Eye and Vision in the SubterraneanRodent Cururo (Spalacopus cyanus,
Octodontidae)
LEO PEICHL,1 ANDRES E. CHAVEZ,2 ADRIAN OCAMPO,2 WILSON MENA,2
FRANCISCO BOZINOVIC,3AND ADRIAN G. PALACIOS2*
1Max Planck Institute for Brain Research, D-60528 Frankfurt, Germany2Centro de Neurociencia de Valparaıso, Departamento de Fisiologıa, Facultad de Ciencias,
Universidad de Valparaıso, Valparaıso, Chile3Centro de Estudios Avanzados en Ecologıa & Biodiversidad, Departamento de Ecologıa,
Facultad de Ciencias Biologicas, Pontificia Universidad Catolica de Chile,Santiago 651-3677, Chile
ABSTRACTSubterranean mammals are generally considered to have reduced eyes and apparent
Indexing terms: cone photoreceptors; rod photoreceptors; spectral sensitivity; ultraviolet vision
Subterranean species have evolved independentlyamong marsupials, insectivores, and, most notably,among rodents (summaries: Nevo, 1999; Lacey et al.,2000). They are rarely, if ever, exposed to light. Manysubterranean mammals show a number of parallel sen-sory adaptations that include reduced eyes and poor orabsent vision (summaries: Burda et al., 1990; Nevo, 1999;Lacey et al., 2000; Kimchi and Terkel, 2002). This has ledto the general assumption that the visual systems of sub-terranean mammals have undergone extensive conver-gent regression. The model species to support such claimsis the muroid blind mole-rat Spalax ehrenbergi, which hasatrophied subcutaneous eyes and retinal deficits (Sanyalet al., 1990; Cooper et al., 1993; Cernuda-Cernuda et al.,2002; David-Grey et al., 2002). On the other hand, recent
studies on subterranean African mole-rats (bathyergid ro-dents) have reported small but well-developed eyes with
Grant sponsor: Fondo Nacional de Desarrollo Cientıfico y Tecnologico (toA.G.P.); Grant sponsor: Fondo de Investigacion Avanzada en Areas Priori-tarias; Grant number: 1501-0001 (to F.B.); Grant sponsor: the University ofValparaiso Graduate Fellowship (to A.E.C.).
*Correspondence to: Adrian G. Palacios, Valparaiso Center for Neuro-science, University of Valparaıso, Faculty of Science, P.O. Box 5030, Val-paraıso, Chile. E-mail: [email protected]
Received 4 August 2004; Revised 6 November 2004; Accepted 2 Decem-ber 2004
DOI 10.1002/cne.20491Published online in Wiley InterScience (www.interscience.wiley.com).
THE JOURNAL OF COMPARATIVE NEUROLOGY 486:197–208 (2005)
photoreceptor properties that suggest visual capabilitiesat higher light levels (Cernuda-Cernuda et al., 2003;Peichl et al., 2004). These findings call for a more differ-entiating view of visual regression in subterranean spe-cies.
The South American rodent family Octodontidae (Cavi-omorpha; phylogeny, see Gallardo and Kirsch, 2001) offersan ideal “experiment of nature” to study correlations be-tween habitat conditions and the visual systems. Octodon-tid species show a large diversity of life modes and habitatuse (Contreras et al., 1987; Honeycutt et al., 2003). Spala-copus cyanus (common name “cururo” or “coruro”) is asubterranean species that inhabits semiarid environ-ments in north-central Chile (Torres-Mura and Contreras,1998). Cururo colonies live in extensive tunnel systemsused for foraging, food storage, and social interactions(Begall and Gallardo, 2000). To test for phylogenetic iner-tia, the cururo’s retinal traits can be compared with thoseof surface-dwelling octodontid species, such as the diurnaldegu, Octodon degus (Chavez et al., 2003; Jacobs et al.,2003), and the nocturnal O. lunatus and O. bridgesi. Tocheck for convergent adaptations to an underground life, acomparison can be made with subterranean species be-longing to different rodent families.
Mammals have a “duplex” retina with rod photorecep-tors for scotopic (night) vision and cone photoreceptors forphotopic (daylight) vision and color vision. The proportionof rods and cones varies considerably across species andcorrelates with the predominant visual lifestyle—diurnal,crepuscular, or nocturnal (review: Ahnelt and Kolb, 2000).The typical mammalian retina contains two spectral conetypes, a majority of middle-to-long-wavelength-sensitive(L-) cones and a minority of short-wavelength-sensitive(S-) cones. Depending on the species, the L-cones havetheir peak sensitivity in the green to yellow part of thespectrum (�max range about 500–560 nm), and the S-conesin the blue to ultraviolet part (�max range about 360–450nm) (review: Jacobs, 1993). The photoreceptors of subter-ranean rodents show some intriguing deviations from thistypical mammalian pattern (see Discussion).
To investigate the degree of visual specialization in S.cyanus, we measured the spectral transmission propertiesof the lens and cornea, assessed the presence, proportion,and retinal distribution of the different photoreceptortypes by immunocytochemical labeling of the visual pig-ments (opsins), and identified the spectral properties ofthese photoreceptors by scotopic and photopic electroreti-nogram (ERG) measurements. In a search for behaviorallyrelevant visual cues in the environment, we measured the
spectral reflectance of cururo fur and urine, because oftheir potential relevance in the case of the degu (Chavez etal., 2003).
MATERIALS AND METHODS
Animals
Six adult male S. cyanus were captured in the wild andkept in a standard animal facility at the University ofValparaıso. Animals were maintained at 18–20°C on a12-hour light / 12-hour dark cycle, with food and water adlibitum before experiments. Animal care and experimen-tal procedures complied with international regulations(NIH Publications No 80-23).
Retinal histology and opsinimmunocytochemistry
For retinal histology, two animals were euthanized byan intraperitoneal lethal dose of ketamine and xylazine.The orientation of the eyes was marked by a ventralperforation of the cornea, the eyes were rapidly enucle-ated, placed in 4% paraformaldehyde in 0.1 M phosphatebuffer (PB, pH 7.4) overnight, and transferred to PB. Afterrecording the eye dimensions, the eyes were completelyopened and the retinae carefully dissected. Pieces of retinawere used for transverse 14-�m cryostat sections; otherpieces from defined retinal regions and two whole retinaewere immunoreacted free-floating.
Immunocytochemistry followed previously describedprotocols (Peichl et al., 2000, 2004). Briefly, adhering re-mains of the retinal pigment epithelium were bleachedand the tissue was preincubated for 1 hour in PB with0.5% Triton X-100 and 10% normal goat serum (NGS), ornormal donkey serum (NDS) when secondary antibodiesfrom donkey were used. Incubation in the primaryantibody/antiserum solution was for 3–4 days (free-floating tissue) or overnight (sections on the slide) at roomtemperature. Rod opsin was detected with the mousemonoclonal antibody rho4D2 (dilution 1:500), kindly pro-vided by R.S. Molday (Hicks and Molday, 1986). TheL-cone opsin was detected with the rabbit antiserum JH492 (dilution 1:2,000–1:4,000), the S-cone opsin with therabbit antiserum JH 455 (dilution 1:5,000) or the goatantiserum sc-14363 (dilution 1:100). The rabbit antiserawere kindly provided by J. Nathans (Wang et al., 1992;Chiu and Nathans, 1994); the goat antiserum was pur-chased from Santa Cruz Biotechnology (Heidelberg, Ger-many). Binding sites of the primary antibodies were de-tected by indirect immunofluorescence, with a 1-hourincubation in Alexa goat antimouse IgG, Alexa goat anti-rabbit IgG, or Alexa donkey antigoat IgG, respectively(dilution 1:500–1:1,000; Molecular Probes, Eugene, OR).Double-labeling for L-cone and S-cone opsin was per-formed by incubating the tissue in a mixture of antiseraJH 492 and sc-14363. In this case, visualization was byincubation with a mixture of Alexa 488-conjugated donkeyantigoat IgG and Cy3-conjugated donkey antirabbit IgG.In the retinae used for the topographical analysis of conedensities, incubation with the primary antisera was fol-lowed by an overnight incubation in goat antirabbit IgG,an overnight incubation in rabbit peroxidase-antiperoxidase (PAP) complex, and visualization with3,3�-diaminobenzidine (DAB) and H2O2. All of the abovevariations of the staining protocol gave consistent results.
Whole retinae and retinal pieces were flattened ontoslides with the photoreceptor side up. All tissue was cov-erslipped with an aqueous mounting medium (AquaPoly/Mount, Polysciences, Warrington, PA). Tissue was ana-lyzed with a Zeiss Axiophot microscope equipped withepifluorescence. Micrographs were taken with a CCD cam-era and the Metaview software (Visitron Systems, Pu-chheim, Germany). The images were adjusted for bright-ness and contrast using Adobe PhotoShop 7.0 (San Jose,CA). Densities of L- and S-cones were assessed in thePAP-DAB-reacted retinae. At sample fields across the ret-inae, cones were counted with a �63 oil immersion objec-tive. At some positions in these retinae rods could also becounted using Nomarski optics and a �100 oil immersionobjective. Photoreceptor densities were not corrected forshrinkage, because shrinkage was negligible in the tissuemounted with the aqueous medium.
A piece of central retina was processed for semithintransverse sections. It was dehydrated with ethanol andpropylene oxide and embedded in Epon 812. With an ul-tramicrotome, 1-�m sections were cut perpendicular tothe retinal surface, collected on slides, and stained withtoluidine blue.
Controls for specificity of antisera
The specificity and the characterization of the mouseantibody rho4D2 and the rabbit antisera JH 492 and JH455 have been described (Hicks and Molday, 1986;Wang et al., 1992; Chiu and Nathans, 1994). They havebeen used in a range of mammals by various laborato-ries and have reliably labeled the respective photore-ceptor types. The goat antiserum sc-14363 has beenraised against a 20 amino acid mapping located withinthe N-terminal 50 amino acids of the human S-opsinand recognizes a single band of appropriate molecularweight in a Western blot analysis of mouse eye tissueextract (Santa Cruz Biotechnology: electronic catalogand pers. commun.).
A first control was to omit the primary antibodies fromthe incubation solution. With all antibodies, this resultedin a complete absence of photoreceptor labeling with ei-ther immunofluorescence or the PAP/DAB reaction. Toassess the specificity of sc-14363 in the cururo retina, wedid preabsorption experiments with the blocking peptidesc-14363P (Santa Cruz Biotechnology) on cryostat sec-tions. A working solution of the primary antiserum wasprepared as above, a 5-fold amount (by weight) of thepreabsorption peptide was added, the mixture was incu-bated for 2 hours at room temperature, then centrifugedfor 6 minutes at 13,000 rpm. The solution was applied tothe sections and all further steps for immunofluorescencewere as described above. After preabsorption, we observedno labeling of any retinal structures. Finally, we diddouble-labeling experiments with sc-14363 and JH 455 toassess whether both antisera exactly labeled the samestructures, i.e., the S-cone outer segments, on cryostatsections. There was complete congruence of both labels,showing that both antisera are reliable markers of cururoS-cones.
Spectral eye transmission and spectralreflectance measurements
Animals were euthanized by an overdose of halothane,decapitated, and the eyes were removed. The isolated lensand cornea were immersed in mineral oil, centered in a
plastic holder with a central aperture of 1.8 mm, and im-mersed in a quartz block transmissive for visible and UVlight. Lens and cornea transmission were measured with acalibrated spectrometer (Thermospectronic, Rochester, NY)at wavelengths from 260–700 nm in 20-nm intervals.
The spectral reflectance of the fur of ventral and dorsalbody parts, and of urine scent marks, was measured fol-lowing the protocol of Chavez et al. (2003). The reflectancecontrast of urine was calculated as the reflectance differ-ence between drops of fresh urine and of distilled water onbrown cardboard.
Electroretinogram (ERG)
To assess the retinal spectral sensitivity, ERGs wererecorded under both scotopic and photopic conditions infour animals. The animals were anesthetized with anintraperitoneal injection of ketamine (120 mg/kg) and xy-lazine (4 mg/kg). A few drops of a local cornea anesthetic(1% lidocaine) and of 1% atropine for pupil dilation wereapplied to the eye before a contact (Ag/AgCl) electrode wasplaced on the cornea. Body temperature was maintainedat 32°C by means of a regulated thermal bed. Most of theexperiments were done early in the afternoon. The proce-dures, the optical system, and the ERG system have beendescribed previously (Chavez et al., 2003). In brief, theoptical system consisted of a stabilized power supply witha quartz lamp (250 W, ORIEL), a monochromator (1200lines/mm grating, ORIEL, 20 nm half-bandwidth), andlong-pass filters (Schott UG11, RG540, RG630) to elimi-nate short-wavelength stray light from the monochroma-tor. An electronic shutter (Uniblitz, Vincent Associates,Rochester, NY) set the flash duration and an opticalquartz wedge (0–4 OD) attenuated the incident number ofphotons. An adapting bright yellow background was ob-tained by adding a long-pass filter (Schott GG435,�0.00001% transmission below 420 nm) to a fiber-opticilluminator (150 W), producing a background illuminationof 500 �W/cm2 per steradian at the cornea measured witha calibrated photocell positioned at the level of the cornea(Optometry S370, UDT Instruments, Baltimore, MD). TheERG was recorded while increasing the number of photonsper flash (with 1–2-second intervals between the flashes)at preset wavelengths. For the photopic (cone) spectralsensitivity measurements, a 20-Hz stimulus was used toreduce a possible rod contribution. The spectral sensitivitywas measured as S� � rpeak / I; where I is the flash photonflux, and rpeak is the b-wave maximal peak average re-sponse (n � 25–50 trials), for dim flashes in the 340–700nm range. The wavelength of maximum sensitivity (�max)was calculated by fitting the experimental data for the440–560 nm range to a template (vertically scaled andshifted in the �max to the best fit) that describes the actionspectrum of vertebrate photoreceptors (Stavenga et al.,1993; Lamb, 1995; Palacios et al., 1998, Govardoskii et al.,2000). For the intensity-response function individual datawere normalized using an � value obtained by r / rmax � i/ i � �, where i is the incident photon number at thecornea and � is the half-saturating response.
RESULTS
Ocular anatomy, lens and cornea spectraltransmission
The cururo eye is rather spherical, with an equato-rial diameter of 5.5–5.8 mm and an axial length of
5.6 – 6.0 mm. This is somewhat smaller than the eyes ofsurface-dwelling octodontids (about 7.5 mm in the diur-nal O. degus and about 8.2 mm in the nocturnal O.bridgesi; our unpubl. obs.), but is well within the rangeof typical rodent eye dimensions (Howland et al., 2004).The colorless lens is 3.6 –3.8 mm in diameter and 3.0 –3.2 mm thick. The pupil is vertically elongated (oval) inthe semiclosed state. Figure 1 shows the lens and cor-nea transmission for two eyes of different individuals.The cururo lens and cornea exhibit a 60 –75% transmis-sion between 380 nm and 740 nm, dropping to 30% at320 nm. Hence, the eye’s optical apparatus is transmis-sive also in the near-UV range (�400 nm).
Retinal histology and photoreceptor types
The overall organization and the layering of the retinawas assessed in transverse sections. Toluidine blue-stained semithin sections reveal that all retinal nuclearand synaptic layers from the photoreceptors to the gan-glion cells show the typical mammalian pattern (Fig. 2A).The thickness of the outer nuclear layer (ONL; photore-ceptor somata) indicates a predominance of rods. Theouter segments of the photoreceptors appear uniform (Fig.2B). In Epon-embedded tissue, outer segments are 18–24�m in length and 1.6–2.0 �m in cross-sectional diameter(not corrected for shrinkage). In the less shrunken flat-mounted tissue, inner and outer segments have somewhatlarger diameters of 2.0–2.8 �m (Fig. 2C).
The presence of rods and spectral cone types was as-sessed by immunocytochemistry with opsin-specific anti-bodies. Rod opsin labeling confirms the presence of a dom-
inant rod population. Their intensely immunoreactiveouter segments form a dense layer (Fig. 2D) and theirmore faintly labeled somata are densely packed across theONL (Fig. 2E). L-opsin labeling reveals a substantial pop-ulation of L-cones (Fig. 2F,H). The label is confined to thelong outer segments. This is also true for a sparser popu-lation of S-cones, revealed by labeling with the S-opsinantisera sc-14363 (Fig. 2G) or JH 455 (Fig. 2I).
A number of mammals possess substantial populationsof cones coexpressing the L- and the S-opsin, predomi-nantly in the ventral retina (“dual pigment” cones; review:Szel et al., 2000). We therefore specifically checked for thepresence of dual pigment cones in the cururo by doubleimmunofluorescence labeling with the L-opsin and theS-opsin, done in a broad strip of retina running from thedorsal to the ventral retinal margin. We found no coex-pression of the two opsins in any cones, each cone waseither an L-cone or an S-cone. This is illustrated in Figure3 for a field in ventral periphery, the most likely region forthe occurrence of dual pigment cones.
The topographic distributions and the proportions ofL-cones and S-cones were assessed in flat-mounted wholeretinae that were single-labeled with the respective anti-sera (PAP-DAB visualization). The staining is shown inFigure 2H,I; the density maps are given in Figure 4. Thecururo L-cone distribution (Fig. 4A) has a shallow centro-peripheral gradient with higher densities of 21,000–26,000/mm2 in central retina. L-cone density is particu-larly low in the dorsal periphery, declining to around10,000/mm2. The S-cone density (Fig. 4B) is overall lower.Superimposed on a shallow centro-peripheral gradient, ithas a pronounced ventro-dorsal gradient, with higher den-sities of up to 6,500/mm2 in central and ventral retina, anddensities below 2,000/mm2 in the dorsal third of the ret-ina. S-cone densities are lowest at the dorsal margin (100–400/mm2). Superposition of the L- and S-cone maps gavean estimate of the local S-cone proportion across the retina(expressed as S-cone percentage of all cones; Fig. 4C). Inthe dorsal third of the retina, there are 1–7% S-cones, inthe middle third 6–21% S-cones, and in the ventral third10–21% S-cones, with one exceptional value of 26% at theventral margin. For example, the field from ventral pe-riphery illustrated in Figure 3 has 18% S-cones. Total conedensities (added L- and S-cone densities) range from29,000–31,000/mm2 in central and ventral retina, to15,000–24,000/mm2 in ventral, temporal, and nasal pe-riphery, and 10,000–15,000/mm2 in dorsal periphery (notillustrated).
Rod densities could be assessed regionally by Nomarskioptics in the above flat-mounted retinae (compare Fig.2C), resulting in a coarse rod density map (Fig. 4D). Roddensities range from 160,000/mm2 to 280,000/mm2, withlarge local scatter but no obvious topographic gradient.The average across all sample fields is 224,000 rods/mm2
(30,000/mm2 SD; n � 38). Comparing the rod and conedensities, roughly 10% of the photoreceptors are cones,with local variations introduced by the rod density scatterand the cone density gradient (average 9.5% 1.9% SD,range 5–12%, n � 32 sample fields). The higher coneproportions (10–12%) are found in central and ventralretina, the lower ones (5–8%) in peripheral retina.
Photoreceptor spectral sensitivity
To assess the change of the eye’s sensitivity during thecourse of dark adaptation, light-adapted animals were
Fig. 1. Average spectral transmission of the lens (open circles) andthe cornea (filled circles) in two individuals (mean SD). Both struc-tures are highly transmissive down to near-UV wavelengths (�400nm).
Fig. 2. Retinal morphology and immunocytochemical identifica-tion of photoreceptors. A: Toluidine blue-stained transverse semithinsection showing well-organized retinal layers. B: Magnified view ofthe photoreceptor layer. C: On-view of the layer of photoreceptor innerand outer segments in a flattened unstained retina (Nomarski optics).All outer segments have relatively large diameters, rods and conescannot be distinguished in B and C. D,E: Transverse cryostat sectionimmunofluorescence labeled for rod opsin. There is a high density ofrod outer segments (D). An overexposed micrograph of the samesection shows the more faintly labeled rod somata, which are denselypacked across the entire ONL (E). F: Section immunofluorescencelabeled for L-cone opsin, showing a substantial population of L-coneouter segments. G: Section immunofluorescence labeled for S-cone
opsin (antiserum sc-14363), showing a sparse population of S-coneouter segments. H: Micrograph from a flat-mounted retina immuno-labeled for L-cone opsin, showing the outer segments of the denseL-cone population; central retina close to optic nerve head, PAP-DABreaction. I: Micrograph from a flat-mounted retina immunolabeled forS-cone opsin (antiserum JH 455), showing the outer segments of thesparser S-cone population; central retina close to optic nerve head,PAP-DAB reaction. RPE, retinal pigment epithelium; OS and IS,photoreceptor outer and inner segments; ONL, outer nuclear layer;OPL, outer plexiform layer; INL, inner nuclear layer; IPL, innerplexiform layer; GCL, ganglion cell layer. Layer labels in E apply toD–G. Scale bars � 50 �m in A; 10 �m in C (applies to B,C); 50 �m inI (applies to D–I).
201EYE AND VISION IN A SUBTERRANEAN RODENT
placed in the dark and the ERG b-wave amplitude evokedby dim flashes was monitored over time. The sensitivityreaches a fully dark-adapted state after 30 minutes,where it is about 2 log units higher than in the light-adapted state (Fig. 5A). Scotopic sensitivity increases lessthan, e.g., in humans or rats. Nonetheless, it can be inter-preted as the transition from cone- to rod-mediated vision(see Discussion).
We further analyzed the rod and cone mechanisms byscotopic and photopic spectral ERG measurements. Thescotopic sensitivity was measured after a 45-minute darkadaptation period. Figure 5B shows the normalizedscotopic response amplitude function for the b-wave inresponse to flashes of different intensities at a fixed wave-length (500 nm). Figure 5C shows the mean (SD)scotopic spectral sensitivity function, measured as theERG response to 10-ms dim flashes (n � 25) at wave-lengths from 340–640 nm in 20-nm steps. The spectralsensitivity function was corrected for the cornea and lensspectral absorption (Fig. 1). To calculate the �max of theunderlying rod (-band) mechanism we used a visual tem-plate (Lamb, 1995, in the modification of Govardovskii etal., 2000) that was varied in the �max position and verti-cally adjusted to provide a best fit. The estimated �max forthe -band was 500 1.6 nm (n � 4, range 498–501 nm).While we observed small deviations from the templatecurve at longer wavelengths, the -band and �max is quitereliable and corresponds well to a standard mammalianrod (dotted line). To extend the analysis into the short-wavelength (�-band) range, an additional template has tobe used to represent the absorption of the retinal itself. Inour calculation we used Stavenga et al. (1993) log-normalparameters for the �-band part of the spectrum. In Figure5C the continuous curve represents the best fit to theexperimental ERG data, using a Lamb -band templatewith �max � 500 nm and a Stavenga �-band template. The�max of the �-band was estimated to be 338 nm using theequation (�-band �max � 123 � 0.429 �max -band), basedon measurements of isolated photoreceptors from severalvertebrates (Palacios et al., 1998). To obtain the best fit,the �-band amplitude was scaled to 10% of the -band.
For measurement of the photopic spectral sensitivityfunctions, a strong light-adapting background and 20 Hzstimuli were used to isolate the cone responses from po-tential rod contamination. Figure 6A shows the meanphotopic spectral sensitivity function, measured with10-ms dim flashes (n � 50) at wavelengths from 340–620nm in 20-nm steps. The spectral sensitivity function wascorrected for the cornea and lens spectral absorption (seeFig. 1). The -band template that best fits the averagesensitivity function (n � 4) has a �max at about 505 nm,indicating a single L-cone pigment (dashed line in Fig.6A). The individual �max in four measured animals rangedfrom 502–509 nm; the mean �max value is 505 3.9 nm.The experimental values show a secondary sensitivity in-crease at wavelengths below 400 nm, which could eitherrepresent the �-band of the L-cone pigment or the pres-ence of a second, short-wavelength (blue/UV-sensitive)cone mechanism. To address that issue we added a �-bandto the -band template. Like for the rods, we used a 10%amplitude for the �-band, assuming that the cone and rodpigments have similar �-band properties (both originatingin the retinal), and a �-band �max of 340 nm. The dottedline in Figure 6A shows the combined - and �-band curvefor the L-cone pigment. At short wavelengths, it runs
Fig. 3. Micrograph from a flat-mounted retina double immunofluo-rescence labeled for the cone opsins. A: L-opsin labeling with antiserumJH 492 (red fluorescence). B: S-opsin labeling with antiserum sc-14363(green fluorescence). C: Merging of the two images shows that none of thecones are double-labeled, there is no coexpression of the two opsins in anycones. In this field located in ventral peripheral retina, total cone densityis 22,500/mm2, and 18% of the cones are S-cones (compare maps Fig. 4).Scale bar � 50 �m in A (applies to A–C).
202 L. PEICHL ET AL.
below the sensitivity of the photopic ERG, suggesting thepresence of a short-wavelength cone. To model the com-plete photopic spectral function we thus used a linearcombination of two templates (L- cone and S-cone; solidline in Fig 6A). The parameters yielding the best fit werean S-cone -band with �max � 365 nm (near UV) and a16% contribution to overall sensitivity and an L-cone tem-plate with the above parameters and a 84% contribution.The fit strongly suggests the presence of a second conetype in the UV range (fine continuous line). To furtherisolate the putative UV-cone from the L-cone mechanism,a bright yellow adapting background was used. The twographs of Figure 6B show such experiments for two ani-mals. In both cases, with yellow adaptation the sensitivitydrops more in the long-wavelength range than in theshort-wavelength range (�400 nm), arguing for separateshort- and long-wavelength mechanisms.
For the animal reported in Figure 6B (bottom), a stan-dard - and �-band template are added, with the L-cone�max at 502 nm (best fit for this individual; dotted lines)and the S-cone �max at 365 nm (fine continuous line). Thewavelength of 365 nm corresponds well to the S-conemechanism present in several rodents, including Octodondegus (Jacobs et al., 1991; Chavez et al., 2003). The solidcurve represents the upper envelope of the two templates,with an S-cone contribution of 75% and an L-cone contri-bution of only 25%, reflecting the L-cone suppression bythe yellow adaptation. It fits the experimental data well. Asimilar result was obtained for the second animal (notillustrated).
Urine and fur reflectance
In a search for behaviorally relevant visual signals,which may have exerted an adaptive pressure on the
Fig. 4. Topographic distribu-tion of the photoreceptors. L-conedensities (A) and S-cone densities(B) were mapped in whole flat-mounted retinae labeled withL-opsin antiserum JH 492 andS-opsin antiserum JH 455, re-spectively. C: Percentage ofS-cones among the cones was as-sessed by superposition of the topmaps and drawn into the outlineof the L-cone map. D: Densities of(unlabeled) rods were assessed byNomarski optics in the retinastained for S-cones. Each dot rep-resents a sample field, and the dotsize the local cell density. D, dor-sal; V, ventral; T, temporal, N, na-sal. All retinae are shown at thesame magnification. Scale bar � 3mm.
203EYE AND VISION IN A SUBTERRANEAN RODENT
cururo’s spectral sensitivity, we measured the spectralreflectance of cururo fur and urine. The fur has less than3% reflectance between 300 and 700 nm at both ventraland dorsal body regions (Fig. 7A). This confirms the fur’ssubjective blackness, providing no spectrally relevantcues. The spectral reflectance contrast of fresh cururourine has a maximum at about 300 nm and decreasesmonotonically towards longer wavelengths (Fig. 7B).Hence, the highly UV-reflecting urine presents a potentialvisual cue for the cururo (see Discussion). After the urinehas dried up (within 2–3 hours under laboratory condi-tions), the remaining dry matter shows no reflectancecontrast against the substrate. This is a difference fromdry degu urine (Chavez et al., 2003).
DISCUSSION
The present analysis of the cururo’s photoreceptors pro-vides a comprehensive account of the first stage of visualprocessing in this subterranean octodontid species. Thewidely held view is that subterranean mammals generallyhave reduced visual systems as a convergent evolution totheir lightless underground microhabitats (see Introduc-tion). It has been hypothesized that selection may havebeen against the metabolic cost of a useless system andthe risk of eye damage during digging (Burda et al., 1990;Borghi et al., 2002), with the normal-sized eyes of thecururo and of, e.g., the burrowing ground squirrels argu-ing against the latter. From this perspective, some of thecururo’s photoreceptor properties are rather unexpected.In search of the potential phylogenetic constraints andadaptive pressures involved, this discussion will comparethe cururo’s photoreceptor properties to those of surface-dwelling nocturnal and diurnal octodontids, and to thoseof other subterranean rodents.
The cururo eye is well within the size range found insighted rodents (cf. Howland et al., 2004) and does notsuggest a loss of visual function. Our observation of astandard, well-developed optical apparatus and retinasupports the view of an eye suited for image processing.The photoreceptor layer shows no signs of regression, thephotoreceptor outer segments are structurally intact, ofsubstantial length, and of larger diameter (2–2.8 �m) thanthose of, e.g., mouse (�2 �m). This results in a lowerpacking density (average 224,000 rods/mm2 vs. 440,000rods/mm2 in mouse; mouse data: Jeon et al., 1998). Thescotopic spectral sensitivity has a peak near 500 nm, sug-gesting a conventional mammalian rod visual pigment.The scotopic spectral sensitivities of other octodontids arein the same range (Chavez et al., 2003; Jacobs et al.,
Fig. 5. Physiology of the dark-adapted cururo eye. A: Course ofdark adaptation, ERG measurements in two individuals (open andfilled circles). The preadaptation light was turned off at time zero. Thestimuli were 10-ms dim flashes of � � 500 nm. B: Scotopic response-amplitude function of the ERG. Each individual intensity-responsefunction was normalized using its � value (four animals, each repre-sented by one symbol). The dimmest 10-ms flash was 0.46 0.23photons/�m2 incident at the cornea. C: Scotopic spectral sensitivityfunction, measured to 10-ms dim flashes (n � 25) at wavelengths from340–640 nm. The calculated �max using a visual template is 500 1.6nm (dotted line). The solid line represents a template with an -bandwith �max � 500 nm and a �-band with �max � 338 nm and 10%amplitude of the -band. For details, see text.
204 L. PEICHL ET AL.
2003). The increase of light sensitivity by about 2 log unitsafter dark adaptation indicates reasonable scotopic, rod-based vision (Fig. 5A). The sensitivity increase is less than
in rat or mouse, but reminiscent of the situation in Oct-odon degus, where Jacobs et al. (2003) suggest that rodscontribute to the ERG at higher light levels than in othermammals. So this may be an octodontid peculiarity.
The cone proportion of about 10% of the photoreceptors,and cone densities up to 31,000/mm2, are surprisinglyhigh for a rodent inhabiting a low-light environment. Noc-turnal surface-dwelling mammals, including rodents,commonly have much lower cone proportions in the rangeof 0.5–3% (for summaries, see, e. g., Peichl and Moutairou,1998; Ahnelt and Kolb, 2000; Peichl et al., 2000). Mostdiurnal and crepuscular mammals have between 5% and20% cones among their photoreceptors. This places thecururo well within the range of diurnal and crepuscularspecies. Here it is instructive to look at the surface-dwelling octodontids. The diurnal degu (Octodon degus)has a considerably higher proportion of about 30% conesand a density maximum above 55,000 cones/mm2 (Jacobset al., 2003). The nocturnal O. bridgesi and O. lunatushave much lower cone proportions of about 2% and densitymaxima below 10,000 cones/mm2 (our unpubl. obs.). Ob-viously, the phylogenetic plan of the octodontid familyallows a wide range of cone-to-rod ratios in adaptation toa diurnal or nocturnal lifestyle. This suggests that thecururo’s photoreceptor populations are adapted to higherlight levels, and not to scotopic conditions or their lightlesssubterranean habitat.
Both the ERG spectral sensitivity measurements andthe opsin immunocytochemistry reveal two spectral conetypes in S. cyanus: a majority of green-sensitive L-cones(�max near 505 nm), and a minority of ultraviolet-sensitiveS-cones (�max near 365 nm). These cone properties are thesame as in O. degus (L-cone majority with �max near500–507 nm, S-cone minority with �max near 360 nm;Chavez et al., 2003; Jacobs et al., 2003). The presence oftwo spectral cone types is a prerequisite for dichromaticcolor vision (review: Jacobs, 1993). It can be assumed thatcururos have this most common form of mammalian colorvision, provided the necessary retinal and cortical colorprocessing circuits are present. The cornea and lens of thecururo eye are transmissive in the near UV, hence theretina can be reached by light that stimulates the UV-
Fig. 6. Spectral sensitivity of the light-adapted cururo eye. A: Pho-topic spectral sensitivity, measured with 10-ms dim flashes (n � 50)at wavelengths from 340–620 nm. The dimmest flash (� � 500 nm, 10ms) was 14.9 16.8 photons/�m2 incident at the cornea. The calcu-lated �max using a Lamb template was 505 3.9 nm (dashed line). Thedotted line represents a template with an -band with �max � 505 nmand a �-band with �max � 340 nm, scaled to 10% amplitude of the-band, representing an L-cone mechanism. The addition of a secondcone mechanism with �max � 365 nm (UV-sensitive S-cone; fine con-tinuous curve) and a contribution of 16% of the L-cone mechanismgives a better fit with the experimental data (solid curve). See text fordetails. B: Photopic spectral sensitivity with white light adaptation(open circles) and yellow light adaptation (filled circles) in two indi-viduals (top and bottom). In the bottom graph, the dotted lines rep-resent an L-cone template with �max � 502 nm (best fit for thisindividual) and a �-band with �max � 338 nm scaled to 10% amplitudeof the -band. The dashed line is a UV-cone template with �max � 365nm, all vertically adjusted to obtain the best fit by eye. The solid linerepresents the sum of the L-cone (�max � 502 nm, 25% contribution)and the S-cone templates (�max � 365 nm, 75% contribution). It fitsthe data points quite well and indicates the presence of an L-cone anda UV-cone mechanism.
205EYE AND VISION IN A SUBTERRANEAN RODENT
sensitive cones. While in most mammals the S-opsin isblue-sensitive (Jacobs, 1993), several rodents have a UV-sensitive S-opsin (e.g., mouse, rat, gerbil; Jacobs et al.,1991). Actually, UV sensitivity of the S-opsin seems tohave been the ancestral condition in mammals (Hunt etal., 2001). The cururo’s cones exclusively contain eitherthe L-opsin or the S-opsin, there are no “dual pigment”cones. This is the most common mammalian pattern. It isnoteworthy because some rodents possess large numbersof dual pigment cones, among them the guinea pig, whichis a caviomorph like the cururo (summary: Szel et al.,2000).
Could some of the cururo’s photoreceptor properties re-flect evolutionary convergence, do they show similaritiesto those in other rodents that have independently con-verted to a subterranean lifestyle? The subterranean“model species” for studying visual system adaptations isthe muroid blind mole-rat Spalax ehrenbergi (Sanyal etal., 1990; Cooper et al., 1993). It has minute subcutaneouseyes and lacks image-forming vision. The retina is rela-tively normal, but less organized than in sighted rodents.Spalax has numerous rods with a functional rod opsin(Sanyal et al., 1990; Janssen et al., 2000; Cernuda-Cernuda et al., 2002). While cones have not been conclu-sively identified by morphology, there is a functionalL-cone opsin but no functional S-cone opsin (David-Grayet al., 1998, 2002). The only role of the Spalax eye is seenin mediating photoperiodicity (Sanyal et al., 1990; Cooperet al., 1993; David-Gray et al., 1998; Avivi et al., 2001).
In contrast, the present findings and recent studies inother species challenge the hypothesis of a general, con-vergent eye reduction in subterranean rodents. Africanmole-rats (rodent family Bathyergidae) have small eyes(1.5–2.5 mm diameter, depending on species) and nor-mally structured retinae (Cernuda-Cernuda et al., 2003;Mills and Catania, 2004; Peichl et al., 2004). Their rodshave short outer segments with large diameters of 3–3.5�m, and a concomitantly low density of 100,000–150,000/mm2 (Peichl et al., 2004). Like cururo, the bathyergidshave a high cone proportion of about 10%, and both S-coneand L-cone opsin is present. Unlike cururo, about 90% ofthe bathyergid cones dominantly express the S-opsin, themajority of these coexpress smaller amounts of L-opsin(dual pigment cones), and only some 10% exclusively con-tain the L-opsin (Peichl et al., 2004). The fossorial pocketgophers (family Geomyidae) have normally sized eyes,about 25% cones, and a low rod density of 55,000-100,000/mm2 (Feldman and Phillips, 1984; Williams et al., 2005).Virtually every cone contains a UV-sensitive S-opsin;practically all cones in dorsal retina coexpress a green-sensitive L-opsin, while there is essentially no L-opsin inventral retina (Williams et al., 2005).
Hence, a surprising common feature of the studied sub-terranean rodents is that they have much higher coneproportions (10–25%) than nocturnal surface-dwelling ro-dents (0.5–3%)—apart from Spalax, where the cone pro-portion is unknown. On the other hand, the cone opsinarrangements show large interspecific differences. It ap-pears that each species has evolved a different cone ar-rangement, suggesting that the crucial adaptive pressuredid not come from the common, subterranean photic con-ditions but from some other, species-specific visual de-mands. Perhaps a complete absence of light does not exertany adaptive pressure, as ever so large increases in sen-sitivity would not enable vision, whereas even sporadicand short episodes of surface activity during light periodscould be relevant factors in molding the visual system.
What are the cururo activity patterns below and aboveground: can we infer their visual needs? Cururos arestrongly subterranean and rarely seen above ground. For-aging occurs primarily underground, but sporadically theyalso forage for leaves in the close vicinity of their burrowopenings (Torres-Mura and Contreras, 1998; Begall andGallardo, 2000). Under laboratory conditions, cururos pre-dominantly show nocturnal activity patterns (Begall et al.,2002; Rezende et al., 2003; Mena et al., 2003). In the wild,cururos appear more active during daytime. A recent ra-
Fig. 7. Reflectance properties. A: Average spectral reflectance ofthe fur on ventral and dorsal body parts (n � 2). B: Spectral reflec-tance contrast of fresh cururo urine. Measurements were obtained atwavelengths from 250–750 nm, average of two animals. The spectralcontrast of urine is highest in the UV.
206 L. PEICHL ET AL.
diotelemetric field study finds that both underground andsurface activity of free-living cururos is largely restrictedto daylight hours (Urrejola et al., 2004). Previous fieldstudies also have reported some daytime surface activity,probably depending on ambient temperature and season(Reig, 1970; Begall et al., 2002; Rezende et al., 2003).However, a detailed account of the frequency and durationof visits to the surface is still lacking, and it remains to beshown whether cururos display visually driven behavior.During diurnal surface activity, cone-based vision cer-tainly would be an advantage. The higher S-cone propor-tion in the upward-looking ventral retina may improvecontrast when viewing objects, e.g., raptors, against sky-light that is rich in short-wavelength components. This“skylight hypothesis” has been proposed to explain thehigh ventral S-cone proportions in a number of mammals,but remains controversial (summary: Szel et al., 2000).For example, the surface-dwelling degu, which certainly ismore threatened by diurnal raptors (Jaksic, 1997), doesnot have high ventral S-cone proportions (Jacobs et al.,2003).
Like degus, cururos are social animals. We have re-cently suggested for O. degus that the UV sensitivity mayenable it to not only smell but also see the highly UV-reflecting fresh urine scent marks that this species uses tomark its communal paths and territory (Chavez et al.,2003). Fresh cururo urine also has a high UV reflectanceand would be visible to the cururo’s S-cones in photopicconditions. The cururo pelage, as well as the soil andtypical vegetation of its habitat (our unpubl. data), have alow UV reflectance and are not conspicuous to the S-cones.Hence, apart from possibly seeing its urine marks, wehave found no obvious advantages for UV-sensitive ratherthan blue-sensitive cones in S. cyanus. It may be that theUV sensitivity merely is an ancestral trait that the cururoshares with the degu and a number of other rodents (Huntet al., 2001).
In conclusion, we suggest that the eye of the cururo hasadapted to visual demands during the short periods ofdaytime surface activity, rather than to the lightless sub-terranean microhabitat. Alternatively, the high cone pro-portion could be the legacy of a diurnal surface-dwellingancestor, and the subterranean specialization could haveoccurred too recently to allow full sensory adaptation yet.This is less likely because, as discussed above, the morerecent divergence of diurnal and nocturnal Octodon spe-cies is accompanied by dramatically different cone/rodratios (for octodontid divergence times, see Gallardo andKirsch, 2001). Either way, this challenges the generalview of evolutionary convergence in eye reduction andblindness among subterranean mammals. Further studieshave to elucidate the specific visual needs and behavioralcapabilities of the cururo, and indeed of other subterra-nean mammals. As an interesting parallel, a recent be-havioral study indicates that in addition to adaptations tosubterranean hearing, cururos have retained some fea-tures useful for above-ground hearing and communication(Begall et al., 2004).
ACKNOWLEDGMENTS
We thank Drs. J. Nathans (Baltimore) and R.S. Molday(Vancouver) for generously providing the opsin antibodies,H. Ahmed for skilled technical assistance. Permission towork on collected specimens was under authorization
#3014 from the Chilean Servicio Agricola y Ganadero(SAG).
LITERATURE CITED
Ahnelt PK, Kolb H. 2000. The mammalian photoreceptor mosaic — adap-tative design. Prog Ret Eye Res 19:711–777.
Avivi A, Albrecht U, Oster H, Joel A, Beiles A, Nevo E. 2001. Biologicalclock in total darkness: the Clock/MOP3 circadian system of the blindsubterranean mole rat. Proc Natl Acad Sci U S A 98:13751–13756.
Begall S, Gallardo MH. 2000. Spalacopus cyanus (Rodentia: Octodontidae):an extremist in tunnel constructing and food storing among subterra-nean mammals. J Zool Lond 251:53–60.
Begall S, Daan S, Burda H, Overkamp GJF. 2002. Activity patterns in asubterranean social rodent, Spalacopus cyanus (Octodontidae). J Mam-mal 83:153–158.
Begall S, Burda H, Schneider B. 2004. Hearing in coruros (Spalacopuscyanus): special audiogram features of a subterranean rodent. J CompPhysiol A: 190:963–969.
Borghi CE, Giannoni SM, Roig VC. 2002. Eye reduction in subterraneanmammals and eye protective behavior in Ctenomys. J NeotropicalMammal 9:123–134.
Burda H, Bruns V, Muller M. 1990. Sensory adaptations in subterraneanmammals. In: Nevo E, Reig O, editors. Evolution of subterraneanmammals at the organismal and molecular levels. New York: Alan R.Liss. p 269–293.
Cernuda-Cernuda R, DeGrip WJ, Cooper HM, Nevo E, Garcıa-FernandezJM. 2002. The retina of Spalax ehrenbergi: novel histologic featuressupportive of a modified photosensory role. Invest Ophthalmol Vis Sci43:2374–2383.
Cernuda-Cernuda R, Garcıa-Fernandez JM, Gordijn MCM, Bovee-GeurtsPHM, DeGrip WJ. 2003. The eye of the African mole-rat Cryptomysanselli: to see or not to see? Eur J Neurosci 17:709–720.
Chavez AE, Bozinovic F, Peichl L, Palacios AG. 2003. Retinal spectralsensitivity, fur coloration and urine reflectance in the genus Octodon(Rodentia): implications for visual ecology. Invest Ophthalmol Vis Sci44:2290–2296.
Chiu MI, Nathans J. 1994. A sequence upstream of the mouse blue visualpigment gene directs blue cone-specific transgene expression in mouseretinas. Vis Neurosci 11:733–780.
Contreras LC, Torres-Mura JC, Yanez JL. 1987. Biogeography of Octodon-tidae rodents: an eco-evolutionary hypothesis. Field Zool 39:401–411.
Cooper HM, Herbin M, Nevo E. 1993. Visual system of a naturally mi-crophthalmic mammal: the blind mole rat, Spalax ehrenbergi. J CompNeurol 328:313–350.
David-Gray ZK, Janssen JWH, DeGrip WJ, Nevo E, Foster RG. 1998. Lightdetection in a ’blind’ mammal. Nat Neurosci 1:655–656.
David-Gray ZK, Bellingham J, Munoz M, Avivi A, Nevo E, Foster RG.2002. Adaptive loss of ultraviolet-sensitive/violet-sensitive (UVS/VS)cone opsin in the blind mole rat (Spalax ehrenbergi). Eur J Neurosci16:1186–1194.
Feldman JL, Phillips CJ. 1984. Comparative retinal pigment epitheliumand photoreceptor ultrastructure in nocturnal and fossorial rodents:the eastern woodrat, Neotoma floridana and the plains pocket gopher,Geomys bursarius. J Mammal 65:231–245.
Govardovskii VI, Fyhrquist N, Reuter T, Kuzmin DG, Donner K. 2000. Insearch of the visual pigment template. Vis Neurosci 17:509–528.
Hicks D, Molday RS. 1986. Differential immunogold-dextran labeling ofbovine and frog rod and cone cells using monoclonal antibodies againstbovine rhodopsin. Exp Eye Res 42:55–71.
Honeycutt RL, Rowe DL, Gallardo MH. 2003. Molecular systematics of theSouth American caviomorph rodents: relationships among species andgenera in the family Octodontidae. Mol Phylogenet Evol 26:476–489.
Howland HC, Merola S, Basarab JR. 2004. The allometry and scaling of thesize of vertebrate eyes. Vis Res 44:2043–2065.
Hunt DM, Wilkie SE, Bowmaker JK, Poopalasundaram S. 2001. Vision inthe ultraviolet. Cell Mol Life Sci 58:1583–1598.
Jacobs GH. 1993. The distribution and nature of colour vision among themammals. Biol Rev 68:413–471
Jacobs GH, Neitz J, Deegan JF. 1991. Retinal receptors in rodents maxi-mally sensitive to ultraviolet light. Nature 353:655–656.
Visual adaptations in a diurnal rodent, Octodon degus. J Comp PhysiolA 189:347–361.
Jaksic FM. 1997. Ecologıa de los Vertebrados de Chile. Santiago de Chile:Ediciones Universidad Catolica de Chile.
Janssen JW, Bovee-Geurts PHM, Peeters ZPA, Bowmaker JK, Cooper HM,David-Gray ZK, Nevo E, DeGrip WJ. 2000. A fully functional rod visualpigment in a blind mammal. A case for adaptive functional reorgani-zation? J Biol Chem 275:38674–38679.
Jeon CJ, Strettoi E, Masland RH. 1998. The major cell populations of themouse retina. J Neurosci 18:8936–8946.
Kimchi T, Terkel J. 2002. Seeing and not seeing. Curr Opin Neurobiol12:728–734.
Lacey EA, Patton JL, Cameron GN. 2000. Life underground: the biology ofsubterranean rodents. Chicago: University of Chicago Press.
Lamb TD. 1995. Photoreceptor spectral sensitivities: common shape in thelong-wavelength region. Vis Res 35:3083–3091.
Mansfield RJW. 1985. Primate photopigments and cone mechanisms. In:Fein A, Levine JS editors. The visual system. New York: Alan R. Liss.p 89–106.
Mena W, Ocampo-Garces A, Palacios AG. 2003. Circadian rhythm of rest-activity in Spacalcopus cyanus. XVIII. Annual Meeting of the ChileanPhysiological Society, Abstract P35, accessed at http://codon.ciencias.uchile.cl/�csfisiol/ValleDorado/P35.pdf.
Mills SL, Catania KC. 2004. Identification of retinal neurons in a regres-sive rodent eye (the naked mole rat). Vis Neurosci 21:107–117.
Nevo E. 1999. Mosaic evolution of subterranean mammals: regression,progression, and global convergence. Oxford: Oxford University Press.
Palacios AG, Varela FJ, Srivastava R, Goldsmith TH. 1998. Spectral sen-sitivity of cones in the goldfish, Carassius auratus. Vis Res 38:2135–2146.
Peichl L, Moutairou K. 1998. Absence of short-wavelength sensitive conesin the retinae of seals (Carnivora) and African giant rats (Rodentia).Eur J Neurosci 10:2586–2594.
Peichl L, Kunzle H, Vogel P. 2000. Photoreceptor types and distributions inthe retinae of insectivores. Vis Neurosci 17:937–948.
Peichl L, Nemec P, Burda H. 2004. Unusual cone and rod properties insubterranean African mole-rats (Rodentia, Bathyergidae). Eur J Neu-rosci 19:1545–1558.
Reig O. 1970. Ecological notes on the fossorial octodont rodent Spalacopuscyanus (Molina). J Mammal 51:592–601.
Rezende EL, Cortes A, Bacigalupe LD, Nespolo RF, Bozinovic F. 2003.Ambient temperature limits above-ground activity of the subterraneanrodent Spalacopus cyanus. J Arid Env 55:63–74.
Sanyal S, Jansen HG, DeGrip WJ, Nevo E, de Jong WW. 1990. The eye ofthe blind mole rat, Spalax ehrenbergi. Rudiment with hidden function?Invest Ophthalmol Vis Sci 31:1398–1404.
Szel A, Lukats A, Fekete T, Szepessy Z, Rohlich P. 2000. Photoreceptordistribution in the retinas of subprimate mammals. J Opt Soc Am A17:568–579.
Wang Y, Macke JP, Merbs SL, Zack DJ, Klaunberg B, Bennett J, GearhartJ, Nathans J. 1992. A locus control region adjacent to the human redand green visual pigment genes. Neuron 9:429–440.
Williams GA, Calderone JB, Jacobs GH. 2005. Photoreceptors and pho-topigments in a subterranean rodent, the pocket gopher (Thomomysbottae). J Comp Physiol A 191:125–134.
