How can population models contribute to contemporary pest management practices?

Abstract

AbstractPopulation models provide a logical knowledge base before conducting laborious and expensive field experiments. Historically, two types of population models have been developed: highly realistic simulations and simple analytical models. Highly realistic simulations comprise a complicated systems model, whereas simple analytical models comprise various analytical models that focus only on the fundamental structure of the target pest population. Although both approaches have contributed to pest management science, each has limitations, poor predictability, and lacks substantial connections to reality. Assimilation by state-space modeling, in which observation and process models are jointly incorporated, is a good compromise between a simple model and reality in nature. In the big data era, artificial intelligence (AI), specifically aimed at high predictability, has recently become popular. If vital physical and biological records are automatically censored in the field with high precision, AI will produce the most plausible predictions, providing the best practical solution given our current knowledge. AI can be a powerful tool in the contemporary world; however, deductive modeling approaches are still important when considering the behavior of AIs and may also provide important insights to detect deficient information in the data.