A data acquisition technique is described, which uses a digital analyser to measure off-line the lengths of intervals between events on two (or more) lines. The method compensates for cumulative flutter of tape recorders; thus the temporal relationships between series of events on different lines are maintained.
On the introduction of a Mormyrus fume into the tank of a resident Gnathonemus petersii, both fishes increase their mean electric organ discharge (EOD) frequencies. Range and relative proportion of interpulse intervals (Fig. 1) as well as the discharge rhythms (Fig. 3a, b) are different from that observed in isolated animals, at rest or while swimming (Figs. 4, 3c--f). During agonistic behaviour, the discharge
rhythm of G. petersii exhibits a high degree of variation, whereas for M. rume the inverse is true. At rest and during swimming, however, an isolated G. petersii displays a marked tendency of at least eleven consecutive intervals to stay all above or below the mean value; at rest, there is a tendency for a long interval to be followed by a shorter one, the next two intervals being again long (or vice versa). With minimal delay, G. petersii tends to regulate the lengths of its discharge intervals inversely in response to concurrent M. rume intervals (Fig. 5). There is no regulation of M. fume intervals in response to G. petersii pulses. Furthermore, G. petersii (~ c~ and a ~) tends to discharge with a latency of approx. 12 ms to the preceding M. fume pulse (Fig. 7). This effect is shown neither to be the result of a particular phase relationship (Fig. 6), nor to be due to the activity of the M. rume (Figs. 8, 9). In five out of six G. petersii, the observed latency distribution differs significantly from what would be expected if the two discharge trains were independent.
G. petersii tends to produce "preferred" latencies in runs of up to twentyone (Fig. 10). Animals which are less aggressive, display a greater number of preferred lateneies (Fig. 11) and longer runs (Fig. 10). Runs of two or more preferred latencies never occur during attack associated burst activity. It is only during these bursts that EOD coincidence was observed. The significance of the preferred latency response is thought firstly to be avoidance of discharge coincidence in intraspecific
social behaviour. Secondly, evidence is discussed which seems to indicate that it is a "hiding" behaviour.