Co-Ordination of Successive Activities in an Aphid. Reciprocal Effects of Settling on Flight

The time course of free-flight activity in Aphis fabae Scopoli, and the excitatory and inhibitory after-effects of flight on the antagonistic group of feeding and parturition responses together called ‘settling’, have been described in three previous papers (Kennedy & Booth, 1963a,b, 1964: hereafter referred to by date only). The excitatory after-effect was termed antagonistic induction (1963 b), and the inhibitory after-effect antagonistic depression (1964), from their resemblance to successive induction (post-inhibitory excitation or rebound) and post-inhibitory depression at lower integrative levels.

In the classical examples of co-ordination by successive induction, such as stepping, the effect is reciprocal. Each of the two antagonistic reflexes primes the other in turn, to generate a rhythmic alternation. The question then arises as to whether reciprocal effects occur also with antagonistic behaviour patterns—activities such as flight and settling—although here the alternation is by no means regular. So far these activities have been considered as a one-way sequence only (flight followed by settling) in aphids as well as other insects (references in 1963b, Introduction). If the effects were reciprocal then a primary role for peripheral feed-back in these changes of responsiveness would become still more improbable and the resemblance to lower reflex co-ordination more striking. This paper deals with the after-effects of undisturbed settling upon flight. The explanation of the depression (instead of priming) of settling that may occur after an interrupted flight, as described in the previous paper (1964), appears to be in these reciprocal effects.

The information comes from the three lengthy flight-chamber experiments designated I, II and III for which the material and methods have already been described in full (1963 a, b, 1964). The various treatments given to the groups of experimental aphids are summarized in tables 1 and 2 of Kennedy & Booth (1964, pp. 806, 811) and will be but briefly specified in giving the results in terms of flight activity here. The tests of statistical significance referred to in this paper are either the ‘two-tailed ‘Mann-Whitney U test for comparing groups of different individuals, or the ‘two-tailed’ Wilcoxon matched-pairs signed-ranks test for comparing the behaviour of the same individuals before and after some treatment (Siegel, 1956).

The terms used in presenting the results are explained in what follows. The intensity of an aphid’s flight activity, a photokinetic response, was measured as the aphid’s rate of climb toward the illuminated window occupying the centre of the flight-chamber roof (1963a, fig. 1). This rate was determined by balancing the aphid’s upward flight against a controlled and calibrated down-draught of air from the same window—an air treadmill. The strength of the flying aphid’s positive phototactic response in the horizontal (yawing) plane was gauged by the extent of the flier’s horizontal excursions away from the centre of the window toward which it repeatedly turned back (1963 a, fig 5). The smaller the excursions, the stronger the phototaxis. If these excursions reach as far as the margins of the window the aphid is said to be ranging; passing beyond the margins is called full ranging; flying still further out until the aphid lands on one of the dark walls of the chamber is ranging out.

An aphid that had taken off for the first time and was kept flying in the flight-chamber showed a progressive weakening of its photokinesis and eventually of its positive phototaxis as well (1963a). The final result was that its rate of climb fell to zero and it ranged out. The settling responses now shown by the aphid on landing on a given surface were much stronger than they had been before it had flown (1963b). This increase of response is referred to as rebound of settling after the flight that temporarily inhibited it; in other words, previous flight had boosted settling. The rebound implies a rise in excitability, which is called priming as a general term. In this special case the priming is brought about by a previous antagonistic response (flight) and such a sequence regarded as a central nervous process was called antagonistic induction of settling by flight (1963b, 1964).

If, however, the surface landed on did not stimulate the settling responses sufficiently strongly, as for instance a leaf from a non-host plant, then the aphid did not settle down in spite of the highly primed state of its settling responses. It took flight again toward the lights and now it was the aphid’s flight responses that showed priming. For the rate of climb now showed rebound, rising above what it had been before the landing. Sometimes the rate of climb even surpassed the rate attained when the aphid took off for the first time (excess rebound). The rate then declined again, usually more rapidly than it had on the first flight, and the whole cycle could recur many times (1963 a; also Fig. 5 below). A rebound of flight activity after a bout of settling behaviour was also observed when the aphid had landed before its rate of climb had fallen to zero. Rebound of flight after a landing is illustrated in Figs. 1 and 2 and below (see also 1963a, fig. 4; 1963b, figs. 2c, 2E, 6C, 7).

The rate of climb immediately after take-off was usually greater than the rate observed during the last minute before landing, but even within 1 min. from take-off it sometimes fell below. This depression of flight after a bout of settling is also illustrated in Figs, 1 and 2. The alternation of rebound and depression is seen as an unstable fluctuation of the rate of climb on re-take-off before the rate becomes steady again or falls to zero. Bouts of ranging often accompanied the fluctuations of rate of climb. This was a repetition of the sequence commonly observed at the start of the first flight (Fig. 1 ; and 1963a, figs. 4, 5).

Thus both positive (rebound, boosting) and negative (depression) after-effects of settling on flight were observed, and they varied in incidence and magnitude. This variation was examined in relation to the pretreatment of the insects in an attempt to discover what governs the type of after-effect that settling has upon flight.

Fig. 2 illustrates how the original flight-chamber records have been condensed in order to present the summarized results below. The strength of the aphid’s settling responses between flights is expressed in the diagrams as a settling score (1964). This indicates, on an arbitrary numerical scale, the farthest stage reached by the aphid at that landing in the sequence of settling responses which, under favourable conditions, culminates in larviposition (1963b): 1, no probe on the exposed upper surface of the leaf or other landing platform used; 2, one probe there; 3, more than one probe there; 4, walking over on to the shaded lower surface and probing there; 5, larviposition.

Fig. 3 embodies the summarized results after first flights of various durations in Expts. I and II. Given only about 1 min. of prior flight the rate of climb after a landing returned to about the same level as before. At the end of longer prior flights the rate of climb had usually decreased somewhat and the settling responses were stronger on landing (1963b); when flight was resumed the rate of climb rose promptly to a higher peak than any in the last minute before landing and maintained a higher average value for at least 60 sec. Even the minimum rate of climb went up after a landing when the prior flight had been continued to the point of full ranging (p. 490).

Although settling responses were of course stronger and stays were longer on host than on non-host leaves (1963b, 1964), this made no significant difference to the flight activity after a single landing in Expts. I and II except in the case of the aphids landing after 6 min, prior flight in Expt. I. After landing on the host, but not after landing on the non-host, the rates of climb of these aphids touched significantly lower minimum levels than before the landing. This depression of their rate of climb was too transient to depress the average rate through the first 60 sec. On the contrary the average and maximum rates were significantly raised after landing on the host leaf, as well as on the non-host.

Fig. 4 summarizes the results of further comparisons of the same kind made in the course of Expt. III. These aphids were more uniform in behaviour (1964) and also more vigorous fliers, as can be seen from their higher average rates of climb in Fig. 4. But they were also rather less flight-mature at take-off and flew more erratically at first (1964). The irregular fluctuation of their rate of climb, with some sharp rises and more sharp drops, is reflected in the combination of higher maximum and lower minimum rates of climb in Fig. 4 when compared with Fig. 3. Associated with this nervous instability there was more evidence throughout this experiment of the flight-depressing effect of landing.

As far as maximum rates of climb are concerned the results in Fig. 4 repeat those in Fig. 3. After a single prior flight of 10 min. or more the rate of climb rose again just after a landing to a peak significantly higher than the highest level attained during the last min. before the landing, on all landing surfaces. The minimum rate of climb, on the other hand, went down after landings on the host leaf, given prior flights of 20, 10 or even 1 min. The resultant 60 sec. average rate of climb also went down a little when the prior flight was 1 min., but did not change significantly when it was longer. Flight was less depressed by landings on the non-host leaf ; and given a prior flight of 18 min. the minimum rate went down only slightly while the average went up. There was no significant flight depression after landings on the card; given 20 min. or more prior flight both the minimum and the average, as well as the maximum, went up.

Sometimes the rate of climb after the second take-off reached a value higher than the highest that the given individual had ever attained during its first flight, even if that flight had been long and the aphid had been flying very weakly by the time it landed. This excess rebound of flight is seen in both records in Fig. 1 and cases of it occurred in all treatment groups. In Expts. I and II aphids given prior uninterrupted flights of 6 min., or up to full ranging, showed excess rebound in about 50% of cases after landing on a host leaf and in 11 % after landing on a non-host leaf. In Expt. III the situation was reversed; excess rebound occurred in 44% of aphids landing on a host leaf and in 67% landing on a non-host, after prior uninterrupted flight of 18–20 min. Even in the non-host group, however, the difference between the peak rates during the first and second flights was not quite significant for the group as a whole, leaving it uncertain whether the cases of excess rebound were more than random variation.

In Expt. II aphids were allowed to fly to full ranging, then to land, take off and fly again to full ranging, repeatedly, until they either settled down to larviposit, or made three flights in succession lasting less than 60 sec., or began but failed to take off three times, or ranged out, or, in a few cases, sank to the floor. Two groups made all their landings on a host leaf (young or mature bean leaf) and the third group on a non-host leaf (potato).

The outstanding feature of the behaviour of these aphids was extreme variability, illustrated by the two individuals in Fig. 5. Some account of the concomitant variability of the settling responses at each landing has been given elsewhere (1963b). No attempt will therefore be made to summarize the flight-by-flight data from each treatment-group for comparison. Rebound of flight was the rule after each landing, in the sense that the rate of climb rose after take-off to a value greater than the rate just before the leaf was presented for landing. This boosting effect did not lessen progressively as the total flying time lengthened. Excess rebound, to a rate of climb greater than the maximum that the given individual had attained during its first minutes of flight, was not uncommon toward the end of long series of flights even when flight could no longer be sustained for as much as 1 min. (Fig. 5). If no host leaf was encountered the series of flights eventually came to an end, not as a rule because flight rebound ceased to occur after each landing, but because it was now followed within seconds by a fall of the rate of climb to zero and ranging out.

Since a single landing was enough to boost flight activity to some extent, even after a long prior flight, it seemed probable that repeated landings would enhance this effect by summation, provided that the intervening bouts of flight were kept brief enough to minimize the reciprocal effect of flight priming settling (1963b, 1964). A series of 1 min. flights and landings were used in Expts. II and III and the results were compared with those from aphids flown uninterruptedly. In Expt. II the brief flights were continued for as long as possible, like the uninterrupted flights above, and the flying came to an end in the same way, with flight rebound still in evidence. The results were even more variable and difficult to summarize than those from the serial uninterrupted flights. Expt. III was designed to reduce or control some of the troublesome individual variation found in Expts. I and II. The aphids were more uniform (p. 493) and the main experimental treatments were given within a standard flight period totalling about 20 min.

In Expt. II the only consistent effect of frequent landings was to prevent the rate of climb from declining as rapidly as it did over a comparably extended period of uninterrupted flight. This was most strikingly demonstrated by giving the same aphid alternating periods of uninterrupted and interrupted flight. Further, the maximum and average rates of climb during one minute’s flight often exceeded those during the previous one and this was often repeated so that the rate continued to mount, instead of declining, thus exhibiting the expected summation. Such a run of 1 min. flights showing mounting rates of climb was common within the first 10–20 flights (see Fig. 9), but diminishing runs were also common and no consistent pattern emerged. Nor did any group show excess rebound consistently enough for the excess over the initial rate of climb to be significant statistically. Nevertheless, cases of excess rebound were more frequent with repeated landings on the host than on the non-host, as observed also with single landings in this same experiment, above. The proportion of individuals, out of all starters, that had shown excess rebound at least once reached 50 % after only two landings on the host and not until after nine landings on the non-host.

In Expt. III one group of aphids (A) were flown uninterruptedly for 20 min. as a control for the other groups. The others were given twenty 1 min. flights punctuated repeatedly by undisturbed landings on the same surface, either a young host leaf (group D), or a non-host leaf, Fuchsia (B), or a green card ‘leaf’ (F) (1964). The results were consistent enough to be summarized in some detail. The minute-by-minute rates of climb of all the members of each group have been averaged to give the composite curves in Fig. 6. The mean rates of climb during the first minute, before there was any difference of treatment, differed somewhat from group to group (by 6 cm./sec. at the most) and in order to facilitate visual comparison of changes with time under the different treatments it is assumed, for the purposes of Fig. 6, that all groups began their flying with exactly the same rate of climb as the control group A; the entire composite curve for each other group has been adjusted up or down accordingly. Further to facilitate running comparison of groups, the maximum, the 60 sec. average and the minimum rates of climb are not shown joined as in Figs. 2-5 but are plotted separately in Fig. 6, although they were obtained from the original records in the way shown in Fig. 2. Only those aphids that completed a total of 20 min. flying are included in Fig. 6. Those that ranged out before that time formed a varying proportion in the different treatment groups and are mentioned later (p. 499).

In the uninterrupted control group A the rate of climb reached its highest and its lowest levels within the first minute. The maximum rate fell in the second minute, steadied for several minutes, and then began its slow decline. Brief drops in the rate, which were characteristic of the initial unstable phase of continuous flight as already described (1963a), became less frequent with successive minutes so that the combined minimum curve rose until the steady cruising phase began after about 5 min. of flying. Between these maximum and minimum values the 60 sec. average rate fell in the second minute, steadied and then slowly declined after cruising set in.

Fig. 6 shows that the sequence of flight behaviour diverged from the uninterrupted pattern in all the interrupted groups. The after-effects of landings varied according to the type of surface landed on and thus according to the strength of the settling responses (1964, fig. 8) between flights.

Settling responses were strongest and stays were longest on the host leaf (group D, Fig. 6). After one landing on the host leaf the rate of climb suffered greater drops in the second minute of flight than did that of the uninterrupted fliers (A) in their second minute, causing a significant depression of the D minimum rate (Wilcoxon test, P < 0·05). But at the same time there were renewed rises in the rate of climb so that the maximum rate fell less in the second minute than it did in group A (Mann-Whitney test: P=0·02). On subsequent flights frequent brief drops in the rate of climb continued in group D and kept the minimum rate of climb for the group as a whole below that of group A throughout the twenty flights—with, however, signs of an upward trend. Both the maximum and the 60 sec. average rates of climb in group D rose progressively for several flights after the second; while in group A those rates remained static. The D rates then remained more or less static at their elevated levels while the A rates declined.

Individual variation was greatest in group D. Nevertheless, the incidence of excess rebound was sufficiently consistent, after the first few summating flights, not to be dismissed as random variation. The curve for group D in Fig. 6 includes only the nine aphids that completed twenty 1 min. flights and landings on the host without settling down (1964), but at least six of these individuals showed excess rebound on every one of the flights from the third to the twentieth inclusive. A majority similarly showed excess rebound of the 60 sec. average rate on every flight starting with the third. The 60 sec. average rate was significantly higher on the twentieth flight considered alone than it had been during each aphid’s first flight (Wilcoxon test : P=0·05), and so also was the mean value of the maximum rate at the tenth flight considered alone (Wilcoxon test: P < 0·01).

Group B, landing repeatedly on the non-host leaf where settling responses were weaker than on the host (1964, fig. 8), followed a combined pattern resembling that of group D with one notable difference. Although the minimum rate of climb was significantly depressed after the first landing (see Fig. 4), it was less depressed than that of group D after the first landing, and ceased to be depressed at all after further landings. By the seventh flight it had risen above the uninterrupted (A) level and it continued its upward trend while A remained static. Accordingly, the 60 sec. average rate of climb of B mounted faster than that of D and exceeded it from the fourth flight on. On every flight after the fourth more than twenty of the thirty B aphids reached a higher 60 sec. average rate than each had attained on its first flight; and the excess of the combined 60 sec. average rate on the twentieth flight, over each individual’s 60 sec. average on its first flight, was highly significant (Wilcoxon test: P < 0·0001).

The course of flight behaviour in group F, landing repeatedly on the card where settling responses were weakest (1964), differed least from that of the uninterrupted group A. The combined maximum rate of climb of group F did not decline like that of group A, nor did it rise like that of groups D and B. The combined minimum rate of F was not depressed on the second flight and was less depressed subsequently than that of D ; it ran slightly below A and finally rose to the same level. The combined 60 sec. average ran with that of A and then slightly above A after the tenth flight.

These results confirmed that summation of flight rebound could occur through a series of flights, and also that landings could depress flight activity as well as boost it, both within seconds of take-off. These opposite after-effects of landings on flight were both increased by using a leaf instead of a card as the landing surface and so increasing the strength of the settling responses between flights.

The balance between the opposite after-effects was not fixed, however. The depressing effect diminished while the boosting effect increased, through the series of flights. Thus, landings on the non-host leaf, which elicited comparatively weak settling, depressed the minimum rate of climb only at the beginning of the series. Later, only landings on the host leaf, which elicited the strongest settling, continued to produce drops in the rate of climb sufficient to hold the group’s minimum rate of climb below that of the uninterrupted fliers. At the same time landings on both leaves were boosting the maximum and 60 sec. average rates of climb well above those of the uninterrupted fliers. This shift in the balance of immediate after-effects from each landing as the series progressed was associated with an overall shift in the relative excitability of flight and settling, in favour of flight. Rates of climb were rising (Fig. 6) ; and although the settling responses were being primed the flight responses were being more strongly primed, with the result that settling activity was at the same time becoming more or less depressed (1964, fig. 8).

In order to obtain a direct comparison of the after-effects of interrupted and uninterrupted flight all four groups were given a test-landing on a standard young host leaf at the end of their 20 min. of flying. Some individuals settled down to larviposit there (1964) and were discarded. The ‘before’ and ‘after’ rates of climb of those that took off again are shown in Fig. 7. On take-off the uninterrupted group A showed both rebound and depression of the rate of climb as already noted (p. 493), but the depression was greater and the net result was lower absolute rates of climb than in the three interrupted groups D, B and F, in which flight was more excitable before the test landing (Mann-Whitney U tests of rates after the test landing: minimum, A ν. B and A ν. D, P < 0·05 ; 60 sec. average, A ν. D, P < 0·05).

Another group, G (also shown in Fig. 4), were given a test-landing on a non-host leaf after 18 min. uninterrupted flight and their rates of climb before and after this landing are shown in Fig. 7 alongside the rates of group B before and after their nineteenth successive landing on the non-host leaf (shown under B′). Again it can be seen that the uninterrupted group (G) did not attain such high rates of climb after the test landing as the group (B′) whose flight had been boosted beforehand by frequent landings (Mann-Whitney U tests: minimum rates, P < 0·01; 60 sec. average rates, P < 0·05).

The test-landing on the host leaf given to group B after their twenty 1 min. flights and landings on the non-host leaf provided a further opportunity to compare the after-effect of an immediate difference in stimulation received at one landing, without any difference in excitatory state beforehand. By the end of their twenty flights the B-group rate of climb had stabilized at a fairly high level (Fig. 6, and B’ in Fig. 7). When these same aphids took off after their next landing, now on the host leaf (B in Fig. 7), their rate of climb sank significantly (Wilcoxon tests: 60 sec. average rate, P < 0·05 ; minimum rate, P < 0·001), confirming the greater depressing effect of the host than of the non-host (cp. Figs. 3 and 4). Group F in Fig. 7 showed a similar (but not significant) depression after landing on the host, when compared with the effects of the previous repeated landings on a card.

Ranging (p. 490), which was nearly always what brought uninterrupted and interrupted serial flights to an end, was liable to occur also during the first minutes of flying (1963a) especially when landings were being made on a host. In Expt. II there had been ten instances of ranging by the fourteenth 1 min. flight between landings on the host, although only seven individuals continued flying for so long. With the non-host, there had been only one instance of ranging by the fourteenth 1 min. flight with twenty aphids still flying.

The incidence of ranging during the first 20 min. of flying in Expt. III is shown in Fig. 8. The percentages in Fig. 8 are based on the total number of aphids that flew for more than 1 min. in each group. A number of aphids that started flying in this experiment ranged out in less than 1 min. and were discarded. These are not included in Figs. 6, 7 or 8 in order to distinguish the effects of the different treatments, which began only after the first minute of flight. Aphids that ranged out (and were discarded) after the first minute are included in Fig. 8 (but not in Figs. 6 and 7), together with other aphids that ranged less widely and could therefore be kept flying.

In group A altogether 38 % of aphids that entered upon a second minute of flight ranged out within the 20 min., over half of them in 2-4 minutes. In groups D and B this overall percentage was 13-14 (rangers-out were not recorded in group F). Fig. 8 shows that ranging of any degree became rare among the uninterrupted A aphids as the minimum rate of climb rose and steady cruising began (Fig. 6). It was more common in group B landing on the non-host leaf, but no more persistent. It was both common and persistent, although irregular, in group D landing on the host, like the depressed minimum rate of climb of this group (Fig. 6).

The incidence of ranging thus paralleled that of depression of the rate of climb, confirming that the phototactic and photokinetic reactions of the aphids are associated (1963a).

In Expt. II the earlier onset of excess rebound during 1 min. flights with landings on the host (p. 495) gave a hint that such landings could boost flight activity more than landings on the non-host; but there was no other good evidence of this during the first twenty flights. During the first twenty 1 min. flights in Expt. III flight activity was boosted more by landings on both leaves than by landings on the card, where the settling responses were weakest (p. 497). The settling responses were stronger on the host than on the non-host leaf but there was again little evidence of any greater boosting effect of landings on the host leaf. But the host leaf, unlike the non-host leaf, was simultaneously exerting some depressing effect on flight activity as shown by the minimum and 60 sec. average rates of climb of group D when compared with group A (Fig. 6). Since this depressing effect of the host seemed to be diminishing it was predictable that landings on the host would eventually boost the rate of climb more than landings on the non-host leaf, if the series of flights and landings were extended so as to reduce the depressing effect of the host still further.

It was difficult to compare the after-effects of long series of flights and landings on each of the two kinds of leaves separately, because the aphids were liable to terminate the series by settling down on the host leaf or by ranging out after taking off from it. It was easier to keep an aphid flying and landing without settling down on the host leaf by interpolating landings on a non-host leaf or card upon which the aphids did not settle down (1963b, 1964) and which had less depressing after-effects upon flight. This procedure had the further advantage, in view of the variation between individuals, of permitting the after-effects of landings on different surfaces to be compared consecutively in the same individual.

During Expt. II eight aphids were each given a first run of 8–14 1 min. flights with landings on a non-host leaf. In most cases this led to excess rebound of flight and in all cases it depressed settling readiness enough to prevent settling down on a host leaf (1964, p. 810). The aphid was then allowed to land on a host leaf after another i min. flight, followed by a run of four more landings on the non-host between 1 min. flights, a second single landing on the host, and so on, as illustrated in Fig. 9.

This particular aphid (Fig. 9) did not usually probe at all on the non-host leaf, taking off again after a quarter of a minute or less. Its rate of climb began high, fell sharply on the second flight and then mounted rather steadily through the next fourteen flights. It landed next on the host leaf, and went below for more than half of the 6 min. that it stayed on the leaf, making many brief probes. On take-off its rate of climb now rose spectacularly. Even the average and minimum rates during that minute were higher than the highest maximum previously attained. There was no trace of the flight-depressing effect of landings on the host which was usual in less flight-excited aphids (e.g. Figs. 3, 4, 7). The rate of climb then declined again with several landings on the non-host, was once again sharply boosted by one landing on the host, and so on for ten similar cycles.

The combined results from the eight aphids treated like the one in Fig. 9 are expressed in Table 1 as the numbers of increases or decreases in the rate of climb from one flight to the next separated by a landing on the stated leaf. Under these conditions the host leaf clearly tended to boost the rate of climb, including even the minimum rate, above the level previously reached as a result of landings on the non-host; whereas after the next following landing, now on the non-host once more the maximum and average rates fell. The minimum rate was about equally likely to rise or fall on that flight, but was more likely to fall after further landings on the non-host as shown in Fig. 9. The differences between the effects of host and non-host in Table 1 were highly significant for the eight aphids combined (² tests: maximum and average rates separately, tach P < 0·001). Note that one individual reacted repeatedly in exactly the opposite way.

In Expt. III some of the aphids that had made twenty landings on the host leaf were kept flying with landings on the host leaf for considerably longer, again by interpolating landings on non-host surfaces. Fig. 10 reproduces part of the original record from one of these aphids while it was being given a succession of approximately 3 min. flights between landings alternately on a host leaf and a non-host leaf. The aphid had already made twenty-two 1 min. flights with twenty-one landings on the host. This had exalted the rate of climb and had depressed settling readiness enough to prevent settling down on the host. There had followed five 3 min. flights alternating between host and non-host before the excerpt in Fig. 10 begins. This aphid was now clearly showing greater flight rebound after the strong settling responses and long stays it usually made on the host (and once also on the non-host, at X in Fig. 10) than after the weak settling responses and brief stays usually made on the non-host (and once also on the host, at Y).

Other aphids were given extended series of 1 min. flights only, with a run of landings on a non-host leaf or green card. The two individual records in Fig. 11 serve to illustrate the range of behaviour obtained. Flight became very excited relative to settling in the aphid of Fig. 11A in the course of its first run of twenty-five flights and landings on the host leaf, and then, during the following run of landings on the non-host leaf, the maximum, the 60 sec. average and even the minimum rate of climb fell slightly. When landings on the host leaf were resumed the maximum and 60 sec. average rates rose again, but this time there were also brief drops which depressed the minimum rate. This whole cycle was then repeated. The behaviour thus reproduced that of the aphids in Figs. 9 and 10 in showing greater flight rebound after landing on a host than on a non-host, but in less extreme form.

The aphid in Fig. 11B started with a lower rate of climb than that in Fig. 11A and its flight did not become so strong nor its settling so weak. After the first run of twenty flights and landings on the card landings on the host did not boost but instead depressed the minimum and 60 sec. average rates of climb, and sometimes even the maximum rate; these rates were boosted again by landings on the card. This cycle was twice repeated. The settling responses to the host were weaker during the second run of landings on it than during the first and the third. Correspondingly, there was less depression of flight and this disappeared again while the landings were still being made on the host leaf. The settling responses to the host were strongest during the third run of landings on it and the depression of flight was then correspondingly severe, with much ranging. Several successive landings on the card were now needed before the rate of climb was brought up to its previous level again. This aphid was thus behaving like the average individual during the first minutes of flight (Fig. 6) in showing greater depression of flight after landing on the host than on a non-host.

The combined results from the four aphids that were given the same treatment as that in Fig. 11B are shown in Table 2. Landings on the host both boosted and depressed the rate of climb significantly more than did landings on the card (Fisher tests (Siegel, 1956, table 1): maximum rates, P = 0·01 ; minimum rates, P = 0·025).

Thus in Expt. III, as in Expt. II, once flight had become sufficiently excitable relative to settling, the rate of climb was boosted more by the strong settling responses elicited on a host leaf than by the weak ones elicited on a non-host.

At first sight the rebound of flight activity after a landing might appear readily explicable in ‘peripheral’ terms without invoking the central nervous interaction of settling and flight. When exhausted and recalcitrant flies are induced to fly again by feeding them with sugar, this is assumed to act peripherally, by refuelling the flight muscles (Wigglesworth, 1949; Clegg & Evans, 1961). But the flight rebound with which we are dealing was a frequent occurrence after landings so brief as to preclude any food intake from a leaf, and even after landings on a card (Figs. 1B, 4, 11B). Wigglesworth’s (1949) observation that an exhausted Drosophila can fly again if simply given a rest, with no food intake but with time for depleted reserves to be mobilized, seems more relevant. Three minutes of further flight was the most he obtained in this way, however, and that only after a rest of an hour. The aphids’ heightened flight activity does not seem at all comparable for it was often observed after a rest of less than 1 min. (e.g. Figs. 1B, 5, 10, 11) and the flight could then continue for many minutes up to an hour or more even in aphids that had already flown to the point of apparent exhaustion (1963a; also Fig. 5 of the present paper).

Evidently free-flying aphids behave as if they were exhausted (rate of climb zero, turning away from lights) well before they are exhausted in the sense in which that term has been applied to tethered, forced fliers, including Aphis fabae (Cockbain, 1961). In nature, this behaviour is indispensable if host plants are to be found, since landings are indiscriminate (Kennedy, Booth & Kershaw, 1959a, b) and hosts will seldom be the first plants landed on. In the flight-chamber free-flying aphids certainly did become exhausted in the usual physical sense, with drastic shortening of flight bouts (1963a; also Fig. 5 of the present paper) and final failure to respond to strong flight stimulation such as being dropped. But the present work is not concerned with behaviour at that extreme stage, although the results do incidentally imply that current physiological interpretations of it may be unduly ‘peripheral’.

Aphids that were not yet even behaving in an exhausted manner also showed flight rebound after a landing (Figs. 3, 4; also 1963b). In Expt. III, following some minutes of uninterrupted flight, landings on all three surfaces boosted subsequent flight activity for some seconds at least (Fig. 4). But landings on the host leaf boosted flight the least, and, within the same minute, depressed it the most (Fig. 4). This is the opposite of what would be expected if the boosting effect were due to the rest gained by a landing, since it was on the host leaf that the aphids rested longest (1964: figs. 4A, 6 G, 7H).

The depression of flight activity after landings on the host leaf was especially note-worthy because these aphids had been, in effect, selected for readiness to fly. A proportion of the group members did not take off from the host leaf again but settled down to larviposit (1964, fig. 4) and were discarded, thus excluding from Fig. 4 of the present paper the individuals that were most ready to settle and least ready to fly. The individuals that were least ready to fly in the other groups, landing on the non-host leaf and card, did not settle down there and are therefore included in Fig. 4. Yet these groups showed less flight depression. Apparently the individuals that took off from the host leaf did so before they were entirely free of the flight-inhibiting influence of such a leaf, which was great enough to prevent some individuals in their group from taking off at all.

The rests gained during frequent interruptions of flight could be expected, at most, to prevent the general rate of climb from decreasing as it did during uninterrupted flight. In fact, an actual summation of flight rebound occurred. The rate of climb mounted through a series of flights to a level significantly higher than any reached by the same aphids during their first minutes of flight (excess rebound: Figs. 6, 9, 11). The individual record in Fig. 9 represents an extreme but not an isolated case of spectacular rebound of flight activity after a strong settling response, and such cases seem quite incompatible with the rest hypothesis of flight rebound.

It appears, therefore, that flight rebound is a nervous after-effect of settling. The immediate effect of settling is to inhibit flight ; and inhibition followed by rebound is the type of succession earlier called antagonistic induction (1963b), referring then to an after-effect of flight upon settling. A like sequence, with the same components occurring in the reverse order (i.e. antagonistic induction of flight by settling) is now equally evident in the after-effects of settling upon flight.

The other after-effect of settling, depression of flight, appears to be a partial carry-over, after take-off, of the inhibition imposed upon flight during settling. This ‘after-inhibition ‘or antagonistic depression of flight by settling is, again, a type of sequence already described with the same components in reverse order, as the antagonistic depression of settling by flight (1964).

The conclusion that flight rebound depends on the prior inhibition of flight during settling is supported by the detailed results. For the after-inhibition (depression) and the rebound were associated in time and in the pattern of their variation. Both after-effects were increased by stronger settling responses, and the one gave place to the other as flight became more excitable relative to settling (p. 498). Thus, landings on the host leaf, which excited settling more than the other surfaces and inhibited flight the longest, not only depressed flight the most afterwards (Figs. 3, 4, 6, 7, 1,1B), but also, when flight had become very excited, boosted it the most (Figs. 9, 10, 11 A: Tables 1, 2). When flight was less excited before a landing the inhibitory after-effect of the landing was more in evidence. The more flight was excited prior to settling, the less depressed and the more boosted it was when released from the inhibition imposed by settling.

These relationships account for the depression of settling after interrupted flight, which was described in the previous paper (1964). Consider the case of two successive landings made on the same surface and separated by a flight of only 1 min. In Expts. I and III it was found that settling was significantly more likely to be depressed (weaker) at the second landing when the first flight (before the first landing) had continued uninterruptedly for a number of minutes than when it had been cut short after one (1964: table 3 and fig. 9). The rate of climb had usually begun to decline after some minutes of uninterrupted flight (1963 a; also Figs. 1 and 6 of the present paper), and the usual result of a landing then was rebound of the rate of climb (Figs. 1, 3, 4). Since the aphid was now allowed to fly for only one more minute before landing again, this second landing occurred before the aphid’s rate of climb had declined again to the level obtaining just before the first landing ; and the priming effect of the second flight on settling was still small. Thus flight was more excitable relative to settling at the second landing than it had been at the first, and the inhibitory after-effect of flight on settling would be expected to be correspondingly enhanced during the second landing. The observed depression of settling at the second landing was presumably the result.

It was further found that settling was more likely to weaken, from the first to the second of two landings on the same surface, when that surface was a non-host than when it was a host (1964, fig. 10). This may now be related to the fact that flight was depressed more after landings on the host (Figs. 3, 4, 6, 7), for this would reduce the inhibitory after-effect of the second flight on settling.

In Expt. III the rate of climb was boosted by a series of 1 min. flights and landings on a leaf, instead of falling as it did during uninterrupted flight (Fig. 6). Accordingly, the settling responses at the final test landing on a standard young host leaf were weaker in the interrupted fliers (B and D) than in the uninterrupted fliers (A) (1964, figs. 4, 5). A like series of flights and landings on a card had less boosting effect upon flight (F in Fig. 6) and, accordingly, little subsequent depressing effect upon settling at the test-landing on the host leaf (1964, fig. 5 and p. 814). The rate of climb reached the highest absolute level as a result of the serial landings on the host leaf (Fig. 7; Fig. 6 shows relative changes only) and, accordingly, the final settling responses at the test-landing on the host leaf were the most depressed (1964, pp. 817–18 and figs. 5 and 8).

The observation that settling was more likely to weaken, from the first to the second of two landings on the same surface, when the settling was strong at the first landing than when it was weak (1964, figs. 9, 10), will be considered elsewhere when further experiments with better control of the settling responses are described.

When the antagonistic induction and depression of settling by flight were described, it was pointed out (1964, p. 823) that these terms referred to two types of change in the excitability of settling, after flight. They were not mere formalized descriptions of behavioural sequences, but lower-order components. Although actual sequences could take either form under appropriate conditions, the same type of co-ordination could be at work when overt behaviour takes a different form, owing to interaction. The same applies to the reverse processes, antagonistic induction and depression of flight by settling, both of which could be observed within a few seconds (e.g. Fig. 10). The immediate after-effect of a landing on the host leaf, after twenty previous landings on the non-host leaf (B in Fig. 7), was depression of the rate of climb; but the continuing boost being given to flight by the previous landings was revealed by the fact that this depressed rate was still significantly higher than that of previously uninter-rupted fliers taking off after a landing on the same leaf (A).

The principal new point that emerges from the present results is the two-way interaction of flight and settling. Each activity has the same positive and negative after-effects on the other. The reciprocity of antagonistic induction recalls lower-order co-ordinations involving central ‘oscillators’, as in stepping (Sherrington, 1906; Hoyle, 1964), or wing-beating (Wilson, 1961 ; Wilson & Gettrup, 1963), although, at the behavioural level, the time-scale is longer and the alternation is not regular, and antagonistic depression is also common. The reciprocity likewise strengthens the previous conclusion (1963a, b; 1964) that antagonistic induction and depression are central nervous events, for there has not yet been any suggestion that known types of peripheral link could act in the required way.

Various types of peripheral link have been postulated (Johnson, 1958; Müller, 1965), but a hormonal one would probably be considered most likely at present. In the pea aphid, Acyrthosiphum pisum (Harris), degeneration of the flight muscles is initiated on the second day after settling down on a host plant (Johnson, 1959), and in Aphis fabae probably on the third day (Johnson, 1957). This eventual outcome of the antagonistic induction of settling by flight is not nervously but peripherally induced, by some blood-borne factor (Johnson, 1959). Similarly, it is thought that release into the blood of neurosecretory material stored in the corpus cardiacum provides the connecting link in certain more rapid and strictly behavioural sequences, where locomotor activity gives way to copulation or oviposition in insects (Highnam, 1962, 1964; Hodgson, 1962). Although these sequences have so far been investigated as one-way processes only, there are other reports of reciprocal effects of neurosecretion and secretion from the corpus allatum, or from corpus allatum and thoracic glands, on behaviour as well as on morphogenesis (Beetsma, de Ruiter & de Wilde, 1962; Engelmann, 1960; Highnam, 1964). Thus, two-way, reversible sequences could, in principle, be co-ordinated by such peripheral mechanisms, given time.

Yet Blest’s (1960) experiments on the induction of settled ‘rocking’ by prior flight in the moth, Automeris, made anything but a central nervous mechanism most im-probable. And in the aphids large changes of responsiveness to flight and settling stimuli can be brought about, either way, within minutes or even seconds by eliciting the antagonistic activity. It seems still more difficult to imagine these rapid and reversible changes as being co-ordinated other than centrally, whatever endocrine and other peripheral mechanisms they may then entrain. Especially so, since the after-effect of one antagonist on the other is not of one single kind but may be excitatory or inhibitory. As we have seen, which kind of after-effect predominates at any moment varies according to the excitability levels built up beforehand and the strength of immediate stimulation.

The painstaking assistance first of Mr C. O. Booth and then of Mrs Lorna Crawley is gratefully acknowledged.