Artificial intelligence can map the best spots to target biological control efforts against the fall armyworm (Spodoptera frugiperda) in Africa.
The moth pest invaded Africa in 2016 and poses a threat a serious threat to food security across the continent, due to it feeding on cereal crops such as maize.
Pesticides are mainly used against the species, but cost and reduced effectiveness due to resistance are major concerns for farmers. A recent review had suggested that there currently is a shortfall in research dedicated to biological control of S. frugiperda —but work led by scientists from the International Centre of Insect Physiology and Ecology has used the latest approaches in data science to better understand pest ‘hot spots’ ripe for sustainable intervention.
Mixed modelling cuts through complexity
The team employed artificial intelligence using a step-by-step modelling approach to firstly understand pest infestation levels in maize production, and then accurately map where would be suitable to release the native parasitoid wasp Cotesia icipe, which was also discovered in Africa in 2016 and attacks the damaging larvae of the moth.
Using AI algorithms helps to grasp how the natural enemies fit into the diverse agroecological zones and farming cultures of Africa, the researchers explained. In this case, they used a hybrid approach which included feeding data into neural networks, where computers are trained to process similarly to the human brain, and a genetic algorithm.
Wide potential applications
Their framework was able to accurately predict maize plants infested with fall armyworm on farms. It also pinpointed areas of high and low suitability for release of the C. icipe across the regions of Kenya, with central Kenya found to be particularly favourable for the natural enemy. Following the initial mapping in Kenya, they extrapolated the results to the whole of Africa.
“These findings highlight the potential for utilizing C. icipe as an effective biological control approach in Sub-Saharan Africa, particularly in areas where maize is extensively cultivated. The results offer valuable insights for decision-makers and stakeholders involved in pest management and agricultural practices in the region,” the scientists wrote in the journal Biological Control.
The work suggests that similar approaches could be used for analysis of biological control release potential across different geographies and pest species, they added.
