ABSTRACT Rolling is an aquatic interaction between two seals which involves head to tail somersaulting, mounting, and twisting and writhing movements as the seals bite each other. It has been alternately explained as male/female mating behavior, play, and dominance establishment; however, it has only been studied from land. This underwater study found that rolling is almost exclusively male, and probably serves to establish dominance and mating priveleges during the breeding season. INTRODUCTION Rolling is an aquatic interaction between two seals involving head to tail somersaulting together, biting, twisting vertically in a head to head position while exchanging blows to the throat and neck, and mounting. Several theories have been developed as to its purpose: dominance establishment (Sullivan 1981); pre-copulatory mating behavior (Venables and Venables 1955, 1957, 1959), (Beier and Wartzok1979); and play between immature seals (Bishop 1967), (Wilson 1974) or between seals of all ages (Renouf and Lawson 1986). However, there has been little agreement about the sexes of the two seals involved or the ages at which rolling occurs; rolling has also never been broken down into types, and descriptions of the behavior sometimes disagree on its components. Much of this confusion is due to the sole study of this behavior from land, from which the identification of the sexes of participants in the water is difficult; it is also almost impossible to accurately describe or time the behavior from the surface, as much of it occurs underwater. P. vitulina is normally very wary of humans; researchers have to construct blinds and observe behavior from great distances with field glasses so as not to frighten the animal. The harbor seal population at Hopkins Marine Station is very unusual in that the animals have become fairly accustomed to people and will allow divers and snorklers to observe them in the water. This allows underwater study of aquatic behavior close at hand, without any of the usual encumberments or distance handicaps of land studies, and provides a special opportunity to observe rolling behavior in the environment in which it occurs, facilitating sex identification and close examination of all types of rolling as well as the behaviors associated with it. MATERIALS AND METHODS This study was conducted from April 23 to May 29, 1991, at Hopkins Marine Station in Pacific Grove, CA. Approximately 37 hours were spent in the water observing seals, in the area between Seal Rocks and Agassiz Beach (see Figure 1). The habitat consists of granite rock and sand substrate; algal growth and surf grass are the prevailing cover. Seal Rocks is one of two major haul-out sites in the refuge for over 100 seals of both sexes and all age classes; the other haul-out is near the adjacent Bird Rocks. The Seal Rocks area was chosen for underwater observation due to accessibility and relative protection from wave action. The observational area has an average depth of about 3 m at high tide, with a maximum depth of 10 m in the deepest part during high high tides. At low tide approximately 50 seals haul out on Seal Rocks; as the tide rises some move into the surrounding waters while others migrate to the Bird Rocks haul-out. This study was carried out by snorkeling throughout the area, following an established pattern and regularly scanning the surface and listening for any signs of rolling in other parts of the study site. Data was recorded on underwater paper mounted on a plexiglass clipboard. In the first portion of the study, observation periods were randomized for time and tidal level; it was soon noted that rolling did not take place at low tide but was frequent at high tide, and from then on most observation was carried out between midtide and high tide. No observations were conducted at night due to personal safety considerations and poor visibility; in a past study, no seals were seen in the area or on the haul-out during night sessions (Shaw 1990). Foreflipper slapping, a behavior in which a seal smacks the surface of the water with the flexor surface of its extended foreflipper, producing a loud sound like a rifle-shot, could occasionally be heard at night; however, because this sound can travel over distances of 100 m it was difficult to determine the location, and it was a not sufficient reason to enter the water. Harbor seals are reportedly night foragers, and would thus be absent from the haul-out; supporting this, seals could be seen returning from the kelp beds early in the morning (Teri Nicholson, pers. comm.). Time, weather, tidal level, surface conditions, water visibility, and estimated depth of the deepest point in the area were recorded during each period of observation in an effort to correlate rolling with physical conditions, as well as to reveal any possible biases in data (such as no rolling viewed in water due to poor visibility). Rolling was defined as any behavior involving tail-chasing, mounting, twisting, and/or biting in a stereotypical fashion. One bout was the time from which two seals began rolling until they paused, even momentarily. An encounter was all bouts seen between a rolling pair until they either separated or were no longer seen rolling. Agonistic behaviors were classified both by location and length. Bites were classified as held (3 seconds or longer), actual, or attempted; bites were also categorized by position into throat bite, neck bite, tail bite, or body bite (any bite on trunk of body) Aggressive behaviors such as head thrusting and foreflipper scratching were also recorded. Location of bout, sex, age, and any individual characteristics were noted. Sex was determined by penile line and opening in the male and lack of penile line and opening and presence of teats (if visible) in the female. Age was determined by size and coloration, as well as sexual development. Individuals seen rolling were characterized by any prominent scars or other identifying marks, as well as grey or brown coloration, sex, and age. RESULTS Conditions: Rolling was observed most often at high tide and during overcast weather. Surface conditions and time did not seem to have any effect on rolling. Rolling occurred when the deepest part of the study site had a depth of at least 3 m, and when visibility was over 1.5 m. Rolling was never seen at low tide, and was observed in all locations; however, it was seen most frequently to the left of Seal Rocks or at the farthest point of the study site towards Cannery Beach. The behavior was observed at all levels in the water column, though some types of rolling were found more frequently at specific levels, as will be discussed later. No trends of increase or decrease were seen in rolling over the five week period. It was seen in 18 out of 33 observation periods, or 54.5%. However, these observation periods were not of equal length, but varied according to presence or absence of rolling. Duration of Bouts: 47 bouts of rolling were recorded (see Table 1). The average duration of a bout was 1 minute 17 seconds (range = 7 seconds to 5 minutes 30 seconds). Adult males were most frequently observed rolling: adult pairs were involved in 36/47 or 77% of all bouts, adult/subadult pairs were seen in 7 bouts, subadult pairs in 1 bout, and subadult/juvenile pairs in 2 bouts. No pups were seen rolling; however, this study ended just as they were being weaned. Adult/subadult bouts lasted almost twice as long as adult bouts, 2 minutes 18 seconds (range= 17 seconds to 5 minutes 30 seconds, n=7). The single subadult bout observed lasted 1 minute 25 seconds, and the two subadult/juvenile bouts averaged 28 seconds (range¬ 7 seconds to 49 seconds, n=2). The Mann-Whitney statistical test for small sample sizes was conducted on this data, and only the difference in time between adult/adult bouts and adult/subadult bouts was found to be statistically significant. Pairs identified as male were involved in 38/47, or 80.9%, of all bouts. Females were seen together in 1 bout, 5 bouts involved males with a seal of unidentified sex, and the remainder of bouts both seals were of unidentified sex. Males and females were never seen rolling together. The average bout between males lasted 1 minute 34 seconds (range= 7 seconds to 5 minutes 30 seconds); the one timed bout between females lasted 49 seconds. When analyzed using Mann-Whitney, the only significant difference in duration was between males and unidentified seals (X= 16 seconds, range= 8 to 24 seconds, n=3, p=0.0191). Duration of Encounters: The mean duration of an encounter was 3 minutes 11 seconds (range=7 seconds to 12 minutes 41 seconds, n=20). Adult males were again most frequent partners. Adult pairs (disregarding sex) were involved in 14719, or 73.7% of all encounters, with an average encounter duration of 3 minutes 14 seconds (range= 8 seconds to 9 minutes 1 second, n=14); the remainder of encounters were between adult/subadult pairs (X=5 minutes 23 seconds, range=17 seconds to 12 minutes 41 seconds, n=3), 1 subadult pair, and 1 subadult/juvenile pair. No significant difference in times was found with Mann-Whitney. Pairs of males rolled in 15/20 encounters, or 75% of all encounters. with a mean time of 3 minutes 55 seconds (range = 7 seconds to 12 minutes 41 seconds). No durations for encounters between females could be calculated due to inaccurate timing (two female/female encounters were observed over the course of the study). Again, Mann-Whitney showed the only significant difference in duration to be between males and unidentified pairs (X=16 seconds, range= 8-24 seconds, n=3. p-0.0270). General Pattern of Rolling: Rolling often began with nose-to-nose contact at the surface, either with the two seals facing one another or with one seal stretching his head back to touch noses with a seal close behind him. In some cases, the seals instead made eye contact at some level in the water column. One of the seals would then lunge at the other and rolling would proceed. Sometimes there was no initial eye or nose contact but one seal would just lunge at the other, with no warning. This was especially common in seals pausing between bouts. For example, in one encounter two seals (male) separated after a bout, one sinking to the bottom, the other rising to the surface and scanning the area. The seal on the surface suddenly swam down and bit its partner on the throat; they then swam off together and resumed rolling. Another way in which rolling may begin is one seal may be swimming past another, and the other seal may lunge at it and initiate rolling; however, it is difficult to determine whether the two seals had been rolling that day and temporarily separated, or whether they are two new partners. Once rolling begins, the two seals rarely change partners. I have only seen two seals rolling change partners once in five weeks of observation; two males were rolling head-to-tail when another seal swam past. One of the pair lunged at the passing seal and began rolling with it; the original partner did not try to resume rolling. Although 1 sometimes saw rolling seals head thrust or snap at other seals that got very close, this is the only time l ever saw partner-switching result from it. During an encounter or even during a bout, a rolling pair often switched between different types of rolling, blending them into a constant flow of movement (except mounting, which involved little movement and usually terminated or made up an entire bout). One of the pair sometimes blew bubbles during or after a bout. Sometimes one of the pair would have his penis extruded or bulge along the penile line during a bout. This was most common with mounting behavior. However, never more than one of a pair extruded his penis or showed this swelling along the penile line during rolling. Äfter a bout, the rolling partners usually stayed together. Sometimes one of the pair would rumble, an underwater vocalization behavior in which a male rolls onto its back, parallel to the bottom, and produces a "loud burst of sound, like a pop, followed by a low frequency rumbling roar (Shaw 1990). Sometimes another male will position himself beside the rumbling male so that their heads are very close, called attending (for a full review of rumbling see Shaw 1990). However, the rolling partner never attended the rumbling seal and the position was not the stereotypical position near the bottom; instead the rumbler swam in a half circle on his back while rumbling. Öften rolling would resume after he rumbled. Partners often swam off together, or went to the surface and scanned the area before resuming rolling. Sometimes they would swim in different directions, however, or just one would swim off. On three occasions when one partner swam off the other foreflipper slapped the water repeatedly. On a fourth occasion, however, a seal foreflipper-slapped the water while his partner looked on. When other seals were seen in the proximity of a rolling pair, they demonstrated two primary reactions: one was to watch the pair at whatever level in the water column they were occupying; the second was to freeze in an upright position and sink down to the bottom, from which the seal either watched or burrowed under eel grass or kelp, whichever was available. Females commonly demonstrated this sinking behavior, as well as smaller seals. If a seal too close to a rolling pair was snapped at, it would swim off, either a little ways or altogether. Types of Rolling: Over the course of the study I was able to identify six types of rolling (see Table 2). Some of them differ only by the plane of rolling or relative body position. They are described below. All bouts of rolling were divided into incidents, and each time a new type of rolling began it was considered a new incident. Head to Head Horizontal Rolling: This type of rolling has never been described in the literature. In head to head horizontal rolling one seal grasps the other by the throat or the back of the head with its teeth and begins spinning rapidly in a horizontal plane, in an extended bite (see Figure 2). The two seals are parallel to the substrate, their heads in close proximity, forming a straight line with their bodies. The seal being grasped does not seem to have any control over its speed of rotation; the grasping seal seems to completely dominate the encounter. I did not see this type after May 7. Head to head horizontal rolling comprised 5 out of 59 incidents of rolling, or 8.47% of all incidents. It was primarily engaged in by adult males: adult pairs comprised 4/5 or 80% of all head to head horizontal rolling; 3 of these pairs were male, while the fourth was a male with a partner of unidentified sex. The fifth incident was an adult/subadult pair of males. Neck biting and throat biting were both common; it was necessary to have one of the two to perform this type of rolling. In one case the seal being grasped was also seen foreflipper-scratching wildly. This type of rolling was roughest in intensity out of all types, due to the speed of revolution and length of biting combined. Head to tail vertical rolling, head to head twisting, and mounting stomach to stomach were all associated with this type of rolling. Head to Head Twisting: This type of rolling is comprised of two seals aligned in fairly close proximity, usually vertically in the water column, who exchange blows to the throat and neck (see Figure 3). As suggested from the description, throat and neck biting are most common. This type of rolling is typically medium in intensity. Head to head twisting constituted 5/59, or 8.47%, of all rolling incidents. It was also most common in adult males: exclusively adult pairs were involved in 475 incidents, three of which were male, the fourth a male with a seal of unidentified sex, and a male adult/subadult pair was responsible for the fifth incident of twisting. This type of rolling was associated with every type, most commonly head to tail vertical rolling and both types of mounting behavior. Head to Tail Horizontal Rolling: This type of rolling is one of two variations involving head to tail somersaulting; I have divided the two by the orientation of the bodies to the substrate. Head to tail horizontal rolling is tail-chasing on a plane parallel to the substrate (see Figure 4). Tail biting is most common, although biting is not integral to the act. Head to tail horizontal rolling was seen 3 out of 59 times, or 5.09%. This was seen exclusively in adult males. It was seen in association with its variant, head to tail vertical rolling, and with mounting back to stomach most frequently, but was also seen with head to head twisting and stomach to stomach mounting. It was sometimes associated with bubble-blowing, a behavior in which the seal emits a cloud of bubbles from its nose or mouth (which orifice of the head that the bubbles came from was undetermined). Head to Tail Vertical Rolling: This type of rolling also involves somersaulting tail-chasing, as in head to tail horizontal rolling, but the seals are instead oriented on a vertical plane (see Figure 5). This type of tail-chasing also involves tail¬ biting, but may also frequently consist of body biting or no biting, as well as head thrusting. Intensity varies widely, from mild(no biting) to rough (frequent and aggressive biting). Head to tail vertical rolling was the most common type of rolling, 23/59 or 38.98% of all rolling. Although it was also seen most commonly in adult males, it was also seen in all other age classes. Adult pairs participated in 13/23 or 56.5% of all incidents; of these incidents, 11 pairs were exclusively male, 1 a male with a seal of unidentified sex, and in one both seals were of unidentified sex. Male adult/subadult pairs were involved in 5 incidents, subadult pairs in two cases (one male, the other male/unidentified), and male subadult/juvenile pairs were observed in two cases. Head to tail vertical rolling was seen in association with all types of rolling, including mounting, and was sometimes associated with rumbling, foreflipper-slapping, and bubble-blowing. Stomach to Stomach Mounting: In stomach to stomach mounting, one seal clutches another ventrally with its foreflippers, often biting its throat (see Figure 6). The two seals are usually aligned on a horizontal plane, although they may be aligned at a slight angle. The seal being grasped usually freezes; rarely it will foreflipper scratch its opponent. Öften the two seals freeze in position and sink down to the bottom substrate, where they may remain for one minute or more. In males, one of the pair often had its penis extruded or showed swelling along its penile line. This interaction was commonly mild in intensity. Stomach to stomach mounting comprised 9/59, or 15.25%, of all incidents. This behavior was most often displayed by pairs of adult males. 7/9, or 78%, of all incidents involved pairs of adults; 4 of these were exclusively male pairs, 2 incidents involved a male with a seal of unidentified sex, and in 1 case the pair was female. 2 incidents of adult/ subadult male interactions were also observed. The female behavior differed from that of the male in that the mounting female did not grasp tightly but scratched at its opponents head with its foreflippers, and no biting was observed. This behavior most strongly resembled that of males by the stomach to stomach alignment of the two females. This behavior was seen in association with every other type of rolling. most commonly head to tail vertical rolling and head to head twisting Back to Stomach Mounting: This behavior was very similar to stomach to stomach mounting. except that the grasping seal and its partner were aligned ventral to dorsal, and neck biting was more common (see Figure 7). The grasped seal froze in position, as in the previous type, and the seals also frequently sank to the bottom substrate. The grasping seal frequently displayed a bulging penile line or an extruded penis, and in one case was actual seen thrusting his penis towards his opponent's stomach (also male). Usually, however, the two seals were almost immobile, and the intensity of the interaction was very mild; sometimes there was not even biting involved. Back to stomach mounting comprised 14/59, or 23.73%, of all incidents. Although it was largely an adult male activity (9/14 of all incidents involved pairs of adult males), adult/subadult male pairs were seen engaged in this behavior 3 times, and subadult/juvenile female pairs were observed twice. The behavior of the female pairs again differed from that of male pairs, and was similar to the female stomach to stomach mounting behavior; the female on top grasped the female on bottom briefly and then began foreflipper-scratching the bottom female's head. All types of rolling were associated with this behavior, most commonly head to tail vertical rolling and head to head twisting. Rumbling, foreflipper-slapping, and bubble-blowing were sometimes associated with back to stomach mounting. DISCUSSION This study coincided with the pupping season, late April to early May (Riedman, 1990) and the beginning of the mating season, 3 - 4 weeks after the birth of the first pup, at the end of lactation. I began to see healthy pups without their mothers in the middle of May, which should mark the beginning of estrus in female harbor seals in California. Unfortunately this study ended soon afterwards, affording little opportunity to witness related male-female interaction. Sullivan(1979), in studies on rolling in harbor seals in Humboldt, CA, reported no male-female rolling was observed until June and July; he also reported rolling was most frequently observed in July and August. However, I never saw any male roll with a known female, and was unable to note any increase in rolling over the course of this study. It would be interesting to continue the study of rolling behavior over a longer period of time, at Hopkins Marine Station, to compare any trends found to those of Sullivan. Sullivan (1981) also found that adult males averaged 5 minutes per bout, and that the "winner" displayed by forefipper-slapping or lobtailing in 33 out of 39 times. I found a mean bout time for adult males of 1 minute 11 seconds, and only saw foreflipper slapping in 4 out of 47 bouts. Of these 4 cases, 3 occurred when one of the seals swam away. determining a "winner"; in the fourth, one seal foreflipper-slapped the water when its rolling partner was still present (no winner determined) However, Sullivan did show similar trends in the ratio of adult males rolling to that of subadult males: he observed a ratio of 39 adult pairs to 5 subadult pairs, while the ratio 1 observed was 30 adult pairs to 1 subadult pair. There are several possibilities for this difference in results. Sullivan found a mean bout time of 2 minutes for pairs of subadult males, which is closer to the mean time l found for pairs of adult males. I determined age by size in the water, which magnifies size; due to this magnification and the small difference in size between age classes, there may be some overlap between age classes. However, Sullivan noted a similar ratio of adults to subadults rolling. It is also possible that my presence in the water with the seals shortened the times of bouts. Although seals at Hopkins Marine Station are more accustomed to people than most harbor seals, they still retain some of their species's notorious wariness, and a rolling pair has sometimes startled and stopped at my approach. Usually, however, the pair has paid very little attention to me, and 1 often had to get out of the way as they rolled towards me. Sullivan studied rolling from land; much of rolling occurs underwater, where he could not see it, so he may have been timing repeated bouts as one due to difficulty in determining an end. This would result in bouts of longer length. Many of the interactions I observed would not be visible from the surface, and would not be included in Sullivan's data pool; this could also account for differences in timing. Wilson (1974) described rolling in P.vitulina vitulina juveniles, which she classified as all seals between six months and 5 years of age. She noted that rolling bouts rarely lasted more that three minutes, and resulted in either separation or more rolling, a pattern 1 noted for both adults and younger individuals. She also noted that in rolling between "adolescents" (subadults in my research), one of the two sometimes had his penis extruded. During rolling l also sometimes noted the same sexual element; this both demonstrates the development of the behavior with age. saw very little juvenile (i.e. 1 - 2 years) rolling over the course of my study; juveniles were more wary of me than any other age group, and might not roll if I was present. It is also possible that juveniles do not roll as frequently as subadults and adults, in which the behavior was well developed. Pups were just being weaned at the termination of my study, and had just begun to form groups; no rolling between pups was observed. Of the types of rolling observed, head to head horizontal rolling had never been described in the literature. Head to head twisting has been described by Beier and Wartzok(1979) as synchronous spiralling in mating behavior of adult P.largha, a close relative of P.vitulina : "Male and female synchronously spiral around a common longitudinal axes as they swim, frequently surfacing". It has also been described by Renouf and Lawson(1986) as occuring for all ages, Venables and Venables (1955, 1957, 1959) as occuring in adults, and Sullivan (1981) as occuring in adults and subadults of P. vitulina, in association with head to tail rolling. The Venables described head to tail vertical rolling and Sullivan described head to tail horizontal rolling; Bishop(1967), Renouf and Lawson, and Wilson (1974) (juveniles) all described head to tail rolling without defining any plane of orientation to separate the two variants. Mounting back to stomach has been observed by Beier and Wartzok, Bishop, Renouf and Lawson, Venables and Venables, and Sullivan, directly associated with head to tail rolling in all but the work of Renouf and Lawson, who labeled head to tail rolling as another type of social play. However, only Beier and Wartzok noted stomach to stomach mounting, in P.largha. Sullivan (1981) and Venables and Venables (1955, 1957, 1959) both described rolling between adult and subadult males and females; Allen (1985) also noted " seals engaged in aquatic, precopulatory play" as described by the Venables and unsuccessful mounting of females by males. Bishop (1967) also saw males and females rolling, but noted male pairs or female pairs in such play with equal frequency, and described all individuals as immature. He did, however, note one pair of seals "apparently copulating" in the water, the larger grasping the smaller with its foreflippers, "rolling languidly“, submerging, surfacing, and drifting. Bishop also witnessed brief, aggressive "breeding encounters“ on land, which he contrasted to rolling. Sullivan saw no copulation in water or on land; Allen observed four copulations over two days, all on land. These land copulations are unusual as harbor seals breed in the water (Riedman 1990). The Venables saw copulation after rolling and bubble-blowing in "calm, clear water". For a review of literature on harbor seal mating, see Thompson (1988). Both Bishop and the Venables listed the seals as "pairs", but it was unclear how sex of each pair member in the water was verified from land. The Venables had no explanation of their methodology for sexing seals in water other than "sometimes one of the rolling pair will have the penis extruded" (Venables and Venables1957); the other member of the pair was assumed to be female. I often observed one member of a pair with its penis exposed in rolling and mounting, but the pairs 1 observed were exclusively male. It is interesting to note that l never saw more than one of the pair in this state at a time. The Venables described coition as the "male" mounting the "female“ and grasping her with his foreflippers, sometimes biting her neck, the pair sinking to the bottom where they remained "in fairly static coition posture for 1 to 3 minutes". I observed this exact behavior, but both members of the pair were male. In one case a male with his penis extruded mounted back to stomach to another male and made thrusting movements with his pelvic region. Beier and Wartzok (1979) demonstrated in their study of captive spotted seal mating behavior the difficulty of determining intromission in field studies of mating; even looking into "calm, clear water", an observer on a cliff face would not be able to determine whether actually mating was occurring, and could misinterpret malelmale behavior as male/female. Backhouse and Hewer (1957) observed homosexual activity in male grey seals during mating season and Johnson and Johnson (1977) suggested rolling was an exclusively male activity, accompanied by mounting and penile extrusion. Sullivan (1981) also noted males rolling and "riding" one another. I only observed male pairs mounting each other in a sexual manner (the female/female mounting observed was not sexual in nature). Rolling has been observed by many researchers, in seals of all ages and sexes, and in many contexts. Many suggestions have been made as to its significance in the social system of the harbor seal, explaining it as dominance establishment, mating behavior, erotic play unrelated to mating, or a combination of dominance establishment and mating behavior. However, all previous studies were done from land, which made identification of the age and sex of seals in the water difficult, as well as description of their aquatic behavior. Underwater observation facilitates both sex identification and close observation, and 1 can verify both the exclusive male activity and the strong resemblance of the mounting between males to copulation. This suggests that rolling is a form of dominance establishment with a pattern of development seen in behavior from juvenile play to adult agonistic encounters. Sullivan (1979) postulated that rolling was both dominance establishment between adult males and a form of pre-copulatory behavioi between females and males which allowed females to further assess a male who had won a rolling bout. However, I did not see any male/female rolling over the course of my study, which leads me to believe that rolling is an exclusively male activity. Sullivan did not observe any rolling between males and females until June and July, whereas this study terminated in June; according to this, I should have observed male/female rolling in the next month of observation. However, it is interesting to note that females were never seen engaged in the play resembling head to tailing rolling or head to head twisting described as mating behavior by Venables and Venables (1955, 1957, 1959) and Sullivan (1981) in P.vitulina, or Beier and Wartzok(1979) in P. largha. Younger animals often play at adult sexual or aggressive behavior; for example, male elephant seal juveniles may be seen holding mock battles and practicing copulatory behaviors upon one another (Riedman 1990). Although males of many species are more frequently observed to practice copulatory behaviors than females, it seems odd that males but not females would rehearse such a complex series of behavior. Wilson (1974) observed rolling play between juveniles and subadults both in the wild and in captivity. She did not note sexual identity of seals in her study of wild P. vitulina, except to note one sometimes had his penis extruded; however, in her study of captive seals the males were observed to roll most often (one of the females would sometimes roll with a male for less than one minute). Wilson had also noted that although both juveniles and adolescents rolled, juvenile rolling had no sexual components, which also suggests a pattern of behavioral development in males. I observed very little rolling in juveniles and subadults, but this could be due to the smaller proportion of younger seals to adults, especially as the Hopkins Marine Station population has just become a breeding population in the past few years, as well as to the general wariness of juveniles. A year-round study of rolling noted that the frequency of rolling increased during the mating season and disappeared at the end of the season (Sullivan 1981). M.A. Bigg (1969) found that testosterone in adult and subadult males increased in March through November, which coincides with the mating season; this increase may both account for the aggression of rolling and the mounting attempts and penile extrusion in adult males seen during this study. Harbor seals are slightly polygynous; since they breed in water, which is boundaryless, they cannot establish territories or control females, and must establish dominance in some other way (Sullivan 1981). Aggressive pinnipeds establish dominance where they breed (Riedman 1990); for example, elephant seals have been well documented as breeding and fighting on land. Harbor seals breed in the water, and should thus establish dominance in the water, where they are mobile and agile, rather than on land, where they are slow and awkward. Rolling is both the most aggressive and the most vigorous behavior in which l ever observed harbor seals participating in the water; the biting and chasing can become very rough. although 1 never saw blood drawn. Although harbor seals mature sexually between 5-7 years of age, they may not become socially mature for several years after that (Riedman 1990). Over the course of this study, a small group of large, scarred males was seen to haul-out daily in the Bird Rocks area, frequently carrying bloody marks in the morning (Teri Nicholson, personal communication). This could mean that dominance is being established in some way other than rolling. However, it is just as likely that these older males are rolling at night or away from the haul-out, and the males 1 observed rolling most often were younger adults which had not established a stable dominance pattern or joined the dominance hierarchy. This could also explain the occasional extrusion of a male's penis during rolling; if these seals have not yet integrated into the dominance hierarchy, they are probably not yet mating, and thus might show homosexual behavior. It is interesting to note that I sometimes saw rolling in association with rumbling, a behavior Shaw (1990) connected with non-violent dominance establishment. Rolling is probably a form of dominance establishment which shows a developmental pattern in males, and provides a way of establishing breeding privledges in the three-dimensional, boundaryless aquatic environment. ACKNOWLEDGEMENTS thank Stuart Thompson for lending me gear so 1 could get in the water; Dennis Powers for lending me The PInnipeds so 1 could know which animal to look at once I was in the water; and Alan Baldridge for sharing his enormous wealth of knowledge of pinnipeds with me so that 1 knew what was looking at once 1 found the animal. I would also like to thank the other "Seal People" for sharing information and interest; my roomates Jen and Edith for their help and understanding; my parents for providing the funds that made this quarter possible; Jim Harvey for advice and information; Nicole Crane for gear, footage, endless camera help, and everything eise 1 forgot; and Chuck Baxter, for his patience, advice, "red marks", and for still taking time for students after so many years of teaching. Most of all, I would like to thank the Hopkins Marine Station seals for making this wonderful experience possible. LITERATURE CITED Allen, S.G. 1985. Mating behavior in the harbor seal. Marine Mammal Science 1(1): 84-87. Backhouse, K.M. and H.R. Hewer. 1957. Behavior of the grey seal (Halichoerus grypus) in the spring. Proceedings of the Zoological Society of London 129: 450. Beier, J.C. and D. Wartzok. 1979. Mating behavior of captive spotted seals (Phoca largha). Animal Behavior 27: 772-781. Bigg, M.A. 1969. The harbor seal in British Columbia. Bulletin of the Fisheries Research Board of Canada 172; 1-33. Bishop, R.H. 1968. Reproduction, age determination, and behavior of the harbor seal, Phoca vitulina L, in the Gulf of Alaska. Master's thesis, University of Alaska. Johnson, B.W. and P. Johnson. 1977. Mating behavior in harbor seals (unpublished abstract). Second conference on the biology of marine mammals. San Diego, CA. Renouf, D. and J.W.Lawson. 1986. Play in harbor seals (Phoça vitulina). Journal of Zoology (London) 208: 73-82. Riedman, M. 1990. The pinnipeds: Seals, sea lions, and walruses. University of California Press, Berkely, CA. Shaw, T. 1990. Underwater behavior of the harbor seal, Phoça vitulina, near the haul-out. Stanford University Hopkins Marine Station Final Papers Biology 175 H 1990. Sullivan, R.M. 1979. Behavior and ecology of harbor seals, Phoça vitulina, along the open coast of northern California. Master's thesis, Humboldt State University. Sullivan, R.M. 1981. Aquatic displays and interactions in harbor seals, Phoca vitulina, with comments on mating systems. Journal of Mammalogy 62(4): 825-831. Thompson, P. 1988. Timing of mating in the common seal (Phoça vitulina). Mammal Review 18: 1-8. Venables, U.M. and L.S.V. Venables. 1955. Observations on a breeding colony of the seal Phoca vitulina in Shetland. Proceedings of the Zoological Society of London 125: 521-532. Venables, U.M. and L.S.V. Venables. 1957. Mating behavior of the seal Phoca vitulina in Shetland. Proceedings of the Zoological Society of London 128: 387-396. Venables, U.M. and L.S.V. Venables. 1959. Vernal coition of the seal Phoca vitulina in Shetland. Proceedings of the Zoological Society of London 132: 665-669. Wilson, S. 1974. Juvenile play of the common seal Phoca vitulina vitulina with comparative notes on the grey seal Halichoerus grypus. Behaviour 48: 37-60. FIGURE LEGEND Figure 1: Map of the study site. Table 1: Mean times of bouts and encounters, both general and divided into interactions by age and by sex. Table 2: Types of rolling interactions divided into general incidents by age and by sex. Figure 2: Typical position of two seals interacting in head to head horizontal rolling. Figure 3: Typical position of two seals interacting in head to head twisting. Figure 4: Typical position of two seals interacting in head to tail horizontal rolling. Figure 5: Typical position of two seals interacting in head to tail vertical rolling. Figure 6: Typical position of two seals interacting in stomach to stomach mounting. Figure 7: Typical postion of two seals interacting in back to stomach mounting. O BIR ROCKS W 2 PAGASSIZ O SEAL ROCKS AGASSE BEACH A 804 WORKS FIGURE KELP FORESI 7 2 CANNERY CEACH F opportunistic observations made here 2 L —0 - - INVN .. .... .... .... 2 .. o- N It U- ONDR g - O 0 .. ..... .... 2 - o - uo o5 1 1 R 11 2 .. .. .. . . .. .. Z.. un U NDNAN 25 50 22 U aaaaaaaa- .. .. .. .. s. .. .. .. 0 a2ö n-o NDN o Noo- 20 — .. .. .. .. . .. .. .. .. o- ooo0 -Ono - .. . .. .. u......... --N-0 oo0 *I D a0 20 325 ap; o- DSUU o — 0 5 Ezu DU - SS O- 5 0 OD De *D oooo ooooo wooo ooooo ooooo ooooo oo ooo ooo00 oo ooooo ooooo ooooo --ooo wood uu oo o onoo 5 5 ooo o — u —u — u — u — U 35 342. 8 . — — 223 — 3 15 52 5. 84. o 9- N-5 --+2 5 ad 31 a 155 5a o Saaa aco -auum 222a oum X21.. §922. D221. -8... -28.. P211. 69 3 o 10 L — 1 .-- . — 1. .—. — W 4 APPENDIX Individuals: Over the course of this study l was able to identify six individuals which I saw rolling at least once, including two females seen in mounting behavior. For the benefit of future research on rolling in this population, shall describe these individuals here: 1)RawChin Male - a grey adult male with a pink spot on his chin where the skin had been rubbed off; if this is not a scar, it will most likely leave one. saw him rolling with a brown subadult male on two separate occasions. 2)Pockhead Male - a brown adult male with a pocked region on the right side of his head, seen in one encounter. 3)VerticalScar Female - a brown adult female with a long vertical scar on the right side of her chest, seen in encounter with ThroatScratch Female. 4) ThroatScratch Female - a grey adult female with several scratches across her throat, visible when she lifts her head back. Seen in one encounter. 5) GrouchoMarx Male - a yellow-grey subadult with dark marks above his eyes, resembling eyebrows. Seen in one encounter. Not to be confused with two raccoon-eyed juveniles in the area (one male, one female). 6) TailScar Male - a grey adult male with 2 long slashes in his penile region, near the base of his tail. Seen rolling on three separate occasions.