Obstract Io study how the species organizes its time, space, and social structure in a semi-natural situation, a group of sik Octopus rubescens were kept in a Sm diameter tank for one month. Observations were made of the win-loss record of each animal in interactions with other members of the group, the time spent by each animal in the different areas of the tank, and the various color patterns displayed by the animals in different situations. A pattern of dominance based on size emerged in which larger animals won encounters with smaller animals. The establishment and defense of a territory by at least one individual in the group sudgested territoriality for the species. Several color patterns were seen that did not fit previously published reports of typical patterns in the species, and imply that the literature on body patterning in octopuses is neither complete nor absolute. Introduction Octopuses are highly evolved invertebrates with a large degree of intelligence, and have been the subject of numerous studies of learning and nervous system function. Despite the extensive literature that has been developed on the learning and physiology of the genus, however, very little is known about the behayior or ecology of octopuses. Examining the ways in which an octopus interacts with others of its species, and how it makes use of its time and living space in a laboratory or semi-natural setting may provide us with clues about the ecology of the species in the field. There is especially little information available on the behavior of Octopus rubescens, a small benthic cephalopod that is found along the Facific coast of Morth America from Baja California to British Columbia. In Monterey Bay, California, where this study was carried out, C. rubescens is known to spend brief period in the plankton after hatching, then settle as juveniles in the kelp forest before migrating further oftshore to sandy mud bottoms (Hochberg and Fields, 1980). The only published research available on behavior in U. rubescens was performed by Warren, Scheier, and Riley (1974), who described the color changes observed during attacks on conditione and unconditioned stimuli. They found many of the same patterns described by Packard and Sanders (1971) for Octopus yulgaris, but defined the color changes during the attack sequence as systemati and independent of background coloration. They concluded that the color changes that occur during feeding are so strongly associated with the animal's motor activity as to be inevitable. The only other study specifically involving Octopus rubescens was detailed in an unpublished thesis by Dorsey (1976). Dorsey studied the natural history and social behavior of the species in Washington and concluded that the animals set up a dominance hierarchy based on size; she found no evidence that the animals established territorities. Because of the small size (12Ocm SOcm icm) of her tanks, however, her results may have been influenced by crowding. More research on the species is clearly needed before the its social organization can be definitively described. Frevious research on social behavior in other Octopus specie: has been limited, but two other studies have examined the types of social orgainization in laboratory octopus populations. Yarnall (1967) observed Octopus cyanea in large, semi-natural holding ponds in Hawaii. He found that the octopuses did not appear to defend territories, but set up a hierarchy in which smaller animals yielded to larger ones. Mather's (1980) study of Octopus joubini examined the use of space by the individuals in the her tank. Occupancy of all areas were not randomly distributed, but Mather concluded that the species was not territorial because the animals did not exclude one another from preferred areas of the tank. She, too, found that the octopuses showed patterns similar to those found in dominance hierarchies. Hanning (1972) defines a territory as 'any defended area. In order to determine whether Octopus rubescens organizes itselt into territories, several aspects of its behavior must be studied: a) how much time does each individual spend in the different areas of the tank? and b) does each animal 'defend' the area in which it spends the most time? Ony laboratory situation, of course, is only a poor substitute for field research, but the use of a semi-natural setting may provide us with a fairly accurate view of octopus behavior. By setting up a large tank with a simulated natural environment, obseryations can be made that may help develup an outline of octopus social organization. Materials and Hethods Octopus rubescens were collected from the sandy bottom of Honterey Bay, California by setting bottle traplines. All animale were collected from within cne-half mile of Monterey harbor, and from a depth of approximately 50 feet. In the laboratory, they were weighed and tagged with colored wire. Sexes were determined by lightly anaesthetizing the animals with a solution of 1.57 ethanol in seawater and inverting the mantle of each animal and examining the gonads. In the case of the two largest males, this process was not necessary because enlarged prowimal suckers and hectocotyluses were easily visible on these animals. Four males and three females were originally placed in the tank; one male crauled out of the tank and died after only one day, and the study included data from only the six animals present in the tank throughout the study. Animals were placed in the observation tank over a three-day period. The tank was round, with a diameter of Sm and a depth of SOcm, and was on a running seawater system. The bottom was coyered with sand in an attempt to mimic the animal's natural Habitat, and bricks, cement blocks, and jars were scattered around the tank for homes; there were always more available homes than there were octopuses. The octopuses were fed small crabs, a type of prey common in their natural environment. They were fed every day, about1.Ssmall crab per octopus per day. Octopus rubescens is nocturnal (Hochberg and Fields, 1780). and for ease of observation the animal's daily light cycle was reversed, with the tank darkened during natural daylight hours (O800 - 1800) and lit with flood lights from 1800 - O800. The 24-hour day was braken up into 2-hour periods, and zeveral perinds of abservations were made every day. Opprokimately 75 hours of observations were made during the cours of the study. Observations were made in the dark, with only a small amount of light entering through the walls of the room -- enough light to enable the observer to identify the animals after dark-adapting for a few minutes, but dark enough that a small flashlight had to be used for taking notes. The observer sat on stool beside the tank, sometimes walking quietly around the room to clusely watch interactions. The octopuses quickly habituated to this situation and showed mo response to the movement of the Ebserver. A written commentary was recorded where each octopus spent its resting time, where and when it moved, how it interactet with ather animals, and its body patterns during activity. Photographs were occasionally taken with a Fentax K-1000 camera with flash. Results Levels of activity For the first two weeks the octopuses were fed at the beginning of the subjective night (0800), and a daily cycle of activity emerged (see Figure 1). Activity levels were calculated by measuring the percentage of each half-hour period that each animal was moving about the tank, then averaging all animals and all observations for that period. When ted at 0800, octopus activity reached a sharp peak during the half-hour of feeding, corresponding with the animals moving about the tank hunting the crabs. Immediately following feeding activity dropped to almost zero for about 1 1/2 hours. presumably corresponding with the time taken to kill and eat one or more crabs. Activity then resumed and was maintained at an elevated level throughout the subjective night, dropped sharply when the lights were turned on at 1800, and remained low throughout the subjective day. For the third and fourth weeks of the study, the animals were fed during their subjective day at 2400 in an attempt to see to what extent the change in time of feeding would disrupt the activity cycle found above. On the new feeding schedule (see Figure 2), the activity level maintained its nocturnal character. The activity pattern associated with feeding was translocated to the new feeding time in the subjective day, but was otherwise unchanged. During the subjective night there was a 2-hour period of high activity following the onset of darkness, then a drop in activity follpwed by an increasing pulse of activity that wae again terminated by the lights turning back on. From this data it appears that the octopuses' daily cycle can be disrupted but not reyersed simply by feeding during their normal resting hours. Interactions Ihree types of interactions were observed: a) approach -- an octopus crauls or swims toward another, and the second octopue crawls or zwims away before contact is made; b) touch -- an actopus crawls or swims to another and touches the second animal with the guter half of one or several arms, and either animal crauls or swims away; and c) fight - an octopus crawls or swims to another and both animals grapple arms, trying to cover each cther with their interbrachial webs, and either animal crawis or suims away. No discrimination was made in the analysis of the different types of interaction; any time one animal ceded space to another it was recorded as an encounter and went into those animals' win-loss records. For each animal a win-loss record was tabulated and compared with the others (Table 1). One animal was a predominant winner, one lost all of its encounters, and the others won every encounter against certain animals and always lost to others. he winner was the largest animal in the tank, and most of the other octopuses never lost an encounter with an animal smaller than itself. Ihis suggests a hierarchy of dominance based on sise, in which larger animals win encounters with smaller animals (Table 2) Only in the two smallest animals, 7 and 3, did this patterr not hold true. This may be due to individual characteristics (e.g. trauma) of these particular animals, to chance because of the small number of encounters between the animals (1), or to the fact that 47 was female and may have been evading a mating attempt by 45, a male. In single-sex interactions, however, a linear pattern of dominance based on size is very clear. Territoriality The tank was divided into 12 sectors, o around the walls of the tank and a corresponding with homes in the center (Figure 3). Ihe amgunt of time spent in each sector of the tank by each anima. was measured (Table 3). No obvious pattern initially emerged: while 1 spent all of his resting time in one sector of the tank, the others distributed their time among several different areas. to, however, spent most of her resting time in sectors VIII and IX, which are adjacent to each other and could, combined, be called a territory. Using only level of occupancy as a criterion for territoriality, it appears that only fi and possibly 46 spend enough time in a single area to term it a territory. Ongther criterion is whether or not the octopus defends the area in which it spends the most time. To measure this, the win loss record and the number of encounters involving the animal in each sector were weighed against the amount of time the animal spent in that sector (Table 4). Again, no clear pattern is shown. It appears that the amount of time spent in an area is not necessarily related to the number or success of encounters in that area. However, almost half of all the encounters in the tank toom place in sectors XI, XII, and VII -- the sector in which fi, the dominant animal, made his home and the two sectors adjacent to it. All of the encounters that occurred in sector XII that involved animal 41 took place when 1 was forcing another animal from his home, or rebutting an attempt by another animal to enter the home while he was occupying it. In addition, although relatively tew encounters occurred in sectors VIII and XI, where (the second dominant octopus) spent most of her time, both initiated and won every encounter that took place there. All of these facts indicate that 1 and t6 may be defending territories. None of the avidence shows, though, that any of the smaller animals were actively defending territories. Color Fatterns Observations of color patterns were largely casual, made during the course of observations of interactions and teeding. Many different patterns were seen, including many of the body patterns cataloqued by Packard and Sanders (1971) in their study of Octopus vulgaris. Particular attention was paid to pattern changes during interaction, but relatively few color changes were obseryed. The octopuses were usually uniformly pale green throughout the encounter, whether it was a simple approach, a touch, or a fight. Occasionally, however, the initiator and eventual winner of an encounter assume an aggressive posture and pattern similar to that pictured in Figure 4a. The posture is upright; the color is uniformly green or mottled with dark bars extending from below each eye to the upper arms. In these encounters, the submissive animal would assume a defensive posture (Figure 4b) with arme curled underneath it; dark rings would sometimes appear around the eyes. These patterns were not consistent with any particular animals and could not be predicted to occur in any particular situation. During feeding, the color change sequence described by Warren et al. (1774) was often observed. In this sequence, the octopus is crawling or sitting in the tank before detection of the prey item, and exhibits one of many patterns. Upon detection and during a free-swimming attack, the animal assumes a color ranging from light orange to grey, then turns completely colorless upon landing. While seising the prey, the octopus is spotted or mottled, then returns to one of a variety of patterns while eating. Sometimes, though, no color change whatsoever occurred during the attack on a prey item, and sometimes the animal flushed a dark red at the moment of attack rather than blanching. Again, the occurence of the pattern sequence was not predictable. Discussion 10 Ihe activity levels found for Octopus rubescens are those one would expect to find in a nocturnal species, and are similar to those found in Octopus vulgaris by Wells, O'Dor, Mangold, and Wells (1983). The results show that the daily light cycle of the animals can be successfully reversed. This is especially interesting in view of the fact that nearly all of the many studies made on learning in Octopus species have been carried out in the daytime in normal daytime light levels, when the animal would normally be fairly inactive. Wells et al. (1983) feel that "behayioural experiments may be underestimating the capacities of the animal by testing its abilities during the wrong time of day." Heversing the daily light cycle of the octopus in the study could dive more accurate results in behavioral or learning researcn without reversing the diurnal cycle of the researcher. In a semi-natural situation, D. rubescens showed interaction patterns that could be called a dominance hierarchy and is probably based on size. This finding is in keeping with Varnalle (1967) obseryation of a dominance hierarchy in O. cyanea and with Hather's (1980) similar findings in O. joubini. A hierarchy in some form may be common to all Octopus species, as Nather suggested, but the strength of the hierarchy may well be ditterent in the field than it is in the semi-crowded conditions in the laboratory. Yarnall (1949) and Mather (1780) found that the species of octopus they studied showed no territoriality; Dorsey concluded 11 that U. rubescens was not territorial, but her results may have been affected by crowding. In the less crowded situation of this study, at least one individual O. rubescens showed clear signs of territoriality and a second showed what could be described as a sott' territory. It's obvious that the animals didn't ewhibit a rigid territoriality in the sense that every animal defends and it dominant in a specific area, but the clear territoriality of one or more animals suggests territoriality for the species as a whole. It should be noted that octopus 1, the dominant and territorial animal in the tank, was also the most active and wide¬ ranging of all the octopuses. He wandered frequently all gyer the tank, winning encounters wherever he went. It may be that his dominance and range of activity had an inhibitory effect on the other animals, preventing them from establishing territories. Ihis possibility should be tested by the removal of the dominant animal trom the tank to see how the use of space by the other animals in the tank is affected. If the territory of animal to becomes more rigid, or if the other animals begin to form territories as the tank becomes less crowded, the species could be said to be fairly strongly territorial, with the activity of the dominant animal having a substantial influence on the territoriality of the smaller animals. In locking for a type of social organization in her study, Mather (1980) assumed that dominance hierarchies and territoriality precluded each other, that one population could not display both types of organization: "...presumably it could be either a territorial pattern, in which each animal defends a piece of space, or a dominance pattern in which animals hold a ranking relative to one another but move in common space." Manning (17/2) states that a dominance hierarchy "relates not to a fixed area, but to a rank order of individuals living in a common area." But the findings in this study suggest that a dominance hierarchy and territoriality can exist in the same population. Again, the territoriality shown here is not rigid and is certainly a secondary structure to the dominance hierarchy, but the two types of organitation seem to be interacting in this situation. It's tempting to project the results found here to the species in its natural environment, but it's entirely possible that the affect of crowding in the laboratory is so great that any social structure found in the lab is a response to the artiticial conditions and not a typical pattern at all. The population density of O. rubescens in the wild is not known; it may be so low that the animals never interact enough to create a need for territories or even heirarchies. Research in the field is clearly needed. Ihe color patterns observed during the study were not well quantified, but the fact that patterns occurred that did not fit the cataloqued repertoire of patterns is significant. It appears that some of the body patterns displayed by the species have not yet been described, and that even those that have been described are not as sterectypical as might be believed. The lack of the systematic color change sequence described by Warren et al.(1774) in at least a few individuals of Octopus rubescens casts doubt on 1 the researchers' hypothesis that the color sequence is inseparably linked to motor activity. Further work is needed to comprehensively document the body patterns of Octopus rubescens. but a point to be made is that it's highly likely that none of the studies that have so far cataloqued color patterns in octopus are truly complete or absolute. Literature Cited Dorsey, E.M. 1976. Natural history and social behavior of Octopus rubescens Berry. Master's thesis, University of Washington: Seattle, Washington. Hochberg, F.G. & Fields, W.G. 1980. Cephalopoda: The squids anc octopuses. Chapter 17 in Intertidal invertebrates of California. Morris, R.D., Abbott, D. 2 Haderlie, E., eds. Stanford, California: Stanford University Fress. Manning, A. 1972. An introduction to animal behavior. Reading, Massachusetts: Oddison-Wesley Fublishing Co., pp. 97-100 and 249-250. Mather, J. 1980. Social organization and use of space by Octopus joubini in a semi-natural situation. Bull. Mar. Sci. 3o: 848-857. Fackard, A. 2 Sanders, G.D. 1971. Body patterns of Octspus vulgamis and maturation of the response to disturbance. Anim. Behav. 19: 780-790. Warren. L.R., Scheier, M.F. & Riley, D.A. 1974. Colour changes of Octopus nubescens during attacks on unconditioned and conditioned stimuli. Anim. Behav. T2: 211-219. Wells, M.J., O'Dor, K.K., Hangold, K. 2 Wells, J. 1783. Diurnal changes in activity and metabolic rate in Octopus vulgaris. Mar. Hehav. Fhysiol. 7: 275-287. Varnall, J.L. 1969. Aspects of the behavior of Octopus cyanea Gray. Anim. Behav. 17: 747-754. Tables Table 1. Win-loss record of each octopus compared with each other octopus. Reads left to right. Iable 2. Hierarchy of dominance set up by the octopuses, with the dominant animal at the top. Hierarchy is linearly based on size within each se. Table J. Fercentage of resting time spent in each sector of the tank by each octopus. Table 4. Fercentage of resting time (TIME), number of encounters (E), and percentage of wins (ZW) by each octopus in the twelve sectors of the tan. O E atataata- pl aa ktatatavava- taa p ktatatata- TABLE 2. DOMINANCE ANIMAL *6 *! *2 * 3 * 7 HIERARCHY 1209 oog 30g 909 459 709 G O 9 6 9 TABLE 3. III 10 VI VII VIII IX XI XII TIME SPENT IN SECTORS OF ONIMAL 15 62 0 5 100 27 TANK 28 44 2 aaao aaaaa- ataaataaataaaaakaaa- n kaataaaaaataatoaaa- aaaoa- aaataaaaataaaata- aaaaakaaaotaaa- aaaaaaaaaa- taaaaaaa- Luvavaavaaav- 8 8 2 2 2 5 Figures Figure 1. Daily cycle of activity for the octopuses when fed at 0800. Dark line denotes subjective night. Figure 2. Daily cycle of activity when fed at 2400. Dark line denotes subjective night. Figure Z. Diagram of the twelve sectors of the tank used in the study. (a) Oggressive posture and pattern sometimes shown by Figure 4. Fosture is upright: the initiator of an encounter. body pattern is uniform or mottled with dark bars below the eyes. (b) Submissive posture sometimes shown by the loser of an entounter. Fosture is defensive, with arms curled beneath the body; color pattern is unitorm with dark rings occasionally appearing around the eyes. 85 E T L P —— T k kaa- 8 5 (u09J00 13437 XLIAIOV I 3 5 5 1uJ0 k- — — L- L — — kkakaa- 115 XI XII SECTORS OF TANK ---- VI 111 — VII IX IV Vin FIGURE 3. .. . . . FIGURE 4. (a) (b) g n 2 Ocknowledgments I'd like to thank my advisor, Chuck Baxter, for his advice, for proyiding access to his seemingly limitless knowledge of things scientific, and for coming through with unsolicited (but much appreciated) help at the most frustrating points in the research process. John Kono was an invaluable ally in the construction and reconstruction of my elaborate tanks. Carol Marzoula came through whenever I needed help, and I appreciate her long hours on the water pulling up octopus pots as well as her long hours in the water looking for octopus pots. Stuart Thompson helped many in the course by buoying spirits when stress was at its highest, and the other professors helped make this quarter my most satisfying at Stanford. I'm grateful, too, to the six charismatic octopuses in my study for allowing me to put them in a tank, mees up their cycles, and force them to make small talk. Ond finally, I'd like to thank that "socially interactive" group, the students of Bio 17SH 1987, for ... whatever it is you thank yeur friends for.