Leveraging computational intelligence to identify and map suitable sites for scaling up augmentative biological control of cereal crop pests

Abstract

Fall armyworm (FAW) Spodoptera frugiperda (J. E. Smith) is a major pest affecting cereal production in Africa. Biological control (BC) technologies are being promoted as a sustainable alternative to chemical control, which can lead to health risks and environmental hazards. However, effective deployment of these technologies requires site-specific recommendations. In this study, we use a step-by-step modelling approach to map suitable sites for BC technologies, focusing on the parasitoid Cotesia icipe using the FAW level of infestation, the ecological niche of the parasitoid, and the FAW host crop. The level of pest infestation was estimated using an evolutionary adaptive Neuro-Fuzzy inference system (R2 > 0.89) while the pest ecological niche was obtained using the maximum entropy algorithm (area under curve, AUC > 0.9). A fuzzy operator was used to combine all fuzzified variables into a single layer that represents the landscape's overall suitability for C. icipe in maize farms. Our computational findings indicate that C. icipe holds substantial promise as a BC agent in maize farms, with suitability levels consistently surpassing 90% throughout maize cropping seasons. The findings demonstrate that the utilization of artificial intelligence, combined with data science and knowledge representation, serves as an effective advisory tool for guiding the deployment of BC agents, such as parasitoids, for the sustainable management of FAW. This approach enables informed decision-making and enhances the efficacy of FAW management strategies by providing valuable insights and recommendations based on data-driven and computer intelligence analyses.

Keywords

• Neuro-Fuzzy
• Maximum Entropy
• Expert System
• Pest Management
• Parasitoid