The common octopus is renowned for its impressive intelligence. For instance, Octopus vulgaris has been observed in captivity to flatten itself and squeeze from its tank, sometimes traveling several meters through open air, in order to hunt fish in adjacent containers . There are reports of puzzle- solving behavior , play , and long-term memory formation  in controlled experiments. In the wild, octopi have been observed engaging in complicated hunting tactics, employing the chromatophore pigment cells in their skin for both camouflage  and for communication during pack hunts . On the coasts of Chile, octopi are known to engage in spear fishing, extending their range of attack by up to a meter . In fact, octopi have adapted several human hunting techniques, whether by observation or opportunity, frequently raiding lobster traps in the Mediterranean  and repurposing fishing nets for trawling the ocean floor .
In the following brief letter, we intend to demonstrate, for the first time, a level of intelligent, social behavior never seen in any species of octopus. We document evidence of complicated communication, hunting and trapping tactics, as well as animal husbandry, in a population of Octopus vulgaris living at ocean depths never before observed. We expect that our results will fundamentally redefine the scientific community’s understanding of the Octopus vulgaris and its intelligent capabilities.
The population we observed is located in the Delphic Gulf, approximately 35 kilometers southeast of Crete, at (34.6106, 26.6061) in latitude-longitude coordinates, at a depth of approximately 5.134 kilometers. We were drawn to the area by several reports from residents of nearby Xerokampos, Crete, of normally solitary octopi traveling in groups of as many as twenty along the beaches there. With the small exception of the larger Pacific striped octopus, which regularly forms groups of upwards of 40 individual octopi living in common dens , octopi are generally solitary and cannibalistic in nature. To find members of Octopus vulgaris behaving in this manner was of high interest to our team and prompted nearly five years of observational study that have produced this Letter, as well as the forthcoming quantitative study.
Our first foray into the field began with a standard tag and release operation: we fit 20 captured adult octopi with subcutaneous tracers and followed their hunting patterns for nearly two years. To our surprise, over the two years we followed our octopi, the tracer in every tagged octopus lost its signal within a 1 kilometer circular area in the depths of the Delphic Trench: a region whose ocean floor lies much deeper than Octopus vulgaris has previously been recorded venturing [11, 12]. Though it was entirely unexpected that all of our tagged octopi died within such a small region, we found no evidence of social behavior among the tagged octopi. Further, during those two years, brief scuba expeditions found no evidence of groups of octopi traveling along the shores of Xerokampos, and the members of Octopus vulgaris we did observe in the wild appeared to behave in every way normally.
Towards the end of our first funding cycle, we decided to perform a survey of the Delphic Trench, as one had not been performed for nearly a decade , and oceanographic equipment had improved considerably in that time. With assistance from the University of Tübingen, we were able to secure and pilot a Deep Sea Rover to the Delphic Trench for several days of analysis. We dedicated the first few days of exploration to standard analyses of temperature, water pressure, oxygenation content, biome composition, etc.; however, the breadth and focus of our exploration shifted considerably upon the discovery of large populations of Hinea fasciata (the striped clusterwink, a species of sea snail), known for its remarkable bioluminescent properties . The density and size of the snail populations we observed were beyond any previously recorded, so we invested several days collecting video recordings of the colonies.
It was during this time that, at depths of approximately 5 kilometers and in near total darkness, we collected upwards of 100 recorded Octopus vulgaris sightings around this population of snails. We have verified, after luring and capturing one of the octopi in question that it was indeed a common octopus—a fact easily considered impossible, given that these octopi were themselves also capable of bioluminescence. For reference, the only known bioluminescent octopi are the stauriteuthids  and the bolitaeninae , both of which are much more commonly found at abyssal ocean depths. We were able to genetically verify the species of a single captured octopus, but it was not able to withstand the rapidly changing pressure caused by the Rover’s ascent, so we have no record other than video evidence of these animals’ hypnotizing visual displays.
We had two key questions upon discovering these octopi: how did they come to exhibit bioluminescent properties? What function, if any, does their bioluminescence subserve? The former answer appears reasonably clear to us, though further validation will clearly be necessary: the octopi likely integrated bioluminescent proteins isolated from the sea snails into their skin. This feature is not unheard of in the animal kingdom, especially among ‘ornamental’ birds in tropical environments, for whom nutritional health is closely correlated with the richness of color and sheen of their feathers. Short of testing our hypothesis by feeding striped clusterwink to common octopi—a study we are currently preparing with our collaborators at the University of Tübingen—we analyzed the spectral profiles of light emitted from the octopi, as well as the snails, and found that the profile of light wavelengths present in the octopi were nearly identical to those in the snails. In contrast, different bioluminescent species normally emit different wavelengths of light .
For birds, the colors of their feathers are used for communication with members of their own species: we have several strong reasons to believe that the octopi we observe are using their bioluminescence for the same function. First, the octopi are able to modulate their luminescence by integrating it with their natural chromatophores. Chromatophores in octopi are normally used for camouflage , though the colored patterns propagating across the skin of both octopi and cuttlefish have recently been associated with complex emotions, planned action, and even dreaming . Chromatophores allow the animal to rapidly change its color in response to its environment, and can be used to either conceal or communicate. We observed between octopi what can only be described as a type of semaphore signaling, where combinations of bioluminescent activation and tentacle gesticulation consistently produced responses from surrounding octopi at distances of up to fifty meters, in a type of call-and- response communication.
The complexity and nature of the code used to communicate between the octopi are difficult to assess. Currently, in collaboration with the University of Essex, we are applying machine learning and computational linguistics tools to uncover the grammatical structure of octopus communication. The challenge is unique, because the grammar combines both light patterns and gestures. We expect this discovery, coupled with hundreds of hours of footage, to provide insight into a fundamentally alien language, not unlike dolphin or whale vocalizations. However, these insights are beyond our current technology, and much research innovation will be necessary for us to take even our first steps.
For now, we must be satisfied by chronicling the many uses we clearly observed the octopi put their language to. Of the twenty unique octopi we have been able to capture, most spent the majority of their time engaging in what can only be termed ‘herding’ behavior. These ‘shepherd’ octopi would steer the snails (numbering in the thousands) with their tentacles. We determined that the octopi had two objectives: to keep the snails together, and to elicit a consistent drift along contours of constant depth, so that the snails would neither ascend too high or descend to depths too great in water pressure for the octopi. We hypothesize that the octopi required the snail population to consistently drift along the edge of the trench in order to graze on detritus on the ocean floor. Plant particles are relatively sparse at abyssal depths, so maintaining such a large population of snails clearly requires considerable coordination and planning.
To accomplish this, octopi appeared to signal their locations with brief pulses of light. Faster, frequent pulses would spur nearby octopi to move to assist the signaler. We were able to decode these few signals because they were clearly tied to the octopus’s behavior; however, it is quite difficult to describe the full richness of the communication protocol employed. For this purpose, we intend to make publicly available several exemplar videos recorded simultaneously in both natural light and infrared. We believe that these recordings contain the first-ever evidence of animal husbandry carried out by aquatic animals.
In addition to the shepherds, we found that approximately five octopi would go ‘dark’ (ie. abstain from phosphorescent communication) and wait at the periphery of the snail colony as the shepherd octopi carried out their duties. Schools of fish or larger, isolated fish are drawn to the colony several times a day, undoubtedly due to the unusually high concentration of phosphorescent activity at the ocean depth. This steady influx turned out to be the octopi’s primary source of food, captured by a stunning display of pack hunting, not unlike those observed in pods of dolphins and humpback whales . Upon sighting an errant school of fish, a single shepherd octopus signals wildly, drawing the other shepherds in a spinning ring about the confused fish. Subsequently, from both above and below, the ‘dark’ octopi descend hidden, suffocating their frenzied and blinded prey by inserting their tentacles into the fish’s gills. In fact, our Rover suffered such an attack from the octopi upon our first approach: from then onward we were forced to record the octopi from distances of roughly 75 meters without engine activations or lights, so as not to agitate them.
This tactic appeared effective even on fish of moderate size, in addition to several common octopi who did not appear to be capable of phosphorescent signaling. This population of aggressive, pack-hunting octopi must be the cause of death of our tagged group of twenty normal octopi. The phosphorescent clusterwink population is visible from hundreds of meters away, clearly and consistently drawing curious fish and octopi from surrounding waters. It is possible that the shepherding behavior exhibited by the octopi that we observed has a cruel intentionality: it may serve as a means to draw unsuspecting prey to the octopi from great distances—as a trap.
On a given day, the ‘dark’ octopi exchange roles with ‘shepherd’ octopi several times. In addition, the group of approximately twenty-five octopi was not constant: we witnessed several instances of octopi entering the group from greater depths, prompting shepherd octopi to leave the group and descend further into the trench. This gives us cause to speculate that the true size of the octopus population may be much larger than we observed: preliminary estimates suggest that a colony of several thousand striped clusterwinks could support the bioluminescent capabilities of several hundred octopi. However, we have no conclusive evidence of such a population of octopi, and what role they could play in the colony is as yet a mystery: we hope that subsequent forays into the Delphic Trench may shed some light on this matter. Currently, as with all research, we feel as though our investigations have produced more questions than answers. Foremost in our minds is the following: why form colonies at all? Life for the solitary Octopus vulgaris at lesser depths is far easier, because depths that support higher rates of photosynthesis result in an abundance of food and biodiversity . Interestingly, the striped clusterwink is unable to live in shallower waters . Thus, we speculate that the clusterwink serves as a ‘forbidden fruit’: octopi must live at greater depths to feed on the clusterwink and allow for phosphorescent signaling, despite the fact that colony life is much more difficult in the depths of the Delphic Trench. One wonders how many of the Octopus vulgaris have abandoned the Delphic colony for the comparative Eden of lesser depths, at the cost of their own communication capabilities.
We hope that readers will find these preliminary results as exciting as we have found them over these past several years. We further hope that publishing this narrative will assist us in our petition to the Greek Conservation Agency to have the Delphic Trench declared a protected environmental zone. Our next step will be to place static cameras along the edges of the trench in order to image the octopi without the clear disruptions the Rover caused in their behavior. With more data and analysis, we hope to deliver to the scientific community a detailed account of one of the most isolated and unique societies documented on the planet.
FUNDING: The Southern Seaboard regulatory committee requires that we disclose all funding sources used to carry out this research. Graduate students and research assistants were funded through the National Oceanographic Scholars training grant; funding for postdoctoral fellows was secured privately on an individual-to-individual basis. The University of Tübingen was kind enough to donate the Deep Sea Rover employed throughout this study, and faculty salaries were paid by the University of Crete through HFRI resource allocation.
 Inglesberger, J. et al. (2007). Case Study: the Foraging Patterns of a Sequestered Octopus vulgaris. Aquatic Life, 9(2), 1428-1434.
 Cowan, W. (1997). Adapting Standard Cognitive Benchmarks for Interspecies Comparison. Frontiers in Cognitive Psychology, 18(45), 560-572.
 Johnson, A., & Sackes, D. (1992). Practice and Play: Rehearsal Mechanisms and Exploratory Learning in Marine Wildlife. Zoological Surveys, 2(19), 220-228.
 He, X. et al. (2017). Working Memory and Facial Recognition in Cephalopods. Oceanic Invertebrates, 11(3), 984-1001.
 Quinn, L., & Maulle, T. (1973). Novel Chromatophore Ethology Identification in the Octopus vulgaris. Journal of Marine Physiology, 14(4), 1822-1834.
 Johan, P., et al. (1999). Analysis of Pack Mentation and Communication in the Octopi of the Peruvian Coast. International Conference on Complex Systems, 18(9), 111-123.
 Agarwal, V., & Threshing, T. (2002). Lancing Techniques: a Comparative Study. Marine Anthropology, 13(2), 556-577.
 Burge, I., & Barnes, O. (2017). Technical Concerns for the CX90 Ultima Entrapment Apparatus. Compiled Fishing Fisheries and Fish Technologies, 17(1), 1442-1446.
 Theophilus, P., et al. (2010). Annual Report from the East Coast United States Fisheries Association. Environmental Protection Agency Public Reports, 34(4), 400-429.
 Sinclair, M., et al. (1998). Gregarious Beasts: Social Habits of the LPSO in Nature and in Enclosure. Zoological Surveys, 3(4), 387-396.
 Morales, J., & Linola, C. (1985). The Life Cycle of Octopus vulgaris: Mating, Dietary Habits, and Senescence. Journal of Aquatic Biodiversity, 2(5), 158-172.
 Hanauer, P., et al. (1994). Case Study: Documenting the Travels of the Common Octopus. Zoological Surveys, 2(5), 155-180.
 Killinger, K., et al. (2010). Subcutaneous Tag and Release of Octopus vulgaris. Ecology, 7(8), 1453- 1512.
 Libriani, G., Ovgana, R. (2013). Centrifugation and Chromatography for Isolating the Luciferase Enzyme from Hinea fasciata. Journal of Applied Biochemistry, 7(13), 1582-1593.
 Ham, L., et al. (2001). Photophore Control in Stauroteuthis Syrtensis. Normative Biophysics, 16(2), 98-114.
 Lundy, O. (1998). Bioluminescence in Pelagic Octopi. Journal of Marine Physiology, 3(4), 345-355.
 Eco, R., & Rinegold, H. (2018). FoURiER: A Novel Spectral Method for Taxonomic Classification of Bioluminescent Organisms. Journal of Applied Statistical Machine Learning Methodology, 5(9), 876-882.
 Libold, C., et al. (2014). Do Cephalopods Dream of Electric Partners? Decoding Latent Mental States from Chromatophore Activation in the Common Octopus. Big Data Reviews, 1(1), 140-180.
 Hellinger, G., & Fuzz, Y. (2004). Studying the Pack Hunting Tactics of Aquatic Mammals. Marine Anthropology, 9(2), 189-198.
 Roget, A., et al. (1989). Predation at Pelagic and Abyssal Depths. Journal of Aquatic Biodiversity, 7(5), 552-586.
 Joun, I., & Sorwild, A. (2003). Habitats and Habits of the Striped Clusterwink. Aquatic Life, 4(8), 662-674.
About the Author
Colin Bredenberg is a graduate student in Neural Science at NYU, studying how the brain learns to process visual information efficiently. He writes extensively, and most enjoys exploring how our perception affects our relationship with the world.