Prediction of breeding regions for the desert locust Schistocerca gregaria in East Africa

The desert locust Schistocerca gregaria, one of about a dozen species of locusts, is a species of swarming short-horned grasshoppers that can migrate great distances during its gregarious phase^1,2,3. As they swarm, they voraciously feed on key staple crops such as maize and sorghum, pastures, and any green vegetation that comes their way, thereby significantly affecting the livelihoods of smallholder farmers and pastoralists^4,5,6. In Africa, the countries of the Sahel region, especially Algeria, Burkina Faso, Chad, Ethiopia, Eritrea, Mauritania, Mali, Niger, Nigeria, Senegal, Somalia, and Sudan, are particularly susceptible to desert locust outbreaks. Until the 1960s, locust outbreaks frequently occurred, however, post-1960s, outbreaks were less frequent and occurred, on average, only once in a decade⁷.

In general, the desert locust breeds extensively in semi-arid zones extending from West Africa through the Middle East to Southwest Asia, threatening livelihoods of the population in over 65 countries. Between 2019–2020, unprecedented locust breeding was observed in Eritrea, Somalia, and Yemen due to unusually heavy rainfall in the horn of Africa between October to mid-November 2019, more than 400% above average⁸. Following this breeding, countries in the horn of Africa, such as Ethiopia, Kenya, and Somalia, are experiencing extraordinary swarms never witnessed during the past 25 years. The current swarm is estimated to consume ~ 1.8 million MT of vegetation per day across 123,200 km², which represents 11% of Ethiopia’s total land area⁹. In Kenya, the locust has spread to approximately 107,000 km² (20% of Kenya) (Kenyan Multi-Agency Team on Desert locust, 2020), and very recently, the locust has invaded Uganda, South Sudan, and Tanzania. It is anticipated to move northward into Sudan and possibly northern part of Chad. The current management strategy of the locust swarm is aerial spraying with chemical pesticides, which has a high cost on humans, livestock, and the environment in addition to its economic burden at the national level biodiversity.

Studies have shown that desert locust has the ability to change its behaviour, ecology, and physiology in response to the changes in climatic conditions¹⁰. In particular, within a few weeks, swarming adults mature, mate, and begin to oviposit in soils at 10–15 cm below ground in suitable environments in the invaded zones^2,11. Suitability for oviposition and subsequent breeding is influenced by factors such as soil type, sand content, soil moisture, surface air temperature, rainfall, and prevalence of vegetation^2,12. The emerging hoppers (nymphs), which are the wingless juvenile stage, can concentrate to become more gregarious and form bands that crawl on the surface over long distances. After several moultings, up to six times, hoppers transition to adults which can come in contact to form a gregarious phase². The time needed for the desert locust to transition from one stage to the other is highly dependent on the weather patterns^13,14. Both the hopper bands and adult swarms can cause significant damage to the vegetation and crops in the invaded zones. Therefore, to prevent catastrophic swarms from maturing hoppers, it is critical to strengthen ground and aerial surveillance efforts to identify potential breeding sites for timely and effective management of hopper bands. However, effective ground and aerial surveillance are constrained by various factors including extensive area of invasion (e.g., 107,000 km² in Kenya), inaccessibility of invasion zones due to poor infrastructure, limited resources, lack of human capacity for monitoring and control, and difficulties in predicting suitable areas for breeding and outbreaks. Such constraints are typical to the currently invaded zones in Kenya, Uganda, and South Sudan, and to other nearby countries at risk.

Previous desert locust outbreaks in the Horn of Africa were observed in 1996–1998, and it affected countries along the Red Sea, with infestations primarily concentrated in Saudi Arabia and, to a lesser extent, in Egypt, Ethiopia, Eritrea, Northern Somalia, Sudan, and Yemen. Countries such as Kenya and Uganda have not experienced the current level of desert locust invasion for more than 70 years, and little or no information is available on the suitability of specific sites for desert locust oviposition and breeding¹³. Such information is urgently needed to strengthen surveillance (ground and aerial) efforts, regional coordination, and preparedness, inform efforts and improve the delivery of preventive measures before the newly emerging hoppers cause damage.

Locust (desert locust and grasshopper) outbreak prediction and monitoring can be modelled using ecological niches (EN) approaches^15,16. A category of EN models apply machine learning algorithms that correlate a set of environmental conditions (e.g., bio-climatic variables) to species presence and absence records to predict its suitable habitats¹⁷. For instance, maximum entropy (MaxEnt), genetic algorithm for rule-set production (GARP), and ecological niche factor analysis (ENFA) are EN tools that predict species suitability using presence-only data^18,19. MaxEnt was revealed to provide a reasonably better result compared to other presence-only models¹⁸. In specific, MaxEnt assumes that the suitable areas for occupancy by species would corroborate to the physics principle of maximum entropy without any environmental restrictions. The model predicts habitat suitability by fitting a probability distribution for the incidence of the species across the whole area. However, MaxEnt often experiences overfitting at low threshold levels than, e.g., GARP models¹⁹.

The objective of this paper is to develop a decision support tool that enables governments and their development partners to control the locust invasion from its breeding sites effectively. The specific objectives are to (1) model the relationship between known desert locust breeding sites around the world with critical bio-climatic (temperature and rainfall) and edaphic (sand and moisture contents) variables using MaxEnt EN model, and (2) validate the model with the existing database, and further develop predictions on potential areas for desert locust oviposition and breeding in Kenya, Uganda, South Sudan, and Sudan.