules are large (up to 4 m m in diameter) and alter the cellshape accordingly

The ectodermal epithelium is equally simple and con-tains only three distinguishable cell types These are themonociliated epithelial cells (figure 1c) a few nemato-cytes and 10ndash15 photoreceptor cells (described in detailbelow) The epithelium is organized as a columnar mono-layer with the cell nuclei close to the epithelial base Theectoderm is thicker in the anterior and posterior end ofthe larva Each cell is ca 2 m m wide and 12 m m tall andits surface is covered with microvilli The apical bordersof neighbouring cells are joined with desmosomes Theepithelial cells contain many mitochondria lysosomes andribosomes (free or as rough endoplasmic reticulum)which suggests a high metabolic rate Otherwise theyresemble typical epithelial cells seen in a wide array ofmetazoans There is no basal membrane or mesogloeabetween the two tissue layers

We did not find any neurons or neuron-like structuresin any part of the body The Tripedalia larvae were col-lected directly from female medusae and may not havebeen fully mature Despite this they swam freely after exci-sion and stayed healthy and active for several hours insmall containers of sea water Our comparison of maturelarvae of Chiropsalmus and Chironex also support the con-clusion that cubozoan larvae do not possess a nervous system

(b) Ocellar structureThere are ca 10ndash15 ocelli in each larva In Tripedalia

these are located in the posterior end compared with thedirection of movement and they are evenly spaced for-ming a regular hexagonal pattern Despite being locatedin the lsquowrongrsquo end the ocelli face forward and stick outas if struggling to see in the forward direction (figure 1a)In Chiropsalmus the ocelli are located around the middle

A simple visual system in jelly sh larvae K Nordstrom and others 2351

and in Chironex in the anterior end In neither do theyprotrude outside the epithelial surface as in Tripedalia

Each Tripedalia ocellus is a single photoreceptor cellsimilar to the other epithelial cells but with fewer ribo-somes (figure 2a) Each photoreceptor cell contains a largelysosome-like granular vesicle located close to the nucleusThe photoreceptor contains a cup of screening pigmentthe cavity of which is filled with microvilli projecting fromthe inside of the cup at all sides of the cavity Becausethese microvilli are located in the cup and represent theonly large membrane area in the cell we can convincinglyargue that they must be the sites of photoreception Thepigment cup forms a 3 m m thick screen of pigment gran-ules surrounded by plasma membranes The opening ofthe cup has a diameter of ca 2 m m and is open to theenvironment The pigment cup restricts the sensitivityangle such as to effectively form a directional light meteraiming laterallyforward The swimming larvae are rotat-ing at ca 2 revolutions s21 around their longitudinal axisAs a consequence the ocelli are performing a continuouscircular scanning of the forward direction

The photoreceptor has a single cilium protruding frominside the pigment cup but as this is practically identicalto the cilia of the other epithelial cells (figure 2bc) weinterpret these as being non-photoreceptive Photorecep-tive cilia are usually heavily modified (Salvini-Plawen ampMayr 1977) but those of the larval ocelli are indis-tinguishable from regular motor cilia complete with a dis-tinct 912 filament structure a basal body with a secondperpendicularly arranged centriole and a striated rootletJudging from the structure and anchoring the cilium mustbe assumed to be motile rather than sensory We did acareful search for microvilli protruding from the cilia onlyto find that all microvillar attachments on the numeroussections we analysed protrude directly from the membraneinside the pigment cup The photoreceptor must thus becategorized as rhabdomeric with a non-sensory motor-cilium

Despite careful studies of numerous specimens andelectron microscopy micrographs we were unable to findanything that could possibly be interpreted as axons Nordid we find any signs of synaptic or electrical (gapjunction) connections between the photoreceptor cells orany other cell in the larvae This is in agreement with thecomplete lack of a nervous system There is thus no struc-tural basis for believing that sensory information is in anyway communicated between cells in the larva

4 DISCUSSION

(a) An extremely simple body planWe have demonstrated the amazing simplicity of

cubozoan planula larvae being composed of only five celltypes (figure 3a) Such a simple larval organization isreminiscent more of that in Porifera than in other Cnida-ria Between the two tissue layers of anthozoan planulaea basal membrane or mesogloea have been described forAcropora millepora (eg Hayward et al 2001) or forPocillopora damicornis (Vandermeulen 1974) but we foundno sign of such structures in cubozoan larvae Instead thetissue layers are in direct contact with each other withoutsupporting or separating structures in between

Proc R Soc Lond B (2003)

(a)

V

N

MV

CR

CR

2 m m

2 m m

1 m m

(b)

(c)

Figure 2 In (a) a cross-section through a photoreceptor cell(ocellus) reveals the protruding pigment cup A largelysosome-like vacuole (V) is located close to the nucleus(N) which is similar to those of other epithelial cells Apartfrom microvilli the pigment cup contains a single cilium(arrow) The close-up in (b) shows the microvillar mass(MV) an abundance of mitochondria and the ciliary rootlet(CR) For comparison the very similar ciliary rootlet (CR)of a monociliated epidermal cell is shown in (c)

2352 K Nordstrom and others A simple visual system in jelly sh larvae

(a)

(b)

Figure 3 Schematic drawings of (a) the planula larva ofTripedalia cystophora and (b) a close-up of an ocellus In (a)the two tissue layers and a total of only five different celltypes reveal the extreme simplicity of the larval body Theepidermal cell types are apart from the normal monociliatedmotor cells the photoreceptors (ocelli) and nematocytesThe gastroderm is a homogeneous mass of two cell typeslsquoyolkrsquo cells and lightly staining cells The ocellus (b) has afully developed kinocilium photoreceptive microvilli and apigment cup all combined in the same cell Also indicatedin the ocellus are the striated ciliary rootlet the nucleus andmitochondria Ocellar morphology together with the absenceof a nervous system indicate that each ocellus operates as anindependent sensoryndashmotor unit

The cubozoan gastroderm is filled with electron-denselarge granules presumably containing proteins and lipidsThese may add to the larvarsquos buoyancy and aid in long-term survival or serve as nutrition during the metamor-phosis to a polyp In Acropora the larval gastroderm isextremely rich in lipids which assures a survival time ofup to months (Muller amp Leitz 2002) and similar-lookinggranules of Pocillopora larvae have been identified as lipid-containing (Vandermeulen 1974) The nutritious contentof the planula gastroderm has illustratively been calledyolk and when the larvae get older the granules have beenshown to get vacuolized as the contents are metabolized(Widersten 1968)

Acropora larvae have an oral pore opening into a gastriccavity (Hayward et al 2002) but we found no such struc-ture in the larvae of Tripedalia Chironex or ChiropsalmusScyphozoan larvae (Aurelia aurita and Cyanea capillata)also lack a gastric cavity (Widersten 1968) Interestinglyin the swimming Acropora planula the oral opening is pos-teriorly compared with the direction of movement

Proc R Soc Lond B (2003)

(Hayward et al 2002) The anteriorndashposterior polarity ofa planula is defined not only by the direction of swimmingbut also by the expression pattern of transcription factorsand the formation of nerve cells (Groger amp Schmid 2001)The development of the Podocoryne carnea larvae showssimilarities to Bilateria (Groger amp Schmid 2001) whereasthe Tripedalia larvae described here have a pure radialsymmetry also with respect to ocellar positions

In Tripedalia planulae we could only identify two gastro-dermal and three ectodermal cell types (figure 3a) Othercnidarian larvae are generally more complex in their cellcomposition (Widersten 1968 Muller amp Leitz 2002) Lar-vae of the scyphozoan Cassiopeia xamachana are the onlyexception they have only two ectodermal and two gastro-dermal cell types (Martin amp Chia 1982) Five ectodermalcell types have been described in hydra and anthozoanlarvae epitheliomuscular cells with a motor-cilium glandcells anteriorly located neurosensory cells ganglionicnerve cells and nematocysts (Vandermeulen 1974Muller amp Leitz 2002) but the Tripedalia larvae have onlytwo of these Because larvae of another cubozoanCarybdea is similarly reported to have a simple epidermis(Okada 1927) this simplicity can be assumed to be gen-eral for cubozoan larvae Pigmented spots on Carybdearastonii and C alata planulae have been reported earlier(Okada 1927 Arneson amp Cutress 1976) but have notbeen identified as photoreceptors As a result of thepresent investigation it is likely that the larval pigmentspots of Carybdea are indeed photoreceptor cells and thatthis condition is general for cubozoan larvae

Surprisingly we could find no nerve cells in the larvaealthough a simple nervous system is known from planulaeof other cnidarian groups In hydrozoan (Martin 2000)and anthozoan (Hayward et al 2001) larvae there is anerve net located between the gastroderm and the ecto-derm Two nerve cell types are found in the hydrozoanPennaria tiarella larval ectoderm sensory nerve cells witha broad base and a narrow neck projecting to the surfaceand ganglionic nerve cells located between the gastrodermand ectoderm (Martin 2000) Similarly in the Acroporalarva Arg-Phe-amide (RFamide)-positive neurons arelocated as a nerve net along the basal membrane withprojections towards the surface (Hayward et al 2001)Sensory RFamide-positive cells first appear in 24ndash30-hour-old hydrozoan Podocoryne larvae and later a differentsubset of tubulin-positive nerve cells appear from theanterior part (Groger amp Schmid 2001) The only reportedcases of cnidarian larvae without a nervous system are thepreviously mentioned Carybdea and Cassiopeia (Okada1927 Martin amp Chia 1982) where the degree of cellulardifferentiation resembles that of Tripedalia

(b) A uniquely simple visual systemThe photoreceptor of the Tripedalia planula has a pig-

ment cup contained within the cell (figure 3b) Larval eyesof similar optical simplicity have been described in the cili-ated larva of the polychaete Platynereis dumerilii but as inmost such examples its pigment cup is formed by a separ-ate pigment cell (Arendt et al 2002) Unicellular photo-receptors and presumed photoreceptors without axonsare indeed known to exist in the animal kingdom (Salvini-Plawen amp Mayr 1977 Blumer et al 1995) but these sim-ple traits are not combined in the same species nor do

A simple visual system in jelly sh larvae K Nordstrom and others 2353

they exist in animals without a nervous system Theunicellular photoreceptors of Tripedalia are unique in theirsimplicity yet they are highly specialized for photorecep-tion

Simple eye cups were probably the only existing eyetype before the Cambrian explosion (Nilsson 1996 Fer-nald 2000) and despite being a simple optical solution itcould clearly provide useful visual information to the ani-mal Simple omni-directional photosensitivity providesinformation on general luminance conditions which arefunctions of depth in the water and time of day Direc-tional photoreceptors such as those reported here can beused to guide the animal to brighter or dimmer parts ofthe environment By the motion of the animal itself it pro-vides the simplest form of spatial vision which althoughcoarse may be enough to seek out suitable habitats

The photoreceptor of Tripedalia larvae has bothmicrovilli and a cilium We believe the microvilli to bethe site of photoreception because the single cilium lacksmembrane extensions for housing large quantities of photo-pigment and is indistinguishable from conventionalmotor-cilia (figure 2) The evolutionary origin of photo-receptors has been suggested to be a general epithelial cellwith a cilium and many microvilli which subsequentlygave rise to all the possible photoreceptors seen today(Salvini-Plawen amp Mayr 1977) The photoreceptordescribed here (figure 3b) is close to such an origin It alsoshows some similarity to the simple photoreceptor of theadult hydromedusa Leuckartia octona claimed to be ciliary(Singla 1974)

Ciliary and rhabdomeric photoreceptors have beenassigned various degrees of evolutionary significance(Salvini-Plawen amp Mayr 1977 Nilsson 1994 Fernald2000) but one should recall that the two types can some-times coexist within the same species and in rare caseseven within the same eye (Eakin amp Brandenberger 1981)Both types appear to have evolved independently severaltimes (Salvini-Plawen amp Mayr 1977 Nilsson 1994) Thestandard 912 pattern of microtubules is usually heavilymodified in ciliary photoreceptors (Salvini-Plawen amp Mayr1977) a modification that receptors in Tripedalia larvaedo not share Typical cnidarian photoreceptors includingthose in the eyes of Tripedalia medusae have ciliary photo-receptors with microvilli projecting from the ciliary mem-brane It appears that cubozoan planulae are the firstexample of cnidarians with rhabdomeric receptors

Discrete photosensory structures of planulae have notbeen described before The negative phototaxis of the scy-phozoan Cyanea larva has been attributed to an anteriorlylocated apical tuft of sensory neurons (Svane amp Dolmer1995) However the apical tuft is generally believed to bemechanosensitive (Widersten 1968 Muller amp Leitz 2002)The cubozoan planulae investigated here have no apicaltuft

Photoreceptors with a normal motor-cilium and theabsence of neural or electrical (gap junction) contact toother epithelial cells are unusual features which mayexplain the functional principle of cubozoan larvae If theocelli are autonomous sensoryndashmotor units they may con-trol the swimming direction of the larva by acting as inde-pendent rudders The ocellar cilia need not propel thelarva as this is done by the normal epidermal cells Bysimply flexing and stretching the ocellar cilia in response

Proc R Soc Lond B (2003)

to light the ocelli may steer the larva towards particularluminance conditions This hypothesis is supported byseveral observations The fact that the ocelli do not con-tain more mitochondria than other epidermal cells sug-gests that propulsion is not a part of ocellar functionBecause photoreception is highly energy-consuming theocelli would need more mitochondria than epithelial cellsto support both systems Neither is there any indicationof a direct involvement of ocellar cilia in photoreceptionbecause the cell has many microvilli for this purpose Thisonly leaves a steering role for the ocellar cilia The anchor-ing of cilia with stout rootlets further points to an effectorfunction of the cilia The lack of a nervous system andthe even spacing and distribution of ocelli are strong indi-cators of ocelli operating as independent units with bothsensory and effector properties

It is particularly interesting to note that the ocelli arespread exclusively across the posterior half of the larvabut still pointing more or less in the anterior directionCubozoan planulae and indeed cnidarian planulae in gen-eral (Widersten 1968) rotate around their long axis dur-ing swimming In Tripedalia this rotation occurs with arate of ca 2 revolutions s21 The outwardndashforward point-ing ocelli will thus continuously scan the envelope of acone in the forward direction If the ocellar cilia stretchor relax in response to light they will consistently causemore drag during a part of the spinning cycle and conse-quently be able to steer the larva The only reason forhaving forward-pointing ocelli in a posterior locationwould be that this is a good place for the effector functionof ocelli in this species If this is correct the behaviour ofcubozoan larvae will depend crucially on the positioningof the ocelli Our comparative material from Chironexfleckerii and Chiropsalmus sp reveals different positions forthe ocelli and it also appears that the ocelli are more flushwith the epidermal surface in these species In yet anothercubozoan genus Carybdea Okada (1927) reports pigmentspots in a central band around the larva It is tempting tobelieve that ocellar placement and morphology is tunedfor finding different settling habitats in different species

The unparalleled simplicity makes cubozoan larvalocelli attractive model organisms for studies of geneticcontrol networks and how these can be modified to resultin different morphologies and behaviours With only twotissue layers five cell types and no nervous system (figure 3a)cubozoan larvae are among the most simply organized ani-mal life-forms Although conspicuous their ocelli (figure 3b)represent the simplest visual system yet described

The authors are grateful to Professor Em Rolf Elofsson andDr Eric Hallberg for comments and suggestions with inter-preting the ultrastructure Thanks to Melissa Coates for assist-ance with collecting larvae in Puerto Rico KN and DNreceived support from the Swedish Research Council grantnos 2432000 and 621-2002-4873

REFERENCES

Arendt D Tessmar K de Campos-Baptista M I Dorres-teijn A amp Wittbrodt J 2002 Development of pigment-cupeyes in the polychaete Platynereis dumerilii and evolutionaryconservation of larval eyes in Bilateria Development 1291143ndash1154

A simple visual system in jelly sh larvae K Nordstrom and others 2351

and in Chironex in the anterior end In neither do theyprotrude outside the epithelial surface as in Tripedalia

Each Tripedalia ocellus is a single photoreceptor cellsimilar to the other epithelial cells but with fewer ribo-somes (figure 2a) Each photoreceptor cell contains a largelysosome-like granular vesicle located close to the nucleusThe photoreceptor contains a cup of screening pigmentthe cavity of which is filled with microvilli projecting fromthe inside of the cup at all sides of the cavity Becausethese microvilli are located in the cup and represent theonly large membrane area in the cell we can convincinglyargue that they must be the sites of photoreception Thepigment cup forms a 3 m m thick screen of pigment gran-ules surrounded by plasma membranes The opening ofthe cup has a diameter of ca 2 m m and is open to theenvironment The pigment cup restricts the sensitivityangle such as to effectively form a directional light meteraiming laterallyforward The swimming larvae are rotat-ing at ca 2 revolutions s21 around their longitudinal axisAs a consequence the ocelli are performing a continuouscircular scanning of the forward direction

The photoreceptor has a single cilium protruding frominside the pigment cup but as this is practically identicalto the cilia of the other epithelial cells (figure 2bc) weinterpret these as being non-photoreceptive Photorecep-tive cilia are usually heavily modified (Salvini-Plawen ampMayr 1977) but those of the larval ocelli are indis-tinguishable from regular motor cilia complete with a dis-tinct 912 filament structure a basal body with a secondperpendicularly arranged centriole and a striated rootletJudging from the structure and anchoring the cilium mustbe assumed to be motile rather than sensory We did acareful search for microvilli protruding from the cilia onlyto find that all microvillar attachments on the numeroussections we analysed protrude directly from the membraneinside the pigment cup The photoreceptor must thus becategorized as rhabdomeric with a non-sensory motor-cilium

Despite careful studies of numerous specimens andelectron microscopy micrographs we were unable to findanything that could possibly be interpreted as axons Nordid we find any signs of synaptic or electrical (gapjunction) connections between the photoreceptor cells orany other cell in the larvae This is in agreement with thecomplete lack of a nervous system There is thus no struc-tural basis for believing that sensory information is in anyway communicated between cells in the larva

4 DISCUSSION

(a) An extremely simple body planWe have demonstrated the amazing simplicity of

cubozoan planula larvae being composed of only five celltypes (figure 3a) Such a simple larval organization isreminiscent more of that in Porifera than in other Cnida-ria Between the two tissue layers of anthozoan planulaea basal membrane or mesogloea have been described forAcropora millepora (eg Hayward et al 2001) or forPocillopora damicornis (Vandermeulen 1974) but we foundno sign of such structures in cubozoan larvae Instead thetissue layers are in direct contact with each other withoutsupporting or separating structures in between

Proc R Soc Lond B (2003)

(a)

V

N

MV

CR

CR

2 m m

2 m m

1 m m

(b)

(c)

Figure 2 In (a) a cross-section through a photoreceptor cell(ocellus) reveals the protruding pigment cup A largelysosome-like vacuole (V) is located close to the nucleus(N) which is similar to those of other epithelial cells Apartfrom microvilli the pigment cup contains a single cilium(arrow) The close-up in (b) shows the microvillar mass(MV) an abundance of mitochondria and the ciliary rootlet(CR) For comparison the very similar ciliary rootlet (CR)of a monociliated epidermal cell is shown in (c)

2352 K Nordstrom and others A simple visual system in jelly sh larvae

(a)

(b)

Figure 3 Schematic drawings of (a) the planula larva ofTripedalia cystophora and (b) a close-up of an ocellus In (a)the two tissue layers and a total of only five different celltypes reveal the extreme simplicity of the larval body Theepidermal cell types are apart from the normal monociliatedmotor cells the photoreceptors (ocelli) and nematocytesThe gastroderm is a homogeneous mass of two cell typeslsquoyolkrsquo cells and lightly staining cells The ocellus (b) has afully developed kinocilium photoreceptive microvilli and apigment cup all combined in the same cell Also indicatedin the ocellus are the striated ciliary rootlet the nucleus andmitochondria Ocellar morphology together with the absenceof a nervous system indicate that each ocellus operates as anindependent sensoryndashmotor unit

The cubozoan gastroderm is filled with electron-denselarge granules presumably containing proteins and lipidsThese may add to the larvarsquos buoyancy and aid in long-term survival or serve as nutrition during the metamor-phosis to a polyp In Acropora the larval gastroderm isextremely rich in lipids which assures a survival time ofup to months (Muller amp Leitz 2002) and similar-lookinggranules of Pocillopora larvae have been identified as lipid-containing (Vandermeulen 1974) The nutritious contentof the planula gastroderm has illustratively been calledyolk and when the larvae get older the granules have beenshown to get vacuolized as the contents are metabolized(Widersten 1968)

Acropora larvae have an oral pore opening into a gastriccavity (Hayward et al 2002) but we found no such struc-ture in the larvae of Tripedalia Chironex or ChiropsalmusScyphozoan larvae (Aurelia aurita and Cyanea capillata)also lack a gastric cavity (Widersten 1968) Interestinglyin the swimming Acropora planula the oral opening is pos-teriorly compared with the direction of movement

Proc R Soc Lond B (2003)

(Hayward et al 2002) The anteriorndashposterior polarity ofa planula is defined not only by the direction of swimmingbut also by the expression pattern of transcription factorsand the formation of nerve cells (Groger amp Schmid 2001)The development of the Podocoryne carnea larvae showssimilarities to Bilateria (Groger amp Schmid 2001) whereasthe Tripedalia larvae described here have a pure radialsymmetry also with respect to ocellar positions

In Tripedalia planulae we could only identify two gastro-dermal and three ectodermal cell types (figure 3a) Othercnidarian larvae are generally more complex in their cellcomposition (Widersten 1968 Muller amp Leitz 2002) Lar-vae of the scyphozoan Cassiopeia xamachana are the onlyexception they have only two ectodermal and two gastro-dermal cell types (Martin amp Chia 1982) Five ectodermalcell types have been described in hydra and anthozoanlarvae epitheliomuscular cells with a motor-cilium glandcells anteriorly located neurosensory cells ganglionicnerve cells and nematocysts (Vandermeulen 1974Muller amp Leitz 2002) but the Tripedalia larvae have onlytwo of these Because larvae of another cubozoanCarybdea is similarly reported to have a simple epidermis(Okada 1927) this simplicity can be assumed to be gen-eral for cubozoan larvae Pigmented spots on Carybdearastonii and C alata planulae have been reported earlier(Okada 1927 Arneson amp Cutress 1976) but have notbeen identified as photoreceptors As a result of thepresent investigation it is likely that the larval pigmentspots of Carybdea are indeed photoreceptor cells and thatthis condition is general for cubozoan larvae

Surprisingly we could find no nerve cells in the larvaealthough a simple nervous system is known from planulaeof other cnidarian groups In hydrozoan (Martin 2000)and anthozoan (Hayward et al 2001) larvae there is anerve net located between the gastroderm and the ecto-derm Two nerve cell types are found in the hydrozoanPennaria tiarella larval ectoderm sensory nerve cells witha broad base and a narrow neck projecting to the surfaceand ganglionic nerve cells located between the gastrodermand ectoderm (Martin 2000) Similarly in the Acroporalarva Arg-Phe-amide (RFamide)-positive neurons arelocated as a nerve net along the basal membrane withprojections towards the surface (Hayward et al 2001)Sensory RFamide-positive cells first appear in 24ndash30-hour-old hydrozoan Podocoryne larvae and later a differentsubset of tubulin-positive nerve cells appear from theanterior part (Groger amp Schmid 2001) The only reportedcases of cnidarian larvae without a nervous system are thepreviously mentioned Carybdea and Cassiopeia (Okada1927 Martin amp Chia 1982) where the degree of cellulardifferentiation resembles that of Tripedalia

(b) A uniquely simple visual systemThe photoreceptor of the Tripedalia planula has a pig-

ment cup contained within the cell (figure 3b) Larval eyesof similar optical simplicity have been described in the cili-ated larva of the polychaete Platynereis dumerilii but as inmost such examples its pigment cup is formed by a separ-ate pigment cell (Arendt et al 2002) Unicellular photo-receptors and presumed photoreceptors without axonsare indeed known to exist in the animal kingdom (Salvini-Plawen amp Mayr 1977 Blumer et al 1995) but these sim-ple traits are not combined in the same species nor do

A simple visual system in jelly sh larvae K Nordstrom and others 2353

they exist in animals without a nervous system Theunicellular photoreceptors of Tripedalia are unique in theirsimplicity yet they are highly specialized for photorecep-tion

Simple eye cups were probably the only existing eyetype before the Cambrian explosion (Nilsson 1996 Fer-nald 2000) and despite being a simple optical solution itcould clearly provide useful visual information to the ani-mal Simple omni-directional photosensitivity providesinformation on general luminance conditions which arefunctions of depth in the water and time of day Direc-tional photoreceptors such as those reported here can beused to guide the animal to brighter or dimmer parts ofthe environment By the motion of the animal itself it pro-vides the simplest form of spatial vision which althoughcoarse may be enough to seek out suitable habitats

The photoreceptor of Tripedalia larvae has bothmicrovilli and a cilium We believe the microvilli to bethe site of photoreception because the single cilium lacksmembrane extensions for housing large quantities of photo-pigment and is indistinguishable from conventionalmotor-cilia (figure 2) The evolutionary origin of photo-receptors has been suggested to be a general epithelial cellwith a cilium and many microvilli which subsequentlygave rise to all the possible photoreceptors seen today(Salvini-Plawen amp Mayr 1977) The photoreceptordescribed here (figure 3b) is close to such an origin It alsoshows some similarity to the simple photoreceptor of theadult hydromedusa Leuckartia octona claimed to be ciliary(Singla 1974)

Ciliary and rhabdomeric photoreceptors have beenassigned various degrees of evolutionary significance(Salvini-Plawen amp Mayr 1977 Nilsson 1994 Fernald2000) but one should recall that the two types can some-times coexist within the same species and in rare caseseven within the same eye (Eakin amp Brandenberger 1981)Both types appear to have evolved independently severaltimes (Salvini-Plawen amp Mayr 1977 Nilsson 1994) Thestandard 912 pattern of microtubules is usually heavilymodified in ciliary photoreceptors (Salvini-Plawen amp Mayr1977) a modification that receptors in Tripedalia larvaedo not share Typical cnidarian photoreceptors includingthose in the eyes of Tripedalia medusae have ciliary photo-receptors with microvilli projecting from the ciliary mem-brane It appears that cubozoan planulae are the firstexample of cnidarians with rhabdomeric receptors

Discrete photosensory structures of planulae have notbeen described before The negative phototaxis of the scy-phozoan Cyanea larva has been attributed to an anteriorlylocated apical tuft of sensory neurons (Svane amp Dolmer1995) However the apical tuft is generally believed to bemechanosensitive (Widersten 1968 Muller amp Leitz 2002)The cubozoan planulae investigated here have no apicaltuft

Photoreceptors with a normal motor-cilium and theabsence of neural or electrical (gap junction) contact toother epithelial cells are unusual features which mayexplain the functional principle of cubozoan larvae If theocelli are autonomous sensoryndashmotor units they may con-trol the swimming direction of the larva by acting as inde-pendent rudders The ocellar cilia need not propel thelarva as this is done by the normal epidermal cells Bysimply flexing and stretching the ocellar cilia in response

Proc R Soc Lond B (2003)

to light the ocelli may steer the larva towards particularluminance conditions This hypothesis is supported byseveral observations The fact that the ocelli do not con-tain more mitochondria than other epidermal cells sug-gests that propulsion is not a part of ocellar functionBecause photoreception is highly energy-consuming theocelli would need more mitochondria than epithelial cellsto support both systems Neither is there any indicationof a direct involvement of ocellar cilia in photoreceptionbecause the cell has many microvilli for this purpose Thisonly leaves a steering role for the ocellar cilia The anchor-ing of cilia with stout rootlets further points to an effectorfunction of the cilia The lack of a nervous system andthe even spacing and distribution of ocelli are strong indi-cators of ocelli operating as independent units with bothsensory and effector properties

It is particularly interesting to note that the ocelli arespread exclusively across the posterior half of the larvabut still pointing more or less in the anterior directionCubozoan planulae and indeed cnidarian planulae in gen-eral (Widersten 1968) rotate around their long axis dur-ing swimming In Tripedalia this rotation occurs with arate of ca 2 revolutions s21 The outwardndashforward point-ing ocelli will thus continuously scan the envelope of acone in the forward direction If the ocellar cilia stretchor relax in response to light they will consistently causemore drag during a part of the spinning cycle and conse-quently be able to steer the larva The only reason forhaving forward-pointing ocelli in a posterior locationwould be that this is a good place for the effector functionof ocelli in this species If this is correct the behaviour ofcubozoan larvae will depend crucially on the positioningof the ocelli Our comparative material from Chironexfleckerii and Chiropsalmus sp reveals different positions forthe ocelli and it also appears that the ocelli are more flushwith the epidermal surface in these species In yet anothercubozoan genus Carybdea Okada (1927) reports pigmentspots in a central band around the larva It is tempting tobelieve that ocellar placement and morphology is tunedfor finding different settling habitats in different species

The unparalleled simplicity makes cubozoan larvalocelli attractive model organisms for studies of geneticcontrol networks and how these can be modified to resultin different morphologies and behaviours With only twotissue layers five cell types and no nervous system (figure 3a)cubozoan larvae are among the most simply organized ani-mal life-forms Although conspicuous their ocelli (figure 3b)represent the simplest visual system yet described

The authors are grateful to Professor Em Rolf Elofsson andDr Eric Hallberg for comments and suggestions with inter-preting the ultrastructure Thanks to Melissa Coates for assist-ance with collecting larvae in Puerto Rico KN and DNreceived support from the Swedish Research Council grantnos 2432000 and 621-2002-4873

REFERENCES

Arendt D Tessmar K de Campos-Baptista M I Dorres-teijn A amp Wittbrodt J 2002 Development of pigment-cupeyes in the polychaete Platynereis dumerilii and evolutionaryconservation of larval eyes in Bilateria Development 1291143ndash1154

2352 K Nordstrom and others A simple visual system in jelly sh larvae

(a)

(b)

Figure 3 Schematic drawings of (a) the planula larva ofTripedalia cystophora and (b) a close-up of an ocellus In (a)the two tissue layers and a total of only five different celltypes reveal the extreme simplicity of the larval body Theepidermal cell types are apart from the normal monociliatedmotor cells the photoreceptors (ocelli) and nematocytesThe gastroderm is a homogeneous mass of two cell typeslsquoyolkrsquo cells and lightly staining cells The ocellus (b) has afully developed kinocilium photoreceptive microvilli and apigment cup all combined in the same cell Also indicatedin the ocellus are the striated ciliary rootlet the nucleus andmitochondria Ocellar morphology together with the absenceof a nervous system indicate that each ocellus operates as anindependent sensoryndashmotor unit

The cubozoan gastroderm is filled with electron-denselarge granules presumably containing proteins and lipidsThese may add to the larvarsquos buoyancy and aid in long-term survival or serve as nutrition during the metamor-phosis to a polyp In Acropora the larval gastroderm isextremely rich in lipids which assures a survival time ofup to months (Muller amp Leitz 2002) and similar-lookinggranules of Pocillopora larvae have been identified as lipid-containing (Vandermeulen 1974) The nutritious contentof the planula gastroderm has illustratively been calledyolk and when the larvae get older the granules have beenshown to get vacuolized as the contents are metabolized(Widersten 1968)

Acropora larvae have an oral pore opening into a gastriccavity (Hayward et al 2002) but we found no such struc-ture in the larvae of Tripedalia Chironex or ChiropsalmusScyphozoan larvae (Aurelia aurita and Cyanea capillata)also lack a gastric cavity (Widersten 1968) Interestinglyin the swimming Acropora planula the oral opening is pos-teriorly compared with the direction of movement

Proc R Soc Lond B (2003)

(Hayward et al 2002) The anteriorndashposterior polarity ofa planula is defined not only by the direction of swimmingbut also by the expression pattern of transcription factorsand the formation of nerve cells (Groger amp Schmid 2001)The development of the Podocoryne carnea larvae showssimilarities to Bilateria (Groger amp Schmid 2001) whereasthe Tripedalia larvae described here have a pure radialsymmetry also with respect to ocellar positions

In Tripedalia planulae we could only identify two gastro-dermal and three ectodermal cell types (figure 3a) Othercnidarian larvae are generally more complex in their cellcomposition (Widersten 1968 Muller amp Leitz 2002) Lar-vae of the scyphozoan Cassiopeia xamachana are the onlyexception they have only two ectodermal and two gastro-dermal cell types (Martin amp Chia 1982) Five ectodermalcell types have been described in hydra and anthozoanlarvae epitheliomuscular cells with a motor-cilium glandcells anteriorly located neurosensory cells ganglionicnerve cells and nematocysts (Vandermeulen 1974Muller amp Leitz 2002) but the Tripedalia larvae have onlytwo of these Because larvae of another cubozoanCarybdea is similarly reported to have a simple epidermis(Okada 1927) this simplicity can be assumed to be gen-eral for cubozoan larvae Pigmented spots on Carybdearastonii and C alata planulae have been reported earlier(Okada 1927 Arneson amp Cutress 1976) but have notbeen identified as photoreceptors As a result of thepresent investigation it is likely that the larval pigmentspots of Carybdea are indeed photoreceptor cells and thatthis condition is general for cubozoan larvae

Surprisingly we could find no nerve cells in the larvaealthough a simple nervous system is known from planulaeof other cnidarian groups In hydrozoan (Martin 2000)and anthozoan (Hayward et al 2001) larvae there is anerve net located between the gastroderm and the ecto-derm Two nerve cell types are found in the hydrozoanPennaria tiarella larval ectoderm sensory nerve cells witha broad base and a narrow neck projecting to the surfaceand ganglionic nerve cells located between the gastrodermand ectoderm (Martin 2000) Similarly in the Acroporalarva Arg-Phe-amide (RFamide)-positive neurons arelocated as a nerve net along the basal membrane withprojections towards the surface (Hayward et al 2001)Sensory RFamide-positive cells first appear in 24ndash30-hour-old hydrozoan Podocoryne larvae and later a differentsubset of tubulin-positive nerve cells appear from theanterior part (Groger amp Schmid 2001) The only reportedcases of cnidarian larvae without a nervous system are thepreviously mentioned Carybdea and Cassiopeia (Okada1927 Martin amp Chia 1982) where the degree of cellulardifferentiation resembles that of Tripedalia

(b) A uniquely simple visual systemThe photoreceptor of the Tripedalia planula has a pig-

ment cup contained within the cell (figure 3b) Larval eyesof similar optical simplicity have been described in the cili-ated larva of the polychaete Platynereis dumerilii but as inmost such examples its pigment cup is formed by a separ-ate pigment cell (Arendt et al 2002) Unicellular photo-receptors and presumed photoreceptors without axonsare indeed known to exist in the animal kingdom (Salvini-Plawen amp Mayr 1977 Blumer et al 1995) but these sim-ple traits are not combined in the same species nor do

A simple visual system in jelly sh larvae K Nordstrom and others 2353

they exist in animals without a nervous system Theunicellular photoreceptors of Tripedalia are unique in theirsimplicity yet they are highly specialized for photorecep-tion

Simple eye cups were probably the only existing eyetype before the Cambrian explosion (Nilsson 1996 Fer-nald 2000) and despite being a simple optical solution itcould clearly provide useful visual information to the ani-mal Simple omni-directional photosensitivity providesinformation on general luminance conditions which arefunctions of depth in the water and time of day Direc-tional photoreceptors such as those reported here can beused to guide the animal to brighter or dimmer parts ofthe environment By the motion of the animal itself it pro-vides the simplest form of spatial vision which althoughcoarse may be enough to seek out suitable habitats

The photoreceptor of Tripedalia larvae has bothmicrovilli and a cilium We believe the microvilli to bethe site of photoreception because the single cilium lacksmembrane extensions for housing large quantities of photo-pigment and is indistinguishable from conventionalmotor-cilia (figure 2) The evolutionary origin of photo-receptors has been suggested to be a general epithelial cellwith a cilium and many microvilli which subsequentlygave rise to all the possible photoreceptors seen today(Salvini-Plawen amp Mayr 1977) The photoreceptordescribed here (figure 3b) is close to such an origin It alsoshows some similarity to the simple photoreceptor of theadult hydromedusa Leuckartia octona claimed to be ciliary(Singla 1974)

Ciliary and rhabdomeric photoreceptors have beenassigned various degrees of evolutionary significance(Salvini-Plawen amp Mayr 1977 Nilsson 1994 Fernald2000) but one should recall that the two types can some-times coexist within the same species and in rare caseseven within the same eye (Eakin amp Brandenberger 1981)Both types appear to have evolved independently severaltimes (Salvini-Plawen amp Mayr 1977 Nilsson 1994) Thestandard 912 pattern of microtubules is usually heavilymodified in ciliary photoreceptors (Salvini-Plawen amp Mayr1977) a modification that receptors in Tripedalia larvaedo not share Typical cnidarian photoreceptors includingthose in the eyes of Tripedalia medusae have ciliary photo-receptors with microvilli projecting from the ciliary mem-brane It appears that cubozoan planulae are the firstexample of cnidarians with rhabdomeric receptors

Discrete photosensory structures of planulae have notbeen described before The negative phototaxis of the scy-phozoan Cyanea larva has been attributed to an anteriorlylocated apical tuft of sensory neurons (Svane amp Dolmer1995) However the apical tuft is generally believed to bemechanosensitive (Widersten 1968 Muller amp Leitz 2002)The cubozoan planulae investigated here have no apicaltuft

Photoreceptors with a normal motor-cilium and theabsence of neural or electrical (gap junction) contact toother epithelial cells are unusual features which mayexplain the functional principle of cubozoan larvae If theocelli are autonomous sensoryndashmotor units they may con-trol the swimming direction of the larva by acting as inde-pendent rudders The ocellar cilia need not propel thelarva as this is done by the normal epidermal cells Bysimply flexing and stretching the ocellar cilia in response

Proc R Soc Lond B (2003)

to light the ocelli may steer the larva towards particularluminance conditions This hypothesis is supported byseveral observations The fact that the ocelli do not con-tain more mitochondria than other epidermal cells sug-gests that propulsion is not a part of ocellar functionBecause photoreception is highly energy-consuming theocelli would need more mitochondria than epithelial cellsto support both systems Neither is there any indicationof a direct involvement of ocellar cilia in photoreceptionbecause the cell has many microvilli for this purpose Thisonly leaves a steering role for the ocellar cilia The anchor-ing of cilia with stout rootlets further points to an effectorfunction of the cilia The lack of a nervous system andthe even spacing and distribution of ocelli are strong indi-cators of ocelli operating as independent units with bothsensory and effector properties

It is particularly interesting to note that the ocelli arespread exclusively across the posterior half of the larvabut still pointing more or less in the anterior directionCubozoan planulae and indeed cnidarian planulae in gen-eral (Widersten 1968) rotate around their long axis dur-ing swimming In Tripedalia this rotation occurs with arate of ca 2 revolutions s21 The outwardndashforward point-ing ocelli will thus continuously scan the envelope of acone in the forward direction If the ocellar cilia stretchor relax in response to light they will consistently causemore drag during a part of the spinning cycle and conse-quently be able to steer the larva The only reason forhaving forward-pointing ocelli in a posterior locationwould be that this is a good place for the effector functionof ocelli in this species If this is correct the behaviour ofcubozoan larvae will depend crucially on the positioningof the ocelli Our comparative material from Chironexfleckerii and Chiropsalmus sp reveals different positions forthe ocelli and it also appears that the ocelli are more flushwith the epidermal surface in these species In yet anothercubozoan genus Carybdea Okada (1927) reports pigmentspots in a central band around the larva It is tempting tobelieve that ocellar placement and morphology is tunedfor finding different settling habitats in different species

The unparalleled simplicity makes cubozoan larvalocelli attractive model organisms for studies of geneticcontrol networks and how these can be modified to resultin different morphologies and behaviours With only twotissue layers five cell types and no nervous system (figure 3a)cubozoan larvae are among the most simply organized ani-mal life-forms Although conspicuous their ocelli (figure 3b)represent the simplest visual system yet described

The authors are grateful to Professor Em Rolf Elofsson andDr Eric Hallberg for comments and suggestions with inter-preting the ultrastructure Thanks to Melissa Coates for assist-ance with collecting larvae in Puerto Rico KN and DNreceived support from the Swedish Research Council grantnos 2432000 and 621-2002-4873

REFERENCES

Arendt D Tessmar K de Campos-Baptista M I Dorres-teijn A amp Wittbrodt J 2002 Development of pigment-cupeyes in the polychaete Platynereis dumerilii and evolutionaryconservation of larval eyes in Bilateria Development 1291143ndash1154

A simple visual system in jelly sh larvae K Nordstrom and others 2353

they exist in animals without a nervous system Theunicellular photoreceptors of Tripedalia are unique in theirsimplicity yet they are highly specialized for photorecep-tion

Simple eye cups were probably the only existing eyetype before the Cambrian explosion (Nilsson 1996 Fer-nald 2000) and despite being a simple optical solution itcould clearly provide useful visual information to the ani-mal Simple omni-directional photosensitivity providesinformation on general luminance conditions which arefunctions of depth in the water and time of day Direc-tional photoreceptors such as those reported here can beused to guide the animal to brighter or dimmer parts ofthe environment By the motion of the animal itself it pro-vides the simplest form of spatial vision which althoughcoarse may be enough to seek out suitable habitats

The photoreceptor of Tripedalia larvae has bothmicrovilli and a cilium We believe the microvilli to bethe site of photoreception because the single cilium lacksmembrane extensions for housing large quantities of photo-pigment and is indistinguishable from conventionalmotor-cilia (figure 2) The evolutionary origin of photo-receptors has been suggested to be a general epithelial cellwith a cilium and many microvilli which subsequentlygave rise to all the possible photoreceptors seen today(Salvini-Plawen amp Mayr 1977) The photoreceptordescribed here (figure 3b) is close to such an origin It alsoshows some similarity to the simple photoreceptor of theadult hydromedusa Leuckartia octona claimed to be ciliary(Singla 1974)

Ciliary and rhabdomeric photoreceptors have beenassigned various degrees of evolutionary significance(Salvini-Plawen amp Mayr 1977 Nilsson 1994 Fernald2000) but one should recall that the two types can some-times coexist within the same species and in rare caseseven within the same eye (Eakin amp Brandenberger 1981)Both types appear to have evolved independently severaltimes (Salvini-Plawen amp Mayr 1977 Nilsson 1994) Thestandard 912 pattern of microtubules is usually heavilymodified in ciliary photoreceptors (Salvini-Plawen amp Mayr1977) a modification that receptors in Tripedalia larvaedo not share Typical cnidarian photoreceptors includingthose in the eyes of Tripedalia medusae have ciliary photo-receptors with microvilli projecting from the ciliary mem-brane It appears that cubozoan planulae are the firstexample of cnidarians with rhabdomeric receptors

Discrete photosensory structures of planulae have notbeen described before The negative phototaxis of the scy-phozoan Cyanea larva has been attributed to an anteriorlylocated apical tuft of sensory neurons (Svane amp Dolmer1995) However the apical tuft is generally believed to bemechanosensitive (Widersten 1968 Muller amp Leitz 2002)The cubozoan planulae investigated here have no apicaltuft

Photoreceptors with a normal motor-cilium and theabsence of neural or electrical (gap junction) contact toother epithelial cells are unusual features which mayexplain the functional principle of cubozoan larvae If theocelli are autonomous sensoryndashmotor units they may con-trol the swimming direction of the larva by acting as inde-pendent rudders The ocellar cilia need not propel thelarva as this is done by the normal epidermal cells Bysimply flexing and stretching the ocellar cilia in response

Proc R Soc Lond B (2003)

to light the ocelli may steer the larva towards particularluminance conditions This hypothesis is supported byseveral observations The fact that the ocelli do not con-tain more mitochondria than other epidermal cells sug-gests that propulsion is not a part of ocellar functionBecause photoreception is highly energy-consuming theocelli would need more mitochondria than epithelial cellsto support both systems Neither is there any indicationof a direct involvement of ocellar cilia in photoreceptionbecause the cell has many microvilli for this purpose Thisonly leaves a steering role for the ocellar cilia The anchor-ing of cilia with stout rootlets further points to an effectorfunction of the cilia The lack of a nervous system andthe even spacing and distribution of ocelli are strong indi-cators of ocelli operating as independent units with bothsensory and effector properties

It is particularly interesting to note that the ocelli arespread exclusively across the posterior half of the larvabut still pointing more or less in the anterior directionCubozoan planulae and indeed cnidarian planulae in gen-eral (Widersten 1968) rotate around their long axis dur-ing swimming In Tripedalia this rotation occurs with arate of ca 2 revolutions s21 The outwardndashforward point-ing ocelli will thus continuously scan the envelope of acone in the forward direction If the ocellar cilia stretchor relax in response to light they will consistently causemore drag during a part of the spinning cycle and conse-quently be able to steer the larva The only reason forhaving forward-pointing ocelli in a posterior locationwould be that this is a good place for the effector functionof ocelli in this species If this is correct the behaviour ofcubozoan larvae will depend crucially on the positioningof the ocelli Our comparative material from Chironexfleckerii and Chiropsalmus sp reveals different positions forthe ocelli and it also appears that the ocelli are more flushwith the epidermal surface in these species In yet anothercubozoan genus Carybdea Okada (1927) reports pigmentspots in a central band around the larva It is tempting tobelieve that ocellar placement and morphology is tunedfor finding different settling habitats in different species

The unparalleled simplicity makes cubozoan larvalocelli attractive model organisms for studies of geneticcontrol networks and how these can be modified to resultin different morphologies and behaviours With only twotissue layers five cell types and no nervous system (figure 3a)cubozoan larvae are among the most simply organized ani-mal life-forms Although conspicuous their ocelli (figure 3b)represent the simplest visual system yet described

The authors are grateful to Professor Em Rolf Elofsson andDr Eric Hallberg for comments and suggestions with inter-preting the ultrastructure Thanks to Melissa Coates for assist-ance with collecting larvae in Puerto Rico KN and DNreceived support from the Swedish Research Council grantnos 2432000 and 621-2002-4873

REFERENCES

Arendt D Tessmar K de Campos-Baptista M I Dorres-teijn A amp Wittbrodt J 2002 Development of pigment-cupeyes in the polychaete Platynereis dumerilii and evolutionaryconservation of larval eyes in Bilateria Development 1291143ndash1154

2354 K Nordstrom and others A simple visual system in jelly sh larvae

Arneson A C amp Cutress C E 1976 Life history of Carybdeaalata Reynaud 1830 (Cubomedusae) In Coelenterate ecologyand behavior (ed G O Mackie) pp 227ndash236 New YorkPlenum

Berrill N J 1949 Developmental analysis of ScyphomedusaeBiol Rev Camb 24 393ndash410

Blumer M J F Salvini-Plawen L V amp Kikinger R 1995Ocelli in a Cnidarian polyp the ultrastructure of the pigmentspots in Stylocoronella riedli (Scyphozoa Saturomedusae)Zoomorphology 115 221ndash227

Bouillon J amp Nielsen M 1974 Etude de quelques organessensoriels de cnidaires Arch Biol Bruxelles 85 307ndash328

Cargo D G 1979 Observations on the settling behavior ofplanular larvae of Chrysaora quinquecirrha Int J InvertebrReprod 1 279ndash287

Eakin R M amp Brandenberger J L 1981 Unique eye of prob-able evolutionary significance Science 211 1189ndash1190

Fernald R D 2000 Evolution of eyes Curr Opin Neurobiol10 444ndash450

Groger H amp Schmid V 2001 Larval development in Cnida-ria a connection to bilateria Genesis 29 110ndash114

Hayward D C Catmull J Reece-Hoyes J S BerghammerH Dodd H Hann S J Miller D J amp Ball E E 2001Gene structure and larval expression of cnox-2Am from thecoral Acropora millepora Dev Genes Evol 211 10ndash19

Hayward D C Samuel G Pontynen P C Catmull JSaint R Miller D J amp Ball E E 2002 Localizedexpression of a dppBMP24 ortholog in a coral embryo ProcNatl Acad Sci USA 99 8106ndash8111

Laska G amp Hundgen M 1982 Morphologie und Ultras-truktur der Lichtsinnesorgan von Tripedalia cystophora Con-ant (Cnidaria Cubozoa) Zool Jb Anat 108 107ndash123

Proc R Soc Lond B (2003)

Martin V J 2000 Reorganization of the nervous system dur-ing metamorphosis of a hydrozoan planula Invertebr Biol199 243ndash253

Martin V J 2002 Photoreceptors of cnidarians Can J Zool80 1703ndash1722

Martin V J amp Chia F-S 1982 Fine structure of a scypho-zoan planula Cassiopeia xamachana Biol Bull 163 320ndash328

Muller W A amp Leitz T 2002 Metamorphosis in the Cnida-ria Can J Zool 80 1755ndash1771

Nilsson D-E 1994 Eyes as optical alarm systems in fanworms and ark clams Phil Trans R Soc Lond B 346195ndash212

Nilsson D-E 1996 Eye ancestry old genes for new eyes CurrBiol 6 39ndash42

Okada Y K 1927 Note sur lrsquoontogenie de Carybdea rastoniiHaacke Bull Biologique de la France 61 241ndash249

Pearse J S amp Pearse V B 1978 Vision in cubomedusanjellyfishes Science 199 458

Salvini-Plawen L V amp Mayr E 1977 On the evolution ofphotoreceptors and eyes Evol Biol 10 207ndash263

Singla C L 1974 Ocelli of Hydromedusae Cell Tissue Res149 413ndash429

Svane I amp Dolmer P 1995 Perception of light at settlementa comparative study of two invertebrate larvae a scyphozoanplanula and a simple ascidian tadpole J Exp Mar BiolEcol 187 51ndash61

Vandermeulen J H 1974 Studies on reef coral II Fine struc-ture of planktonic planula larva of Pocillopora damicornis withemphasis on the aboral epidermis Mar Biol 27 239ndash249

Widersten B 1968 On the morphology and development insome cnidarian larvae Zoologiska Bidrag fraEcirc n Uppsala 37139ndash182