The Fluctuating Distributions of the Desert Locust in Relation to the Strategy of Control
By Miss Z. V. WALOFF
Anti-Locust Research Centre, London
Some thirty years ago, at the time when a plague of Desert Locusts, Schistocerca gregaria (Forsk.), was assuming serious proportions and threatening the agriculture of many countries, Dr. B. P. Uvarov first began the systematic assembly and analysis of distribution and migration records of this species. Over the years such records have provided valuable material for biogeographical studies by a number of investigators, and reference is made in this note both to the earlier published data, and to some recent analyses carried out at the Anti-Locust Research Centre by the author, Miss E. Betts and Miss J. Laszlo. Before dealing with this species, however, a brief reference must be made to two other plague locusts, whose biogeography has been studied at the Centre-i.e., the Tropical Migratory Locust, Locusta migratoria migratorioides (R. & F.) and the Red Locust, Nomadacris septemfasciata. (Serv.)
The plague of Locusta, which broke out in 1928, and in the course of a few years spread over some 7 million square miles of the African continent (Uvarov, 1933; 1933a; 1934) has been shown by cartographical and field investigations by O. B. Lean (1931, 1936) to have originated in a restricted outbreak area, less than 30,000 sq. miles in extent, situated in the flood plains of the river Niger in the Soudan Republic. Similarly the main outbreak areas of Nomadacris, which infested some 4 million square miles of the southern half of the African continent during its last plague beginning in 1930 (Uvarov, I.c.), have been shown by Allan (1931), Harris (1933), and Bredo (1938), to be restricted to two localities in Northern Rhodesia and Tanganyika, i.e., Mweru Wantipa and Rukwa Valley, which between them cover less than 40,000 sq. miles. Subsequently, subsidiary outbreak areas of Nomadacris have been found in the Malagarasi plains in Tanganyika (Gunn, 1952), and, more recently, within the Niger outbreak area of Locusta, where populations of Nomadacris built up to a dangerous level in 1947 (Roblot, 1951) and again in 1959. The last major plagues of these two species came to an end in the forties, and all their subsequent outbreaks have been contained within the outbreak areas by the International African Migratory Locust Organisation and the International Red Locust Control Service, with headquarters at, respectively, Kara and Abercorn.
The plague dynamics of the Desert Locust, whose huge invasion area covers some eleven million square miles in Africa and south-western Asia, have been found to differ in some important respects from those of the other two species. Fig. 1 shows the year-by-year fluctuations in the number of territories infested by swarms of Locusta, Nomadacris, and Schistocerca during the period 1887 to 1958.* In Locusta and Nomadacris there had been only two major, and in both cases partly concurrent, plagues during this period, lasting some fifteen and eighteen years, and separated by plague-free intervals of about twenty years in the case of Nomadacris, and twenty-four years in the case of Locusta. A very different picture is provided by the graph of the Desert Locust infestations, which shows much shorter-term fluctuations, with peaks, in the last fifty years, in 1915, 1930, 1944 and 1953-55. With the gradual improvement in reporting, moreover, the troughs between the peaks have become progressively narrower and more shallow, indicating that swarms may occur in a considerable number of territories even during the recessions of the major plague. No fixed outbreak areas have been found for the Desert Locust, and it appears probable that one of the contributing factors to the rapid recrudescences of its plagues may be the building up of the gregarious populations which persist during the troughs. The possible importance in plague dynamics of such residual populations was first suggested by a French entomologist, Pasquier (1942), and our detailed examination of historical material confirms his hypothesis. Another mechanism of plague recrudescence, suggested by Stephenson and Rainey (in Rainey, 1951), may be the concentration by convergent air-flow of mobile scattered populations of ph. solitaria into areas which may become temporarily favourable for successful breeding and gregarisation.
* The graphs for Schistocerca and for Locusta are from, respectively, Waloff and Rainey (in preparation) and Betts (in preparation); the graph for Nomadacris is based on manuscript and published records on the files and in the library of the Anti-Locust Research Centre.
Fig. 1. Year-by-year fluctuations in areas infested by swarms.
From the systematic cartographical analysis of the Desert Locust records it soon became apparent that its infestations are markedly seasonal in nature over most of the invasion area. This characteristic is illustrated in Maps 1, 2 and 3, Fig. 2, showing the widely separated seasonal breeding belts in which Desert Locusts breed in the spring and in the summer seasons, and the very restricted area over which breeding occurs in winter. The close relationship between seasonal breeding and the rainfall regimes which has been discussed in a series of Anti-Locust Memoirs (Waloff, 1946; Donelly, 1947; Davies, 1952, and Fortescue-Foulkes, 1953) is further illustrated in Map 4, Fig. 2, which shows the rainfall regimes at some representative stations in the central and western parts of the invasion area, and the number of years, during the twenty-year period from 1939 to 1958 in which layings have been recorded in any given month within areas covering two one-degree squares containing and adjoining the station. It will be seen that on the Somali Peninsula, for example, the two laying seasons correspond closely to the two seasons of rain (cf. conditions at Belet Uen), while in the summer breeding belt running across the African continent to south of the Sahara there is only a single laying season per year, coinciding with a single period of rains (cf. histograms for Boghé, Goundam and Khartoum). In central Arabia the single laying season occurs in spring, in accordance with the rainfall regime in that area (cf. conditions at Riyadh), while the incidence of breeding in winter, spring and summer, on the Red Sea coastal plain in the vicinity of Massawa is related to the incidence of rains in all these seasons in the southern Red Sea area. The close relationship between the rainfall and laying seasons is less apparent only near the northern margins of the distribution area-e.g., in north-western Africa, where, in spite of the arrival of the swarms from October onwards, and the beginning of the Mediterranean winter-spring rains in the last quarter of the year, the maturation of locusts and the onset of laying are delayed for some months, probably by relatively low temperatures (cf. histograms for Marrakesh).
Fig. 2.-Maps 1, 2, 3. Frequencies of Desert Locust hopper infestations. Map 4. Laying in relation to rainfall.
The seasonal occurrence is characteristic not only of Desert Locust breeding but of its infestations in general, for over most of the seasonal breeding belts the locusts lay soon after their arrival, and the newly formed swarms fly out of the areas in which they are produced soon after the locusts reach the winged stage. The transit areas between the widely separated breeding belts are traversed by long-range migrations, with swarms often covering distances of 1,000-2,000 miles, or longer, between their source areas and the areas in which they breed. As frequently recurring examples, may be quoted the movements of summer generation swarms produced south of the Sahara to the spring-breeding areas in north-western Africa, or in the Middle East, and the return migrations of spring generation swarms to the summer areas; other examples are provided by movements of winter swarms originating on the Somali Peninsula to the Middle East and Pakistan, or the migrations of summer swarms from India to Arabia and the Somali Peninsula. All parts of the invasion area may be linked (either directly, or with the intervention of a breeding season) by long-range movements, which bind its component parts into an immense, but essentially single, unit (Waloff, 1959).
Many of the seasonal movements have been found to agree with prevailing winds, and to change with them (Rao, 1942; Waloff, 1946), suggesting that the direction of swarm displacement is determined by wind; more recently Rainey and Sayer (1953) have demonstrated by aircraft observations on hour-to-hour trajectories of individual swarms that they moved directly down the corresponding mean wind between the ground and the height of the swarm. As could be expected from the downwind displacement of swarms, their distributions and movements were found to be closely related to current synoptic situations and wind-fields. In particular, as pointed out by Kraus (1958), the direction in which the newly-formed swarms move out of their source areas may often depend on the position, at the time, of seasonal low-level divergence areas, in which the outflow of air exceeds the inflow, and which form kinds of aerial " watersheds " from which air may flow in two opposite directions; for instance, the position of the winter divergence area which becomes established on the Somali Peninsula after the cessation of winter rains determines whether most of the winter swarms produced there will move southwards to Kenya and Tanganyika, or northwards towards Arabia. The ultimate destinations of the swarms, on the other hand, have been shown by Rainey (1951) to be areas of low-level convergence, in which surface winds show a net excess of inflowing air over the outflow, and which tend to be areas of precipitation, which is essential for the successful breeding of the Desert Locust in its mostly arid distribution area. Convergence areas of great importance in the annual cycle of the Desert Locust occur within the Inter-Tropical Convergence Zone when it is near its northern limit in the northern summer, and becomes a reception and breeding area for swarms produced in the preceding spring both to the north and to the south of this belt (Rainey, l.c.).
Since the seasonal synoptic situations in different parts of the invasion area are not identical from year to year, corresponding variations are shown by Desert Locust movements and distributions. The changes in the extent of infestation arising in the course of a plague are illustrated in a graph of month-by-month fluctuations in the number of one-degree squares infested by swarms and hoppers during the ten years from 1949 to 1958 (fig. 3A) and by the maps in Fig. 3B, which provide examples of widespread and restricted infestations occurring at the same seasons in different years. It will be seen that in addition to the long-term fluctuations illustrated in Fig. 1, the distribution of this highly mobile species is subject to continuous short-term changes, with the infested areas expanding and contracting with the seasons, and the seasonal distributions varying from year to year. Of special interest, from the practical point of view, are the occasions when the curves of the swarm and hopper infestations both descend to a low level, and the actual distributions show a compact pattern, indicating that the bulk of the swarming populations are temporarily concentrated in a relatively small area. Such a situation arose in the winter of 1951-1952 when the infested area was largely restricted to the Somali Peninsula-in contrast to the more typical widespread infestation at this season exemplified by December 1954; another example of a relatively restricted distribution is provided by the summer of 1956, in marked contrast to very widespread infestation in summer 1954.
Fig. 3A. Schistocerca gregaria Forsk. Month-by-month fluctuations in infested areas.
Fig. 3B. Schistocerca gregaria Forsk. Contrasts in distribution.
Such marked variations in distributions raised the problem of relative liability to infestation of different countries invaded by the Desert Locust, and this aspect was investigated by the preparation of maps showing the frequency of infestations over a given period of time in different parts of the invasion area. This method appears to have been introduced by Cook (1929) in his studies of the infestations of the Army Worm, Cirphis unipuncta Haw., in the United States; in locust studies it was first used by Morant (1947) to illustrate the frequency of breeding in different parts of the Red Locust invasion area. For the Desert Locust the compilation of detailed maps of swarm and hopper infestation frequencies was begun in the fifties, after some fifteen years in which the reporting has been sufficiently reliable and uniform over the greater part of the invasion area; following a suggestion for which we are indebted to Dr. R. C. Rainey, these maps have been constructed with a one-degree square as the unit of area. Examples of monthly frequency maps of hopper infestations are provided by Fig. 2; other maps have been constructed for the swarm and hopper frequencies during each quarter of the year, and have suggested that the concentrations of large proportions of swarming populations into relatively restricted areas may occur particularly frequently during the third quarter of the year when swarms from different parts of the distribution area have moved into the summer breeding belt (Waloff, 1959).
It will have become apparent from the preceding remarks that the control of the Desert Locust presents very different problems from the control of Locusta and Nomadacris, and it may be useful, before concluding, to recapitulate briefly those aspects of its biogeography which appear particularly relevant to the strategy of control. Because of the great mobility of swarms and the wide geographical separation of the seasonal breeding belts, many parts of the Desert Locust distribution area are invaded by swarms from very distant sources, and in the majority of countries subject to invasions the first manifestation of a plague is the appearance of swarms originating some hundreds-or thousands-of miles beyond their boundaries. At the same time, owing to seasonal changes in the location of infestation most parts of the invasion area become seasonally free for periods of months even at the height of a plague, while the variations in the extent of seasonal infestations may, on occasion, result in the clearance of even high-frequency areas for periods of years. It follows that Desert Locust control organisations confining their resources to their own territories are likely to be occupied for only part of their time, and to remain idle while swarms which will invade them in the near future are being produced in another area, where control resources may be extended to the full and inadequate to cope with the infestation.
The rational control strategy thus appears to require the readiness by most territories affected by the Desert Locust to direct a large part of their control effort to areas outside their boundaries, and the creation of control units which could emulate the Desert Locust in their mobility. Only such an approach could make full use of opportunities provided by the high-frequency areas, and by those occasions when large proportions of the total swarming populations become temporarily concentrated into restricted areas, where their effective control could greatly modify, and even arrest, the further development of a plague.