Rod Contributions to the Electroretinogram of the Dark-Adapted Developing ZebrafishJOSEPH BILOTTA,1 SHANNON SASZIK,1,2 AND SARAH E. SUTHERLAND1,3

1Department of Psychology and Biotechnology Center, Western Kentucky University, Bowling Green, Kentucky2College of Optometry, University of Houston, Houston, Texas3Southern College of Optometry, Memphis, Tennessee

ABSTRACT Anatomical studies of the devel-oping zebrafish retina have shown that rods ap-proach maturity at about 15 days postfertilization(dpf). Past work has examined the photopic spec-tral sensitivity function of the developing ze-brafish, but not spectral sensitivity under dark-adapted conditions. This study examined rodcontributions to the dark-adapted spectral sensi-tivity function of the ERG b-wave component indeveloping zebrafish. ERG responses to stimuli ofvarious wavelengths and irradiances were ob-tained from dark-adapted fish at 6–8, 13–15, 21–24,and 27–29 dpf. The results show that dark-adaptedspectral sensitivity varied with age. Spectral sen-sitivity functions of the 6–8 and 13–15 dpf groupsappeared to be cone dominated and contained lit-tle or no rod contributions. Spectral sensitivityfunctions of the 21–24 and 27–29 dpf groups ap-peared to have both rod and cone contributions.Even at the oldest age group tested, the dark-adapted spectral sensitivity function did notmatch the adult function. Thus, consistent withanatomical findings, the rod contributions to theERG spectral sensitivity function appear to de-velop with age; however, these contributions arestill not adult-like by 29 dpf, which is contrary toanatomical work. These results illustrate that thezebrafish is an excellent model for visual develop-ment. © 2001 Wiley-Liss, Inc.

Key words: Danio rerio; dark-adapted spectralsensitivity; electroretinogram; roddevelopment

INTRODUCTION

Anatomical studies of the developing zebrafish retinahave shown that rods continue to develop long afterhatch. Branchek and Bremiller (1984) found that by 8days postfertilization (dpf) there were still no apparentsigns of rod outer segments, and that by 12 dpf, rodouter segment length was only 25% that of adult rods.However, by 15–20 dpf, rods appeared to approachanatomical maturity (Branchek and Bremiller, 1984).Behavioral work has reported that zebrafish rod func-tion begins at about 12 dpf. Clark (1981) used theoptomotor response to measure visual acuity and foundthat at 7 dpf there were no differences in acuity values

obtained under dark- and light-adapted conditions. Heconcluded there were only cone photoreceptors contrib-uting to the behavioral response. However, by 14 dpf,subjects were more sensitive under dark-adapted thanlight-adapted conditions and, thus, had significantlyhigher acuity values than those tested under light-adapted conditions. Clark (1981) reasoned that the dif-ference in acuity between the two light conditions re-flected the changing contributions of the developingrods with age.

Physiological studies using the electroretinogram(ERG) response have demonstrated a rod/cone break inthe flicker fusion frequency vs. intensity function inzebrafish as young as 15 dpf (Branchek, 1984). Prior tothis age, there did not appear to be a break in thefunction. However, although the stimulus intensity atwhich this break occurred was similar to the valueobtained from adult zebrafish, the slopes of this bipha-sic function differed from the adult function. Thus, it isnot clear whether the break in the young subjects’function represents adult-like processing or not. Also,it is not certain whether the break represents a shiftfrom pure cone contributions to pure rod or mixedrod-cone contributions. It also is possible that thebreak may not represent a change from cone to rodfunction at all. Green and Siegel (1975) showed thatthe ERG response of the rod-only skate retina produceda biphasic fusion frequency vs. intensity function.Thus, although the previous work has demonstrated adevelopmental shift in zebrafish visual processing, inorder to determine the nature of the photoreceptorcontributions in the developing zebrafish, spectral sen-sitivity functions must be obtained and compared tothe photoreceptor spectra.

Recent physiological work has examined adult ze-brafish spectral sensitivity under both light-adapted(Hughes et al., 1998) and dark-adapted (Saszik andBilotta, 1999) conditions. Under photopic conditions,the spectral sensitivity derived from the b-wave re-sponse of the ERG appears to receive contributions

Grant sponsor: Kentucky NSF/EPSCoR; Grant number:OSR-9452895.

*Correspondence to: Dr. Joseph Bilotta, Department of Psychology,Western Kentucky University, 1 Big Red Way, Bowling Green KY42101. E-mail: joseph.bilotta@wku.edu

Received 25 May 2001; Accepted 12 July 2001Published online 29 August 2001

DEVELOPMENTAL DYNAMICS 222:564–570 (2001)

© 2001 WILEY-LISS, INC.DOI 10.1002/dvdy.1188

from the four known cone types (U-, S-, M-, and L-cones; lmax 5 362, 415, 480, and 570 nm, respectively(Robinson et al., 1993) and appears to be the result ofboth nonopponent and opponent interactions (Hugheset al., 1998). Under dark-adapted conditions, the spec-tral sensitivity function of the adult zebrafish appearsto receive input from rod photoreceptors containingrhodopsin as well as from U-cones (Saszik and Bilotta,1999).

In addition, photopic spectral sensitivity has beenexamined in the developing zebrafish (Saszik et al.,1999). Spectral sensitivity functions were derived fromthe b-wave response from zebrafish at four ages (4–5,6–8, 13–15, and 21–24 dpf). It was found that thespectral sensitivity functions varied with age. At theultraviolet wavelengths, spectral sensitivity was simi-lar across all age groups. However, at the longer wave-lengths, the functions appeared to shift upward withage, indicating an increase in sensitivity. Also, therewas no evidence of any opponent mechanism contrib-uting to the response in the younger groups as seen inthe adult data. Therefore, zebrafish retinal processingin subjects as old as 24 dpf does not appear adult-likeunder light-adapted conditions (Saszik et al., 1999).The purpose of this study was to examine the rodcontributions to the dark-adapted spectral sensitivityfunction of the developing zebrafish. ERG responseswere obtained from dark-adapted zebrafish at variousages. Spectral sensitivity was derived and the photore-ceptor contributions to the spectral sensitivity data

were assessed quantitatively and compared acrossages.

RESULTS

Each ERG waveform was digitally filtered for 60 Hznoise. Figure 1 shows sample ERG waveforms from twosubjects of different ages. The raised horizontal bardepicts stimulus onset and termination. Figure 1a andb are responses from an 8 dpf subject to 360 and 580nm stimuli, respectively. In Figure 1a, the ERG re-sponse to the 360 nm stimulus consisted of a largeinitial voltage-negative component, the a-wave, fol-lowed by the voltage-positive b-wave. On the otherhand, the ERG response to a 580-nm stimulus (Fig. 1b)produced a small a-wave component and a large b-wavecomponent. By 23 dpf (Fig. 1c,d), only a small a-waveresponse, with a dominant b-wave component are ap-parent.

Spectral sensitivity functions were obtained from theb-wave component of the ERG waveform. To derive aspectral sensitivity function, the reciprocal of the logstimulus irradiance that produced a 50 mV criterionresponse was determined for each stimulus wavelength(Saszik and Bilotta, 1999). To derive sensitivity values,log irradiance vs. log response functions were gener-ated for each stimulus wavelength and linear regres-sion was used to interpolate to find the stimulus irra-diance that produced the criterion response (Hughes etal., 1998). The 50-mV criterion response was chosenbecause it fell within the linear portion of the log irra-

Fig. 1. Sample ERG waveformsof dark-adapted 8 and 23 dpf sub-jects. The horizontal line illustratesthe onset and termination of the 200ms stimulus. a: ERG response of an8 dpf subject to a 360 nm stimuluswith an irradiance of 12 log quanta/s/cm2. b: ERG response of an 8 dpfsubject to a 580 nm stimulus with anirradiance of 14.5 log quanta/s/cm2.c: ERG response of a 23 dpf subjectto a 360 nm stimulus with an irradi-ance of 11 log quanta/s/cm2. d:ERG response of a 23 dpf subject toa 580 nm stimulus with an irradianceof 11 log quanta/s/cm2.

565DEVELOPING ROD CONTRIBUTIONS

diance-log response functions. Using a different re-sponse criterion, within the linear portion of the func-tion, did not alter the shape of the spectral sensitivityfunction (data not shown). In addition, this criterionresponse was used to derive dark-adapted spectral sen-sitivity functions from adult zebrafish (Saszik andBilotta, 1999). Figure 2 compares the mean dark-adapted spectral sensitivity functions of the 6–8(squares), 13–15 (triangles), 21–24 (circles), and 27–29(diamonds) dpf subjects. Error bars represent 6 1 stan-dard error of the mean (S.E.M.). Although the meanspectral sensitivity function of the 13–15 dpf subjectsappears slightly more sensitive overall compared to the6–8 dpf group, the shapes of the two functions aresimilar. However, the mean spectral sensitivity func-tion of the 13–15 dpf subjects appears to be differentfrom the function of the 21–24 dpf subjects. The 21–24dpf subjects appear to be much more sensitive to mid-dle wavelengths (from 460 to 580 nm) compared to the13–15 dpf subjects. Finally, there is very little differ-ence in the shape or absolute sensitivity between thespectral sensitivity functions of the 21–24 and 27–29dpf subjects.

Because the spectral sensitivity functions from the6–8 and 13–15 dpf were similar, the data were com-bined to form a 6–15 dpf age group. For similar rea-sons, the data from the 21–24 and 27–29 dpf subjectswere combined to create a 21–29 dpf age group. Figure3 compares the mean dark-adapted spectral sensitivityfunctions of the 6–15 (triangles) and 21–29 (squares)dpf subjects. In addition, the mean spectral sensitivityfunction for dark-adapted adult zebrafish (circles; datafrom Saszik and Bilotta, 1999) is plotted for compari-son. The curve associated with the adult data is the

rhodopsin spectra (Dartnall, 1953) normalized to thedata at 500 nm. In general, the spectral sensitivityfunction appears more “rod-like” with age. In addition,all functions have a peak in sensitivity at about 360nm, near the lmax of the U-cones. The 21–29 dpf sub-jects are more sensitive to middle wavelengths (from460 to 580 nm) than are the 6–15 dpf subjects. Errorbars represent 6 1 S.E.M.

To determine the photoreceptor contribution to thedark-adapted spectral sensitivity function, a linear-additive model that has been used successfully withadult zebrafish dark-adapted functions was used(Saszik and Bilotta, 1999). The linear-additive modelhas the following form:

Sl 5 Oi 5 1

w

~Kx 3 Ax! (1)

where Sl is the sensitivity at wavelength l, Ax is theabsorptance of photoreceptor x at wavelength l, Kx isthe input weight of photoreceptor x, and w equals thenumber of photoreceptor types contributing to themodel. Cone spectra were obtained from templatesbased on recordings from giant danio cones (Palacios etal., 1996) placed on a normalized frequency axis usingthe lmax from zebrafish microspectrophotometric data(Robinson et al., 1993); the rhodopsin absorption curvewas obtained from Dartnall (1953). Sensitivity valueswere converted to proportions and normalized to theirmaximum value. A simplex algorithm was used to de-termine the photoreceptor weights that yielded thebest-fit model to the sensitivity data. Similar proce-

Fig. 2. Averaged dark-adapted spectral sensitivityfunction of 6–8 (n 5 11; squares), 13–15 (n 5 16;triangles), 21–24 (n 5 12; circles), and 27–29 (n 5 6;diamonds) dpf subjects. Log absolute sensitivity (logquanta/s/cm2) is defined as the reciprocal of the logstimulus irradiance that produced a b-wave criterionresponse of 50 mV. Error bars represent 6 1 standarderror of the mean (S.E.M.).

566 BILOTTA ET AL.

dures have been used to derive the photoreceptor con-tributions to the spectral sensitivity functions of light-adapted and dark-adapted adult zebrafish ERG(Hughes et al., 1998; Saszik and Bilotta, 1999).

Table 1 shows the resulting weights from the modelincorporating all four cone types and rods across ages.For the 6–15 dpf subjects, there was very little rodinput to the response. The majority of the input ap-pears to be from the U-cones; this is found in 6–8 dpfzebrafish spectral sensitivity functions under photopicconditions as well (Saszik et al., 1999). For the 21–29dpf subjects, there appear to be both cone and rodcontributions to the response. On the other hand, foradults, the largest input contributing to the responseappears to be from the rods with some U-cone contri-butions. The fairly large M-cone weight found across allage groups may be an artifact of the modeling proce-dures because the M-cone and rod spectra are similar(lmax 5 480 and 500 nm, respectively).

To address the fact that the M-cone and rod inputsare similar, two additional models were employed. Onemodel was restricted to cone inputs (S, M, and L), whilethe other was restricted to rod input. Since U-conesappear to contribute under both dark and photopicconditions, they were excluded from both models and

only data from 400 to 640 nm stimuli were examined inthe models. To determine how well each model fit thedata, a chi-square goodness of fit test was calculatedcomparing the final model to the data. Figure 4 showsthe results of the chi-square analysis on the rod andcone models. For 6–15 dpf subjects, the cone modelappears to fit the data very well; the chi-square value is

Fig. 3. Averaged dark-adapted spectral sensitiv-ity function of 6–15 (n 5 27; triangles), 21–29 (n 518; squares) dpf, and adult (n 5 10; circles) subjects.The rhodopsin spectra (line) was normalized withrespect to the adult data at 500 nm. Log absolutesensitivity (log quanta/s/cm2) is defined as the recip-rocal of the log stimulus irradiance that produced ab-wave criterion response of 50 mV. Error bars rep-resent 6 1 S.E.M. The adult data were obtained fromSaszik and Bilotta (1999) and the rhodopsin spectrawas obtained from Dartnall (1953).

TABLE 1. Linear-Additive Model Weights

Age (dpf) U-cones S-cones M-cones L-cones Rods6–15 10.79 10.35 10.22 10.15 20.0721–29 10.64 10.07 10.57 10.18 10.30Adult 10.22 10.04 10.19 10.03 10.72

Fig. 4. Goodness of fit (chi-square) values of two restricted linear-additive models for spectral sensitivity data for 6–15, 21–29 dpf, andadult subjects. One model (cone model) incorporated three cone types(S-, M-, and L-cones); the other model (rod model) used only rod input tomodel the data. For both models, the analysis was restricted to data from400 to 640 nm; this was done to exclude the U-cones from both models.See text for details.

567DEVELOPING ROD CONTRIBUTIONS

very small. For the adult subjects, the rod model fitsthe data very well. Finally, for the 21–29 dpf subjects,although the cone and rod models fit the data similarly,neither model provides a good fit.

DISCUSSION

The results of this study support past anatomical,physiological, and behavioral studies that show thatthe zebrafish retina develops with age. This was thecase for both the characteristics of the ERG waveformas well as the photoreceptor inputs to the dark-adaptedspectral sensitivity function.

ERG Waveform

In this study, it was shown that the characteristics ofthe ERG waveform of the dark-adapted zebrafishchanged with age. At 8 dpf, the ERG waveform char-acteristics depended on stimulus wavelength. At ultra-violet wavelengths, the ERG waveform was dominatedby a large a-wave component, while at middle- to long-wavelengths, the waveform was dominated by a largeb-wave component. With age, the large a-wave thatwas found in the younger subjects’ response to ultravi-olet wavelengths was no longer apparent. The ERGresponse characteristics of the older subjects did notvary as a function of stimulus wavelength.

These changes with age correspond with those foundin zebrafish under light-adapted conditions as well.Saszik et al. (1999) reported that the ERG responses ofyoung zebrafish under photopic conditions were wave-length dependent. Like the present study, they foundthat the young age groups (4–5 and 6–8 dpf) possesseda large a-wave component in response to ultravioletwavelength stimuli and a predominant b-wave in re-sponse to the other wavelengths. The large a-wavecomponent diminished with age and the waveform ap-peared adult-like by 21–24 dpf across all stimuluswavelengths. Thus, the fact that under dark-adaptedconditions, the a-wave component diminishes with ageis not surprising. This is supported by the results ofSaszik and Bilotta (1999) who showed that the ERGwaveform of dark-adapted adult zebrafish was domi-nated by the b-wave component of the ERG across allwavelengths. In addition, they found that the dark-adapted responses were much slower than those ob-tained under light-adapted conditions. In the presentstudy, the latencies of the dark-adapted developingzebrafish also were somewhat slower than under light-adapted conditions (Saszik et al., 1999).

Spectral Sensitivity

The results of the present study also showed that rodcontributions to the dark-adapted zebrafish variedwith age. This was supported by the developmentalchanges in the spectral sensitivity functions as well asthe specific photoreceptor contributions based on thelinear-additive modeling.

The spectral sensitivity function obtained from theyounger fish (6–15 dpf) appeared to be primarily the

result of cone input. Absolute sensitivity was highest inthe ultraviolet to short-wavelength stimulus rangecompared to the sensitivity to the middle- and long-wavelength stimuli. These findings are similar to thosereported in the developing zebrafish under photopicconditions. Saszik et al (1999) found that sensitivity toultraviolet and short-wavelength stimuli was higherthan to the middle- and long-wavelength stimuli. Thiswas attributed to the strong contributions of the U- andS-cones to the ERG response. With age, sensitivity tothe middle- and long-wavelength stimuli increased, butwas never higher in sensitivity than at the ultravioletand short-wavelength stimuli. Thus, it is very likelythat the spectral sensitivity functions obtained underdark-adapted conditions reflect only cone contribu-tions. The shapes of the functions from the youngersubjects obtained under light- and dark-adapted func-tions are qualitatively similar. This also is supportedby the fact that the best-fit model of the spectral sen-sitivity data of the young group in this study reflectsonly cone contributions (see Fig. 4). These results aredifferent from the best-fit model of the spectral sensi-tivity data of the adult dark-adapted zebrafish, whichreflects only rod contributions from wavelengths rang-ing from 400 to 640 nm (see Fig. 4).

On the other hand, the spectral sensitivity functionobtained from the older fish (21–29 dpf) did not appearto be the result of only cone input. Absolute sensitivitywas highest in the middle-wavelength stimulus range,near the peak sensitivity of the rod spectra, as opposedto sensitivity in the ultraviolet and short-wavelengthstimulus range. Also, the results of the photoreceptormodeling based on cone input did not fit the data verywell. However, the results of the photoreceptor model-ing based on the rod only input did not fit the data atthis age very well either. In fact, both the cone only androd only models were poor fits to the spectral sensitiv-ity data. It also is apparent that the spectral sensitivityfunction of this age group is not similar to the adultfunction, which is most likely the result of only rodinput. The 21–29 dpf group was less sensitive than theadults and the shape of the function was different fromthe adult data. Thus, the 21–29 dpf dark-adapted func-tion appears to be the result of both rod and coneinputs. This would explain why the slopes of the bipha-sic functions of the older subjects (15–24 dpf) obtainedby Branchek (1984) did not match the slopes of theadult function. Under dark-adapted conditions, theadult zebrafish function appears to be solely a productof rod function. However, the function of the 21–29 dpfsubjects appears to be the result of a combination ofboth rod and cone input.

The fact that under dark-adapted conditions, ze-brafish visual processing is not adult-like is consistentwith the same findings under photopic conditions. Un-der photopic conditions, the spectral sensitivity func-tion of the 21–24 dpf subjects was not adult-like. Thesesubjects were much less sensitive to middle- and long-wavelength stimuli than were adults. In addition, the

568 BILOTTA ET AL.

light-adapted spectral sensitivity functions of the21–24 dpf subjects did not display any indication ofopponent mechanisms. The spectral sensitivity func-tion of adults under photopic conditions possess severaldips or “notches” in sensitivity (Hughes et al., 1998),suggesting the presence of opponent mechanisms(Sperling and Harwerth, 1971). Thus, under both light-and dark-adapted conditions, the 21–24 dpf zebrafishretina is still not functioning as an adult, even thoughanatomically fish this age appear to possess all of thephotoreceptor types. This suggests that further ana-tomical development, perhaps in the form of additionalsynaptic connections, is required before the zebrafishbecomes adult-like in its visual processing. The factthat rod photoreceptors develop later than cones is notunique to zebrafish. Complete rod development takesplace later than cones in other fish species (Powers andRaymond, 1990), as well as in rodents (Cepko et al.,1996). In addition, ERG responses from human infantsappear to reflect rod photoreceptor immaturities (Bre-ton et al., 1995).

It is interesting to note that the present study sup-ports the idea that the U-cones are a very importantcomponent of the zebrafish visual system. Althoughseveral studies have shown that rod and cone opsinexpression appears as early as 40–50 hrs postfertiliza-tion, well before the physical characteristics of the pho-toreceptors become apparent (Larison and Bremiller,1990; Raymond et al., 1995), the U-cones appear to bethe first cone type to develop (i.e., the short-single conetype; Branchek and Bremiller, 1984), and they appearto be a dominant cone type in the ERG response inadults, under both light-adapted (Hughes et al., 1998)and dark-adapted (Saszik and Bilotta, 1999) condi-tions. In addition, the U-cones appear to be the domi-nant cone type in the developing retina. Under pho-topic conditions, even at the earliest ages examined(4–5 dpf), the U-cones appear to have the strongestinput to the spectral sensitivity function (Saszik et al.,1999). With age, the contributions of the U-cones re-main constant while the contributions of the other conetypes increase. In fact, it appears that the U-cones arethe only part of the zebrafish visual system that ap-pears adult-like in function at the earliest ages. Theresults of the present study also support this notion.

In sum, consistent with past anatomical, physiolog-ical, and behavioral studies, the rod contributions tothe zebrafish ERG spectral sensitivity function appearto develop with age. At very young ages (6–15 dpf), thedark-adapted spectral sensitivity function can best beexplained by cone inputs. At older ages (21–29 dpf), thefunction appears to be the result of both rod and coneinputs. However, even by 21–29 dpf, the dark-adaptedspectral sensitivity function is not adult-like becausethe adult function can be best explained by contribu-tions of solely rod input from 400 to 640 nm. Thus,zebrafish visual function clearly develops after hatch-ing, and even by 29 dpf, its ERG response is still notadult-like. Finally, the results of this and similar work

illustrate that the zebrafish is an excellent model forvisual development since it provides the opportunity tostudy and compare behavior, physiology, anatomy, andgenetics in one system.

EXPERIMENTAL PROCEDURES

Subjects

The participants were larvae zebrafish (Danio rerio)bred using standard procedures from adult zebrafish(wild type) obtained from a local pet store (Bilotta etal., 1999). They were raised in a 14-hr light on/10-hrlight off cycle with a water temperature of 28–30°C. Atthis temperature, adult zebrafish ERG spectral sensi-tivity, under dark-adapted conditions, is best fit by therhodopsin spectra (Saszik and Bilotta, 1999). Four agegroups were tested: 6–8, 13–15, 21–24, and 27–29 dpf.

ApparatusOptical system. Details of the optical system can be

found elsewhere (Hughes et al., 1998; Saszik andBilotta, 1999). The light source consisted of a 150-Wxenon arc lamp. The light from this source was colli-mated and then focused onto an optical shutter, whichwas controlled by the laboratory computer. The lightwas collimated again and then passed through an in-terference filter, which controlled stimulus wave-length. The interference filters had a peak transmit-tance ranging from 320 to 640 nm with a 10-nm half-bandwidth. Stimulus irradiance was controlled byplacing quartz neutral density filters in the collimatedpath attenuating the irradiance over a 9-log unit range.The collimated light was focused onto one end of a5-mm diameter liquid light guide; the other end wasplaced in front of the subject’s eye. The light guide waspositioned close enough so that the entire pupil (theaverage adult zebrafish pupil diameter is 0.98 mm;unpublished observation) was filled with a uniformdistribution of light irradiance. Light measurementswere obtained with a radiometer sensitive to ultravio-let and visible wavelengths (International Light, New-buryport, MA, IL1400) and converted to quanta/s/cm2.

Electrophysiological apparatus. Electroretino-grams (ERGs) were recorded with electrodes consistingof glass pipettes with a 10-mm diameter tip filled witha teleost saline solution and a 36-gauge chlorided silverwire. The recording electrode was placed on the eyewhile the reference electrode was placed on the bodynear the tail. For the older subjects (i.e., 21–29 dpf), therecording electrode was placed in the vitreal chamber.Pilot data showed that electrode placement did notalter the shape of the ERG waveform. Electrical signalswere differentially amplified (WPI, Sarasota, FL,DAM-50) with a band-pass of 0.1 to 100 Hz and sentsimultaneously to a digital oscilloscope and the labora-tory computer. The computer data acquisition rate was4 ms.

569DEVELOPING ROD CONTRIBUTIONS

Procedures

Prior to the experiment, subjects were dark-adaptedfor at least 1 hr. Under dim light, subjects then wereanesthetized by submersion into a solution of tricainemethanesulfonate. The dose of the anesthesia (0.01,0.02, or 0.04%) varied with subject age (6–8, 13–15,and 21–29 dpf, respectively). Once anesthetized, sub-jects were placed on a cotton pad, moistened with ananesthetic solution, located on a petri dish, and placedinto a light-tight Faraday cage. Subjects were thendark-adapted for an additional 5 min prior to testing.

Each stimulus consisted of a 200-ms stimulus of aparticular wavelength and irradiance. For each stimu-lus wavelength, trials began at an irradiance belowthreshold and then increased until a 50-mV b-waveresponse was produced. Stimulus irradiances that pro-duced responses larger than the criterion responsewere avoided to eliminate the possibility of adaptationeffects to the stimulus wavelength. In addition, tomaintain the dark-adapted state, at least 15 sec sepa-rated stimulus presentations. Seventeen wavelengthsfrom 320 to 640 nm in 20-nm steps were used. Wave-length presentation was staggered in 40-nm stepsstarting at either 320 or 640 nm; once the initial seriesof wavelengths were presented, the remaining wave-lengths were filled in. This procedure was done to en-sure that there were no order effects. Preliminary anal-ysis comparing the sensitivity values between the twopresentation series found no systematic order effects.

ACKNOWLEDGMENTS

This project was supported by Kentucky NSF/EPS-CoR grant OSR-9452895 (J.B.). We thank Carla Givinfor assistance in data collection and analysis.

REFERENCES

Bilotta J, Saszik S, DeLorenzo AD, Hardesty HR. 1999. Establishingand maintaining a low-cost zebrafish breeding and behavioral re-search facility. Behav Res Methods Instrum Comput 31:178–184.

Branchek T. 1984. The development of photoreceptors in the ze-brafish, Brachydanio rerio. II. Function. J Comp Neurol 224:116–122.

Branchek T, Bremiller R. 1984. The development of photoreceptors inthe zebrafish, Brachydanio rerio. I. Structure. J Comp Neurol 224:107–115.

Breton ME, Quinn GE, Schueller AW. 1995. Development of electro-retinogram and rod phototransduction response in human infants.Invest Ophthalmol Visual Sci 36:1588–1602.

Cepko CL, Austin CP, Yang X, Alexiades M, Ezzeddine D. 1996. Cellfate determination in the vertebrate retina. Proc Natl Acad Sci USA93:589–595.

Clark DT. 1981. Visual responses in developing zebrafish (Brachy-danio rerio). Ph.D. dissertation. Eugene, Oregon: University of Or-egon. p 138.

Dartnall HJA. 1953. The interpretation of spectral sensitivity curves.Br Med Bull 9:24–30.

Green DG, Siegel IM. 1975. Double branched flicker fusion curvesfrom the all-rod skate retina. Science 188:1120–1122.

Hughes A, Saszik S, Bilotta J, DeMarco PJ, Jr., Patterson WF, II.1998. Cone contributions to the photopic spectral sensitivity func-tion of the zebrafish ERG. Vis Neurosci 15:1029–1037.

Larison KD, Bremiller R. 1990. Early onset of phenotype and cellpatterning in the embryonic zebrafish retina. Development 109:567–576.

Palacios AG, Goldsmith TH, Bernard GD. 1996. Sensitivity of conesfrom a cyprinid fish (Danio aequipinnatus) to ultraviolet and visiblelight. Vis Neurosci 13:411–421.

Powers, MK, Raymond, PA. 1990. Development of the visual system.In: Douglas RH, Djamgoz BA, editors. The visual system of fish.London: Chapman and Hall. p 419–441.

Raymond PA, Barthel LK, Curran, GA. 1995. Developmental pattern-ing of rod and cone photoreceptors in embryonic zebrafish. J CompNeurol 359:537–550

Robinson J, Schmitt EA, Harosi FI, Reece RJ, Dowling JE. 1993.Zebrafish ultraviolet visual pigment: Absorption spectrum, se-quence, and localization. Proc Natl Acad Sci USA 90:6009–6012.

Saszik S, Bilotta J. 1999. The effects of temperature on the dark-adapted spectral sensitivity function of the adult zebrafish. VisionRes 39:1051–1058.

Saszik S, Bilotta J, Givin CM. 1999. ERG assessment of zebrafishretinal development. Vis Neurosci 16:881–888.

Sperling HG, Harwerth RS. 1971. Red-green cone interactions in theincrement-threshold spectral sensitivity of primates. Science 172:180–184.

570 BILOTTA ET AL.