Prediction Model of the Temporal Dynamics of Severe Pest Cashew Anacampsis phytomiella Using Artificial Neural Networks

Abstract

The cashew tree (Anacardium occidentale) is a tropical plant that yields the world's most consumed nut. Anacampsis phytomiella (Lepidoptera: Gelechiidae) is the primary pest that infests cashew nuts, leading to losses of up to 80%. The comprehension of the spatiotemporal dynamics of pests enables enhanced planning of sampling and control methods. Artificial neural networks (ANNs) are computational models in machine learning with high predictive capability. Thus, the objective of this study was to determine a seasonal dynamic model of A. phytomiella using ANNs. Over 3 years, the pest attack intensity and climatic elements were monitored in two cashew orchards. A total of 1716 ANN models were determined. The model predictors included the time of plant fruiting, average air temperature, dew point, atmospheric pressure and rainfall. The temperature, atmospheric pressure and time of plant fruiting had a positive effect on pest infestation, while the opposite occurred with rainfall. The pest infestation curve in relation to the dew point exhibited a point of maximum. The model successfully predicted the intensity of A. phytomiella infestation across different years, plant fruiting stages and pest densities. Therefore, the ANN model determined in this study is promising for predicting the intensity of A. phytomiella infestation in cashew orchards.