EXPERIMENT 2: THE CIRCADIAN STRUCTURE OF
BEHAVIOR AND RESPIRATION*
[*The following section is based largely on the text of a paper delivered at the 1976 annual meeting of the ,Animal Behavior Society, Boulder. That paper was originally coauthored and titled: Ray, R. D., White, M. J., Coleman, V. L., & Upson, J. D. "Bio-behavioral oscillatory rhythms in a killer whale (Orcinus orca)."]
A special opportunity was afforded to us for addressing the questions just raised. A local oceanarium, desiring to further research exotic marine mammals, agreed with our interests in bio-behavioral rhythms. The choice of a particular species was relatively open. Only singular and isolated studies of biological rhythms exist for any marine mammals, and comparatively little descriptive behavioral work is available (Caldwell & Caldwell, 1972). These facts, plus current ecological interests and specimen rarity, eventually suggested to us the killer whale (Orcinus orca).
Existent physiological work on marine mammals predominantly has emphasized the unique properties of respiration dynamics. As Ridgway (1972) has stated, "respiratory rate can be taken as a prime example of the remarkable degree of integration of the physiological adjustments achieved in aquatic adaptation" (p. 592). However, in studies of virtually all marine animals, any potentially rhythmic variations in respiration rate have been hidden by the use of statistical averaging (e.g., Ridgway, 1972; Schmidt-Nielsen, 1972; Scholander, 1963; Spencer, Gornall & Poulter, 1967).
These observations, coupled with the comparative ease of visually and/or audially determining respiratory inspiration-expiration in the killer whale, led us to investigate the relations among categorical behavior, respiration, and circadian time periods. We conducted this experiment in two stages. The first stage established appropriate behavioral categories for describing activities of a killer whale in an oceanarium environment. Since we anticipated that behavioral structures might vary across the day, we designed sampling procedures accordingly. The second stage focused directly on the circadian rhythmicity of behaviors and respiration.
Method and Procedure: Stage I
Subject. A 12-yr.-old male killer whale (Orcinus orca), "Ramu," was the subject for these studies. At the time of this first stage (October, 1975), he had been in captivity 8 years and 7 months. He weighed 2925 kg and was 5.88 in in length. The subject had been a trained performer for 8 yr., had been housed in this particular park and tank facility for 6 mo., and was performing 20 min. in each of three to four daily performances. The animal was observed in a circular holding tank (10.6 m diameter, 4.6 m depth) at the Whale and Dolphin Arena, Sea World of Florida, Inc., in Orlando. Water temperature was maintained within a range of 14.4-16.1 C.
The video data were viewed for the purpose of listing all the different behavioral descriptions which could be identified. From an inclusive list of such categories, a running record account was made by additional viewings of the video recordings; this allowed a kinematic sequencing analysis of behavioral patterning.
Separate kinematic analyses for morning and afternoon behavioral sequencing patterns are given in Figure 4. To simplify the kinematic diagram, all single-occurrence behavioral sequences have been eliminated, leaving a description of only the more frequent patterns of behavioral organization. Clearly, the a.m. behavioral organization is more complex than the p.m. organization. There are 37 different pattern variations in the morning, but only 12 in the afternoon. Thus, a preliminary confirmation of temporal variations in behavioral organization is established.
Figure 4. A kinematic analysis of killer whale micro-behavior sequencing patterns for composite morning and composite afternoon sessions. The connecting arrows have proportional widths depicting the probability of occurrence of each type of behavioral sequence.
The kinematic illustration reveals certain high-probability clusters, or integrated networks, of multiple behavior sequences. These clusters suggest general macro states of behavior which are consonant with observed general behavioral states and their purposeful functions. Thus, various micro categories concentrate into naturally and functionally integrated interbehavioral networks to form macro behavioral states, quite similar to our previous demonstrations of differentially integrated behavioral patterning resulting from various experimentally contrived situations (Ray & Brown, 1975, 1976; Ray & Ray, 1976).In the a.m. graph a cyclic "surface breathing-surface floating" pattern is prominent. This pattern occasionally included subtle head scans and effortless glides from one tank location to another. We have labeled these inclusive sequences as Surface-Float Breathing.
A frequent shift from this macro state to a second definable macro state occurred when the animal submerged. Upon submerging, the animal most typically rolled 901 and assumed a position of floating on his side. This submerged side floating nearly always involved one specific environmental element, a water supply stream originating from a single outlet in the side of the tank. This group of behaviors is called Submerged Floating.A third probable sequence involves active swim patterns. These swim patterns show occasional submerging, subsequent surfaces for breathing, and continued swimming. These sequences often involved rubs against tank objects as well, and we have categorized them as Free Swim Activity.
Inspection of the p.m. kinematics reveals quite a different emergent pattern. Here the most frequent behavioral sequences involved movement into the submerged floating pattern, as described for the a.m. patterns; however, the behavioral category from which entry into this submerged floating category was made was a component of a new macro state. This new state involved a recurrent "surfacing-head-bob breathing-submerging-surfacing-head-bob breathing" cycle of activity. We thus have labeled these activities Head-Bob Breathing.
The four macro states that emerge from the kinematic analysis of micro behavioral categories largely correspond with the more general accounts of whale activity available in naturalistic literature. Thus our surface-float breathing seems to correspond to accounts of surface floating and breathing activities observed in both open ocean and captive environments (Spencer et al., 1967). What we have identified as submerged floating is perhaps a captive environmental adaptation of open ocean sounding (Slijper, 1962). The fact that the animal quite frequently sequenced from submerged floating into head-bob breathing patterns suggests that this particular activity corresponds with open ocean porpoising breath patterns noted by Ridgway (1972), Slijper (1962), and Spencer et al. (1967). Finally, the free swim activity presently described seems in line with a variety of behavioral descriptions of many marine mammals in socially isolated circumstances, whether captive (e.g., Caldwell & Caldwell, 1972) or in open ocean environments (e.g., Payne, 1976; Herman, Note 1).Having thus established behavioral categories suitable for describing killer whales in this particular environment, we proceeded to Stage 11. In this second stage, we focused on the potential relationships among the parameters of macro behavioral categories, respiration, and circadian rhythms.
Method and Procedure: Stage II
Conditions
for Observation
The general procedure
in this second stage was to observe the animal across 24-hr. periods for a
total of 5 days. During daylight hours, observers were positioned on the observation
stadium roof approximately 10 m above the surface of the water and with an
observation angle of approximately 350 from vertical. From this position,
observers had virtually complete view of the animal at all times, both in
the show and holding tanks. During hours of darkness, observers were repositioned
onto the roof of a stage-prop building. This position placed them considerably
closer to the animal's holding tank and at an approximate height of 4 m above
the water. The viewing angle from this position was somewhat sharper and included
an approximate range of from 50 to 450 from vertical, depending on the position
of the animal in the tank. Visibility was quite sufficient from this vantage
point even during the darkest of nights. For the 5 days and nights of the
study, the animal was kept in the holding tank at all times except for show
performances or training periods. This did not appear to alter representative
behavior patterns normally observed in this environment.
Observations were made during the five continuous 24-hr. periods (midnight to midnight) on the alternate weekdays of Monday, Wednesday, Friday, and the following Monday and Wednesday, during January 5-14, 1976. Hourly temperature variations, barometric pressure variations, cloud cover and relative humidity for the general metropolitan area were obtained from the local U.S. weather service and were assessed as additional setting factors.
Respiration and Behavior Assessments
Respiration. Each interval between expirations was timed to the nearest second by recording digital clock readouts at the time expiration was heard and/or observed visually. These records were made in conjunction with macro-behavioral recordings, thus making it possible to ascertain breath intervals within each specific behavioral state.
Behavior. The four macro categories of behavior as defined by Stage I, plus show performance as an additional category, were timed from beginning to end in order of occurrence. Thus, a continuous temporal-behavioral record for each 24-h r. period was made. In addition to these timed observations, continuous video recordings were made for the hour between 8-9 a.m. and 4-5 p.m. on 2 days (Days I and 3). On a third day (Day 5), a continuous 9-hr. video record was made for the hours between 8 a.m. and 5 p.m. It was possible, therefore, to conduct a micro-behavioral analysis of these time periods for these 3 days.
Results and Discussion: Stage II
Respiration. Figure 5a illustrates that averaged January respiration frequency varied systematically within a daily range of 58 to 120 breaths/hr. This variation is organized within the boundary of a 24-hr. cycle. Starting from the lowest frequency occurring between 2 and 5 a.m., respiration frequency increased beginning at dawn, momentarily dropped, increased substantially for a sustained period, and then gradually declined until the 25 a.m. minimum was reached once again. From this point, the cycle continued anew. This rhythm was stable and apparent across all 5 days of observation. Weather factors (i.e., temperature, relative humidity, barometric pressure, and percent of cloud cover) were statistically ruled out as being significantly correlated with the respiratory variations. The October plots for Ramu suggest virtually the same variation pattern, but with a slightly lower overall respiration rate. These plots are averaged data obtained from 3 partial days of observation during Stage I experiments.
Figure 5. (A) Mean breaths/hr. for each hour of observation of Ramu during the 5 combined experimental days in January and during sampled periods taken over a 3-day period in October. (13) Mean breaths/hr. for each hour of observation of Toki and Hugo during 2 combined days in May.
We were able eventually to make 48 continuous hours of observation on two additional killer whales, Hugo and Toki, housed together at another oceanarium.3 Both conformed to the essential patterns of Ramu (see Figure 5b). Importantly, these data on Hugo and Toki also expand our generalizations to a female killer whale (Toki), thus ruling out sex differences. [Details concerning these two additional whales are: Tokitae: female, approximately 8 years of age, 2835 kg, 5.92 m, placed in oceanarium in 1970. Hugo: male, 4050 kg, 6.8 m in length, approximately I I years of age, placed in oceanarium in 1968. Miami Seaquariu's whale facility houses both animals together in an oval tank 24.4 m x 18.3 m, 6.7 m deep, 1,900,000 1, 17.2 C. We wish to note our thanks to Wometco Corporation's Miami Seaquarium and their staff for making possible the observations of these two animals.]Differences in absolute respiration rates illustrated in Figure 5 suggest a possible seasonal variation: Ramu's October (i.e., 3 mo. prior to January) respiration rate samples, illustrated in Figure 5a, are lower than his January rate, but are consonant with Hugo's and Toki's respiration rates early in May (i.e., 31/2 mo. after January).
Respiration and behavioral specificity. Questioning whether specific behavioral relations might exist relative to respiration activity, we assessed Ramu's respiration rates within each macro-behavioral category across successive 24-hr. periods. Figure 6 shows averaged hourly respiration rates observed for each different behavior category. A striking feature of these results is not only the difference in breathing rate for each macrobehavioral category, but also the constancy of breathing rate within each behavior across the 24-hr. period (i.e., the absence of any rhythmic pattern in the variance), Further, show performance evidenced little difference from normal levels of free swim respiration rate. Only the difference between show performance and free swim respiration means is not statistically reliable (p > .05, 148 df).
Figure 6. Five-day mean hourly respiration frequencies within each specific macro-behavioral category having inspiration/expiration cycles.
The observation of
a characteristic respiration rate for each behavior, along with the circadian
rhythmicity in total breaths/hr., suggests that rhythmic variation in respiration
must be accounted for behaviorally. To evaluate this position, total min./hr.
spent in each macro-behavioral condition were calculated (see Figure 7).
Figure
7. Five-day mean total min./hr. spent in each macro-behavioral state. Open
circles in Free Swim Activity illustrate actual time in show performance. Closed
circles in all behaviors for those hours which include show performance have
been adjusted based on actual proportional time in the holding tank.
Submerged floating behavior reached a stable maximum
"total time in behavior" throughout the evening hours, but was unimodal within
the circadian period. Surface-float breathing and free swim activity likewise
were unimodal. However, each of the three behaviors had independent phase
relations relative to the 24-hr. period. Importantly, free swim activity
peaked just after dawn, reflecting increased activity near this zeitgeber similar
to that known to exist for terrestrial mammals. Finally, head-bob breathing
peaked at two different times each day, therefore being bimodal. In summary,
the four behaviors demonstrated independently phased and sometimes complex rhythmic
variations across the 24-hr. period of a day.
These behavioral measures, based on total time/hr., may tend to mask component variations in frequency of behavioral initiations and/or differences in behavioral durations. These additional measures are depicted in Figures 8 and 9. The rhythmic variations in frequency and duration shown in Figures 8 and 9 are entirely different from those of total time spent in each behavior, as was illustrated in Figure 7.
Figure 8. Five-day mean frequency of occurrence/hr. for each macro-behavioral category. Hours which included show performance have been proportionally adjusted based on actual time in the holding tank.
Free swim activity peaked in frequency near the evening twilight, but because the lowest duration occurred at this time, the "total time" measure does not reflect accurately this character for the morning-evening twilight influences. Since normal activity counters used to measure terrestrial animals' activity usually are cumulative micro- switch recordings, they too would tend to mask such important and distinct behavioral dynamics. It is noteworthy that durational measures of activity in the killer whale actually peaked approximately 2-3 hr. before dawn, and 2-3 hr. after evening twilight.
Figure
9. Five-day mean duration in sec. for each hour and each macro-behavior
category.
The importance of simultaneous accounts of behavioral frequencies and behavioral duration generally has been overlooked by behavioral researchers. These two measures are not completely independent (i.e., long durations within fixed time samples decrease available time for additional frequencies); yet the power of simultaneous interpretations of these measures in accounting for certain variances in behavioral/physiological measures is particularly well illustrated by the present data. As Figure 10a illustrates, when individual behavioral frequencies are multiplied by corresponding behavioral durations, the total time spent in each behavior is derived. These idealized curves approximate actual "moving average" curves based on N = 3, where each N is the 5-day mean plots depicted in Figures 7, 8, and 9. In addition, each curve is plotted to depict the rhythms for 2 successive days, thus allowing for a better visual display of the degree of phase synchrony among the various behavioral rhythms.
Figure 10. Smoothed curves across 2 consecutive days demonstrating: (A) (hourly frequency) X (hourly duration) = total min./hr. spent in each of the macro-behavioral states. Curves are derived by smoothing "moving average" plots based on N=3. (B) (specific and constant respiration rates for each of the macro-behavioral categories) X (the fraction of an hour [plotted in min.] spent in respective categories). Their products are summated to give the original total breaths/hr. circadian rhythm illustrated fro Ramu, January in Figure 5a.
When total time for each specific behavior is, in turn, multiplied by its appropriate respiration frequency constant and subsequently is summated with similar specific products, the observed circadian rhythm in total respiration frequency is obtained (Figure 10b). Thus, one may account for physiological circadian variation (i.e., respiration) almost entirely by an analysis of macro-behavioral variation dynamics.