Feel the light: sight-independent negative phototactic response in octopus arms

The octopus has eight long arms that are void of rigid skeletal elements. The control of motion in such a long and flexible appendage is extremely complex, as it involves control of organs with an extremely large degree of freedom. Not surprisingly, the octopus possesses poor proprioceptive sensation of its arms motion and position in space (Wells, 1978; Zullo et al., 2009), since such a system would require a huge computational load. These features have forced special evolutionary innovations of exceptional and sophisticated mechanisms that allow the octopus efficient control of its body (Levy et al., 2016; Nesher et al., 2014; Sumbre et al., 2005, 2006). Here, we report on a phototactic response in the octopus arm, which is mediated by the brain in an extraocular photoreception manner. This could allow the octopus to retract and protract its arms when exposed to danger, even when it is minimally or not at all aware of their position and posture. Extraocular photoreception is a well-documented phenomenon and has been observed in many and diverse species (Cronin and Johnsen, 2016). Florey (1966) was the first to report that increased illumination on the squid skin evokes chromatophore expansion. Later, Packard and Brancato (1993) reported the same observation in octopus skin, and recently the mechanism was investigated in Octopus bimaculoides skin by Ramirez and Oakley (2015). They termed this behavior ‘light-activated chromatophore expansion’ (LACE). Their work on O. bimaculoides skin indicates that LACE is induced through similar opsins, G-protein coupled receptors (GPCRs), to those found in the octopus eye. It has been suggested that the function of LACE is peripheral, local fine tuning of the camouflage to environmental luminance and/or that the skin photoreceptors serve as supplementary sensory organs to the eyes, sending the brain information about the surrounding environment (Buresch et al., 2015; Kingston et al., 2015). We have found an additional body response to light in Octopus vulgaris, which is sight independent and might reveal mechanosensation properties. In this work, we performed a set of behavioral experiments and surgical manipulations to elucidate and characterize the strong phototactic response of the octopus arm tip to illumination.

Octopus vulgaris Cuvier 1797 specimens (female and male, 300–1000 g) were collected by fishermen from the Israeli coast of the Mediterranean Sea. In the lab, each octopus was maintained in an individual 130 l tank connected to a semi-open running sea water (SW) system with controlled temperature of 20±2°C and 12 h:12 h light:dark cycle (illuminated by T5 fluorescent lamps). The ambient light in the aquarium area (indoor system) measured with a HOBO MX2202 light sensor, was around ∼30 lx (equivalent to 0.1 μmol m⁻² s⁻¹). The animals were fed raw defrosted fish meat three times a week. All animal handling and experimental procedures were in accordance with the ethics and regulations applicable in the Israel academy and nature conservation authorities.

In order to test the effect of prolonged change in environmental illumination conditions on the phototactic response, the aquarium was covered with a black opaque plastic sheet (Fig. 1), the ambient light in the aquarium was below the sensitivity of the light meter ∼0 lx. Apart from the illumination conditions, all other biotic and abiotic variables were unchanged. The octopuses were held under dark conditions, except for very short periods of 1–2 min, every other day, during feeding. The measurements of the phototactic response took about 30 min and were performed under regular illumination.

The influence of direct light on octopus arms was tested by illuminating the arms with a LED flashlight (LED 5000 lm, XML-T6). The flashlight was manually attached perpendicular (±10 deg) to the aquarium glass and the arm was directly illuminated when it was adjacent to the glass (Movie 1). The octopuses were tested throughout the day and no correlation was found between the time of examination and the characteristics of the response.

For the sensitivity examination, the light intensity was measured and adjusted before each experiment using a HOBO MX2202 light sensor; the logger was placed in the aquarium adjusted to the glass and was illuminated through the glass of the aquarium, in a similar manner to the arm illumination. The light measurement was logged at 1 s intervals and simultaneously transmitted through the Bluetooth of a handheld cellular phone, for online measurements. The measured units (in lux) were converted to PPFD (photosynthetic photon flux density), according to a conversion curve performed with a MQ-200 quantum meter and HOBO MX2202 light sensor. The adjustment of light intensity was achieved by incorporating a light dimmer into the flashlight electrical circuits.

To test whether the phototaxis is sight independent, the aquarium was covered with black opaque plastic sheet on top and an open tube on the lid which allowed the octopus to extend an arm out of the aquarium and reach for the food outside the tank (Fig. 1A). The arm was illuminated from a distance of ∼20 cm, while foraging arm on the covered lid was out of the octopus' eyesight (Fig. 1B). In a further experiment, the food on the lid was illuminated and arm response was examined when it entered the illuminated area (Movie 2).

All surgical procedures were performed under deep anesthetization according to a protocol developed by Shomrat, et al. (2008). Octopuses were immersed in 2 l seawater (SW) supplemented with 55 mmol l⁻¹ MgCl₂ and 1% v/v ethanol (96%) for 30–45 min. Octopuses were considered deeply anesthetized and ready for surgical procedures when they exhibited the following: pale skin, no voluntary movements, and limited or inability of the suckers to attach.

Arm amputations were performed as described by Nesher et al. (2014). In brief, the distal third of the arm was amputated by a single surgical cut, leaving the proximal portion undamaged. Then, the isolated arm section was rinsed in fresh SW and the octopus was returned to the aquarium for recovery. All the animals recuperated within a few minutes after awakening from anesthesia and exhibited normal behavior. Arm amputation is followed by regeneration of a functioning arm within several weeks (Fossati et al., 2013).

The octopuses were anesthetized during the entire surgical procedure. Both procedures were done under a dissecting microscope, with fine surgical scissors (Fig. 2). The incision site was designated at the point of a distal third of the arm. We severed the arm's main nerve cord with minimal damage to the surrounding muscle tissue. A sudden white appearance of the skin distal to the cut indicated a successful transection of the cord (Fig. 2B). Peripheral incision was achieved by cutting around the circumference of the arm while leaving the nerve cord intact (Fig. 2C). Following these procedures, the octopus recovered within a few minutes after being removed from the anesthetic solution and returned to the aquarium. The octopus's treated arm was then illuminated and tested for a phototactic response. Following this test, the section distal to the cut was removed in order to promote regeneration. The arms healed in less than a day, and initiation of regeneration was clearly observed after approximately 2 weeks.

In order to minimize the number of octopuses killed, examination of the involvement of the supraoesophageal brain mass and the optic lobes in LACE and the phototaxis mechanism were done on six octopuses that were dedicated for a different study on the neurophysiology of learning and memory (which is dependent on the supraoesophageal brain mass). Under the dissecting microscope, the optic lobes were disconnected and the supraoesophageal brain mass was carefully removed by cutting the brain tissue on the sides of the esophagus (for more details, see Shomrat et al., 2008, 2011). The octopuses, with only the suboesophageal brain remaining, were returned to the aquarium and recovered within a few minutes after being removed from the anesthetic solution. At the end of the short examination, the octopuses were euthanized under deep anesthesia.

Behavioral experiments were captured by video camera, and the movies were analyzed off-line by a ‘blind observer’ for both motoric reaction and arm behavior. Statistical analysis was performed using JMP Pro, version 15 (SAS Institute Inc.). Sight-independent tests, determining the effective luminance intensity 50 (ELI50) were examined in a binary manner (response versus no response). Two-tailed Fisher's exact test was used as goodness-of-fit test. Differences among datasets for phototactic response latency were verified by repeated measure ANOVA, interquartile range method was implemented for deduction of outliers, and post hoc Tukey HSD multiple comparison test was implemented when d.f. was higher than one.

ELI50 results were summarized, normal distribution was evaluated using the Shapiro–Wilk test, and difference among datasets were verified by two-tailed paired t-test.

While investigating the effect of light on octopus behavior, we noticed a curious phenomenon. Illuminating the arm tips of freely behaving octopuses elicited a negative phototactic response, meaning the arms withdrew from the illuminated area (Movie 1). First, we tested whether this behavior is sight dependent. Four octopuses were trained to obtain food by extending an arm outside the tank through an open tube in the cover of the aquarium (Fig. 1). The octopuses were ready for the experiment after 2 days of training. The top of the aquarium was covered with an opaque plastic sheet to prevent the octopus from seeing the light projected on top of the cover while their arm was searching for food (Fig. 1A). When the foraging arm reached outside of the aquarium it was directly illuminated by ultra-bright LED flashlight (mean±s.e.m.: 439.27±4.36 μmol m⁻² s⁻¹) on its tip, once per time the arm reached out (Fig. 1B). Negative phototactic response occurred in 84% (n=74) of the tip illuminations. In this experiment no food was actually offered, and the octopus's foraging behavior was a result of the anticipation of food according to the training.

Next, we tested whether sensing the strong illumination forced the octopus's arm to avoid the food. We placed a piece of fish on the covered lid out of the octopuses' sight (Movie 2). The food was continuously illuminated. Except for the change in illumination, no change in temperature or other abiotic condition at the food area were detected. We tested whether the foraging arm would sense the change in illumination around the feeding area and avoid the food. Indeed, when octopuses approached with their arms, they avoided the illuminated food in 88% of the trials (n=26).

In a further experiment we examined the sensitivity to light illumination in different sites along the arm (n=6 octopuses). Tip illumination resulted in a negative phototactic response in 92.3% of cases (n=26). This was significantly different to the 62.3% response (n=26) following illumination of the middle section of the arms (d.f.=1, P=0.039, N=52; two-tailed Fisher's exact test). Then, we monitored the latency of the response as an assay for the phototactic response reactivity. The latency was defined as the time from the initial moment of illumination until the beginning of the illuminated arm movement. The response latency was significantly shorter at the tip of the arm (n=13, median=0.77 s) compared with the middle (n=13, median=1.57 s), (d.f.=1, F=14.086, P=0.001, N=26; repeated-measure ANOVA). No phototactic response was observed at the base of the arms.

Two sets of experiments were done in order to test the adaptability of the responses to changes in environmental illumination and examine the long-term effect of maintaining the octopus under minimal illumination. In both experiments the octopuses were maintained in darkness by completely covering the aquariums with opaque nylon sheeting. In the first experiment we initially ruled out short term modifications in the response reactivity, by comparing the latency of the arms response to white light projection in three octopuses before (n=47, median=0.74 s) and after they were kept in darkness for 1 h (n=50, median=0.7 s). Then, the octopuses were maintained under darkness for 1 week, except for very short periods of 1–2 min, every other day, during feeding. At the end of this week the latency of the arm response was tested again (n=54, median=0.57 s; Fig. S1). Overall, the results showed a significant change in the latency of response (d.f.=2, F=14.443, P<0.0001, N=151; repeated-measure ANOVA) while no significant differences were detected after 1 h. In the second examination we tested the reversibility of the mechanism. First, the response latency of four additional octopuses was measured (n=276, median=0.81 s). Then, the octopuses were maintained under dark conditions (using the procedure described above). The response latency was measured after 1 week of darkness (n=114, median=0.68 s) and again after an additional week in darkness (n=153, median=0.68 s). After these 2 weeks, three of the octopuses were examined again 1 month after they were returned to regular conditions of 12 h:12 h light:dark cycle (one of the four octopuses died). As depicted in Fig. 3A, the results reveal a significant decrease in the latency of the phototactic response after the darkness period (d.f.=3, F=23.647, P<0.0001, N=644; repeated-measure ANOVA). Measurements of latency a month after returning to regular light conditions revealed a return to the initial value range (n=101, median=0.74 s; Fig. 3A). In an additional complementary experiment, we focused on the response sensitivity. We determined and compared the white light intensity that triggered a phototactic response in 50% of the illuminations, i.e. the ELI50 (see Materials and Methods for more details) in four octopuses before and after 1 week in darkness. The experiments were recorded and analyzed off-line by a trained blind observer familiar with octopus behavior. The results revealed a significant increase in the ELI50 after darkness suggest a decrease in octopus sensitivity after the dark period (d.f.=3, P=0.016, N=4; paired t-test, two-tailed; Fig. 3B).

For preliminary characterization of the neural pathway and control of the phototactic response and to reveal the central or peripheral nervous system involvement, we carried out top-down examinations through surgical and behavioral procedures (Fig. 2). We examined the findings in comparison to the LACE response. LACE and the arm phototactic response can be induced in octopuses by direct light projection. In an anesthetized octopus that exhibited pale skin and lack of voluntary movement, no phototactic response and only negligible LACE could be induced, while mild aversive stimulation of the arm induced local movement and color change in the skin. Complete removal of the supraoesophageal section of the brain along with disconnection of the optic lobe (suggested to be a pivotal motor control for chromatophore patterns; Liu and Chiao, 2017; Messenger, 2001) prevents the phototactic response, but has no effect on LACE which can even be activated in isolated skin tissue. We preceded these experiments with a set of surgical procedures at the arm level (Fig. 2). After transection of the arm's nerve cord (Fig. 2B) the arm become pale and both phototactic response and LACE distal to the surgery site are functionally disabled. Interestingly, although transection of the nerve cord prevented LACE in octopus, following amputation the same arm again showed LACE but not the phototactic response. Peripheral incision of the arm (Fig. 2C), i.e. around the axon tract while leaving the axon tract intact, eliminates the phototactic response while preserving the LACE. The portion distal to the peripheral incision showed quite normal motor behavior relative to the rest of the animal's intact arms (e.g. typical crawling movement and stretching). In order to test whether the peripheral incision prevents the phototactic response as a result of the traumatic procedure, we did a sham control with a peripheral incision of the arm circumference skin tissue, while leaving the muscle mass intact. We found that there was no effect on either the LACE or phototactic response in this sham control.

Here, we describe a phenomenon that has not been previously reported, of an extraocular photoreception mechanism in the octopus arm. We found that illuminating the tip of the arm causes a negative phototaxis response (Movie 1). Using behavioral experiments and surgical manipulations, we characterized this newly disclosed behavior and examined the findings in comparison to the LACE response, as both are extraocular photoreception phenomena.

The first set of behavioral experiments examined the initialization of the phototactic response while the arm was out of the octopuses' sight. The results of these experiments clearly confirmed that the phototactic response of the arms is sight independent, and that it is sufficient to modify behavior such as food gathering. In a further set of behavioral experiments, we aimed to examine the basic properties of this mechanism. By illuminating different locations along the arm while counting the successful elicited response and their latency, we showed that the tip of the arm is much more sensitive to illumination compared with the middle of the arm. These results could be due to differences in photoreceptor density. Another explanation could be that the phototaxis mechanism is restricted to the tip of the arm and the responses counted after illuminating the middle part of the arm were due to scattered light that stimulated the tip. Importantly, these results are in contrast to LACE, which can be initiated at different locations along the arm.

Next, we tested the adaptability of the phototactic response to prolonged change in the environmental illumination. Maintaining the octopuses in minimal light condition for 1–2 weeks resulted in decrease of the response latency (Fig. 3A). The decrease in response latency did not occur after 1 h in darkness (Fig. S1), which ruled out short term modifications. In addition, measurements of latency 1 month after returning to regular light conditions revealed reversibility of the mechanism as it returned to the initial value (Fig. 3A). In contrast to the decrease of the response latency (greater reactivity) due to prolonged darkness, there is a decrease in sensitivity (the light intensity that elicited response) as reflected by the ELI50 test (Fig. 3B). These dichotomous effects of a decrease in sensitivity along with an increase in reactivity, suggest a separate source for each mechanism. The decrease in sensitivity may occur due to alterations at the receptor level. For instance, there are several reports mainly from studies on invertebrates (Bloom and Atwood, 1981; Eguchi and Waterman, 1979; Meyer-Rochow, 2001; Röhlich and Tar, 1968; Zhukov et al., 2006) and also some vertebrate studies (Vistamehr and Tian, 2004) on degradation of the retina components, including photoreceptors, owing to prolonged darkness conditions. The increase in response reactivity may be the result of modifications at a higher level of the neural network processing. Therefore, more light will be needed in order to reach the threshold for activation of the response, but once the threshold is achieved, the latency, which is carried out by the nervous system, is shorter. Overall, this set of results revealed that the response behavior may adapt to environmental light conditions, thus suggesting an eco-physiological role that may be important for the octopus's ability to cope with dynamic environmental light conditions.

In order to reveal any nervous system involvement, we carried out top-down examinations of the phototactic versus LACE responses, through surgical and behavioral procedures (Fig. 2). The results (summarized in Table 1) point toward a reflex arc-like control configuration that involves the brain for the phototactic response, in contrast to the LACE local activation. The surgical manipulations revealed two interesting findings. The first one is the whitening of the distal part of the arm after transection of the arm's nerve cord (Fig. 2B), whereas amputating the arm re-demonstrated the pigmentation and LACE response. The second interesting finding was the ablation of the phototactic responses following a peripheral incision (leaving the arm nerve cord intact; Fig. 2C). Both these observations might imply that there is an essential neuronal or myogenic conduction pathway (backward or forward) passing through the muscle mass of the arm. Such a conduction pathway has not been described in octopus to date. However, the presence of musculature gap junctions has previously been described in several works, as in the cuttlefish stomach and between chromatophore muscle cells (Bone et al., 1995). Notably, a simple arm withdrawal reflex can be activated in an isolated arm (Hague et al., 2013); therefore, the lack of phototactic response is not due to impaired motor capabilities in an isolated arm.

In summary, our results revealed a CNS-mediated motoric response in the arm tip that is triggered by photo-sensation. This reflex-like behavior adapts to environmental light conditions and appears to play a functional role for the octopus. Yet, the question is what could be the role of this light sensitive response in a natural setting?

Full awareness of eight long and exceptionally flexible arms requires a huge and impractical workload on the nervous system. It is accepted that octopuses possess limited proprioceptive sensation for their arms (Wells, 1978; Zullo et al., 2009). Consequently, the octopus is poorly or not at all aware of its arms' location and posture when they are out of sight (Gutnick et al., 2011) Although, recently, Gutnick et al. (2020) broadened this view. Previous works have revealed unique and efficient strategies evolved in the octopus that allow handling of the arms, independently or with minimum involvement of the CNS (Hochner, 2012; Levy et al., 2016). For example, goal-directed movement (Sumbre et al., 2005, 2001), withdrawal response (Hague et al., 2013) and decision making-like behavior (Altman, 1971; Nesher et al., 2014).

During observations of octopus in our laboratory we noticed a recurring behavior that points to the physiological relevancy of arm tip phototaxis mechanism. During the daytime, octopuses kept theirs arms folded (Movie 3) while at night their arms are spread and fully extended (Movie 3, at 50 s). Several public domain videos (e.g. YouTube) of octopuses in the wild that were captured during daytime or night, also demonstrated this same alternating behavior.

We cannot completely discount the possibility that the phototactic response is a side effect of strong LACE that causes an aversive sensation due to drastic chromatophore expansion. However, the results and observations presented here, led us to hypothesize that this reflex-like behavior is actually a functional mechanism that might prevent the tip of the arms from being exposed to daytime predators such as fish and crabs that could identify the worm-like tips of the arms as food (such as the worm-like lure of the tasseled anglerfish). The fact that in contrast to LACE, the phototactic response is controlled by the CNS, allows the octopus to over-ride the instinctive response and to use its arms tips when needed in a goal-directed movement. Indeed, our observations revealed that when the octopus identified food by sight, in some cases (depending on the strength of illumination and the state of the octopus) it will extend its arm tip and reach for the food although it is being illuminated.

Interestingly, Ramirez and Oakley (2015) identified the LACE-activated opsin GPCRs of the octopus, on ciliated sensory cells that have been suggested to be mechanoreceptors, homologous to those in the fish lateral line system (Bleckmann et al., 1991). It is worth mentioning that recently, non-visual and light-independent functions such as temperature sensation (Leung and Montell, 2017), mechanosensation for hearing (Senthilan et al., 2012) and ciliated proprioceptors (Zanini et al., 2018) have been identified for opsin in the fruit fly Drosophila melanogaster. Therefore, it is not inconceivable that the phototactic response is initiated by opsin receptors stimulated by light to elicit a neural circuit that ends with a motoric response.

In conclusion, this work revealed another extraocular photoreception mechanism that may provide a further piece in the puzzle of how the octopus controls its arms using a cost-effective computational load and prevents its exposure to risk given its limited proprioceptive abilities.

We thank Tal Eyal, Eden Goldfarb, Ivgeni Tsigalnitski and Roi Siegelman for their preliminary contribution in the framework of annual project during their BSc studies. We thank Michael Apfelbaum for the drawings in the paper in Figs 1 and 2. We thank Mai Sadeh and Maria Roubanov for help with movie analysis. We would also like to thank Rafi Yavetz and Arik Weinberger from the Maritime Aquaculture Department at Ramot-Yam High School, for their help with octopus maintenance and the experimental apparatus design and setup. We thank Roxanne Halper for editorial assistance. Lastly, we would like to thank Benny Hochner for fruitful discussions.