Gecko Vision—Retinal Organization, Foveae and Implications for Binocular Vision

Home , Gonatodes, Phelsuma, Phelsuma dubia, Quedenfeldtia, Sphaerodactylus, Uroplatus phantasticus

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

Vision Research 41 (2001) 2043–2056 www.elsevier.com/locate/visres

Gecko vision—retinal organization, foveae and implications for binocular vision

Beate Ro¨ll*

Lehrstuhl fu¨r Tierphysiologie, Fakulta¨t fu¨r Biologie, Ruhr-Uni6ersita¨t Bochum, D-44780 Bochum, Germany

Received 11 August 2000; received in revised form 16 February 2001

Abstract

Geckos comprise both nocturnal and diurnal genera, and between these categories there are several transitions. As their retinae have deﬁnitely to be classiﬁed as pure cone retinae, they provide an especially attractive model for comparison of organization and regional specializations adapted to very different photic environments. While the visual cells themselves show clear adaptations to nocturnal or diurnal lifestyles, the overall retinal organization is more related to that of diurnal vertebrates. Nocturnal geckos have lost any foveae of their diurnal ancestors, but they have retained a low convergence ratio and a high visual cell density. To enhance visual sensitivity, they exploit binocular — but not necessarily stereoscopic — vision. Diurnal species have retained binocular vision. Most diurnal species have developed new foveae, which are consequently located not in the central but in the temporal region of the retina. © 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Geckos; Retinal organization; Binocular vision; Foveae; Evolution

1. Introduction of the retinal neurons (Rodieck, 1973). Typically, these retinae are provided with regional specializations, e.g. The basic structure of vertebrate retinae comprises visual streaks, areae or foveae, which show higher three cellular and three neural layers. From sclerad to densities of visual and/or ganglion cells compared with vitread the nuclei of the visual cells, which form the the unspecialized regions. outer nuclear layer, are followed by the outer plexi- Most lizards are strictly diurnal. Their retinae usually form, inner nuclear, inner plexiform, ganglion cell and lack rods and are generally characterized by centrally optic nerve fiber layers. However, both the visual cells located foveae. Pure-cone retinae appear to be common and the other retinal neurons tend to vary in size, in lizards like iguanids, chameleonids, agamids, scin- shape, number and arrangement among vertebrate cids, lacertids, anguids, pygopodids and varanids (De- groups. These variations reflect habits and habitats of twiler & Laurens, 1920; Verrier, 1933; Underwood, the different groups. Most striking are retinal adapta- 1951, 1957, 1970; Anh, 1968, 1969; Crescitelli, 1972; tions to nocturnality and diurnality. Young, 1977; Armengol, Prada, Ambrosiani, & Genis- Visual cells of vertebrates are usually categorized into Galvez, 1988; Ro¨ll, 1998; Ro¨ll & Horn, 1999). Their rods and cones, which operate maximally at low and foveae are convexiclivate in eyes of chameleonids, high light intensities, respectively. Accordingly, retinae iguanids and agamids or more concaviclivate or shal- of strictly nocturnal vertebrates are dominated by rods. low in eyes of lacertids, scincids and varanids (Detwiler In contrast, retinae of diurnal vertebrates have high & Laurens, 1920; Verrier, 1933; Walls, 1942; Vilter, numbers of cones, while the population of rods is small 1949; Williams, Colley, Anderson, Farber, & Fisher, or completely absent. Furthermore, retinae of diurnal 1986; Lanuza, Martıńez-Marcos, Font, & Martıńez- animals are characterized by thick inner nuclear and Garcıá, 1996; Ro¨ll, 1998). The eyes of anoline iguanids inner plexiform layers due to the connectivity patterns are unique among reptiles in that they are bifoveate with both a convexiclivate central and a shallow tempo- * Tel.: +49-234-3224327; fax: +49-234-3214185. ral fovea (Makaretz & Levine, 1980; Fite & Lister, E-mail address: [email protected] (B. Ro¨ll). 1981).

However, the majority of the Gekkonidae, the sec- species and the ultrastructure of foveal visual cells is ond largest extant lizard family, is nocturnal. Geckos unknown. In this study, retinae of different diurnal and are small to moderate-sized, agile lizards occurring nocturnal geckos are investigated with emphasis on worldwide in tropical and subtropical regions. Only 15 retinal organization and occurrence and structure of genera out of approximately 90 are diurnal, and be- foveae. tween the categories ‘nocturnal’ and ‘diurnal’ there are several transitions. With regard to their evolution, the Gekkonidae are unique among lizards. According to 2. Materials and methods Walls (1934, 1942), nocturnal geckos are supposed to have descended from primarily diurnal lizard ancestors 2.1. Materials with pure cone-retinae possessing coloured oil droplets. Thus, the rod-like visual cells of extant nocturnal Animals were housed in glass terraria under a 12 h geckos have transmuted from cones. However, they light/dark cycle; temperatures ranged from 21–23°Cat generally lack oil droplets (Walls, 1942; Underwood, night to 28–34°C during the day. The geckos were fed 1951, 1970; Ro¨ll, 2000a). Regarding nocturnality as a a variety of insects and their larvae; drinking water was derived character, Walls designated geckos as ‘secon- enriched with calcium, phosphate and vitamins. Several darily’ nocturnal. Several genera have reverted from species were reared in the laboratory; of a few species nocturnal to diurnal habits again, designated as ‘tertiar- only one or two specimens have been available. ily’ diurnal by Walls. Their visual cells have undergone Retinae of the following species were examined: (1) a second transmutation from superficially rod-like vi- diurnal species (Gonatodes albogularis, G. 6ittatus, sual cells back to cones, most of which also lack oil Lygodactylus picturatus, L. chobiensis, Phelsuma an- droplets. However, the visual cells of both nocturnal damanensis, P. dubia, Pristurus rupestris, Quedenfeldtia and diurnal geckos exhibit characteristics of cones at all trachyblepharus, Rhoptropus barnardi ); crepusculo-diur- levels of their ultrastructure (Ro¨ll, 2000b). The bio- nal species (Sphaerodactylus elegans, S. glaucus); noc- chemical compositions of gekkonid eye lenses and ge- turno-diurnal species (Phelsuma guentheri ); (2) netics of the visual pigments strongly support this nocturnal species (Paroedura pictus, Phyllopezus polli- notion. The lenses of diurnal geckos contain a specific caris, Stenodactylus sthenodactylus, Tropiocolotes nat- ultraviolet filter consisting of vitamin A2 bound to tereri, Uroplatus phantasticus) and diurno-nocturnal i-crystallin, which is identical with the cellular retinol- species (Ailuronyx seychellensis, Lepidodactylus lugu- binding protein type I (Ro¨ll, Amons, & de Jong, 1996; bris). The experiments were performed in compliance Ro¨ll & Schwemer, 1999). The visual pigments both of with the current laws of the Federal Republic of nocturnal and diurnal geckos show greatest similarities Germany. to those of cone opsins (Kojima, Okano, Fukada, Shichida, Yoshizawa, & Ebrey, 1992; Taniguchi, 2.2. Electron microscopy Hisatomi, Yoshida, & Tokunaga, 1999). Geckos apparently have a trichromatic visual system with absorption Light- and/or dark-adapted (minimum of 4 h) ani- maxima in the violet, blue and green wavelength range mals were anaesthetized by chilling to 4°C, killed by (e.g. Crescitelli, 1972; Ro¨ll, 1996a; Loew, Govardoskii, decapitation and enucleated. The eyes were either fixed Ro¨hlich, & Sze´l, 1996). whole or after removal of cornea, iris, lens and vitreous. Geckos are a distinct phylogenetic group of lizards Both whole eyes and eye cups were marked by small depending mainly on their visual sense for prey capture, incisions at the dorsal and nasal retinal margins for regardless of their activity periods. As their retinae are reconstruction of the orientation of the preparations. pure cone retinae, both scotopic and photopic vision in Samples were fixed for 2–4 h in 2% paraformaldehyde, geckos is performed by one basic visual cell type. Thus, 2.5% glutaraldehyde in phosphate buffered saline at pH their retinae provide an especially attractive model for 7.2. Some of the specimens were fixed for 4 h in 4% comparison of organization and regional specializations paraformaldehyde, 3% glutaraldehyde, 6% acrolein, adapted to very different photic environments. 10% dimethylsulphoxide at pH 7.2 in 0.1 M cacodylate Up to now, regional specializations typical for reti- buffer. After rinsing, the preparations were postfixed in nae of predatory vertebrates have been mentioned in 2% osmium tetroxide for 2 h and dehydrated through a the eyes of the neotropical diurnal gekkonid genera graded ethanol series (range 30–100%). The samples Gonatodes and Sphaerodactylus and in one representa- were passed through propyleneoxide or acetone and tive of the Madagascan genus Phelsuma: retinae of embedded in Epon or Spurr’s epoxy resin, respectively. these genera are provided with foveae (Underwood, Eye cups were carefully oriented before polymerization 1951; Tansley, 1964). The fovea of the Phelsuma species so that longitudinal or tangentional sections through is described as being shallow and poorly developed. the fovea could be obtained. Serial semithin sections Both the morphology of the foveae of other diurnal were stained with toluidene blue and examined and B. Ro¨ll / Vision Research 41 (2001) 2043–2056 2045 photographed with a ZEISS axioplane microscope. Ul- 3.1. Extrafo6eal retinal organization trathin sections were double-stained with uranyl acetate and lead citrate and examined with a ZEISS EM 109 Thicknesses of the different retinal layers of the transmission electron microscope at 80 kV. extrafoveal regions of gekkonid retinae are summarized in Table 1. Thicknesses of the visual cell layers differ 2.3. Measurements considerably between nocturnal and diurnal geckos and vary from 21mm to 117 mm. This difference is mainly Thicknesses of the retinal layers were determined on due to the significantly different lengths of the photore- light and electron micrographs of longitudinal sections ceptor outer segments (Fig. 1; Table 1). of whole retinae. Measurements of lengths and diame- A cell type which traverses through the whole retina ters of photoreceptor outer segments are based on from the inner limiting membrane to the vitreal parts of electron micrographs and are quoted as mean and the photoreceptor inner segments is the neuroglial standard deviations. On longitudinal sections, only vi- Mu¨ller cell. Their small nuclei with evenly distributed sual cells which have been cut centrally through chromatin are found in the vitreal part of the inner paraboloid, ellipsoid and outer segments have been nuclear layer, generally between the cell bodies of the chosen for measurements. However, it cannot be ex- amacrine cells. Their cytoplasm is granulated and cluded that some outer segments have not been cut darker than that of the other retinal cells (Fig. 1a–d). along their entire lengths. Thus, the average lengths of Strands of microtubules and filaments are usually seen the outer segments may be underestimated. The vol- in Mu¨ller cells; mitochondria accumulate in the scleral umes of the more conical outer segments of visual cells cytoplasmic regions closely beneath the outer limiting of diurnal geckos were calculated using the formula for membrane. Endfeet of Mu¨ller cells of all retinal regions y 2 circular cones (V= /3·r ·h). Data were not corrected are devoid of mitochondria. for slight shrinkage during embedding. Visual cell The outer nuclear layer is 5–12 mm thick and consists counts and density calculations were done on cross generally of one row of visual cell nuclei; occasionally sections of the different retinal regions. The conver- two visual cell nuclei are lying one upon the other (Fig. gence ratio (number of receptor cells to ganglion cells) 1a–d, Table 1). Typically, the visual cell nuclei are was estimated by counting cell nuclei on longitudinal circular or oval in shape and lie closely beneath the sections. outer limiting membrane or sometimes penetrate it. Vitread to the visual cell nuclei displaced neurons of the inner nuclear layer are found in all retinal regions. 3. Results These are considerably denser stained than those of the visual cells and resemble nuclei of the bipolar cells The retinae of all geckos investigated are composed of the visual cell layer (excluding the nuclei), two (displaced bipolar cells) (Fig. 1e–f). Displaced bipolar plexiform layers, two nuclear layers, a ganglion cell cells are commonly found in non-mammalian retinae layer and a nerve fiber layer, as is typical for vertebrates (Rodieck, 1973). In retinae of Pristurus rupestris the (Fig. 1). In the retinae of most diurnal genera, four displaced neurons have lobed nuclei. different regions can be distinguished: peripheral, ex- In some sections, the displaced neurons have delicate trafoveal, parafoveal and foveal region. In the retinae processes, which point sclerad but do not reach the of nocturnal and diurno-nocturnal genera, the differen- outer limiting membrane. These processes could be tiation is restricted to peripheral and extrafoveal re- called Landolt’s clubs. However, the processes of bipo- gions. The foveal region is defined as the region of the lar cells of geckos differ from those of other lower foveal pit. In retinae of those species which do not have vertebrates, e.g. of fishes (Locket, 1970, 1973), by the a distinct pit, it is defined as the region of the highest absence of ciliary structures and mitochondria. In the visual cell density. The parafoveal region is an annular retinae of all species investigated, the ratio of displaced region surrounding the fovea, in which either the thick- neurons to visual cell nuclei ranges from about 0.3 to nesses of the inner nuclear and the inner plexiform 0.9 depending on the retinal region. Generally, the layer increase to a maximum or where the visual cell frequencies of displaced neurons decrease towards the densities are lower than in the fovea, but significantly periphery. higher than in the extrafoveal region. The extrafoveal The outer plexiform layer is relatively thin and region is defined as the region beween the parafovea ranges between 5 mmand17mm (Table 1). It consists of and the periphery. Because of the position of the fovea a mixture of processes of horizontal and bipolar cells in gekkonid eyes, the extrafoveal region between the and the cone pedicles. Extrafoveally, the latter are fovea and the temporal region of the ora serrata is very arranged in a single row along the distal margin of the small; here, parafoveal and extrafoveal regions are not outer plexiform layer, isolated from each other by differentiated. Mu¨ller cell extensions. 2046 B. Ro¨ll / Vision Research 41 (2001) 2043–2056

Fig. 1. Retinal organization of geckos. (a–d) Longitudinal semithin sections of retinae of Paroedura pictus (a), Ailuronyx seychellensis (b); Phelsuma dubia (c); Quedenfeldtia trachyblepharus (d). White arrowheads indicate outer limiting membrane and black arrowheads Mu¨ller cell fibers. C, choroid; PE, pigment epithelium; VCL, visual cell layer; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer; NFL, optic nerve fiber layer. (e–g) Longitudinal ultrathin sections of retinae. (e) Inner nuclear region of Rhoptropella ocellata illustrating localization and morphology of horizontal (H), bipolar (B) and amacrine (A) cell nuclei. White arrows in (e, f) point to displaced bipolar cells. (f) Retinal layers of the parafoveal region of Lygodactylus picturatus. Black arrowheads indicate inner fibers that have been cut transversially. (g) Enlarged view of displaced bipolar cells of Gonatodes albogularis. B, bipolar cell; DB, displaced bipolar cell; S, synaptic pedicle. All other abbreviations as in (a–d). Scale bars: 50 mm(a–d), 10 mm (e, f), 5 mm (g). B. Ro¨ll / Vision Research 41 (2001) 2043–2056 2047 m% 5 6.4 8 7.8 4 3.9 6 4.6 1512 10.8 8.6 m bre layer. fi membrane; ILM, inner 4.9 4.5 7.1 7.6 7.9 % 4.6 m m yer; NFL, nerve m% m m% m m% 9 m 17 % 6.2 6.4 5 5.7 20 25.5 38 48.4 6 6.36.8 11 7.8 38 28.3 57 42.4 6 8.6 12 8.6 46 32.8 48 34.3 10 5.9 7 6.9 35 34.3 41 40.2 5 6.9 10 9.4 33 32.5 40 39.4 8 5.3 9 6.9 44 33.6 59 45.0 6 m 5 6 5 7 ONL OPL INL IPL GCL NFL 12 m 100 100 100 100 100 12 8.3 7 4.8 49 33.8 61 42.1 9 6.2 7 4.8 % m 102 139 131 145 m m 22 59 69 135 100 9 21 102 100 7 25 31 78 100 26 117 162 100 10 102 103 100 7 m a Activity OLM to ILM N 11.0 50 31.0 63 39.0 9 5.3 13 8.0 N5079 N 8.3 31 29.8 44 42.4 7 6.8 6 5.8 D-N D D-N D N-D 6.4 7 9.0 21 26.9 36 46.2 4 5.2 5 6.4 D D D, diurnal; N-D, nocturno-diurnal; D-N, diurno-nocturnal; N, nocturnal; VCL, visual cell layer (tips of outer segments to OLM); OLM, outer limiting — Activity a Species VCL Table 1 Thicknesses of retinal layers of various geckos Uroplatus phantasticus Tropiocolotes nattereri Ailuronyx seychellensis Paroedura pictus Quedenfeldtia trachyblepharus Lepidodactylus lugubris Phelsuma guentheri Pristurus rupestris Phelsuma dubia Lygodactylus picturatus limiting membrane; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell la 2048 B. Ro¨ll / Vision Research 41 (2001) 2043–2056

The inner nuclear layer is 20–50 mm thick. In noctur- nasally (Fig. 2e). It contains both myelinated and un- nal and diurno-nocturnal genera and in the nocturno- myelinated axons of strikingly variable diameters (Fig. diurnal species Phelsuma guentheri it consists of about 2d). Occasionally, mitochondria are present in the ax- five to six rows. In retinae of strictly diurnal species, the ons. In all species, chromatophores with melanin gran- inner nuclear layer is composed of about seven to eight ules are found within the nerve. These closely resemble rows, in Rhoptropella ocellata of up to 10 rows of nuclei those of the conus papillaris originating on top of the of horizontal, bipolar and amacrine cells interspersed optic nerve (Fig. 2e). Blood vessels, which supply the between the radial fibers of the Mu¨ller cells (Fig. 1a–d). conus, are also found within the optic nerve. In con- Different types of neurons can be distinguished on the trast, the retinal layers of all species investigated are basis of location, nuclear staining, their sizes and completely devoid of blood vessels. shapes (Fig. 1e). The horizontal cells appear poorly Measurements of thicknesses of retinal layers were stained and lie in an irregular row at the scleral side of based on micrographs of longitudinal semithin and/or the inner nuclear layer. Unambiguous identification of longitudinal ultrathin sections of extrafoveal regions. the horizontal cells is somewhat difficult as they are As some longitudinal sections may be slightly oblique, more horizontally extended. The tightly packed nuclei the absolute values may be overestimated. Thus, for of the bipolar cells are smaller in size and densely comparison the thicknesses of the different retinal lay- stained. They are predominantly located in the distal ers were calculated as percentages of the retinal region inner nuclear layer and from three to five rows like the between the outer and inner limiting membranes (Table amacrine cells, which have larger and less stained nu- 1). The region between the limiting membranes was clei, much more cytoplasm and are arranged at the chosen because differences in thicknesses of the whole vitreal border of the inner nuclear layer. Small bundles retina are largely due to the different lengths of the of axons lie between the retinal neurons; the axons are photoreceptor outer segments (Fig. 1; Table 1). separated from each other by Mu¨ller cell extensions The absolute thicknesses from the outer to the inner (Fig. 2a). limiting membranes of the retinae of the various species The inner plexiform layer consists of a complex range from 78 mmto162mm (Table 1). The relative meshwork of processes of the neurons situated in the values for outer nuclear, outer plexiform, inner plexi- inner nuclear layer and in the ganglion cell layer. Its form and ganglion cell layers vary between species; thickness ranges from 36 mmto61mm between species however, they do not differ between retinae of noctur- and does not depend on lifestyle (Table 1). nal and diurnal species. The corresponding values for The ganglion cell layer is approximately 4–10 mm the inner nuclear layers change only slightly comparing thick and contains a single cell layer of relatively large retinae of species with different activity period. The somata, separated by Mu¨ller cell endfeet (Fig. 1a–d, values for nocturnal, diurno-nocturnal and nocturno- 2c). In the retinae of some gekkonid species, ganglion diurnal range from 26 to 33%, whereas those for strictly cell nuclei are occasionally displaced to the inner plexi- diurnal species range from 33% to 34%. This slight form layer. As these are extremely rare, they are re- difference is not significant. garded as misplaced ganglion cells. The ganglion cells have relative abundant cytoplasm containing rough en- 3.2. Occurrence of fo6eae doplasmic reticulum (Nissl substance), mitochondria and Golgi complexes. Their nuclei have prominent In order to localize foveal regions and to examine nucleoli. The morphology of the somata — without their ultrastructure, whole eyes or eye cup preparations dendritic staining — does not permit a clear differenti- of 14 species were investigated with light and electron ation of ganglion cell types and/or displaced amacrine microscopy on serial semithin and ultrathin sections cells. The convergence ratio (visual cell nuclei to nuclei (Paroedura pictus, Phyllopezus pollicaris, Tropiocolotes in the ganglion cell layer) reaches a value of approxi- nattereri, Uroplatus phantasticus, Lepidodactylus lugu- mately 1 in retinae of all species independent of their bris, Gonatodes albogularis, G. 6ittatus, Lygodactylus activity periods and remains constant in the extrafoveal picturatus, L. chobiensis, Phelsuma andamanensis, P. regions. dubia, Rhoptropus barnardi, Sphaerodactylus elegans, S. The unmyelinated axons of the ganglion cells form glaucus). the nerve fiber layer. They often run in bundles which Eyes of all nocturnal and diurno-nocturnal species are surrounded by branches of Mu¨ller cell endfeet (Fig. completely lack foveae. In contrast, eyes of diurnal 2c). Thickness of the nerve fiber layer is dependent on species — except R. barnardi — possess a single fovea. the retinal regions; the layer is thickest in the central Although being strictly diurnal, R. barnardi apparently retina near the optic nerve, and thinnest in the periph- lacks foveae. eral regions of the retina. In all species provided with a fovea, this is located in In all species investigated the optic nerve arises in the the temporal region of the peripheral retina, near the central part of the retina, slightly displaced ventro- level of the eye equator (Fig. 2e, f). B. Ro¨ll / Vision Research 41 (2001) 2043–2056 2049

Fig. 2. (a–d) Ultrastructure of retinal layers. Longitudinal ultrathin sections of the inner nuclear layer of Tropiocolotes nattereri (a), of the inner plexiform layer of Phelsuma guentheri (b), of ganglion and nerve ﬁber layer of Lepidodactylus lugubris (c) and transversal ultrathin section of the optic nerve of Sphaerodactylus glaucus (d). B, bipolar cell; C, chromatophore; G, ganglion cell; M, Mu¨ller cell. Black arrowheads in (c) point to the inner limiting membrane, white arrowheads in (c) mark bundles of small axons surrounded by Mu¨ller cell endfeet. Stars in (d) mark myelinated axons. (e, f) Position of optic nerve and of fovea in the eye of Lygodactylus picturatus. Horizontal semithin sections of a whole eye. The spectacle has been removed. C, cornea; CP, conus papillaris; I, iris; L, lens; ON, optic nerve; R, retina; n, nasal; t, temporal. The arrowhead points to the temporal fovea. Scale bars: 5 mm(a–d), 500 mm (e, f). 2050 B. Ro¨ll / Vision Research 41 (2001) 2043–2056

3.3. Structure of fo6eae consists foveally of three to maximally four rows of nuclei. The pedicles of the visual cells comprise up to two The foveae of the genera Lygodactylus and Gonatodes rows and are displaced laterally. The neurons of the inner are well-developed with a clearly defined pit, which is nuclear layer are banked in up to ten rows; the number relatively deep, but not funnel-shaped as in eyes of some of rows decreases towards the periphery where this layer other lizards or of birds. Thus, the foveae appear more consists of about two rows. The nuclei of the ganglion concaviclivate in shape according to Walls’ (1942) defin- cells are arranged in vertical stacks (consisting of three ition. The foveal pits of Lygodactylus and Gonatodes are to four nuclei) which are separated from each other by not free of inner retinal layers. Due to the high visual cell Mu¨ller cell endfeet. In comparison to the structure of the density, the outer nuclear layer of the pit comprises four foveae of the genera Gonatodes, Lygodactylus and to five nuclei layers in Lygodactylus species and three to Sphaerodactylus, there is no pronounced displacement of four rows in Gonatodes species. In the parafoveal region the inner retinal layers. of Lygodactylus and Gonatodes the outer nuclear layer Phelsuma guentheri differs from its congeners in being consists of up to five to six and eight to nine rows of nocturno-diurnal. Unfortunately, there was not sufficient nuclei, respectively. Here, the visual cell nuclei become retinal material of P. guentheri for examination of considerably elongated and spindle shaped. They are not regional specializations. directly juxtaposed to the outer limiting membrane, but In all diurnal species provided with a fovea, displaced lie deeper, thus leaving a narrow zone between the outer neurons from the inner nuclear layer are also present in limiting membrane and the nuclei occupied by the outer the foveal and parafoveal regions; their nuclei lie in one fibers. The synaptic terminals (pedicles) are aligned in row, occasionally in two rows, between the transversally one row foveally and in two to three rows parafoveally. cut outer fibers and the pedicles (Fig. 1f). They are laterally displaced so that in longitudinal sections the inner fibers are usually cut transversially 3.4. Visual cell densities (Fig. 1f). The main parts of the inner nuclear and the In the retinae of all diurnal species — except for inner plexiform layer are displaced sideways resulting in Rhoptropus barnardi — the visual cells become slender a pit with steep slopes (Fig. 3a). In the foveal pit the and less tapered from the periphery towards the fovea, ganglion cells are double-layered, whereas in the so that the highest visual cell density is reached in the parafoveal region the ganglion cell layer comprises four foveal pit (Table 2). The visual cell densities in the to five and three to four layers in retinae of Lygodactylus extrafoveal regions of diurnal geckos — including the and Gonatodes, respectively. Here and also in the afoveate R. barnardi — range from about 40000 cells/ parafovea, the nuclei of the ganglion cells lie one upon mm2 to about 51000 cells/mm2. They increase towards the other resulting in vertical layers which are isolated the fovea about four to fivefold reaching maxima of from each other by Mu¨ller cell endfeet (Fig. 3a). about 160000–250000 cells/mm2 depending on the Parafoveally, the ganglion cell layer consists of five to six genus. In contrast, the visual cell densities in retinae of (Lygodactylus) and eight to nine (Gonatodes) rows of nocturnal species are markedly lower and range from nuclei. about 10000 cells/mm2 to 36000 cells/mm2. This number The foveae of Sphaerodactylus species are also well-de- is almost constant in the whole retina except for the veloped and concaviclivate, but with a broader and more region of the ora serrata, where the visual cell density is rounded pit than in Lygodactylus and Gonatodes (Fig. reduced. 3b). In the foveal pit the outer nuclear layer is arranged in up to 6 layers and is thickest in this region. The cone 3.5. Fo6eal cones pedicles can be seen as a single bright row within the darker stained nuclei (Fig. 3b). As in foveae of Lygo- In retinae of diurnal foveate species the lengths of their dactylus and of Gonatodes, the main parts of the inner photoreceptor outer segments increase significantly from nuclear and the inner plexiform layers are displaced the extrafoveal region towards the foveal pit, where they laterally. Foveally, the inner nuclear layer consists of two are at least twice as long as those outside the fovea (Table to three rows of nuclei, and parafoveally, of up to 12 3). As the outer segments of the cones are oriented in the rows. Within the foveal pit the ganglion cells are ar- direction of the foveal pit, their actual maximum lengths ranged in an ill-defined row, while in the parafoveal may be even longer. At the same time, their basal region they are distributed in two to three rows. diameters decrease to about half, resulting in a slight The foveae of the strictly diurnal Phelsuma species are decrease of the average volumes of outer segments from not so well marked: they are only shallow depressions in extrafoveal regions towards the fovea. The diameters of the retina (Fig. 4a). A foveal pit comparable to that of the paraboloids — generally the widest parts of reptilian the other genera was not detected. The fovea can be visual cells — decrease by about 0.5- to 0.8-fold from identified by the increased visual cell density and hence the extrafoveal to the foveal region, depending on the by the thickening of the outer nuclear layer, which cone type (Table 3). B. Ro¨ll / Vision Research 41 (2001) 2043–2056 2051

Fig. 3. Concaviclivate foveae of Lygodactylus picturatus and Sphaerodactylus glaucus. (a, b) Longitudinal semithin sections of foveae of L. picturatus (a) and S. glaucus (b). Arrowheads in (a) mark the outer limiting membrane; arrows in (b) indicate the row of pedicles. VCL, visual cell layer; PE, pigment epithelium; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. (c, d) Tangentional ultrathin sections of visual cells at the level of their outer segments in extrafoveal (c) and foveal (d) regions of S. glaucus. Triangles in (c–f) indicate members of double visual cells; white stars in (d) mark members of a triplet visual cell. (e, f) Longitudinal ultrathin sections of visual cells of extrafoveal (e) and foveal (f) regions of L. picturatus. Scale bars: 50 mm (a, b), 1 mm (c, d), 5 mm (e, f).

The foveal cones of all species investigated lack oil droplets — if present — are found in the minor droplets, although the visual cells in the extrafoveal members of type B double visual cells only (Fig. 4b, c). regions of retinae of Gonatodes and Phelsuma possess Here, they are situated in the most scleral part of the oil droplets (Figs. 3e, f and 4b–f). Extrafoveally, oil ellipsoids. The double visual cell of type B consists of 2052 B. Ro¨ll / Vision Research 41 (2001) 2043–2056 two cells of unequal sizes, a major (larger) member 4. Discussion which is rather voluminous and possesses a prominent paraboloid, and a minor (smaller) member, which usu- The inner layers of gecko retinae are organized ac- ally contains an oil droplet. Additionally, the gecko cording to the basic plan of vertebrates. Except for the retina contains a second type of double visual cell (type visual cell layers, the retinal organization does not C), which consists of two members of approximately differ markedly between diurnal geckos on the one equal sizes. hand and nocturnal geckos on the other hand. The The foveae of all genera contain single and double whole thickness of the extrafoveal retina between the cells; in the fovea of Sphaerodactylus even triplet cells outer and the inner limiting membranes (inner retina) have been found (Figs. 3d, f and 4e). However, in varies between the different species, but there is no clear foveal regions, the minor members of double cones of relationship between thickness and activity period. For type B deﬁnitely lack an oil droplet (Fig. 4c, e). In example, the inner retinal layers of the strictly diurnal parafoveal regions, there are double cells of type B with genera Pristurus and Quedenfeldtia are relatively thin and without oil droplets. like those of the strictly nocturnal Paroedura. Also, the

Fig. 4. Shallow fovea of Phelsuma andamanensis (ultrathin sections). Longitudinal sections of the fovea (a) and of different retinal regions (b–f): peripheral region (b); region between the fovea and the temporal ora serrata (c); retinal region between fovea and optic nerve (d); parafoveal region (e); foveal region (f). Arrowheads mark outer limiting membrane; white arrows indicate oil droplets. The major members of double cones type B lying in front of the minor ones have been cut at the levels of their paraboloids and the vitreal part of their ellipsoids (asterisks in b, c, e). White circles in (c) and (e) mark minor members of type B double visual cells lacking an oil droplet. Scale bars: 50 mm (a), 5 mm(b–f). B. Ro¨ll / Vision Research 41 (2001) 2043–2056 2053

Table 2 retinae of strictly diurnal vertebrates (Walls, 1942). −2 a Visual cell densities (mm ) in eyes of various geckos Nocturnal animals tend to have a considerably thinner SpeciesActivity Retinal region inner nuclear layer. Thus, it is remarkable that a con- spicuous difference in thickness of the inner nuclear Extrafoveal Foveal layer of retinae of nocturnal and diurnal geckos has not been found. Gonatodes albogularis D 51 000 250 000 A different result — concerning the thickness of Lygodactylus picturatus D48 000 240 000 Phelsuma andamanensis D 40 000 160 000 retinal layers — was reported for representatives of the Rhoptropus barnardi D 40 000 – nocturnal gekkonid genera Homopholis, Pachydactylus Sphaerodactylus glaucus C-D 45 000 220 000 and Tarentola, as compared with two other species of Ailuronyx seychellensis D-N 10 000 – the diurnal genus Phelsuma (P. guimbeaui, P. madagas- Lepidodactylus lugubris D-N 36 000 – cariensis) (Dieterich, Dieterich, & Hildebrand, 1976). Paroedura pictus N 12 000 – The absolute values of the thickness of the region Phyllopezus pollicaris N 20 000 – ‘between the inner limiting membrane and the inner Stenodactylus sthenodactylus N 13 000 – nuclear layer’ of these nocturnal geckos differ signifi- Tropiocolotes nattereri N 30 000 – Uroplatus phantasticus N 12 000 – cantly from the corresponding retinal region of the diurnal Phelsuma species. However, when the data are a Activity — D, diurnal; C-D, crepusculo-diurnal; N-D, nocturno- transformed to relative values, they correspond to the diurnal; D-N, diurno-nocturnal; N, nocturnal. values found in the present study. The scleral part of the inner nuclear layer of thicknesses of inner retinae are not markedly influenced gekkonid retinae is occupied by an ill-defined row of by absolute eye size, as that of the inner retinae of horizontal cells. In vertebrate retinae, the number of Paroedura and Uroplatus with similar eye sizes differ horizontal cell rows varies with the type and proportion considerably. Furthermore, the inner retina of the small of photoreceptors present. In retinae with an equal species Lepidodactylus lugubris is thicker than that of number of rods and cones there are generally two rows. the larger species. Both size and number of horizontal cells increase with The most sclerad layer beneath the outer limiting the proportion of rods: in retinae with a high ratio of membrane is the outer nuclear layer formed collectively rods there may be four rows of horizontal cells (Ali & by the visual cell nuclei. In all gekkonid species investi- Klyne, 1985). In the pure cone retina of the iguanid gated, this layer consists generally of one row; this is lizard Anolis carolinensis, the horizontal cells are irregu- characteristic for cone retinae. larly arranged in more or less one row (Sherry & The inner nuclear layer of geckos differs in number Ulshafer, 1992). Thus, the distribution of horizontal of rows of nuclei between the main categories diurnal cells in all gecko retinae is similar to that in retinae of and nocturnal. In retinae of nocturnal, diurno-noctur- diurnal animals. nal and even nocturno-diurnal species, it consists of In all gekkonid species investigated here, the inner about five to six rows. In retinae of strictly diurnal plexiform layer contributes significantly to the thickness species, the inner nuclear layer is usually composed of of the inner retina. There is no obvious relationship about seven to eight and maximally of up to ten rows between activity period and thickness of the inner plex- of nuclei. iform retina. The presence of a relatively thick inner nuclear layer Counts of cell nuclei of the ganglion cell layer and consisting of eight to nine cells is characteristic for the visual cell layer in extrafoveal regions result in a

Table 3 Measurements of photoreceptor elements of geckos

Gonatodes Lygodactylus picturatus Phelsuma andamanensis albogularis

Average length of outer segments Extrafoveal mm 9.191.6; n=17 6.891.3; n=21 6.592.2; n=39 Foveal mm 15.592.5; n=16 11.592.0; n=17 18.692.9; n=8 Average diameter of outer segment basesExtrafovealmm 2.190.3; n=8 1.990.2; n=18 2.090.3; n=15 Foveal mm 1.490.2; n=11 1.190.2; n=27 1.190.2; n=9 Average volume of outer segmentsExtrafoveal mm3 11 6 7 Foveal mm3 84 6 Average diameter of paraboloids: Extrafoveal mm 5.890.6; n=24 5.590.7; n=11 6.990.6; n=17 Major member of type B double coneFoveal mm 3.390.5; n=7 3.490.3; n=6 3.590.2; n=6 Average diameter of paraboloids: Extrafoveal mm 2.490.3; n=15 2.590.4; n=14 4.890.5; n=20 Double cones of type C and single cones Foveal mm 2.090.2; n=7 2.090.4; n=5 2.690.5; n=8 2054 B. Ro¨ll / Vision Research 41 (2001) 2043–2056 convergence ratio (visual cells to ganglion cells) of while retaining a low convergence ratio and a relatively approximately 1 in retinae of all gekkonid species. A high visual cell density. That means that the develop- similar ratio — even with a surplus of ganglion cells — ment of binocularity allowed the geckos to become was calculated for the diurnal species ‘Phelsuma inun- nocturnal retaining the general retinal structures of guis’ (synonym of P. cepediana) and the nocturnal their diurnal ancestors. species Gekko gecko (Pedler & Tilly, 1964). Binocular vision in nocturnal geckos does not imply Probably, this ratio is underestimated as the retinal necessarily that they use stereoscopic vision. The ability ganglion cell layer of all vertebrates contains displaced to estimate distance is of the same importance for amacrine cells. However, without staining of their den- nocturnal geckos as for diurnal predatory lizards, drites and axons a clear differentiation between ‘true’ which generally have laterally placed eyes and centrally ganglion cells and displaced amacrine cells is generally located foveae. However, for estimation of distance the impossible. In the retina of the diurnal agamid lizard latter can use monocular clues only, e.g. movement Ctenophorus ornatus, displaced amacrine cells represent parallax or accommodation values, while nocturnal approximately 10% of the ganglion cell numbers (Beaz- geckos may be able to use binocular clues, e.g. asymme- ley, Sherad, Tennant, Starac, & Dunlop, 1997). If a try of the images on the two retinae which depend on similar percentage is assumed in gekkonid retinae, the the relation of the object to the sagittal plane of the convergence ratio increases only slightly. animal (Locket, 1977). Additionally, in the retina of the nocturnal gecko Chameleons are an exception among lizards. They h have a single, centrally located fovea as most diurnal Hemidactylus frenatus one type of ganglion cell ( a) has a displaced soma within the inner nuclear layer (Cook lizards. But because of the extreme mobility of their & Noden, 1998). However, as this type represents only eyes they can fixate binocularly. Chameleons use ac- 0.38% of the estimated total number of ganglion cells, it commodation cues when measuring the distance to can be neglected with regard to the convergence ratio. their prey. As in geckos, there is no hint of the usage of Apparently, in the retina of both nocturnal and stereoscopic vision (Harkness, 1977; Ott, Schaeffel, & Kirmse, 1998). diurnal geckos one or at most a few photoreceptors are Eyes of both the strictly nocturnal geckos species linked in average to one ganglion cell. Generally, in (Paroedura pictus, Phyllopezus pollicaris, Tropiocolotes retinae of nocturnal vertebrates many rods converge on nattereri, Uroplatus phantasticus) and the diurno-noc- one ganglion cell, increasing visual sensitivity by high turnal species (Ailuronyx seychellensis, Lepidodactylus spatial summation which is reflected in poorly devel- lugubris) investigated here completely lack foveae. All oped inner layers of the retina. In contrast, with respect other nocturnal species investigated up to now are also to the structural features of the retina of nocturnal afoveate (Tansley, 1959, 1964; Underwood, 1970). Ob- geckos and their low convergence ratio a high degree of viously, they have lost the (presumably centrally lo- summation cannot be expected. cated) fovea of their ancestors, as a fovea with a high Another mechanism of spatial summation is photore- visual cell density allows high visual acuity in high but ceptor coupling. It generally improves the signal-to- not in low ambient light conditions. noise ratio even under a wider variety of conditions When starting to occupy diurnal niches again, geckos than receptor/ganglion cell convergence as the photore- retained binocular vision from their nocturnal ances- ceptor output synapses are highly non-linear (Rowe, tors. Additionally, most diurnal and crepusculo-diurnal 1991). Among reptiles, photoreceptor coupling has species — except R. barnardi — developed a new fovea been observed between cones of turtles (Kolb & Jones, with high visual cell densities located temporally in the 1985). However, it is not known if this mechanism peripheral retina. In laterally positioned eyes which can occurs in gekkonid retinae (Peterson, 1991). be directed frontally to some extent, this location corre- While the eyes of nocturnal geckos are placed later- sponds to a region which represents a point in front of ally, they are highly movable either independently of the gecko, within the binocular field. Once a prey item each other or coordinated under a transparent spectacle has been detected, both the head and the eyes are which covers the eyes of most geckos. Monocular vi- aligned toward it and the target is fixated binocularly as sion allows the gecko a panoramic view for detection of in nocturnal geckos. Thus, diurnal geckos use the same prey objects or predators. Furthermore, the eyes can be binocular clues — symmetry and relative positions of directed forward to some extent resulting in a binocular the images on the retinae — for distance estimation. It field of view and subsequently in binocular fixation of can be assumed that images of objects within attack prey. The advantages of binocular vision may become range are projected right onto the foveae. decisive in low ambient light intensities. These include In eyes of the diurnal R. barnardi no definite fovea increased signal-to-noise ratio and improved contrast could be detected. Geckos of the genus Rhoptropus detection (Lythgoe, 1979; Pettigrew, 1991). Thus, it differ from other diurnal genera in having visual cells seems possible that nocturnal geckos achieve higher which are intermediate between the typical cones of visual sensitivity by using a binocular field of view diurnal geckos and those of nocturnal geckos (Ro¨ll, B. Ro¨ll / Vision Research 41 (2001) 2043–2056 2055

2000b). Additionally, R. barnardi — like its congeners of potentially harmful shortwave radiation. In eyes of — is able to reduce the amount of light reaching the diurnal geckos this function has been taken over by the retina by closing its iris, while the aperture of the pupils yellow lens, the colour of which is caused by retinoids of the other diurnal species is virtually fixed (Ro¨ll, and/or carotenoids bound to lens crystallins (Ro¨ll & 1996b). Thus, geckos of the genus Rhoptropus appear to Schwemer, 1999; Ro¨ll, 2000c). be just on the verge of evolving visual adaptations to Geckos have developed unexpected adaptations for strict diurnality. visually guided prey capture both in dim and in bright In the genera investigated, foveation reaches its light on the basis of pure cone retinae. Besides morpho- highest development in Gonatodes, Lygodactylus and logical modifications of the visual cells and biochemical Sphaerodactylus. Their foveae have a clearly defined pit changes in the lens crystallin composition, the use of and show displacement of the inner retinal layers, re- their movable eyes for binocular vision and the devel- sembling the concaviclivate foveae of lacertids and scin- opment of temporally located foveae in diurnal species cids. Towards the fovea the cones gradually become may in fact have contributed largely to the successful more and more slender and cylindrical in form reaching radiation of this group. a maximal density of about 220000–250000 cells/mm2. This is somewhat lower than the visual cell density in the central foveae of the scincid lizard Cryptoblepharus Acknowledgements boutonii and of the iguanid lizard Anolis carolinensis, where densities of up to 290000 visual cells/mm2 were I am grateful to Professor Dr K. Hamdorf for critical observed (Makaretz & Levine, 1980; Ro¨ll, 1998). reading and discussion of the manuscript. For provid- The shallow fovea of the genus Phelsuma is less ing specimens of some species I would like to thank T. specialized. It is only a depression in the retina, and Kowalski, Dr G. Osadnik, E. Schro¨der and K. Stu¨rzen- there is no pronounced displacement of the inner retinal hofecker. Special thanks are due to S. Tonge, Wildlife layers. Still, it is characterized by an increasing visual Preservation Trust, Channel Islands, for permission to cell density reaching 160000 cells/mm2, which is, how- use the eyes of an aged female of Phelsuma guentheri ever, considerably lower than that of the foveae of the for ultrastructural and biochemical studies. other diurnal geckos and of the central foveae of Cryp- toblepharus and Anolis. In shape, the fovea of Phelsuma References resembles that of the temporal fovea of anoline lizards. However, the visual cell density of the temporal (sec- Ali, M. A., & Klyne, M. A. (1985). Phylogeny and functional ond) fovea of Anolis carolinensis reaches only 16000 morphology of vertebrate retina. In R. Fleischer, & A. Fleischer, cells/mm2; this comparatively low value is obviously Vertebrate morphology, fortschritte der zoologie, vol. 30 (pp. 633– due to the large diameters of the photoreceptor inner 648). Stuttgart: Gustav Fischer Verlag. Anh, J. N. H. (1968). Ultrastructure des rećepteurs visuels de la re´tine segments (Makaretz & Levine, 1980), and it is greatly de Lacerta viridis. Bulletin of the Association of Anatomy, 53, surpassed even by the extrafoveal visual cell density of 1247–1259. all diurnal geckos and indeed of some nocturnal Anh, J. N. H. (1969). Ultrastructure des rećepteurs visuels chez les geckos. verte´bre´s. Archi6es of Ophthalmology, 29, 795–822. The extrafoveal visual cells of the genera Gonatodes Armengol, J. A., Prada, F., Ambrosiani, J., & Genis-Galvez, J. M. (1988). The photoreceptors of the chameleon retina (Chameleo and Phelsuma possess oil droplets which are found in chameleo). A golgi study. Journal fu¨r Hirnforschung, 29, 403–409. the minor members of type B double visual cells in the Beazley, L. D., Sherad, P. W., Tennant, M., Starac, D., & Dunlop, S. scleral part of the ellipsoids. The colourless oil droplets A. (1997). Optic nerve regenerates but does not restore topo- in gekkonid visual cells are supposed to function as graphic projections in the lizard Ctenophorus ornatus. Journal of 6 microlenses focussing light on the outer segments, pro- Comparati e Neurology, 377, 105–120. Cook, J. E., & Noden, A. J. (1998). Somatic and dendritic mosaics vided that the diameters of the oil droplets are larger formed by large ganglion cells in the retina of the common house than those of the outer segment bases (Ro¨ll, 2000b). gecko (Hemidactylus frenatus). Brain, Beha6ior and E6olution, 51, Though within the fovea both single visual cells and 263–283. double visual cells are present, the double cells lack oil Crescitelli, F. (1972). The visual cells and the visual pigments of the 6 droplets. It is assumed that in the densely packed foveal vertebrate eye. In H. J. A. Dartnall, Photochemistry of ision, handbook of sensory physiology,vol.VII/I (pp. 245–363). New cones the necessary size ratio cannot be maintained and York: Springer–Verlag. therefore any light funneling properties decrease signifi- Detwiler, S. R., & Laurens, H. (1920). Studies on the retina. The cantly. This inturn may have led to the loss of foveal oil structure of the retina of Phrynosoma cornutum. Journal of Com- droplets. parati6e Neurology, 32, 347–356. On the other hand, single and double cones in the Dieterich, C. E., Dieterich, H. J., & Hildebrand, R. (1976). Compar- ative electron-microscopic studies on the conus papillaris and its fovea of the scincid lizard C. boutonii retain their oil relationship to the retina in night and day active geckos. Albrecht droplets (Ro¨ll, 1998). However, here the main function 6on Graefes Archi6 fu¨r klinische und experimentelle Ophthalmolo- of these yellow oil droplets apparently is the absorption gie, 200, 279–292. 2056 B. Ro¨ll / Vision Research 41 (2001) 2043–2056

Fite, K. V., & Lister, B. C. (1981). Bifoveal vision in Anolis lizards. Ro¨ll, B. (2000a). Characterization of retinal oil droplets in diurnal Brain, Beha6ior and E6olution, 19, 144–154. geckos (Reptilia, Gekkonidae). Journal of Experimental Zoology, Harkness, L. (1977). Chameleons use accommodation cues to judge 287, 467–476. distance. Nature, 267, 346–351. Ro¨ll, B. (2000b). Gecko vision—visual cells, evolution, and ecologi- Kojima, D., Okano, T., Fukada, Y., Shichida, Y., Yoshizawa, T., & cal constraints. Journal of Neurocytology, 29, 471–484. Ebrey, E. B. (1992). Cone visual pigments are present in gecko Ro¨ll, B. (2000c). Carotenoid and retinoid—two pigments in a gecko rod cells. Proceedings of the National Academy of Science of the eye lens. Comparati6e Biochemistry and Physiology A, 125, 105– USA, 92, 2780–2784. 112. Kolb, H., & Jones, J. (1985). Electron microscopy of golgi-impreg- Ro¨ll, B., & Horn, H. G. (1999). The structure of the eye of the nated photoreceptors reveals connections between red and green monitor lizard Varanus griseus caspius (Reptilia, Varanidae). In: cones in the turtle retina. Journal of Neurophysiology, 54, 1445– H. G. Horn & W. Bo¨hme (Eds.), Ad6ances in monitor research II, 6 1458. Mertensiella, ol. 11, Rheinbach: Deutsche Gesellschaft fu¨r Her- Lanuza, E., Mart´ınez-Marcos, A., Font, C., & Mart´ınez-Garc´ıa, S. petologie und Terrarienkunden¸ pp. 291–306. i (1996). Catecholaminergic interplexiform cells in the retina of Ro¨ll, B., & Schwemer, J. (1999). -Crystallin and vitamin A2-isomers 6 lizards. Vision Research, 36, 1349–1355. in lenses of diurnal geckos. Journal of Comparati e Physiology A, Locket, N. A. (1970). Landolt’s club in the retina of the African 185,51–58. lungfish, Protopterus aethiopicus, Heckel. Vision Research, 10, Ro¨ll, B., Amons, R., & de Jong, W. W. (1996). Vitamin A2 bound to 299–306. cellular retinol-binding protein as ultraviolet filter in the eye lens Locket, N. A. (1973). Retinal structure in Latimeria chalumnae. of the gecko Lygodactylus picturatus. Journal of Biological Chem- Philosphical Transactions of the Royal Society of London. Series B, istry, 271, 10437–10440. 266, 493–521. Rowe, M. H. (1991). Functional organization of the retina. In B. Dreher, & S. R. Robinson, Neuranatomy of the 6isual pathways Locket, N. A. (1977). Adaptations to deep-sea environment. In F. and their de6elopment. Vision and 6isual dysfunction, vol. 3 (pp. Crescitelli, The 6isual system in 6ertebrates, handbook of sensory 1–68). London: MacMillan Press Ltd. physiology,vol.VII/5(pp.67–192). Berlin: Springer Verlag. Sherry, D. M., & Ulshafer, R. J. (1992). Neurotransmitter-specific Loew, E. R., Govardoskii, V. I., Ro¨hlich, P., & Sze´l, A´ . (1996). identification and characterization of neurons in the all-cone Microspectrophotometric and immunocytochemical identification retina of Anolis carolinensis, I: Gamma-aminobutyric acid. Visual of ultraviolet photoreceptors in geckos. Visual Neuroscience, 13, Neuroscience, 8, 515–529. 247–256. Taniguchi, Y., Hisatomi, O., Yoshida, M., & Tokunaga, F. (1999). Lythgoe, J. N. (1979). The ecology of 6ision. Oxford: Clarendon Press. Evolution of visual pigments in geckos. FEBS Letters, 445,36– Makaretz, M., & Levine, R. L. (1980). A light microscopic study of 40. the bifoveate retina in the lizard Anolis carolinensis: general Tansley, K. (1959). The retina of two nocturnal geckos Hemidactylus observations and convergence ratios. Vision Research, 20, 679– turcicus and Tarentola mauritanica. Pflu¨gers Archi6es, 268, 213– 686. 220. Ott, M., Schaeffel, F., & Kirmse, W. (1998). Binocular vision and Tansley, K. (1964). The gecko retina. Vision Research, 4,33–37. acommodation in prey-catching chameleons. Journal of Compara- Underwood, G. (1951). Reptilian retinas. Nature, 167, 183–185. 6 ti e Physiology A, 182, 319–330. Underwood, G. (1957). On lizards of the family Pygopodidae. Journal Pedler, C., & Tilly, R. (1964). The nature of the gecko visual cell. A of Morphology, 100, 207–268. light and electron microscopic study. Vision Research, 4, 499–510. Underwood, G. (1970). The eye. In C. Gans, Morphology B, biology Peterson, E. H. (1991). Retinal structure. In C. C. Gans, Neurology C of the Reptilia, vol. 2 (pp. 1–97). New York: Academic Press. sensorimotor integration, biology of the reptilia, vol. 17 (pp. 1– Verrier, M. L. (1933). Recherches sur la vision des reptiles. E´ tude 135). London: University of Chicago Press Ltd. compareé de la morphologie des cellules visuelles et du pouvoir Pettigrew, J. D. (1991). Evolution of binocular vision. In J. R. se´parateur de la re´tine. Bulletin Biologique de France et Belgique, 6 6 Cronley-Dillon, & R. L. Gregory, E olution of the eye and isual 67, 350–370. system, 6ision and 6isual dysfunction, vol. 2 (pp. 271–282). Lon- Vilter, V. (1949). Nouvelle conception de relations neuronales dans la don: MacMillan Press Ltd. foveá. Comptes Rendus de la Societe´ Biologique, 143,784–785. Rodieck, R. W. (1973). The 6ertebrate retina. Principles of structure Walls, G. (1934). The reptilian retina. I. A new concept of visual cell and function. San Francisco: W.H. Freeman and Company. evolution. American Journal of Ophthalmology, 17, 892–915. Ro¨ll, B. (1996a). Sehfarbstoffe nacht- und tagaktiver Geckos (Rep- Walls, G. (1942). The 6ertebrate eye and its adapti6e radiation. tilia). Verhandlungen der Deutschen Zool Gesellschaft, 89.1, 291. Bloomfield Hills: Cranbrook Institute of Science. Ro¨ll, B. (1996b). Rhoptropus barnardi Hewitt. Sauria Suppl., 20, Williams, D. S., Colley, N. J., Anderson, D. H., Farber, D. B., & 427–430. Fisher, S. K. (1986). In vitro maintenance of a pure-cone retina. Ro¨ll, B. (1998). Extrafoveal and foveal visual cells in the retina of In6estigati6e Ophthalmology and Visual Sciences, 27, 666–673. Bouton’s skink (Reptilia). In N. Elsner, & R. Wehner, New Young, R. W. (1977). The daily rhythm of shedding and degradation neuroethology on the mo6e. Proceedings of 26th Go¨ttingen Neurobi- of cone outer segment membranes in the lizard retina. Journal of ology Conference, vol. II (p. 435). Stuttgart: Thieme. Ultrastructure Research, 61, 172–185.