go to previous section of this publication

Assessments of the losses in hopper populations due to predation and other causes were made on four occasions.

Three of the studies were made on first- and second-instar hoppers and in these cases the number of hoppers expected at hatching (based on assessments of the viable egg population) was compared with the number present a few days after hatching.

At El Rago site C, observations began on the laying swarm, which was not very dense and had probably laid before. The largest concentration of laying females observed comprised 14. The soil was uniformly wet over the whole site, the exact limits of which could not be determined since laying took place during a single night. Inspection of the site the next day, when probe holes were still obvious, indicated that laying had been fairly heavy over an area of some 200,000 sq. yd. To allow study in greater detail, only a proportion of the egg field was surveyed. This was an area of 200 x 200 sq. yd., in what we believed was the central part of the egg field. Samples, each of one square yard, were taken at random according to the pattern shown in Fig. 13. One quarter of the area was sampled more intensively than the rest, but shortage of time made it impossible to continue sampling on this scale. The frequency distribution of the 180 samples for egg pods is shown in Fig. 15 and, as is usual for locust egg fields was not a normal distribution. The mean number of pods per square-yard sample was 2.472 ± 0.249 and from this the total number of eggs laid in the study area of 40,000 sq. yd. was calculated. Using the data from Table 15 an assessment was made of the number of hoppers expected to hatch (Table 19).

Fig. 15-Frequency distributions of the numbers of pods found per sample (A) and the number of dead first-instar hoppers per sample (B) at El Rago site C. Sample areas were one sq. yard each. (See also Fig. 13.)

Table 19
Studies on mortality in first three instars of the Desert Locust and estimates of percentage survival. (Figures in brackets give fiducial limits.)

The original plan had been to kill the hoppers, by sprinkling them with petrol, as soon as they became concentrated into well defined groups, then to collect them and estimate the number by volume. There was very little movement of hoppers because of continuous overcast weather and concentration was still incomplete three days after hatching began. It was decided, therefore, to kill all the hoppers by liberally applying poison bait to the area of 40,000 sq. yd. and to estimate the number of hoppers by counting those within one-square-yard sample areas, distributed in a similar way to those for the egg-pod counts. Hatching was completed some 20 hours before killing took place. In part of the area, covering 2,500 sq. yd., the dead hoppers were in fairly well defined groups. These were measured and sampled separately, using one-square-foot samples. A single check on this method was made. This involved sampling an area of dead hoppers (16 x 23 sq. ft.) and then collecting and counting them. From the sampling, the estimated total was 69,552 (fiducial limits, 24,228-114,816) and from the total count 62,853. The frequency distribution of the number of dead hoppers per square-yard sample for the rest of the 40,000 sq. yd. is shown in Fig. 15; the mean was 38.24 ± 4.30 hoppers per sq. yd.

Comparison of the number of hoppers expected on the basis of the egg counts with the number found 54 hours after hatching began showed that there had been a loss of some 1,500,000 hoppers or 42% of the population (Table 19).

The egg field at El Rago site B was laid by a low-density swarm, which had been part of a large and dense swarm that laid on the other side of a ridge of low hills. Laying began at 20.00 hours, with the females in small groups of not more than 20, and continued during the night. The following morning a general survey was made of the egg field. Two methods of sampling for egg pods were used. In the first, 22 samples, each of 10 x 10 sq. ft., were chosen at random within the egg field (Fig. 14). Pods were found in only three of the sample areas. In the second method, two trenches one foot wide were dug at right angles to each other across the egg field; all the dug soil was examined for pods, and in all, 1,202 sq. ft. of ground was examined. This method proved very laborious. Some of the data for the egg-pod sampling are shown in Table 20.

Table 20
The frequency distribution of numbers of pods per sample for three egg fields.

The positions of the groups of newly hatched hoppers and their subsequent history were carefully recorded and mapped. Four days after hatching was first observed, a survey of the site was made with a view to assessing the total number of hoppers present. The numbers roosting in groups were estimated by comparison with similar groups on the Dagabur egg field, where all the hoppers were killed, collected and counted. The numbers in 'basking' groups were obtained from careful measurement of the areas of these groups and from observations as to whether the hoppers were one or more deep; sampling had shown that the mean densities of first-instar hoppers per sq. yd. were 679.6 ± 82.1 when they were one deep and 1,746 ± 82.6when they were on top of each other. The survey was completed in the period from dawn to the end of morning ground-grouping.

Table 19 shows that the sampling errors for the estimation of the egg populations were very large, but both methods of sampling seemed to indicate that there had been a considerable loss amongst the first-instar hoppers. The estimates suggested that at least 80% of the population had been lost, during the first four days after hatching.

At Dagabur, the parent swarm was described by experienced observers as one of low density. Assessment of the egg population was carried out by sampling, using a system of five randomly chosen lattices, each of 100 sq. ft.; one-square-foot samples were taken at 10 ft. intervals (Table 20). A count of the subsequent hopper population on the same site was made by killing collecting and counting the hoppers as the groups were located in the fairly dense bush. Hatching began on the 23rd May and the last hoppers were killed on the 3rd June when they had reached the second instar.

The results of this study are given in Table 19, which shows that there was a considerable discrepancy between the numbers of hoppers expected on the basis of the egg sampling and those actually counted. .

It was estimated that since hatching there had been a loss of up to four million hoppers, or 97% of the population.

In the final study made on the Eritrean coast, three estimates of the same hopper population were made at intervals of a few days. The most reliable method of estimating the numbers of hoppers living in bands is that which we have called the area-density method (p. 20). Briefly, this depends on the tendency of hoppers in bands exhibiting the same general behaviour pattern to have approximately the same mean density.

In this Eritrean study during February 1957, the hopper population of a single egg field (one of four similar egg fields in an area of 12 sq. miles) was kept under constant observation throughout the period of its existence. The position and size of each band were mapped at intervals, and the results are illustrated in Fig. 16. The estimates of the numbers of hoppers present were obtained by the area-density method (Table 11). At the end of the first instar it was estimated that the 17 hopper bands contained 15 million hoppers. Five days later there had been an estimated natural loss of over 3 million hoppers or 20% of the original population. Ten days later again, there had been a further loss of just over 5 million hoppers so that the total loss due to natural causes was 55% of the original population (Table 21).

Fig. 16-Map showing results of successive surveys of a gregarious hopper population on the Eritrean coastal plain in February, 1957.

Table 21
Showing the reduction in the number of hoppers in a small gregarious population of the Desert Locust on the Eritrean Coast-1957

In our experience, mortality due to disease was non-existent. One hopper parasite, the nemestrinid (Symmictus costatus Loew) was found in a few hopper bands, where its maximum incidence was 34% (Shulov 1948; Greathead 1958b). This parasite does not appear to kill the hoppers until they have reached the fourth or fifth instar.

Our various observations have indicated that natural losses in gregarious hopper populations are due mainly to cannibalism and predators. Sometimes unfavourable environmental conditions, such as excessive heat and low humidity, may cause death; this will be considered in relation to cannibalism.

In the laboratory, newly moulted hoppers are often eaten by others. In the field, newly hatched hoppers are frequently eaten by those that have hatched earlier. This may be illustrated by the sequence of events at El Rago site B. Although the layings had been scattered, pods were stir in groups of up to 20. Vegetation on the site could be described as luxuriant (Figs. 16 and 42 in Ellis & Ashall, 1957) and there were plenty of food plants such as Melhania and Blepharis (lc., Table 19) for the young hoppers. Humidity was fairly high, and heavy morning dews were recorded.

During our field observations we have noticed that Desert Locust hoppers hatch before or within three hours after sunrise. Predtechenskii (1935) also found that hatching was limited to the period two hours before to about three hours after sunrise. It may be that rising temperatures during the morning prevent hatching until the next day. Provided that the embryos are fury developed, it seems that hatching itself may be induced by lowering the temperature. It is the practice of one of us (P.E.), when requiring newly hatched hoppers in the laboratory, to transfer egg pods due to hatch from a room where the temperature is 28°C to one where it is 15-25°C; some hoppers invariably hatch within an hour. This same procedure was not successful with another species, Locusta migratoria migratorioides (R. & F.). It seems likely that the hatching of Desert Locust eggs in the field is triggered off by a fall in temperature during the night. Locust eggs incubated under the same conditions have the same rate of development and this accounts for the fact that hatching on any one egg field is generally completed on two or three consecutive mornings.

On site B at El Rago, on the first day of hatching, hoppers were observed emerging from the soil half-an-hour before sunrise. Hatching continued for three hours after sunrise. It took 2-3 hours for the cuticle of the hoppers to darken. Feeding, on seedling plants, was first observed 3.5 hours after sunrise and, a little later, some recently hatched hoppers, whilst wriggling out of their intermediate moult skins, were seized and eaten by the hoppers that had hatched earlier that morning. During this first day there was little movement of the hoppers other than wanderings between basking and roosting sites. On the second day hatching began before sunrise, but did not continue for long afterwards, even though the air temperature at the time was only 18.5°C and that of the soil surface 19.5°C. Just before sunrise, some of the hoppers that had emerged on the previous day left their roosts and began to devour the newly hatched ones. During this second day small hopper group.-, joined to form bands (Fig. 18 in Ellis & Ashall, 1957). A little more hatching took place on the third morning at dawn, when most of the newly emerging hoppers were eaten by the older ones.

On this egg field it seemed most unlikely that there was any shortage of plant food to account for the cannibalism which occurred. On the four occasions when egg fields have been under almost continuous observation during the hatching period, cannibalism of the kind described above was seen. It was most striking and occurred on the largest scale when the rainfall had been insufficient to provide a good supply of suitable tender plant food. The first of such occasions was at Lina in northern Arabia in April 1952. Here laying had been dense and hoppers hatched in large numbers. Densities at hatching of over 700 hoppers per sq. ft. were recorded. Vegetation suitable as food for first-instar hoppers was almost non-existent. Hatching began on the 3rd April and on five successive mornings newly emerging hoppers were observed being voraciously eaten by hoppers from the previous days' hatchings. In some cases hatchlings were pulled out of the ground as they neared the surface. It is considered that such cannibalism was a factor which largely contributed, on this occasion, to the virtual extinction of the population by the time the second instar was reached.

A similar, though less extreme, case occurred at Wajir in northern Kenya in December 1954, where again the rainfall had been light and annual vegetation was sparse. No cannibalism was seen on the first day of hatching, but here no hoppers emerged later than 1.5 hours after sunrise, when the air temperature had reached 26°C. On the second and third mornings hatching was restricted to the dawn period, when the older hoppers were seen to be feeding on the newly emerging ones and, as at Lina, to be dragging them from the ground. At Wajir, different parts of the egg field hatched on different days, and on the third day almost all the hoppers were eaten by the older ones which marched into the area where hatching was taking place. Total losses due to cannibalism were estimated at over 50%. Cannibalism was also observed at Scillave, Ogaden, in November 1955, but it was not on such a severe scale; some annual vegetation was present.

Cannibalism has occasionally been seen amongst older hoppers. One striking example occurred at Wajir in 1954-55, where the bush was rapidly drying out by the time the hoppers were in the fifth instar. Under these conditions it seems likely that the hoppers were short of water. On several occasions, when bands were actively marching, individuals that slipped into cracks in the soil were set upon by the other hoppers and eaten. At Lina, in 1952, in addition to the severe cannibalism amongst hatchlings, older individuals in bands were often seen to be attacked and eaten by the hoppers marching along with them.

The Desert Locust inhabits areas of generally low, uncertain and unevenly distributed rainfall. Its survival and its capacity for increase depend on its being able to utilise the rainfall which does occur. Rainey & Waloff (1948, 1951) and Rainey (1951) have provided an explanation of how adult locusts and rain arrive in the same places at the same time. The actual amount of rain that fall-, is then a most important factor affecting the size of the increase in the new generation of locusts.

Following good rains (about 5-9 in. distributed over 6-8 weeks) in these semi-desert areas, there is rapid growth of grass and herbaceous plants and these, together with the leaves of the trees and shrubs, remain green for about two months. This is just sufficient time for the locust eggs to incubate and the resulting hoppers to become adult. Such conditions as these are necessary for maximal increase in the population. There are several ways in which rainfall deficiency affects the increase. Sometimes there is complete failure of the rains and this prevents sexual maturation so that no eggs are laid. At other times there may be sufficient rain to induce laying, but not to provide completely suitable oviposition sites. Oviposition may occur, but the eggs shrivel and die because there is not enough moisture in the soil for them to incubate (Shulov 1952; Popov 1958).

On other occasions, as at Lina in 1952, oviposition is followed by successful hatching, but the rain is not sufficient to provide a supply of young annual plants for the hoppers to eat. This, coupled with high temperatures and low humidity (averages at Lina were: maximum air temperature 92.9°F, minimum relative humidity 5%), leads to the death of the hoppers by desiccation. On several days at Lina groups of dead hoppers were found in the afternoon beneath the shrubs where they had sought shade during the heat of the day. On one occasion several hundred hoppers shaken from a bush at midday died within 10 min. of reaching the ground surface. Loss of water by evaporation was probably aggravated by wind-blown sand that could easily damage the wax covering of the exocuticle (Beament 1959). It is clear that under such conditions cannibalism provides the hoppers with much-needed water and probably other nutrients also. Part of the population survives for several days and is able to take advantage of more favourable conditions, should they supervene. On occasions like that at Wajir in 1954, the rainfall had been sufficient to provide some plant food for the hoppers, but there was a lack of the tender annual plants that first-instar hoppers appear to need. In such cases cannibalism enables a part of the population to survive until the hoppers are old enough to eat the generally tougher leaves of the trees and shrubs.

It is clear therefore that cannibalism can, in certain circumstances, have direct survival value, since it enables part of the population to reach the adult stage when large-scale migrations become possible. It is not easy to explain cannibalism amongst hatchlings that apparently have an abundance of moisture and young green plants. Perhaps it is part of a spontaneous pattern of behaviour (cf. marching, Ellis 1951). Alternatively it may merely be that other hatchlings are an easily available source of some necessary food material, such as protein. Whatever its cause, cannibalism appears to be common in first-instar hoppers of the Desert Locust, even though it has received little attention in the published literature. There is little doubt that it is sometimes responsible for very large losses in locust populations, even when there is plenty of suitable green food.

Returning to the population studies summarised in Table 19, most of the loss at El Rago site C was due to cannibalism. Some predation went on, but was not on a large scale. There was ample food for the hatchlings and no reason to believe that any were killed by unfavourable conditions such as excessive drought. It seems reasonable to assume that at least 30% of hoppers were lost through cannibalism on this site.

At El Rago site B, cannibalism was observed, but here, as at Dagabur, the losses were only partly due to it. At least half of the recorded losses were due to predation.

Locusts provide suitable food for any animal that includes animal protein in its diet, so there is a continuous loss in locust populations due to predators. Very little quantitative information has been published, and detailed and prolonged studies are greatly needed. Our field observations on gregarious hopper populations over a number of years have led us to conclude that most of the loss is due to birds. The number eaten per meal obviously depends on the sizes of the hoppers and the bird. Small species like warblers may eat only 40 fourth-instar hoppers a day (Hudleston 1958), whereas storks may eat several hundred adult locusts in a meal (Smith & Popov 1953). The number of birds present throughout hopper life will affect the relative importance of the various species. Some are chance predators, attacking locusts which they come across whilst they are themselves migrating. A thorough study may show that resident birds like ravens and hornbills do more damage, on the whole, than chance visitors which eat very large numbers of locusts in one meal.

Mammals undoubtedly take a steady toll of locusts, especially aard-varks, jackals, hyaenas, hedgehogs, mongooses and various species of rodent. The last may store locusts in their nests, as British species have been found to do (unpublished), and so may kill far more than they eat.

Of the smaller animals that attack hoppers, scorpions, spiders and ants are probably the most important. However, we have never had any evidence that they are in sufficient numbers in the Horn of Africa or in Arabia to cause large losses in locust populations. Lizards and snakes eat locust hoppers readily, but they too probably have little effect on locust populations.

At El Rago site B, predation was mainly by small birds, many of which were nesting in the area, and jackals. Five days after hatching a small roosting group of about 1,000 hoppers was marked; by the next morning no trace of them could be found, but there were many jackal tracks in the vicinity of their roost. Eleven days after hatching, when the other hoppers of the locality were in the second instar, predation had further reduced the population to a few scattered hoppers that assumed a green, non-gregarious coloration after moulting to the third instar.

At Dagabur, predators were more numerous. Those seen on the site were a small flock of Abdim's storks, an occasional Marabou stork and numerous smaller birds, as well as one greater bustard, a jackal and an ant bear. A Carabid beetle (Graphopterus cineraceus), a black ant (Paltothyreus tarsatus) and a wasp (Tachytes observabilis) were also seen preying on the locusts. Some of the hoppers lost from this site would be accounted for by cannibalism (probably not more than 30%). The predators present might be expected to have accounted for the rest. In this same area, two hopper bands were kept under constant observation. One of these, which contained about 30,000 hoppers in the first instar, was reduced to only 100 hoppers in the fifth instar. The other, estimated to have originally contained 15,000 first-instar hoppers, had by the fourth instar been reduced to about one third of this number; it then completely disappeared during one night and tracks of small mammals (probably hedgehogs) were seen in the vicinity the following morning.

The reduction in the population on the Eritrean coast was due to predators (except for 2 million hoppers which were killed by human agency) and mainly to birds, which were seen continually attacking the hoppers and, being visible at a distance, were extremely useful as indicators of the whereabouts of bands. By the end of the third instar, only scattered hoppers and a few small groups remained and a striking feature of the landscape at this time was the kites, sitting like sentinels on the tops of bushes containing small groups of hoppers and feeding on the locusts that emerged from cover. Such a concentration of birds as was seen in this area was considered to be rather unusual by the several experienced field observers present. It does represent, in our experience, the maximum scale of loss due to birds which occurs in eastern Africa.

A record was kept of the number of birds seen on the site. The figures given in Table 22 are minimal, since continuous observation could not be carried out on all days. Table 23 gives some data on the numbers of hoppers found in the alimentary tracts of shot birds. As some of the birds were shot while still actively feeding, it seems reasonable to take the maximum recorded for each species for calculations, i.e. 600 for a kite and 3,000 for an Abdim's stork. It is assumed that a Glossy Ibis and a vulture eat at least as many hoppers as a kite and that a European stork eats as many as an Abdim's stork, Johnston (in Johnston & Buxton 1949) discussed the difficulties of assessing the numbers of locusts eaten per day by birds from the contents of the alimentary tracts of shot birds. There is no appropriate information on the rate at which food passes through the alimentary tract of birds. The observations that Hudleston (1958) used as a basis for his paper, also showed that individual birds continued to feed at intervals of less than 1.5 hours throughout the day (Fig. 17). This is similar to two captive rooks (Corvus frugilegus frugilegus L.) which have been observed to feed 7-9 times in a 12-hour period. Further observation on these same rooks suggests that locusts pass completely through the gnat in 4.5 to 7.5 hours (unpublished). It therefore seems reasonable to assume that the larger bird predators consume at least twice as many locusts per day as the maximum found in the whole of the alimentary tract of shot individuals.

Table 22
The number of birds observed feeding on the Desert Locust population recorded in Table 21. There was an estimated loss of 8 million hoppers in 14 days. (The number of birds recorded is a minimum. It is assumed that birds fill the alimentary tract with hoppers twice a day.)

Table 23
Counts of undigested hoppers in the alimentary tract of various predatory birds. (Those marked with an asterisk were shot while still feeding.)

Fig. 17-The feeding activities of birds during the day. The birds, which were under continuous observation, comprised one wheatear (Oenanthe phillipsi Shelley) feeding on first-instar hoppers, two warblers (Eremomela sp.) feeding on fourth-instar hoppers and one parent and one young hornbill (Torkus erythrorhynchus Teem.) feeding on fourth-instar hoppers. Maximum air temperatures 92-93°F.

Hudleston (1958) showed that warblers ate just over 40 fourth-instar hoppers per day. It is assumed that wagtails could be expected to eat at least 80 second-instar hoppers per day. These assumptions were made in calculating the total number of hoppers eaten by birds actually seen and recorded on the site in Eritrea (Table 22). Since the total number of birds on the site would certainly exceed the numbers recorded it seems reasonable to assume that birds accounted for the majority of the losses observed.

As opportunity offered, we killed specimens of bird predators and counted the number of hoppers in the alimentary tract. Very few insects other than locusts, or food materials of other kinds, were found in the tracts of birds that had been feeding on locusts (cf. Johnston & Buxton 1949). Some of the data obtained by our unit as the result of continuous observation on predators have already been published by Hudleston (1958).

When locust populations were very large, the predators were too few to make any appreciable difference to the population. An example of this occurred near Scillave in 1957. Some 2,000 Marabou and European storks were feeding on bands of fourth- and fifth-instar hoppers for a week. Even if each bird ate only 6,000 hoppers a day, this would represent nearly 84 million hoppers in seven days. In spite of this loss, large-scale control measures were necessary to prevent adult swarms being produced from this area, and in this connection it is relevant to remember that 84 million locusts constitute only a small swarm, covering about one sq. mile (Rainey 1954).

go to next section of this publication