The Mechanism of Orientation of Roach, Rutilus Rutilus L. In an Odour Gradient

There has been no comprehensive analysis of the orientation of fish to non-visual environmental stimuli. Individual cases have been fitted more or less easily into the existing classification of orientation mechanisms that derives primarily from work on invertebrates, especially insects (Fraenkel & Gunn, 1961). Jones (1960), in reviewing the reactions of fish to stimuli, points out that chemical stimuli, unless associated with water currents, are non-directional. Thus theoretically the only available methods of orientation are the ortho-kinesis and klino-kinesis, klino-taxis and possibly tropotaxis. For orientation by simultaneous comparison of intensities at paired receptor organs (tropotaxis) the receptors must be widely separated or very sensitive unless the gradient is very steep. One case where it is possible that a tropotaxis occurs is in the orientation of the hammerhead shark (Hasler, 1957; Jones, 1960). Klino-taxis, by definition, involves a characteristic lateral swinging of the receptors, usually on the head of the organism. Head swinging is not a significant part of fish locomotion or behaviour, except in dogfish which do swing the head quite markedly; Parker (1914) described dogfish (Mustelus cams) showing circus movements when one of the nares was blocked, suggesting orientation was by a taxis. This finding was criticized by Hasler (1957) who suggested that the fish might equally well have been turning away from the irritant plug. Irrespective of whether the circus movements were positive towards the odour or negative away from the irritant, Parker’s work does suggest that the pairing of olfactory organs may have some functional significance. All ‘fish’ other than cyclo-stomes have paired olfactory organs. Evidence thus suggests that kinetic mechanisms of orientation are most likely in fish. The work described in this paper investigated the mechanisms of orientation to species odour in a horizontal linear odour gradient situation.

The lower limiting value of C is 1·0 when fish turn only on meeting end walls, and the highest recorded value was 7·5.

The biological nature of the odour excreted or secreted by Rutilus was not investigated and no assay of its concentration was attempted. A chemical model of the concentration gradient was used to indicate the probable distribution of odour in the experimental tank. A volume of hydrochloric acid was added to the water of the inlet tank to simulate the odour concentration built up during the 1 hr. period that four fish were in the inlet tank prior to the olfactory experiments. A burette was set to run more acid into the inlet tank at a nearly constant rate during the course of an experiment. When calculating the volume of acid initially added, and the rate at which it subsequently dripped into the inlet tank, it was assumed that the rate of production of odour by the fish was constant. The test unit of a Pye ‘Master’ pH meter and millivoltmeter was water-proofed to form a probe for use in the experimental tank. Readings were taken at six sampling positions with respect to the length of the tank and at different depths and times. The mean result from six chemical model runs is given in Fig. 2 in which the concentration in the experimental tank (H+)_e is presented as a percentage of the concentration in the inlet tank (H+)_i. The table in Fig. 2 shows the actual concentrations in the inlet tank, from which it is clear that if the data were presented as actual concentrations the curves would be closer together. Relative values were more important because the 100 % level of the inlet tank, being the concentration entering the experimental tank, was that which a fish would perceive if it ‘sniffed’ at the perforated inlet pipes through which water entered the tank. The scatter of points is most probably due to turbulence caused by the repeated insertion of the probe to determine the concentration. Although the acid was very dilute, a consistent small difference in pH readings at different depths in any one region occurred, suggesting that density differences were causing the variation. However, density measurements in the two inlet tanks, with a standard hydrometer reading specific gravity to the nearest 0·001, showed no observable difference so the variation of pH with depth in the experimental tank may have been due to its flow characteristics.

The term ‘concentration gradient’ is in common usage to describe the distribution of concentration in space around chemical sources. The word ‘gradient’ should strictly be used only for linear ratios between the unit rise and the unit slope, i.e. the sine of the angle of rise. Fraenkel & Gunn (1961) point out that theoretically the concentration about a point source of chemical approximates to an inverse square law relationship assuming only diffusion. They retain the term ‘concentration gradient’ when in fact the distribution is non-linear. The result in Fig. 2 shows that the concentration distribution in the present apparatus is more nearly a straight line on a semi-log plot (base 10). Thus it is dangerous to assume that the method of orientation shown by a fish at some distance from the source is necessarily the same as occurs close to it.

It was assumed that the nature of the odour gradient changed in the course of each 20 min. experiment in the same way as in the chemical model. The response of the fish appeared not to be to a critical concentration threshold but to the position of maximum stimulation even at the end of experiments. Fig. 3,a−d shows the mean change in response for the four 5 min. recording periods from ten experiments, and Fig. 3 e—h shows the expected result if the fish were responding to a critical relative concentration of 0·1 % derived from the chemical-model results, but behaving randomly at concentrations higher than threshold. Woodhead (1957) reported that minnows in a light gradient remained in regions where illumination was below a critical threshold. It appears that Rutilus responded to the whole of the odour gradient, not just to a threshold concentration.

A close inverse correlation existed between the activity of fish in the odour-gradient situation and the strength of their response to it. Response strength was defined as the sum of the deviations, from expected random, of the three end regions of the experimental tank, i.e. those with the highest stimulus concentration. A Spearman rank correlation coefficient of activity and response strength from the series of experiments gave r_s = 0·87 with P < 0·0005 and N = 37 (Siegel, 1956). Clearly the fact that fish showing a strong response to odour were inactive and that active ones showed a lower response implied an orthokinesis. However, there was no evidence that speeds were directly related to the concentration of odour in the experimental tank. Fig. 4 shows the speed of movement in the odour-gradient situation compared with speeds in control experiments with tap water running from each end. In the control situation there was no significant difference between the swimming speeds, whereas in the gradient the speed when swimming away from the odour was significantly less than when swimming towards it; but this did not differ significantly from the mean control situation. Two points must be remembered in the interpretation of Fig. 4 a: first, the maximum frequency of positional recordings and therefore most data on speed occurred at the high concentration end; and secondly, the close inverse correlation which existed between response strength and activity. These two facts mean that the data on speed at the low concentration end were derived from comparatively few recordings which tended to be from more active fish not showing a strong response to the odour gradient.

Whilst recording the movement of fish in the experimental tank it was frequently possible to see fish swimming slowly away from the stimulus end and then at some point turning and swimming much faster back again. These results suggest that although significant speed changes occur in the odour-gradient situation these are not closely correlated with stimulus concentration and therefore the results do not support a straightforward orthokinesis.

Orientation by klino-kinesis requires that turning is related to the stimulus intensity. Fig. 5 shows the regional turning frequencies in different regions of the experimental tank for odour and control experiments. The top two histograms shows the frequency of ‘up-gradient’ turns where a fish, after swimming away from the stimulus maximum, turns and swims towards it; and the second row shows turns in the opposite direction. Control results show turns in the same directions, there being no odour present during the experiments. The bottom histograms are each the sum of the two above giving the total turning frequency. It is clear that the greatest frequency of turning in olfactory experiments occurred around the end of maximum stimulation and the distribution of turns was random over the remaining three-quarters of the experimental tank. In control experiments turning was symmetrically distributed about the middle of the tank; the end frequencies in these control histograms show the recording error concerning turning, and are lower than expected. This was due to fish having moved into and out of the end region within the 5 sec. interval between recordings.

The mean turning frequency of fish was higher in the control situation but this was associated with their higher activity increasing the number of compulsory turns at the ends of the experimental tank. Comparison of the turning coefficients shows that when compulsory turning is excluded the fish did turn more in the odour-gradient situation. Fig. 6a shows the values of turning frequency plotted against the turning coefficient for odour-gradient and control experiments. A Mann Whitney U test showed there was a significant difference between frequencies but not between coefficients. However, the figure does show clearly the tendency for high turning coefficients and low turning frequencies in the odour gradient and the reverse in control experiments. The wide range of turning coefficients indicates the variation in the response of the fish to the odour stimulus. Fig. 6b shows the values of the turning coefficient plotted against the response strength in gradient experiments. Thus those fish which showed a strong response to the odour turned relatively more than those that showed a low response.

Two short series of control experiments were carried out using the same five fish in a uniform odour concentration provided by four fish in each inlet tank and a normal control with tapwater. If speed and/or turning frequency were simply related to stimulus concentration then some difference might have been expected. Table 1 shows the results of comparing by a ‘t’ test all data collected for these experiments. There was clearly no significant difference in activity and turning in the two concentrations.

The behaviour of Rutilus in the odour gradient can be considered as appetitive behaviour which would under normal circumstances have led the individual to one or more others to form a school. The variability of individual behaviour was not open to analysis, but could be due either to lowering of the response to a perceived odour stimulus, or to the stimulus falling below an effective threshold. The first possibility appears more likely, due to the wide range of odour concentration in the experimental tank.

In a simple orthokinesis the rate of movement is related to stimulus intensity, and as pointed out by Kennedy (1945) a negative kinesis (negative correlation between stimulus intensity and activity) is required for aggregation at high stimulus values. Keenleyside’s (1955) study of the behaviour in an odour gradient of temporarily blinded rudd Scardimus, which are very closely related to Rutilus, suggested that a simple ortho-kinesis was the means of orientation. The measurements on Rutilus of the speeds of movement in different regions of the experimental tank do not support this conclusion although there was a significant difference between the speeds of movement in different directions.

Klino-kinesis was first demonstrated by Ullyott (1936) using the platyhelminth Dendrocoelum. He showed that abruptly increasing the light intensity resulted in an increase of the rate of change of direction (R.C.D.), but if the intensity remained constant, adaptation to the basal R.C.D. occurred. By this means animals moving about in a light gradient and therefore subject to changing light intensities gradually collected at the darkest part of the apparatus. Thus positive correlation of stimulus intensity and R.C.D. resulted in aggregation at low stimulus intensities. This clearly belongs in the class ‘positive kinesis’ according to Kennedy’s (1945) modification of the original scheme (Fraenkel & Gunn, 1961). The R.C.D. of Ullyott is directly comparable to the turning coefficient in the present work; both are rates of turning per unit time. Where an organism varies speed and turning in relation to stimulus intensity it is essential to define whether the R.C.D. represents a rate of change per unit time or per unit distance. If an organism shows no ortho-kinetic effect the rates must be the same. The experimental results described above show that the maximum frequency of turning of Rutilus occurred where the odour concentration is high, but in contrast to Dendrocoelum the aggregation occurred at high stimulus concentration. Thus the mechanism of orientation to an odour gradient shown by Rutilus cannot be a simple klino-kinesis, although there is a change in the rate of turning which is related in some way to the stimulus intensity. The results suggest the following method of orientation. Individual fish spend much time at high odour concentrations where they are relatively inactive. One or two factors may cause it to move away, either adaptation to the stimulus or failure to achieve a consummatory situation i.e. being a unit in a fish school. The fish swims slowly away thus experiencing lower stimulus concentrations, and at some point usually not too far removed, it turns and swims back much more rapidly to higher concentrations of odour. This must involve a spatial and temporal comparison of intensities. It may be that when swimming into higher odour concentrations the increase in attractant odour acts as an immediate positive reinforcement resulting in relatively faster swimming.

The major drawback of the linear gradient apparatus used in this study is that turning is effectively restricted to the length of the tank and therefore is related in an inflexible way to the odour gradient. Confirmatory work is now required in a circular tank in which an intensity ‘gradient’ can be set up radially.

Financial support for this work, which forms part of a Ph.D. thesis submitted to the University of Cambridge, was received from the Development Commission in the form of a Fishery Research Training Grant. I am indebted to Prof. C. F. A. Pantin for accommodating me in the Department of Zoology, and I would like to thank Dr H. W. Lissmann for his help and supervision during the work, and Dr Lissmann, Dr F. R. H. Jones and Mr B. B. Parrish for suggestions at the manuscript stage.