Kennedy, Playford and Beck (1944) concluded, after experiments on dusting settled locusts from aircraft, that the possibility of using liquid spray should be investigated (see Kennedy, Ainsworth and Toms, 1948) and that insufficient was known about the behaviour of swarms at the roosts; before migration begins in the morning. The present part of the work (cited as Gunn, Perry, et al. 1948) consists essentially of a study of behaviour in relation to aircraft attack on settled swarms. Some other observations on swarm migration and settling, and on swarm densities and locust numbers, were made incidentally and are also recorded and discussed here. The subsequent aircraft spraying experiments are described in a separate publication by Gunn, Graham, et al. (1948).

Observations were made in 1945 in the Kenya Highlands on winged but sexually immature locusts on 17 days, starting before or soon after sunrise and continuing until the swarm had departed. Ten of these sets of observations refer to a large swarm, or to parts of it, which was followed for 18 days; methods of locating and following swarms are described by Gunn, Graham, et al. (1948). The following kinds of observation were made in relation to locust behaviour- air temperature, humidity, locust body temperature, radiation, wind speed, and wind direction.

The locusts always roosted for the night in trees or bushes, except on one occasion when many of them were on tall grass and herbs. Some time after sunrise a descent from the trees to the ground would begin, but many of the locusts would remain in the trees until they flew away. On the ground, principally on bare patches, they at first orientated their bodies at right angles to the sun's rays; this basking generally continued until the swarm streamed away, but on several occasions some or many of the locusts had re-orientated in line with the rays before stream-away occurred. From the time when the progressive descent from the trees had become established, there were always locusts flying, and there was a progressive increase in the number flying about, as distinct from flying down from the trees. At a time which was usually determinable within ± 10 minutes, migration would begin with a wide-spread taking to flight and streaming away in one direction; the whole swarm generally took a considerable time to evacuate the site completely.

From the point of view of spraying from the air, the half-hour of growing light before sunrise is the most unfavourable time, for at this time a very large proportion of the locusts are generally inside the trees and so sheltered from the spray by the vegetation. If there were available a method of spraying against which the locusts could not be shielded by one another and by vegetation, this early hour would be most favourable for spraying. Once the sun appears, there is a movement to the eastern sides of trees and eventually a progressive descent to the basking places on the ground. Some extension of the swarm will generally occur then, but basking occurs mostly within a few score metres of the roosts and largely within the original roosting area, in patches where there are no trees. Consequently there is a progressive increase in the number of locusts arranged ideally for attack, basking in some patches as densely as several hundreds to the square metre. At the same time, however, there is a progressive increase in the number of locusts in flight; a proportion of these will, by random movements, avoid the spray, while eventually the flying locusts become hazardous to low-flying aircraft. There is, therefore, no moment at which all the locusts are ideally placed for spraying. In the absence of methods of estimating densities of locusts on trees and bushes, it is not possible to make a quantitative statement about the relative numbers spread out on the ground and remaining in the trees at any moment, but it could be seen, by comparison with the migrating swarm, that the number flying about before stream-away was never a high proportion of the whole. Consequently the best time for spraying would appear to be as late as possible; provided time is allowed to complete the operation before the flying locusts endanger the aircraft.

In considering the time available for spraying operations, it must be remembered that roosting locusts are not easily visible from conventional aircraft and would not be visible even from a helicopter until basking is well established. They become visible at certain angles in sunshine only when flying down densely from the trees. Consequently a ground party is required to delimit and to demarcate the swarm (see Gunn, Graham, et al. 1948), processes which can hardly be started in darkness or completed before sunrise. The earliest time at which spraying can begin is determined by the necessity for demarcation, and the time taken to complete this work must depend on size of swarm, terrain, etc. The time taken to complete the spraying will also depend on size of swarm, etc. Our behaviour observations indicate the times which would have been available for these operations with the swarms we studied. The descent from the trees began at 20 to 180 minutes after sunrise- (average 1 ½hours), thus allowing from 1 to nearly 4 hours (average 2 hours) of daylight. Stream-away began 80 to 200 minutes (average 2 ½hours) after sunrise, or 2 to 4 hours (average 3 hours) after first light. The interval between the descent from the trees and streamaway varied from 20 minutes to nearly 2 hours (average 1 hour). Flying locusts became dangerous to low-flying aircraft between 80 and 200 minutes (average 130 minutes) after sunrise; on several occasions this was about half-an-hour before stream-away and sometimes it was half-an-hour after stream-away. Clearly, useful prediction of the time available for operations could not be made purely on the basis of the previously observed times, and some correlation with environmental factors is desirable.

Full consideration has been given to correlating the time of stream-away with weather factors, while the beginning of the descent from the trees has been less fully considered. Neither stage of the behaviour regime seemed to be related to air humidity, or wind speed and direction, at all. There was no simple direct connection with direct solar radiation; total radiation, including reflexion, was much greater, but could not be suitably measured. There was no satisfactory consistency in body temperatures at these stages. The most consistent data were for air temperature; this was 15-19°C. (average 17°C.) at the beginning of the descent from the trees and 17-23°C. (average 20°C.) at the beginning of stream-away. These temperature ranges are rather wide, certainly too wide for any precise prediction, and suggest that the variations in time were not only due to variations in weather, but also to real variations in behaviour.

It is conceivable that the density of a swarm may affect the readiness with which it begins to stream-away, but there seem to be no substantial grounds for regarding density as a factor likely to alter the air temperature at which stream-away begins; nevertheless, it remains a possible source of variation in our data.

Careful consideration of body temperature data reveals no consistent relation between them and stream-away, except that body temperature was always above the permissive temperature for flight (20°C.) at stream-away. The warmest locusts caught ranged from 27° to 41 °C. at that time on different days. Since stream-away occurs as a result of sustained flight, as distinct from the short flights commonly seen earlier in the day, and since representative body temperatures from locusts in sustained flight could not be obtained, it remains possible that there is a relation between body temperature in flight and stream-away.

Detailed consideration of the air temperature data indicates that when the general weather was warmer, the locusts streamed away at a higher air temperature. A change of 1 °C. in the previous day's maximum air temperature corresponded to a change of about 0.5°C. in the air temperature at which stream-away occurred. Extrapolating from the data, with a daily maximum of 13 °C. stream-away would fail altogether, while with one of 40°C. it would start at 27°C. It is not clear how a locust which is much warmer than the air can react in this way to air temperature, but appendages carrying receptors may be little if any warmer than the air. Precisely this kind of adaptation appears not to have been found previously in animals, and further data are required from a wider range of conditions.

The direction of migration at the time of stream-away was with the wind when its speed was 5 feet per second or over, and against or across it at lower speeds. To understand the behaviour of the locusts in migration, it is necessary to obtain velocity data for both the locusts and the wind. We could find no simple correlation between the wind and the direction of migration of a swarm which was followed for 18 days, but our wind data were inadequate. This swarm kept over ground at a fairly constant altitude of about 6,000 feet until it descended to the plains, where egg laying began. Another swarm which was observed during cool weather moved roughly as would be expected if carried passively by the wind.

A few data indicated that swarm settling occurred at air temperatures of 20-23°C., both during the day and in the evening.

Photographic methods are described. of finding swarm densities of locusts flying near the ground, and of locusts basking on flat ground and in certain other simple situations. By drenching with insecticide strips across a settled swarm, it should be possible to estimate the total number of locusts in the swarm. On two occasions densities of about 660,000 locusts (1 ton) per acre were found and once the density may have approached 700 tons per acre.

In their account of aircraft dusting experiments against settled Desert Locusts, Kennedy, Playford and Beck (1944) pointed out that the planning of such aircraft operations required a greater knowledge of locust behaviour than then existed. In particular, one must be able to predict the time available- for an attack before the departure of a swarm. This report describes our attempt to find an adequate basis for such predictions. At the same time, work on liquid insecticides for dispersal on locusts from aircraft was going on in England (see Kennedy, Ainsworth and Toms, 1948) and it was hoped to make sufficient progress on both lines to justify further field experiments within the year (see Gunn, Graham, et al, 1948); the objective of our field party was therefore to get some information quickly, and to extend the enquiry if time allowed.

The most important requirement from the field party was to establish the time available for attack before a swarm leaves the roosting area. In approaching the problem, the first postulate was that the departure of a swarm from the roosts was determined by observable phenomena, e.g., of weather or locust condition; otherwise prediction of departure would be impossible. It was to be expected that there would also be many small factors affecting the time of departure, so that prediction could never be perfect, even if locust condition could be specified and the weather could be perfectly predicted. Enough information was required to enable the season and place of the aircraft experiments to be chosen so as to give them the best possible chance of success.

The study of roosting swarms in the early morning required the development of methods of finding and following migrating swarms. These are .described by Gunn, Graham, et al. (1948). When the scientific party began to work on the spraying experiments at the end of September, 1945, their practical experience of reconnaissance and familiarity .with the country were vitally important. The areas in which investigations were made are indicated in Table 1.

Key to movements of the scientific party and index numbers of observations on adult locust swarms

go to next section of this publication