Vision in the Ctenid Spider Cupiennius Salei: Spectral Range and Absolute Sensitivity

Mainly from the classical work of Homann (1928, 1931, 1951, 1971) and from the landmark studies of Land (1985) and Blest (1985), we know that spider eyes are far from being simple or having uniform properties throughout the taxon. There are masterpieces of eye design among the jumping spiders (Salticidae) and dinopids (Dinopidae). It is mostly these cases that are quoted as examples of perfection: a spatial resolution of 2.4′ in the anterior median eye of Portia (Salticidae; Williams and McIntyre, 1980); the light sensitivity of the posterior median eyes of Dinopis, which reach half of their maximum response at corneal fluxes corresponding to an intensity halfway between those of starlight and moonlight (Laughlin et al. 1980) and have a lens with an F-number of 0.58 (Blest and Land, 1977); or the tiered retina in the anterior median eyes of jumping spiders, which potentially serve to deal with chromatic aberration by being located in the best image planes for light of different wavelengths (Land, 1969a, 1985). Another fascinating aspect of spider eyes, in particular those of jumping spiders, is that the retinae of the anterior median or main eyes are moved relative to the lens to scan the image produced by the lens (Land, 1969b).

Eyes with such impressive properties and forming good images all belong to hunting spiders, such as salticids, lycosids, thomisids and sparassids. Orb weavers, in contrast, are generally considered to have comparatively poor vision with poor spatial resolution and with the image often out of focus at the photoreceptor level (Land, 1985; Yamashita, 1985).

In this paper we examine vision in Cupiennius salei, a ctenid that is closely related both taxonomically and behaviourally to the lycosids. Cupiennius salei is a nocturnal hunting spider and has been the subject of neurobiological research for a long time. Much of its behaviour, for example prey capture and courtship, has been shown to be guided predominantly by mechanosensory systems (Barth, 1985a,b, 1990; Barth and Schmitt, 1991; Seyfarth, 1985; Eckweiler and Seyfarth, 1988). This seems to make sense in a nocturnal species, whose members are active when very little light is available (Barth and Seyfarth, 1979; Seyfarth, 1980; Schmitt et al. 1990). There are several reasons, however, for taking a closer look at the visual capabilities of Cupiennius. One of these is that the optics of the eyes of this spider have recently been demonstrated to be of very good quality, with the F-numbers of the lenses being between 0.58 and 0.74, which suggests bright images and a reasonably well developed spatial resolution with angular separation of the receptor cells between 0.9° and 3.6° (along the rows they form). The impression of rather well-developed vision is further strengthened by neuroanatomical studies on the visual neuropiles in the central nervous system, first analyzed in spiders in Cupiennius salei in some detail (Strausfeld and Barth, 1993; Strausfeld et al. 1993). Finally, one of the similarities between the Ctenidae, to which the genus Cupiennius belongs (Lachmuth et al. 1985), and the Lycosidae, is their eye structure (Homann, 1961, 1966, 1971; Land and Barth, 1992). Since at least some of the lycosids have good vision (Land, 1985), a similar capacity may be present in Cupiennius.

With our previous neuroethological work on Cupiennius in mind, and following the idea of relating its visual capabilities to its behaviour, the prime objective of the present study is not so much visual physiology as such, nor the discovery of adaptations as sensational as those quoted for salticids and dinopids. Instead, we strive to determine whether C. salei is likely to see at night in the field and how much its four pairs of eyes differ in this regard. At the present early stage of this analysis, measurements of the electroretinogram (ERG) seemed to be both appropriate and sufficient.

Adult females and a few males of Cupiennius salei Keyserling from our laboratory stock in Vienna were taken to Japan and kept in climatized boxes either under natural daylight conditions or in artificial light:dark cycles (12h:12h) at 25°C. When kept under artificial light conditions (illuminance in the spider jars approximately 500–1100lx), the spiders were used for an experiment not earlier than 3 weeks after having been first exposed to these conditions. As shown by Seyfarth (1980), a few days suffice to change the activity rhythm of the animals according to the new L:D cycle. When comparing the responses of the spider in its subjective day and night (see Fig. 7), the animal in its day state was kept in the dark for 30min before the measurements to ensure dark adaptation.

A xenon arc lamp (500W, Ushio UXL 500 D-0) was used as a source of white light ranging in wavelength from ultraviolet to infrared. Quartz lenses formed a parallel beam of light, which passed, in the case of monochromatic stimuli, through one of 22 narrowband interference colour filters (Vacuum Optics Corp., Japan, IF-S) covering the wavelength range between 290nm and 700nm. With the aid of a quartz neutral density filter (optical wedge), the various monochromatic lights were adjusted to contain an equal number of photons.

The end of the quartz light guide that led the light to the preparation was mounted on a manipulator which allowed the light source to be positioned in the middle of the visual field of the eye under examination (Land and Barth, 1992). For stimulation with monochromatic light, the light-emitting end of the light guide (diameter 3mm) was 3mm away from the eye surface. The maximum intensity available (logI=0) was 3.11×10¹⁴ photonscm⁻² s⁻¹ as determined with a radiometer (K.K. Sanso S.S., model 4700). In experiments with white light, the distance of the light guide from the eye was 100mm. This guaranteed a patch of light large enough to be measured with a digital illuminance meter (Topcon IM-3, Tokyo Optical Co. Ltd). The maximum intensity (illuminance) available in our experiments (logI=0) was 900lx.

Stimulus duration was 15ms. The interval between stimuli was 20s in the case of spectral response measurements and intensity curve measurements. The precise timing of an experiment was controlled by using a computer.

ERG responses were recorded by placing the tip of a glass microelectrode (inner diameter of tip 78.3±20.6 μm, S.D.; resistance 0.55±0.83 MD, S.D.; N=10) very close to the eye’s surface so that the electrolyte bridged the small gap between the electrode and the lens and provided electrical contact. Polyvinylpyrolidone was added to the electrolyte (spider or butterfly Ringer’s solution, with no difference in recording quality) in order to reduce crystallization. The indifferent electrode was a silver wire slightly inserted into the opisthosoma of the spider. Conventional equipment was used to amplify, measure and record the ERGs.

Although the recording procedure of an ERG as such is straightforward, a particular problem in this case is the spider’s tendency to move its prosomal tergum with its powerful musculature. These movements result in a change of distance between the electrode and the eye. To circumvent this problem, the legs of the animal were tightly fixed to the support with adhesive tape and, in addition, the prosoma was embedded in dental wax and its tergum firmly glued to a rod attached to a mechanically separate manipulator. The spiders survived the experiments and could be used repeatedly.

In some of our graphs variation is given as S.E. of the mean. We use N and N′ to indicate the numbers of the spiders and eyes, respectively, used for an experiment and n to indicate the number of measurements.

The chromophore of photopigment contained in the eyes was identified using HPLC and following the methods described by Suzuki and Makino-Tasaka (1983). Chromophores extracted from the crayfish eyes (Procambarus clarki) were used as standards for 3-dehydroretinoids, chromophores extracted from the eye of Papilio as a standard for 3-hydroxyretinoid (Seki et al. 1987). A standard for retinoids was available commercially (Sigma).

Photographs of male and female spiders of Cupiennius salei and C. coccineus were taken using the following material: film, Kodak Tri-x; lens, 50% transmittance at 360nm and 81% transmittance at 380nm; filter, Kodak 18A, 320–390nm.

Like most spiders, Cupiennius salei has four pairs of eyes. These are called the anterior median (AM), anterior lateral (AL), posterior median (PM) and posterior lateral (PL) eyes (Fig. 1). Features characteristic of the genus and most valuable for quick identification in the field are (i) the specific arrangement of the eyes on the prosoma, (ii) the round circumference of the lenses of all eyes and (iii) a size relationship between the lens diameters of AM:AL:PM:PL eyes of roughly 1:0.5:1.5–2:1.5–2 (Lachmuth et al. 1985).

The ERGs recorded from all of the eyes are simple cornea-negative waves (Fig. 1) very similar to the monophasic ERGs long known for many arthropods (Autrum, 1958). Such ERGs are generally considered to represent responses of the retina only, not of the optic ganglia. This is likely to apply for C. salei as well, whose optic neuropiles are a considerable distance away from the eyes (Babu and Barth, 1984; Strausfeld and Barth, 1993).

In a few instances, a positive-going second wave was observed in the secondary eyes in their night state. No experiments were performed to trace the origin of this hyperpolarization.

The absolute values of the ERG potentials at maximum illumination intensity with white light reached 10mV, the PL eyes usually showing the highest values among all eyes. In most cases, however, the ERG amplitude measured only around 5mV. With monochromatic light stimulation at the maximum intensity possible in our setup, ERGs were mostly close to 3mV.

A general feature of the spectral response curves of all four pairs of eyes is the presence of a green peak at 520 or 540nm and of a smaller second shoulder in the ultraviolet at approximately 380nm in the posterior eyes and at around 340nm and 360nm in the anterior eyes. The response level in the spectral range of the ultraviolet shoulder is between 65% and 80% of that of the green peak and stretches from about 340nm to 400nm (Fig. 2). According to a comparison of the spectral sensitivity curves taken during the day and at night, the spectral position of the peaks remains unchanged, whereas sensitivity increases during the dark phase. This change is bigger in the median than in the lateral eyes (Fig. 2). The relative amplitude of the ultraviolet peak increases significantly at night compared with the day in the case of AM and PM eyes. The corresponding change is not significant in AL and PL eyes, however (Table 1).

Selective adaptation of the eyes to either ultraviolet or green light was used to get an indication of whether the spectral response curves reflect the presence of one or two, or indeed several, types of photoreceptors. To this end, the eyes were preadapted to either ultraviolet light at 364nm or to green light at 522nm for 30min before the measurement started. The intensity of the preadaptation light was set to elicit 10% of the maximum response of the respective eye achieved with white light. The preadaptation light was kept on during the measurement of the spectral response.

As shown by the examples given in Fig. 3, no evidence in favour of the presence of more than one spectral type of photoreceptor could be derived from these experiments. Absolute response values drop after selective adaptation, but the shape of the curve remains largely the same as before. There is no significant difference between the ratios of ultraviolet/green between the measurements taken after ultraviolet and green adaptation (t-test).

Fig. 4A presents a spectral sensitivity (SS) curve obtained from the conversion of a spectral response curve using the eye’s V–logI curves. When comparing individual values of this curve with values obtained for the corresponding wavelengths from V–logI curves for monochromatic light (such as that shown in Fig. 6), a very good coincidence is seen (Fig. 4A). This finding points to identical slopes of all the V–logI curves for monochromatic light and for white light. Consequently, V–logI curves for any wavelength or for white light can be used for the conversion of spectral response curves into SS curves.

Dartnall nomograms plotted onto SS curves (Fig. 4) fit the SS curves quite well at wavelengths longer than 530nm in all four pairs of eyes. We conclude that a single visual pigment is responsible for the response in the spectral range between 530nm and 650nm. In the range from 450nm to 500nm, the Dartnall curve clearly deviates from the SS curve. The SS curves of all eyes have a small shoulder at 480nm, which indicates the possible existence of a visual pigment maximally absorbing in the blue. Since there is also a peak in the ultraviolet, trichromatic colour vision is a possibility. However, sensitivity in the ultraviolet may also be due to the secondary peaks of green receptors, as described in insect compound eyes such as those of dragonfly (Eguchi, 1971), locust (Bennett et al. 1964), fruitfly (Kirschfeld et al. 1977) and butterfly (Bandai et al. 1992).

In all eyes, an ERG response is seen at stimulus intensities as small as 6–7logunits down from the maximum intensity possible in the experiments (logI=0), which was 900lx. With the luxmeter available, we could measure illuminance values down to –4.75logI, where they amounted to 0.01lx. Thresholds were clearly below this value.

The log–linear response range of the eyes covers a stimulus intensity range of 4 logunits. When plotting the original absolute response values, also used for Fig. 5, it can be seen that the slope of the intensity curve is steepest in the PL eye and flattest in the AM eye. This difference is statistically significant (t-test; P=0.017). It implies a smaller dynamic range in the PL than in the AM eyes. The PM and the AL eyes show values for the slope which are intermediate between those of the AM and PL eyes. According to statistical tests (Table 2), the PL eyes differ significantly from all other eyes regarding slope (m), whereas m does not differ for AM, AL and PM eyes. The statistical tests also show that the values for V_max/2 (K), referred to as relative sensitivity, do not differ significantly among any of the eyes.

Stimulation with monochromatic light at six different wavelengths was used to determine whether there are obvious differences in slope of the intensity curves. As shown by the representative example in Fig. 6, there are no such differences. This implies that, despite differences in threshold and relative sensitivity, the spectral sensitivity function S(A) can be expected to remain the same for different response levels.

Cupiennius salei is a nocturnal spider whose locomotor activity starts within less than an hour after sundown (Barth and Seyfarth, 1979; Seyfarth, 1980; Schmitt et al. 1990). We find that the responses of the eyes change with the time of day. Sensitivity in the AM eyes increases at night compared with daytime. Taking K as a measure of sensitivity, the night-time eye is up to 0.81logunits (1.97–1.16) more sensitive than the daytime eye. Such an increase in sensitivity may be brought about by a decrease in m (which is typical of invertebrates), by a shift of the curve towards lower intensity values (which is typical of vertebrates) or, indeed, by both of these mechanisms (Eguchi and Horikoshi, 1984). The average value of m representing the slope for nine AM eyes was 0.47 for the day and 0.42 for the night. However, in addition to a decrease in the slope, the curve also shifts towards lower values at night. In other cases of AM eyes, the shift was more prominent and indeed the main reason for the observed increase in sensitivity. Fig. 7A,B shows V–logI curves for an AM eye and gives the relative and the absolute response values for both the day and the night. In addition, the corresponding theoretical curves derived from the Naka–Rushton equation are presented. It can be clearly seen that the increase in sensitivity at night in this case is mainly due to a horizontal shift of the daytime curve towards lower stimulus intensity values but not to a decrease in the slopes.

Contrasting with the AM eyes, some of the PM eyes examined showed the same sensitivity (light intensity at V_max/2) at night and during the day. The average slopes (N=6) of the intensity curves were only slightly smaller at night (0.43) than during the day (0.45) (Fig. 7C,D).

The absolute responses of the AM eyes recorded at night are about twice as large as those recorded during the day (Fig. 7A). Threshold illumination shifts by about 1logunit towards lower values.

Using HPLC techniques, the chromophore of the visual pigment of all the eyes of Cupiennius salei could be identified as 11-cis retinal (Fig. 8A,B). According to the results shown in Fig. 8C,D, there is no 3-hydroxyretinal in any of the eyes of this spider species. In addition, no dehydroretinal was detected, using crayfish (Procambarus clarki) chromophore as a standard. The presence of nearly the same amounts of 11-cis and all-trans isomers as well as the presence of 13-cis retinal (as seen in Fig. 8A) is noted for light-adapted eyes. A high peak of 11-cis retinol is also typical of these eyes.

There are both ERG studies and intracellular recordings of the spectral sensitivity of spider eyes. The data so far available are from representatives of four spider families vastly differing both in life style and with regard to the behavioural significance of their vision: orb weavers (Argiopidae), net-casting spiders (Dinopidae), hunting spiders (Lycosidae) and jumping spiders (Salticidae) (see review by Yamashita, 1985). As one would expect, spectral sensitivities differ with regard to both the taxonomic affiliation and the type of eye (principal or secondary) considered. Cupiennius is a genus closely related to the wolf spiders (Lycosidae), both taxonomically (Homann, 1961, 1966, 1971; Lachmuth et al. 1985) and behaviourally. The various species are hunting or wandering spiders. In the principal eyes (AM) of wolf spiders (Lycosa baltimoriana, L. lenta, L. miami), only one type of receptor cell was found (De Voe, 1972). It responded maximally at 510nm (green) and at 360–370nm (ultraviolet). Intracellular recordings from photoreceptor cells of a secondary eye (AL) demonstrated the presence of only one type of cell, with the main response maximum at 510nm and a secondary maximum at 380nm (De Voe, 1972). This is in good agreement with earlier ERG recordings in the same species from all three secondary eyes (AL, PM, PL) and with the absence of selective adaptation effects (De Voe et al. 1969). In summary, the data presently available point to the existence of only one population of receptor cells in all of the lycosid eyes and, consequently, to the absence of colour vision.

There are obvious similarities between these findings in lycosids and our data on the ctenid Cupiennius salei. (i) One similarity is the spectral range (approximately 300–680 nm) eliciting a response. The behavioural significance of the sensitivity of the eyes of C. salei in the ultraviolet is not clear. Photographs of male and female C. salei and C. coccineus made with ultraviolet light did not disclose any ultraviolet reflection patterns. Light coming directly from the moon contains a relatively large amount of short wavelengths below 450nm (Menzel, 1979). It is not known, however, whether C. salei makes use of those wavelengths behaviourally, e.g. for orientation. Sensitivity in the green, in contrast, is more easily understood since light reflected from leaves and the soil is dominated by wavelengths longer than 450nm, i.e. green and yellow (Menzel, 1979). (ii) Another similarity to the lycosids is the dominance of the green peak (520 or 540 nm) in the spectral response curve and the existence of a shoulder in the ultraviolet (PM, PL: approximately 380nm; AM, AL: approximately 360nm and 340nm). We never found a dominant maximum in the ultraviolet, however, as was the case in the lycosid AM eyes. Also, the green peak is slightly shifted towards longer wavelengths compared with the peak in the lycosids. (iii) The position of the maximum relative sensitivities at the above-mentioned wavelengths remained unchanged by chromatic adaptation.

The conclusion from these observations is that there is no compelling evidence from our ERG study in favour of colour vision in C. salei, either in the principal or the secondary eyes. We note, however, that the deviation of the Dartnall curve from the SS curves between 450nm and 500nm and a small but consistent shoulder on the SS curves in the blue at 480nm leave the possibility of the existence of more than one spectral type of photoreceptor cell.

Considering the predominantly nocturnal activity of Cupiennius salei (Barth and Seyfarth, 1979; Schmitt et al. 1990), a lack of colour vision would not be surprising. So far, however, too little is known about the behavioural role of vision in Cupiennius salei, which has well-developed eyes (Land and Barth, 1992) and optic ganglia (Strausfeld and Barth, 1993; Strausfeld et al. 1993). In Dinopis subrufus, the enormously enlarged PM eyes, which are specialized for extreme sensitivity and used to capture prey at night at very low luminance levels, have only green receptors (Laughlin et al. 1980). The same is true for all other secondary eyes so far studied in any spider and has been taken to indicate colour blindness. The only known exceptions are the PL eyes of several species of orb-weaving Argiope. They have ultraviolet, green and blue receptors, not unlike the principal eyes (AM) of the same species and the principal eyes of salticids, which have di-, tri- or tetrachromatic vision and are thought to be capable of colour vision (see review by Yamashita, 1985).

Our present ERG data strongly support the possibility that Cupiennius salei does indeed use its visual capabilities, despite being a largely nocturnal species and therefore most active at times of very poor illumination conditions. Our results also point to differences in functional specialization among the four pairs of eyes. Judging from the V–logI curves (Fig. 5) and from the sensitivity values K (representing V_max/2) (Table 2), all eyes are about equally sensitive at night. There are some differences between the eyes, however. The slope of the V–logI curve is steepest in the PL eyes. The AM eyes, in contrast, seem to be the only ones capable of adjusting to daytime light conditions by shifting their V–logI curve towards higher values of light intensity (Fig. 7). In this regard it should be remembered that the AM eyes are the only ones not having a reflecting tapetum and that it was for them that the lowest K value (Table 2) was found.

According to field observations of Cupiennius salei in its natural habitat in Guatemala (Barth and Seyfarth, 1979; Seyfarth, 1980), the spiders do not leave their retreats until nightfall. After sunset, at illumination intensities of about 15lx, Cupiennius salei turns around abruptly above its retreat so that its prosoma points upwards. It then remains motionless for about half an hour until darkness is complete (for the human eye) and starts moving slowly up a leaf to wait for prey at illumination levels below 0.1lx. The ERG threshold intensities found in this study using white light stimuli were below 0.01lx, i.e. well within these ranges. Similar field observations are now available for C. coccineus in Costa Rica, a close relative of Cupiennius salei with very similar behaviour and habitat (F. G. Barth, D. Baurecht and A. Schmitt, in preparation). According to a recent study of the optical performance of the eyes of Cupiennius salei (Land and Barth, 1992), F-numbers of the lenses (f/D; f, focal length; D, lens diameter) are all between 0.58 and 0.74, i.e. impressively low and providing bright images. In addition, all secondary eyes have a gridiron tapetum. The tapetum doubles the effective length of the rhabdoms, which receive the light both directly and after reflection from the tapetum. Both these circumstances are thought to contribute to the sensitivity of the eyes of Cupiennius salei.

A behavioural role of vision must be taken into account not only shortly after sundown, however. A corneal illuminance of 3×10¹² photonscm⁻² s⁻¹ at half-maximal response to monochromatic light (522nm) and one of 3×10⁹ photonscm⁻² s⁻¹ at threshold indicates a sensitivity high enough to permit vision under conditions of moonlight and clear sky. The illuminance of the corneal surface of a spider eye that is irradiated by moonlight reflected from a white card (diameter 3mm and assumed to correspond to the tip diameter of the light guide used in our experiments) can be calculated from the luminance data and equations given by Land (1981) as 7.85×10⁹ photonscm⁻² s⁻¹. The equivalent value for reflected starlight is 7.85×10⁷ photonscm⁻² s⁻¹. Thus, under optimal conditions, the threshold intensity found for Cupiennius eyes suggests that vision is possible in moonlight but not in starlight. The fact that the values compared refer to 555nm (moonlight) and 522nm (λ_max, the most effective wavelength of the spider eye), respectively, does not affect the argument, since the corresponding change in sensitivity amounts to only about 5%. In the end, however, it is the retinal illuminance that determines the visual response. According to Land (1981), retinal illuminance J_R=J_S(D²×A_p)/(A₁×f²), (where J_S is surface illuminance, A_p is the area of the pupil, D is the diameter of the lens, A₁ is the area of the tip of the light guide and f is the distance between pupil and retina. Values for A_p and f are taken from Land and Barth, 1992). Retinal illuminance at threshold can be calculated to be 5.9×10⁹ photonscm⁻² s⁻¹, i.e. roughly twice as high as the illumination at the corneal surface. This does not change our argument. Absolute sensitivity may turn out to be even higher when measured with more sensitive methods, such as single-cell recording. The values now available are about 10⁷ times higher than those given for the PM eyes of Dinopis (Laughlin et al. 1980) and 10² times higher than those given for the PL eyes of the diurnal jumping spider Plexippus (Hardie and Duelli, 1978). Thus, the nocturnal net-casting spider Dinopis clearly remains the champion.

In our search for chromophores of the photopigments in the eight eyes of Cupiennius salei we found only retinal; there was no dehydroretinal or hydroxyretinal. Accordingly, retinal is taken to be the chromophore in both the principal and the secondary eyes. 11-cis retinal was also reported for another spider, Cryophoeca montana, by Smith and Goldsmith (1990). As shown in the chromatogram of Fig. 8A, a peak of 11-cis retinol is prominent and higher than those of 11-cis retinal and all-trans retinal. A large amount of 11-cis retinol in the retina probably indicates the storage of a precursor of 11-cis retinal. It may also suggest that retinol is a sensitizing pigment responsible for the ultraviolet sensitivity (secondary peak) like that found for a green receptor in the fly eye (Kirschfeld et al. 1977; Vogt and Kirschfeld, 1983).

The experimental work was carried out in the laboratory of E.E. It was begun while F.G.B. held a much appreciated guest professorship at the Department of Biology of Yokohama City University and could be finished there thanks to the financial support granted under the ‘Japan–Austria memorandum of understanding’ (BMWF, Wien; JSPS, Tokyo). Y. Shimazaki kindly provided the chromatograms. We are also grateful to H. Halbritter for help with some of the figures, to M. Wieser for perfect secretarial work and to two referees for valuable criticism.