State-dependent, visually guided behaviors in the nudibranch Berghia stephanieae

Gastropod mollusks have been shown to have visual responses ranging from phototaxis (Matsuo et al., 2014; Zieger et al., 2009) to high-resolution spatial vision (Irwin et al., 2022; Land, 1982). Moreover, gastropod species display a wide diversity of eye types ranging from open pit eyes to simple and complex lens eyes (Serb and Eernisse, 2008; Zieger and Meyer-Rochow, 2008). Nudibranchs have relatively simple lens eyes, whose behavioral functions are not known. Studying visually guided behaviors in nudibranchs has been challenging because animals are often wild-caught, limiting control over the animal's life history and internal state. To better understand the role of nudibranch eyes, we examined visually guided behaviors of a laboratory-raised aeolid nudibranch, Berghia stephanieae.

Nudibranch eyes are located beneath the integument near the brain. Many nudibranchs lack epithelial pigment over the eye, allowing it to be visible as a small black spot (Hughes, 1970) (Fig. 1). Each eye contains a spherical lens that is covered by a cellular cornea (Chase, 1974; Eakin et al., 1967; Hughes, 1970). Several pigment-producing cells shield light from entering the eye from behind. The eyes of adult nudibranchs possess only three to five photoreceptor cells (Chase, 1974; Eakin et al., 1967; Hughes, 1970), which is fewer than for other gastropods, which can have hundreds or thousands of photoreceptor cells forming an organized retina (Bobkova et al., 2004; Jacklet, 1969; Meyer-Rochow and Bobkova, 2001; Zhukov et al., 2002). Nonetheless, the positioning and neural connectivity of the photoreceptors in another aeolid nudibranch, Hermissenda crassicornis, suggest that they could support spatial vision (Stensaas et al., 1969; Tabata and Alkon, 1982).

Although spatial vision has not been demonstrated in nudibranchs, they have been shown to have phototactic responses to light. For example, the dorid Onchidoris bilamellata spends more time in the dark when given a choice between light and dark (Barbeau et al., 2004), whereas another dorid, Chromodoris zebra, and H. crassicornis spend more time in illuminated areas and approach light sources (Crozier and Arey, 1919; Lederhendler et al., 1980). Furthermore, when H. crassicornis encounter a shadow in an otherwise illuminated environment, they stop moving forward and return to the light (Lederhendler et al., 1980).

Anecdotal observations of B. stephanieae suggest that they spend most of their time in dark environments, such as underneath objects or in dark crevices. Here, we tested the responses of B. stephanieae to visual stimuli to gain insights into the visual behaviors and capabilities of these nudibranchs. We found that B. stephanieae exhibit visually guided behaviors, and we provide evidence of low-resolution spatial vision. Furthermore, we tested animals under different conditions and found that visually guided behaviors are state and context dependent.

Specimens of Berghia stephanieae (Valdés 2005) were initially obtained from Salty Underground (Crestwood, MO, USA) and Reeftown (Boynton Beach, FL, USA). They were propagated in the laboratory by placing an egg mass into a plastic Petri dish and incubating at 30°C. Artificial seawater (ASW; Instant Ocean, Blacksburg, VA, USA) was maintained at a specific gravity of 1.020–1.022, temperature of 22–26°C and pH of 8.0–8.5. ASW was exchanged twice weekly through manual pipetting. Late-stage juvenile B. stephanieae were transferred in groups of 10 to 1 gallon (∼3.8 l) acrylic aquariums filled with ASW and kept on a 12 h:12 h light:dark cycle. Unless otherwise indicated, animals were fed twice weekly with the sea anemone Exaiptasia diaphana (Carolina Biological Supply Co., Burlington, NC, USA). Exaiptasia diaphana were kept in glass aquariums filled with ASW maintained at the above conditions, and were fed brine shrimp (Artemia nauplii, Carolina Biological Supply Co.) twice per week.

Individual B. stephanieae were video-recorded while freely moving inside a circular arena, which consisted of a 9.5 cm diameter glass dish filled with 240 ml ASW. The dish was placed in the center of a 11.5 cm diameter white PVC pipe with a height of 9.5 cm. White cardstock paper was inserted between the pipe and the glass dish. Visual stimuli were printed onto the paper using a Color Laser Jet Pro M454dw (HP, Palo Alto, CA, USA).

An LED tracing board (tiktecklab) was fixed 15.25 cm above the testing apparatus to illuminate the arena (Fig. 2A). The light was turned off during dark conditions. To shade half of the arena, black cardstock paper was placed on top and on one side of the arena. The hemisphere that was shaded was rotated between trials. An 850 nm infrared light (CMVision) was used to illuminate the dish from above at a 30 deg vertical angle. Berghia stephanieae is insensitive to infrared light (Bui, 2021). A USB infrared-sensing camera with OV2710 CMOS sensor (webcamera_usb) was fixed 16 cm below the dish, and videos were recorded at 30 frames s⁻¹. When animals were tested with a visual target, infrared light was not used, and videos were recorded at 2 frames s⁻¹. The arena was rotated between trials.

All animals used were reproductive adults (1–2 cm length) and were tested at least 12 weeks post-hatching. Like other nudibranchs, B. stephanieae is hermaphroditic. Each animal was used only once, except when paired testing was performed as indicated. All experiments were performed during the daytime. Unless otherwise indicated, animals were tested 24–48 h after being fed. For experiments on food-deprived animals, testing was performed 5–6 days after their last feeding. To create conditioned ASW for food odor, six E. diaphana were kept in 200 ml ASW for 24 h; 10 ml of conditioned ASW was diluted with 230 ml ASW to provide food odor. The arena was cleaned thoroughly with 70% ethanol between trials.

For each trial, a single B. stephanieae was gently pipetted into the center of the glass dish. For tests in partially or uniformly illuminated arenas without a visual target, animals were given 5 min to acclimate to the arena, after which they were recorded for 10 min. Sample size was chosen using the resource equation approach, which suggested 11–21 animals for within-subjects repeated measures. In the half-shaded arena, 15 animals were tested for each feeding and odor condition (60 animals total). When animals were tested in arenas that were completely illuminated or darkened, the order of the light and dark trial was counterbalanced, and 15 animals were tested.

When testing animals with a visual target, 18 animals were tested for each stimulus type. After being pipetted into the arena, animals landed on the bottom of the arena and began the trial facing random directions with respect to the arena. Individuals were excluded if they did not right themselves immediately upon being pipetted into the arena. No acclimation period was used, and animals were recorded until they reached the edge of the arena or until 6 min elapsed. Animals that did not reach the edge were tracked and plotted but excluded from analyses, unless indicated.

The location of each individual within the arena was tracked for the duration of the trial using the markerless pose estimation software DeepLabCut (Nath et al., 2019). Networks were trained to detect B. stephanieae using training datasets in which animals were manually marked posterior to the first ceratal row. A different network and training dataset were used for each behavioral assay. The trajectories of each animal were exported into CSV files, after which they were analyzed using custom MATLAB scripts. Incorrectly labeled points were removed using criteria such as a likelihood score and the maximum possible distance to travel between frames. Arena boundaries were determined by manual segmentation using makesense.ai (https://github.com/SkalskiP/make-sense/).

To determine whether B. stephanieae approached visual targets, the distribution of the locations where B. stephanieae touched the wall of the dish were analyzed. Videos were trimmed to include the time from when the animal righted itself to when it contacted with the wall. The trajectories of animals were adjusted so that the first coordinate of each trace was located at the origin. The location that each animal traveled 95% of the distance from the center to the wall was identified as the point where each animal approached the wall. For each visual target tested, a maximum likelihood analysis using the R package CircMLE (Fitak and Johnsen, 2017) was used to compare how well uniform and unimodal models of animal orientation describe the distribution of the locations where animals approached the wall with respect to the visual target (Schnute and Groot, 1992).

The fit to each of three models of animal orientation was assessed: uniform, unimodal and modified unimodal. Model parameters of mean direction (MD), 95% confidence interval of the mean direction, concentration parameter (CP) and distribution size (DS) are reported for each model tested. For modified unimodal models, distribution size represents the proportion of individuals that oriented toward the mean direction in the given model, while the rest oriented randomly. The corrected Akaike information criterion (AICc) was used to compare models, and the model with the lowest AICc value was reported as the best fitting model. For each model, the relative difference in AICc value (ΔAICc) was calculated by subtracting the AICc of the best model from the AICc of a given model. Models with ΔAICc <2 were considered equally supported. There was strong evidence for additional models if ΔAICc was <4. Additionally, we considered models with ΔAICc between 4 and 7 to have moderate evidence if they also had an evidence ratio (ER) <10. The results of all models tested are reported in Tables 1 and 2. Berghia stephanieae was said to approach a visual target if there was strong evidence in support of a unimodal or modified unimodal distribution with a confidence interval that overlapped with the visual target. Confidence intervals that overlapped with the entirety of the arena were not considered. To compare how reliably animals approached different visual targets, we calculated the percentage of animals that approached the edge within the quadrant containing the visual target.

Behavioral measures such as mean speed, straightness and mean heading change were calculated. Mean speed was calculated as the total distance traveled divided by the elapsed time. Straightness index was calculated by dividing the radius of the arena by the total distance traveled from the center to the wall. Mean heading change was calculated as the mean of the absolute values of the angular differences between successive pairs of points. The percentage increase of mean speed and mean heading change in the dark compared with the light were calculated for each individual.

Statistical testing was performed in MATLAB unless otherwise indicated. To determine whether animals spent significantly more time in the dark than in the light, a one-sample Wilcoxon signed rank test was used to test whether the proportion of time spent in the dark was significantly different from random chance (50%). Paired t-tests were used to compare behavioral measures between the same individuals tested in the light and dark. Two-sample t-tests were used to assess differences in behavioral measures between independent groups. One-way ANOVA was used to compare straightness index between different visual targets (Tables S1 and S2). A two-way aligned rank transform ANOVA was performed using the R package ARTool (Wobbrock et al., 2011) to analyze the effect of food deprivation and the presence of food odor on the proportion of time spent in the dark (Table S3). Two-way ANOVA was performed to analyze the effect of food deprivation and the presence of food odor on the increase in mean speed and mean heading change in the dark (Table S4). Three-way ANOVA was used to analyze the effect of the presence of a visual target, food deprivation and the presence of food odor on mean speed and mean heading change (Table S5). Follow-up comparisons were made using Tukey's honestly significant difference procedure as indicated by ANOVA testing (Table S6).

When placed in an environment that was half-shaded, animals spent most of their time in the dark. The movements of 15 animals were tracked for 10 min in an arena that was half-illuminated and half-shaded (Fig. 2A,B). Following a 5 min acclimation period, 13 of 15 animals (87%) were located in the dark side of the arena. During the 10 min trial, all animals spent the majority of their time in the dark (Fig. 2C). On average, animals spent 83.6±14.2% of their time in the dark half of the arena, which was statistically higher than random chance of 50% (W=120, n=15, P<0.001).

While in the dark half of the arena, animals turned frequently and did not stay on the edge. In contrast, when animals were in the illuminated half of the arena, their paths appeared straighter, and they tended to stay near the edge. Fig. 2Di shows an example of an individual that started in the light half of the arena, moving along the edge, but once it reached the darkened side, it moved away from the edge and increased the frequency of turns (Fig. 2Di). Fig. 2Dii shows a different individual that started on the dark side, but after entering the light side, promptly turned around and re-entered the dark. Thus, B. stephanieae had a strong preference for being in the dark and showed notable differences in behavior between the light and dark sides.

To test whether differences in B. stephanieae’s behavior could be attributed to the level of ambient lighting, the movements of another group of 15 animals were tracked for 10 min in a circular arena that was darkened and one that was illuminated (Fig. 3A). The trajectories of individual animals were more consistent under both conditions than when in a half-illuminated arena. For example, an individual that circled the perimeter of the arena did so under both the dark and the light conditions (Fig. 3Bi), and an individual that entered the interior of the arena did so in both conditions (Fig. 3Bii). However, in the completely darkened arena, the animals rarely came as close to the edge as they did under uniformly illuminated conditions, as can be seen in the individual trajectories (Fig. 3B) as well as the density plots (Fig. 3Ci).

Under uniform illumination, animals frequently made contact with the edge of the dish and spent most of their time within a body's length (about 1 cm) of the edge of the 9.5 cm diameter dish (Fig. 3Ci). However, the proportion of time spent within a body's length of the edge in the lighted side of a partially illuminated arena (92%) was much higher than that in a uniformly illuminated one (68%; Fig. 3Ci,ii).

There were other behavioral differences between animals in uniformly and partially illuminated arenas. Animals crawled around 59% faster in a uniformly darkened arena than in one that was uniformly illuminated (Fig. 3D). However, in a partially illuminated arena, animals crawled over 200% faster in the dark side compared with the light side (Fig. 3D). Animals did not exhibit significantly different turning behavior in uniformly darkened and uniformly illuminated arenas (Fig. 3E). However, there was a significant decrease in the mean heading change when animals were on the light side of a partially illuminated arena, indicating that they turned less in the light (Fig. 3E). This is likely due to the increase in thigmotaxis in the light. Thus, B. stephanieae behaved differently in uniformly and partially illuminated arenas, suggesting that they may be responding to visual features of the environment and not just ambient light levels.

To test whether B. stephanieae approached visual targets, animals were placed in a uniformly illuminated environment with or without a single vertical stripe on the wall outside the arena (Fig. 4A). Animals placed in the center of an arena with no external markings typically changed direction several times before approaching the edge, and 5 of 17 animals (30%) did not reach the wall at all (Fig. 4B).

Without a visual target, the locations where animals approached the edge were best described by a uniform distribution (Table 1). There was also strong evidence that animals approached the edge with a weak unimodal distribution directed toward one side of the arena even though there was no visual target, suggesting that the animals may be influenced by other cues (Table 1). However, with a black stripe that extended 45 deg around the arena, every animal reached the wall and most animals approached the wall near the stripe, either moving directly toward it or making a large orienting turn before moving in a straight path toward the stripe (Fig. 4C). The locations where animals approached the edge were best described by a unimodal distribution directed toward the black stripe (Table 1). There was also strong evidence that animals approached the edge with a modified unimodal distribution with a distribution size of 0.52, indicating that 52% of animals approached the stripe and the rest followed a uniform distribution (Table 1).

When tested with a white stripe on a black background, all animals approached the black part of the wall rather than the stripe (Fig. 4D). These locations were best described by a uniform distribution or a unimodal distribution directed to the wall opposite to the stripe (Table 1). While over 70% of animals approached the quadrant with a 45 deg black stripe, only around 10% of animals approached the quadrant containing a 45 deg white stripe (Fig. 4G). However, over 75% of animals approached the quadrant directly opposite to the white stripe (Fig. 4G). In contrast, 25% of animals approached a quadrant in an arena without a visual target (Fig. 4G).

Animals were tested with stripes of various widths (Fig. 4F). When tested with black stripes between 15 and 45 deg, the locations where animals approached the edge were best described by a unimodal distribution directed toward the stripe (Table 1). When tested with black stripes larger than 45 deg, the locations where animals approached the edge were best described by a modified unimodal distribution with less than half of animals directed toward the stripe while the rest followed a uniform distribution (Table 1). Black stripes of 15 deg or wider were approached by over 50% of animals (Fig. 4E,G). Animals had the highest approach rate for a black stripe of 45 deg (Fig. 4G).

Animals traveled in direct paths when a visual target was present even when they did not approach it. The straightness of each animal's path was quantified using the straightness index (Fig. 4H). Animals traveled in straighter paths with a 25 or 45 deg black stripe compared with animals tested without a visual target (Fig. 4H). Although only about half of the animals approached a 25 deg stripe, all of those animals traveled in a straight path to the edge of the arena (Fig. 4F). Similarly, although a stripe that was 5 deg did not elicit approach, several individuals traveled in direct paths to the edge (Fig. 4F).

Animals also traveled in straight paths when presented with a gray stripe at 25%, 50% or 75% contrast, even though the majority of animals did not approach these visual targets (Fig. 5A). Furthermore, when comparing the paths of animals tested with a black stripe of at least 15 deg, straightness was not significantly different between animals that approached and did not approach a visual target (t₉₅=−0.84, 95% CI=[−0.16, 0.06], P=0.40). Berghia stephanieae moved in straighter paths when a visual target was present even when they did not approach it, suggesting that they use visual landmarks as a navigational aid.

Berghia stephanieae could be approaching visual targets through non-visual phototaxis or by using coarse spatial vision. Spatial vision is required for the detection of a visual target when it is isoluminant with the background. An isoluminant visual target was created by surrounding a 25 deg black stripe with two 12.5 deg white stripes on a 50% gray background, so that the average luminance over the 50 deg was the same as that of the rest of the arena. When tested with the isoluminant visual target, the locations where animals approached the edge were best described by a unimodal and a modified unimodal distribution directed toward the stripe (Table 1, Fig. 5C). Animals traveled significantly straighter when tested with an isoluminant visual target compared with animals tested without a visual target (Fig. 5B). Straightness was not significantly different between animals tested with an isoluminant visual target and animals tested with a 45 deg black stripe (Fig. 5B), indicating that animals were responding to contrast rather than luminance. This suggests that B. stephanieae has spatial vision rather than just sensing light and dark.

Berghia stephanieae seemed to have a preference for dark regardless of food deprivation or whether food odor was present. Animals were tracked for 10 min in a half-shaded arena following food deprivation (Fig. 6A, left), the addition of food odor to the water (Fig. 6A, middle), or both food deprivation and food odor (Fig. 6A, right). On average, animals spent over 80% of their time in the dark following food deprivation or when tested with food odor. Food-deprived animals that were tested with food odor spent around 65% of their time in the dark, and 4 of 15 animals (27%) spent the majority of their time in the light. However, when the preference for dark was compared between groups, there was not a significant effect of food deprivation or food odor, or a significant interaction between food deprivation and food odor on the proportion of time spent in the dark (Fig. 6B).

There were notable differences in the behavior of animals that were tested with food odor. Although animals frequently followed along the edge of the arena in the light (Figs 2A and 3A), this behavior was less common in animals that were tested in the presence of food odor. Without food odor, 11 fed animals and 6 food-deprived animals moved around the edge of the arena. In contrast, only 3 fed animals and 2 food-deprived animals followed along the wall with food odor (Fig. 6A).

All animals crawled faster in the dark side than the light side, regardless of food deprivation or whether food odor was present (Fig. 6C). This increase in speed in the dark was not impacted by food deprivation or the presence of food odor (Fig. 6C). In addition to increasing their speed in the dark, all animals had an increased mean heading change in the dark side compared with the light side (Fig. 6D). However, animals had a lower increase in mean heading change in the dark when tested with food odor (Fig. 6D). When mean heading change was compared in the light side for animals tested with and without food odor, animals had a significantly higher mean heading change in the light when food odor was present (Fig. 6E). Thus, animals tested with food odor were less likely to follow along the wall of the arena or straighten their path in the light.

Both food deprivation and sensing food odor impacted B. stephanieae’s propensity to approach a stripe (Fig. 7A). When tested with a black stripe, the locations where food-deprived animals approached the edge were best described by a uniform distribution (Table 2). There was also strong evidence for a unimodal distribution directed toward the stripe. When tested with a white stripe, the locations where food-deprived animals approached the edge were best described by a uniform distribution (Table 2, Fig. 7B). When given a food odor, the locations where food-deprived animals approached the edge of an arena with a black stripe were best described by a uniform distribution (Table 2, Fig. 7A). There was also strong evidence for a unimodal distribution directed opposite to the stripe. This suggests that food-deprived animals were less likely to approach visual targets.

When fed animals were given food odor, the locations where animals approached the edge were best described by a uniform distribution and a modified unimodal distribution directed toward the stripe (Table 2). There was also strong evidence for a unimodal distribution directed toward the stripe. Just over half of the animals approached the quadrant with the stripe, while 22% did not approach the edge at all, suggesting that there was a reduction in the propensity to approach the stripe when animals were given food odor (Fig. 7C). Additionally, fed animals that were given food odor made notably sharper turns and sometimes reversed direction completely rather than traveling directly to the edge; however, this was not observed in food-deprived animals that were given food odor (Fig. 7A).

Unlike fed animals, food-deprived animals did not straighten their paths when a visual target was present. To examine how food deprivation and food odor impacted B. stephanieae’s responses to visual targets, we analyzed the effect of a visual target, food deprivation and food odor on straightness index. There was a significant interaction between the presence of a visual target and food deprivation on straightness index (Fig. 7D), suggesting that responses to a visual target changed with food deprivation. Follow-up comparisons revealed that fed animals traveled in straighter paths when a visual target was present; however, this was not true for food-deprived animals (Fig. 7D). There was also a significant effect of food odor on straightness index (Fig. 7D), suggesting that animals traveled in more convoluted paths when food odor was present.

Overall, food-deprived animals were less responsive to visual cues than fed animals. We analyzed the effect of a visual target, food deprivation and food odor on mean speed. There was a significant interaction between the presence of a visual target and food deprivation on mean speed (Fig. 7E), suggesting that visual responses were impacted by food deprivation. In particular, fed animals traveled faster when a visual target was present, whereas food-deprived animals traveled with similar speeds with or without a visual target (Fig. 7E). There was also a significant interaction between food deprivation and the presence of food odor on mean speed (Fig. 7E). In particular, food-deprived animals moved slower when tested with food odor, whereas fed animals moved with similar speeds with or without food odor (Fig. 7E).

We found that B. stephanieae exhibits visually guided behaviors. Animals spent more time in dark environments and approached a contrasting visual target. When a visual target was present, animals crawled in straighter paths even when they did not approach it, suggesting that visual cues are important for navigation. Animals that were food deprived or given food odor had a reduction in behavioral responses to visual stimuli, demonstrating that visual responses are state and context dependent. Additionally, there was an even stronger reduction in behavioral responses when animals were hungry and encountered food odor, indicating an interaction between visual information, olfactory information and hunger state.

When given a choice between light and dark areas, B. stephanieae spent most of its time in the dark. Berghia stephanieae had distinct behaviors in the light versus the dark. For example, animals consistently followed along the edge of the arena in ambient light, but rarely in the dark. Following a physical surface after coming into contact with it is commonly referred to as thigmotaxis. Thigmotaxis is a spatial navigation strategy that has been observed in other animals, including other gastropods (Moisez and Seuront, 2020), insects (Jin et al., 2020), fish (Champagne et al., 2010; Sharma et al., 2009), rodents (Simon et al., 1994; Treit and Fundytus, 1988) and humans (Kallai et al., 2005; 2007). Animals are thought to perform thigmotaxis when they are trying to avoid or escape an environment. Berghia stephanieae used thigmotaxis only in illuminated environments and ceased this behavior in the dark, suggesting that it may use thigmotaxis to navigate back to dark areas following displacement.

Berghia stephanieae approached a dark vertical stripe on a light background. Although other gastropods have similarly been shown to approach dark vertical stripes (Chiussi and Díaz, 2002; Hamilton and Winter, 1982, 1984; Shepeleva, 2013; Vakolyuk and Zhukov, 2000), this is the first demonstration of this behavior in a nudibranch. Similar responses in other gastropods have been suggested to be related to seeking shelter (Chiussi and Díaz, 2002) or habitat selection (Hamilton and Winter, 1982; Shepeleva, 2013). Anecdotal observations in the laboratory suggest that B. stephanieae prefers to spend most of its time dark areas, such as in dark crevices and underneath objects. Additionally, Berghia feeds on anemones that are found in shaded areas on the roots of mangrove trees (Bedgood et al., 2020; Bellis et al., 2018). Thus, it is likely that Berghia approaches visual targets to seek out dark habitats that provide food and shelter.

Animals traveled in a straight path when a visual target was present even if they did not approach it, suggesting that B. stephanieae uses visual landmarks to navigate its environment. External cues are indispensable in allowing animals to navigate in a straight line (Cheung et al., 2007, 2008). In addition to approaching objects, moving in a straight path allows animals to navigate to new locations, whereas convoluted paths may lead animals to re-enter previously explored areas. In the absence of directional sensory information, even humans fail to navigate in a straight path (Dacke and el Jundi, 2018). When B. stephanieae was placed into an illuminated arena without any visual targets, animals changed direction several times before reaching the edge. The tortuosity of B. stephanieae’s path could be a result of the arena being void of directional olfactory or visual information.

In this study, we provide evidence that B. stephanieae is capable of low-resolution spatial vision. Differences in B. stephanieae’s behavior in the light and dark were stronger when light in the environment varied spatially than when it was uniformly illuminated. Berghia stephanieae most effectively approached a black stripe subtending an arc of 45 deg around the arena while thinner or wider stripes were approached less, suggesting that B. stephanieae is not simply moving toward darkness. Additionally, animals approached a 25 deg stripe that was isoluminant with the background, which suggests the detection of contrast rather than light intensity. It is therefore likely that B. stephanieae uses spatial vision to detect objects in the environment.

Studies of the anatomy and electrophysiology of nudibranch eyes provide potential neural mechanisms that could underlie spatial vision. Although nudibranchs lack an organized retina, the microvillous regions of photoreceptors in nudibranchs form distinct areas within the eye (Dennis, 1967; Hughes, 1970; Stensaas et al., 1969). Further, photoreceptor cells in Hermissenda crassicornis have been shown to have distinct receptive fields (Dennis, 1967). Additionally, there are inhibitory connections between the five photoreceptor cells in each eye (Crow and Tian, 2003; Detwiler and Alkon, 1973). Photoreceptor cells also inhibit neurons in the contralateral optic ganglion, demonstrating a convergence of visual information between the two eyes (Alkon, 1973). It was suggested that inhibition between photoreceptors or contralateral optic ganglia may support the detection of contrast (Alkon, 1973). Thus, nudibranchs may have the necessary components for spatial vision, and the results from the current study provide behavioral evidence to support this.

Visually guided behaviors in B. stephanieae were influenced by hunger state. Both fed and food-deprived animals had a preference for being in the dark. However, following food deprivation, animals were less likely to approach a black stripe or move away from a white stripe. Unlike fed animals, food-deprived animals did not move in straighter paths or increase their speed when a visual target was present. These results suggest that food deprivation in B. stephanieae reduces responses to visual cues.

Visually guided behaviors are also influenced by the presence of food odor. When water was conditioned with B. stephanieae’s prey, a sea anemone, animals still preferred dark environments, but they had a slight reduction in their propensity to approach a black stripe. Additionally, they showed changes in the style of locomotion, with animals performing sharper rather than smooth turns when food odor was present. Animals tested with food odor also rarely performed thigmotaxis. Additionally, fed animals that were tested with food odor often sharply reversed direction, while this never occurred in food-deprived animals that were given food odor. Together, these results indicate that there are interactions between hunger state, olfactory information and visual information that lead to changes in B. stephanieae’s behavior.

Visual responses in other gastropods are also dependent on internal state or the presence of olfactory information. Similar to B. stephanieae, the sea snail Nerita fulgarans approaches dark visual targets. When presented with a predator odor, N. fulgarans avoids rather than approaches visual stimuli (Chiussi and Díaz, 2002). The dorid nudibranch Chromodoris zebra ceased orienting to light when in the presence of conspecifics (Crozier and Arey, 1919). The sea slug Pleurobranchaea californica responds to food preferentially to light (Davis and Mpitsos, 1971). The aeolid nudibranch Hermissenda crassicornis changes its preference for light according to the time of day, with bright light being preferred during the day but not during the night (Lederhendler et al., 1980). When given a choice, H. crassicornis approaches a light over a food source, except when hungry (Alkon et al., 1978). Furthermore, stimulation of tentacular chemoreceptors inhibits the responses of photoreceptor cells in H. crassicornis (Alkon et al., 1978). In these gastropods, as well as B. stephanieae, visual information seems to be ignored during other behaviors such as foraging or seeking mates.

Although vision was not previously considered an important sensory modality for nudibranchs, the current study provides behavioral evidence that nudibranchs respond to visual features of their environment. Our findings demonstrate that B. stephanieae has visually guided behaviors that are influenced by hunger state and odors. It is likely that B. stephanieae uses its eyes for low-resolution visual tasks such as seeking dark habitats, approaching objects and navigating its environment.

We thank Amanda Cho for assistance with designing the half-shaded arena and collecting pilot data. We also thank Niah Holtz and Jackson Southard for assistance with data collection. Additionally, we thank Thi Bui for determining B. stephanieae’s sensitivity to long-wavelength light.