The Relationship between Cardiac and Respiratory Rhythms in the Dogfish, Scyliorhinus Canicula L

It has been known for a long time that a relationship exists between the rhythms of the heart and ventilatory movements in fishes. Earlier workers suggested that the ratio was always a simple one, taking the form 1:2, 1:3, 1:4, etc. (Lyon, 1926; Lutz, 1930; Satchell, 1960). In some cases it may approach 1:1 or even 2:1 or 3:1 (Hughes, 1961; Hughes & Umezawa, 1968a). With the use of modern recording equipment more detailed attention has been paid to this relationship. In the dogfish (Squalus lebruni), Satchell (1960, 1968) concluded that the two rhythms were closely coupled and suggested a possible reflex mechanism whereby this co-ordination was achieved. Simultaneous recordings in trout of the electrocardiogram and pressure changes in the respiratory cavities clearly showed that these rhythms were closely coupled at certain stages, while at other times in the same animal the rhythms seemed to slide past one another and to be completely independent (Hughes, 1961, 1964). Shelton & Randall (1962) found similar relationships for the tench, with frequent periods of synchrony with the mouth-open phase, especially under light anaesthesia. Subsequently it was suggested that the coupling tended to be closer in fish under MS 222 anaesthesia or when the of the inspired water was reduced (Randall & Smith, 1967). This conclusion for rainbow trout was not supported by observations on the dragonet (Callionymus lyra), where great variation in the degree of coupling seemed relatively independent of the presence or absence of MS 222 or of hypoxia (Hughes & Umezawa, 1968a). In a study of cardiac and ventilatory frequencies during exercise, Sutterlin (1969) seldom observed perfect synchrony in trout, pumpkinseed or bullheads except during the period of recovery from the anaesthetic. A tendency for increased cardio-ventilatory synchrony has also been observed in rainbow trout under temperature-stress (Heath & Hughes, 1971).

These studies have all emphasized the major experimental problem that co-ordination between these rhythms is sensitive to many environmental conditions (Serfaty & Raynaud, 1957; Labat, 1966; Roberts, 1968). Furthermore, the analysis of sufficient data has been both difficult and laborious. Even with long recordings, the measurement of frequencies and timing of the two rhythms with respect to one another does not give sufficient data. In the study an averaging computer has been used so that data over long periods from different individuals could be analysed. This report, although based largely on the dogfish, developed from comparative observations on some marine teleost fish. A preliminary report has already been published (Hughes, 1971).

Recently caught dogfish (S. canicula) weighing between 500 and 900 g were anaesthetized with MS 222 (1 in 10000) and a polyethylene cannula was fixed dorsally into their orobranchial cavities using a method similar to that of Hughes & Roberts (1970). The fish were replaced in laboratory holding tanks for at least 24 h before an experiment. The temperature of the sea water circulating through the holding tanks varied about 1 ° C during any 24 h period and the overall variation during the time when the experiments were carried out was 15–17 ° C. In early experiments a few dogfish were subjected to changes in temperature of the circulating sea water, but these were discontinued in order to concentrate on the normal rhythms. Before the start of an experiment the fish was lightly anaesthetized once more and insulated stainless-steel wires inserted, one anterior and another posterior to the heart, in order to record the electro-cardiograms. The fish was placed in a specially constructed perspex box which formed part of a closed circulation of air-saturated water (Fig. 1) with its temperature controlled to be the same as in the holding tanks. All experiments discussed in this paper were carried out under normoxic conditions. In order to keep the fish at a more or less constant level of quietness once the anaesthetic had worn off, the box was painted black and the fish was covered by a more closely fitting perspex ‘restrainer ‘which restricted its movements and also kept its ventral surface in contact with the floor of the main box. The fish was by no means fixed by this arrangement, which was necessary because of the facility with which a dogfish can turn within its own length. The ECG leads were taken to a Tektronix 122 pre-amplifier, and the orobranchial pressure was recorded using a Sanborn 268 pressure transducer and 350-Series amplifier. The amplifier outputs were observed on an oscilloscope, fed into a Devices pen-recorder and recorded simultaneously on a Thermionic T 3000 taperecorder. Recordings were taken over at least a 24 h period.

Recordings were not usually started until at least 6 h after the dogfish had been placed in the water circulation, by which time any effects of the anaesthetic had worn off. In general, a recording was made for at least 5 min every 30 or 60 min.

The results were analysed by inspecting the pen recordings and in more detail by the use ofthe tape recordings and a Biomac 1000 computer (Figs. 2,3). The oscilloscope (Tektronix 565) sweep was triggered by the pressure waveform and the time-base output was used to trigger the Biomac. The ECG was displayed on another oscilloscope and was also led into a Schmitt trigger circuit, whose output was used as the event signal. Either signal could be used with the computer operated in the interval histogram mode. Histograms were obtained for the intervals between heart beats and between ventilatory pressures ; overall frequencies were obtained from the pen-recordings. Average pressure wave-forms and event correlograms were plotted using the same triggering point on the pressure wave-form, so that the position of the QRS wave of the ECG was obtained with respect to the oro-branchial pressure cycle. From the calibration of the correlogram, the number of heart beats which occurred in each tenth or eighth of a ventilation cycle was calculated. In this way compensation was made for individual variations in ventilatory frequency, and the distributions obtained were analysed as discussed below (see Appendix). In the early stages of the work repeated checks were made between the distribution of ECGs within the ten phases of a ventilation cycle derived from the event correlogram, and those obtained by direct analysis of the pen recordings. There was very good agreement between the two sets of data, with a mean error of about 2%. Constant checks were made during the Biomac analyses to ensure that no cardiac or ventilation cycles were omitted.

Observations on many different individuals over relatively long periods of time have shown that there are quite wide variations in the respiratory and cardiac frequencies. As far as can be judged, these differences must be regarded as due to individual variation, for even in the holding tanks undisturbed dogfish showed differences in ventilatory frequency. For all the observations made on dogfish (temperature 14–22 ° C), the ventilatory frequency range was 22–80/min whereas the cardiac frequency range was 11–64/min. For similar body sizes (500–900 g) but over a more restricted temperature range (15–17 ° C) the frequencies were less variable, ventilatory frequency being 30–70/min. Table 1 gives examples from four individuals at similar temperatures and illustrates the wide range in ventilatory (40–58/min) and cardiac (20–46/min) frequencies. These observations emphasize the need to establish the normal frequencies for a particular individual before carrying out any experimental interference. Variations for individual dogfish were generally less than the overall range, and for short periods were very small. Hughes & Umezawa (1968) noted a similar constancy for the same species, usually at a frequency of 50/min, which is close to the overall mean frequency (Table 2).

In general, the interval histograms indicate that the duration of a ventilation cycle is less variable than the intervals between successive heart beats (Fig. 4). Sometimes, however, both rhythms vary by about the same amount, and this may be related to the closeness of their overall frequency and the interaction between them which results in periods of locking (Fig. 4). In general, the heart rate varied from being about the same frequency as ventilation but often approached half the ventilatory rate.

As mentioned above, in most cases the interval between heart beats is less regular than that between ventilation cycles, but Fig. 5 (C), for example, shows the mean peak of the ECG interval histogram at 1·55 sec, the variation being over a total range of 0·033 sec (1·517–1·583 sec). For the orobranchial pressure, the mean is at 1·400 sec with a total variation of 0·067 sec (i.e intervals of from 1·33 to 1·467 sec). It is apparent from the ECG interval histogram, however, that many individual cycles have intervals outside the main peak, and this is typical of all specimens. The ventilatory frequency in the dogfish is more constant than in many teleost fish. No clear-cut relationship between body size and ventilatory frequency has been established. With increasing water temperature there is an increase in ventilatory frequency, as is known for the trout (Hughes & Roberts, 1970) and other fish.

The relationship between the frequencies of the two rhythms was very constant in some specimens, as shown in Fig. 8, where the cardiac to ventilatory frequency ratio (H/V) remained almost constant about 0·7 with a whole number ratio of 3:2. In the second dogfish, for which data are plotted in Fig. 8, the rhythms were not so constant and showed a gradual increase in the H/V ratio. It is notable that here there was no tendency for a whole-number ratio to be observed. The effect of increased temperature is also shown; after an initial rise in the H/V ratio it was markedly decreased, mainly because of a significant bradycardia. Plots such as this illustrate the common observation that whole-number ratios are not maintained for long periods. Although detailed studies were not continued on the effect of temperature on the relationship between the two rhythms, several experiments showed clear-cut results (Fig. 9) and illustrate the usefulness of the histogram method of plotting the results. The effect of increased temperature was to decrease the interval between successive ventilatory cycles. Associated with this there was a distinct tendency in this particular specimen for the original random distribution of the phase in which the ECG occurred to shift to one in which at temperatures of 20 ° C the coupling became very distinct, although the particular phase varied from time to time. However, not all preparations showed such a clear-cut result as that illustrated in Fig. 9.

Some correlation between the two rhythms has appeared in about 70 % of the pen recordings analysed for 13 dogfish, at least at some stage. By a ‘correlation’ is meant that there seems to be a particular phase of the respiratory cycle when the heart beat (ECG) tends to occur most often. This particular relationship and a discussion of different degrees of correlation or coupling will be analysed in more detail later. The further analysis confirmed this general impression in that 60% of the recordings showed significant coupling at the 1 % level.

Tape recordings taken from the same specimens were analysed for the event correlogram ; in less than half of them a clear correlation was found. The phase at which the heart tends to beat shows much individual variation and for a given specimen it may remain more or less constant for long periods. Nevertheless there are variations equal to at least half a cycle. Examples of some different types of relationship are illustrated in Figs. 5–7. In one of these (Fig. 5 A) the interval between individual ECG’s and between successive ventilation cycles have a similar duration, but there is no preferred phase at which the electrocardiogram tends to occur. In Fig. 5 E, F, however, the two rhythms were almost identical in frequency and a very close correlation between the two rhythms is found, the heart beat tending to occur either shortly before or shortly after an increase in pressure within the orobranchial cavity. In Fig. 5 A an example is given where the event correlogram suggests a tendency for the heart to beat at two phases in the cycle.

In order to visualize temporal changes in the phasing of the two rhythms an alternative mode of display was employed and automatically provides similar plots to those for Callionymus (Hughes & Umezawa, 1968a). One example (Fig. 5 A) is chosen to show ‘free-running’ of the ECG with respect to the ventilation cycles, as indicated by the constant increment in the vertical signal which represents the interval between the trigger and event signals of the Biomac. When the ECG is synchronous with a particular phase the signal has a constant voltage (Fig. 5E) for prolonged periods but then ‘slips ‘out of synchrony.

Information gained from the event correlogram was analysed further using the method described in the Appendix. Essentially this analysis takes account of ventilation being a cyclical phenomenon and makes use of polar co-ordinates. The 360° are divided into ten equal parts and the number of occurrences of the heart beat in each sector (36°) is indicated around the circumference of the circle. The overall nature of this distribution is found by establishing what is equivalent to the ‘centre of gravity’ of the distributed occurrences. Thus with rigid synchrony in one particular phase, all of them would occur in a single sector or phase and hence the centre of gravity would be at the circumference of that sector. Correspondingly, if the occurrences are equally distributed then the centre of gravity will lie at the centre of the circle. Distance from the centre therefore indicates the degree of non-uniformity of the distribution and is a measure of the coupling between the heart and ventilation in a particular phase relationship, which is the angle given between the fixed point (V) of the ventilation cycle and the centre of gravity.

In Fig. 10 four such distributions are plotted in detail. In one case the coupling was 76·5 %, which was the highest coupling obtained. On the other hand, recordings from the same specimen at other times show a very low coupling (9·24%) but at a similar phase angle, 155·5° and 158-9° respectively. An example of the type of variability in coupling obtained for another specimen is shown in Fig. 10D, which has been selected because the phase angle is greater than 18o°.

The distributions for a single dogfish which was investigated in greater detail are shown in Fig. 11. Varying degrees of coupling were obtained and with phase relationships ranging from 334° to 162°. All distributions were tested statistically and in most instances couplings of less than 10% are not statistically significant at the 5% level. The data so far analysed for eight different specimens is summarized in Fig. 12. The wide variation in the nature of these couplings can be seen at almost all possible phase angles. Moreover, there does not seem to be any relationship between percentage coupling and the phase angle. The largest number of occurrences (52) was found in the two quadrants from o to 180°, whereas between i8o° and 360° the total number of points is 19. Of the 71 points, 24 are not statistically significant (5 % level), but all those where the coupling is greater than 15 % are very significant. The overall nature of these results confirms the impression gained from inspection of the pen recordings and correlograms. It has the great advantage of expressing the different ‘degrees of coupling’ in a quantitative way.

The data obtained from this analysis can be used to compare the coupling with the ratio between the cardiac and ventilatory frequencies. A whole-number ratio is considered to occur when the ratio falls within 5 % of the expected ratio as indicated in Fig. 13. Coupling between the two rhythms may be quite high, although the ratio of their frequencies falls outside a whole-number ratio. It is also of interest to observe that a 1 : 1 ratio can occur even when the percentage coupling is extremely low. Thus the results obtained with this of vector analysis contrast somewhat with results obtained by methods often used in work of this kind, such as whole-number ratios, and emphasizes the need for cycle-by-cycle analysis rather than mere dependence upon overall frequencies.

It is clear that there are a number of ways in which a 1:1 ratio can be obtained. The heart beat may occur at random within the ventilatory cycles, but so long as one beat occurs in each inter-breath interval, then the overall whole-number ratio will be the same. Correspondingly, it is possible to have a close coupling as defined in the present context in a number of different ways. These include (a) rigid coupling in a particular phase for a certain time, followed by a period of free-running between the two rhythms before synchrony is re-established; (b) free-running over a restricted phase range. This raises the question as to whether the phase of an ECG in a given ventilatory cycle influences the position of the heart beat in the next cycle. Analyses of this kind are clearly relevant to gaining further understanding of the co-ordination mechanism.

The extent of the data used in the present work is an important factor supporting the results that have been obtained. The need to collect so much data has arisen because of the observed variations both between individuals and for a given individual at different times. This main conclusion regarding the resting conditions emphasizes the need to establish some of the basic respiratory and cardiovascular parameters of individual dogfish before further experimentation. The fish were kept in conditions that were as restful as possible and yet allowed continuous recording from cannulated respiratory cavities and ECG leads. Telemetry would be the most obvious technical improvement. The normal variability, at least in ventilatory frequencies, was confirmed by observations on dogfish in holding tanks. Each individual showed a different frequency, all within the range recorded in our experiments. Some of the latter frequencies are lower than those recorded in other physiological experiments and perhaps this reflects the more excitable conditions of fish in such experiments. In general it seems that the respiratory rhythm is the more constant and that the interval between successive heart beats is the main variable. Under environmental stresses such as hypoxia, this relationship was further confirmed.

Theoretically there are several possibilities regarding the relationship between the two rhythms, including:

1. Absolute synchrony which is the same in all individuals, i.e. whenever the heart beats it will be at the same phase of the ventilatory cycle.

2. Absolute synchrony but at different phases in different individuals.

3. Fixed phasing (synchrony) but phase is variable at different times.

4. Fixed phasing over short periods between which the phase changes with time. The rate of change could vary between individuals. Depending on the duration of the period of synchrony, so variations would be found in the degree of coupling by the vector analysis.

5. Random phasing of the heart beat, which might be associated with a greater constancy of the interval between ECG’s and/or ventilation cycles. Hence, if there is a slight difference in frequency of the rhythms, all phase relationships would be found.

The main result of analysing the correlation between two rhythms is to reveal the absence of any fixed overall relationship which would describe all cases. There is no particular phase with which the heart beat tends to become locked to the ventilatory rhythm (Table 3). Even when the coupling is well developed, the phase may vary within a given individual, although in some cases it is also true that the range of phases with locking is much less than in other specimens. A similar conclusion has been reached by Taylor & Butler (1971). Accepting this conclusion, it is of interest to explore the possible significance in relation to the flows of water and blood across the dogfish gill.

Gas exchange at the surface of the secondary lamellae is the main function in which cardiac and ventilatory functions interact. Thus there will be a relationship between the volume of oxygen available in the water at the secondary lamellae and the volume taken up by the blood perfusing the gills (Hughes, 1964). Because of the greater solubility of oxygen in the blood, the volume flowing through the gills in unit time (Q+̇) is less than the water flow ⁠. The volume flowing on both sides of the exchanger is produced by pumps which can vary in frequency and stroke volume. Hence, it is possible for stroke volume to vary and compensate for changes in frequency. Clearly an arrangement whereby the maximum blood flow through the gills occurs at times when maximum water is flowing would be advantageous. There is, however, relatively little knowledge about the detailed nature of the flow past the secondary lamellae. Measurements of pressure changes on both sides of the gills (Hughes, 1960; Grigg, 1970) indicate that the bulk flow of water is continuous but probably at different velocities at different phases of each ventilatory cycle. Counter-current flow seems almost certain (Grigg, 1970; Hills & Hughes, 1970).

The absence of a preferred phase relationship strongly suggests that the overall ventilation/perfusion relationship is a very flexible one. Moreover it seems probable from a number of recent observations that the ventilation/perfusion ratio varies not only in different gill slits but also in different regions of a given gill arch or filament at any one time (Piiper & Schumann, 1967; Cameron, Randall & Davies, 1971; G. M. Hughes, unpublished). However, under conditions of stress when presumably the whole system is maximally ventilated and perfused, the relationship might be expected to become more definite. This in fact seems to be true for both teleosts and elasmobranchs, although some unpublished observations on the effects of hypoxia on dogfish are not so convincing (in this respect) as had been expected.

An important factor which might be involved in the interactions between the two pumps is the variation of the resistance of the gills to flows of water and blood. For example, if the resistance to blood flow changes during the ventilatory cycle then the output per beat would vary in relation to the phase in which it occurs and could be involved in maintaining functionally important patterns of blood flow relative to water flow across the gills. Part of the importance of this would be to ensure that the blood residence time in the gill was related to the rates of water flow. Differences in recruitment of effective secondary lamellar units are also thought to occur (Hughes, 1972). Hence the various couplings probably reflect corresponding variations in the precise conditions for flow at the exchange surface and thus invite more detailed study. Technically, however, investigations at this level are going to be extremely difficult, for not only can we expect differences of the type already indicated but also these would not be the same in different parts of the gill sieve.

The main conclusions to be drawn from the present study are:

(1) Variations in coupling between the cardiac and ventilatory rhythms give the impression of being random in phasing, but when analysed in detail it appears that given individuals may show synchrony with a particular phase over quite long periods. However, this phase is not always the same.

(2) The methods at present available for analysis of the relationships between the rhythms have a number of disadvantages. The whole-number ratios studied for such a long time and still used in mammalian physiology (Weiss & Salzano, 1970) take little account of the relationships within a given cycle. The method of estimating the coupling and phase angle used here has advantages over previous methods of plotting the distribution of ECG’s within a ventilatory cycle. It expresses quantitatively different degrees of synchrony which are the most commonly found. As a method for summarizing the data it seems satisfactory, but for further analysis it would be necessary to have a method which takes into account the effect of one cycle upon another.

This work was supported by a grant from the Natural Environment Research Council. The experimental work was carried out at the Plymouth Marine Laboratory and it is a pleasure to thank Dr J. E. Smith and his staff for the space and facilities that were placed at our disposal. I thank Mr B. Knights and Mr D. W. Williams for their excellent assistance with the experimental work. I also wish to thank Dr Clifford for suggesting the method of analysis and writing the Appendix.

BY P. CLIFFORD

School of Mathematics, University of Bristol

As time progresses the occurrence of events of two types, A and B, is observed; for example, events associated with the heart and ventilation cycle. The frequency with which events of type A occur is assumed to be higher than for events of type B. If t_i is the time between the ith occurrence of event A and the next occurrence of A and s_i is the time between ith occurrence of A and the next occurrence of B, then θ_i= 360s_i /t_i is defined to be the phase angle between B and A. If we display the points θ_i on the circumference of a circle with unit radius then to each θ_i there is associated its rectangular co-ordinates x_i = cos θ_i and y_i = sin θ_i. Suppose we have n points, then the centre of gravity of the points is ⁠, where and ⁠. θ = arctan is defined to be the mean phase angle and from ⁠, the distance of the centre of gravity from the centre of the circle, we obtain the percentage coupling defined to be 100r.

If we divide the circle into ten sections, 0–36°, 36–72°, etc., and call the number of points falling in each, n₁n₂…n₁₀, then x̄ and have a particularly simple form.

where n = n₁ + n₁+… +n₁₀.

If it is assumed that the phase angle θ_I is independent of the previous phase angle θ_i-1 then we can compute a test statistic which tests the null hypothesis that phase angles are uniformly distributed through 360° against the alternative that they tend to be concentrated about a particular angle. The quantity T — 2nr² is distributed approximately as chi-square with two degrees of freedom under the null hypothesis, so we reject the hypothesis if T > 6 at 5 % level and if T > 9·2 at 1 % level.

For a similar treatment arising in an entirely different context the reader is referred to Fisher (1953).