PesTwin: a biology-informed Digital Twin for enabling precision farming

Abstract

In a context of growing agricultural demand and new challenges related to food security and accessibility, boosting agricultural productivity is more important than ever. Reducing the damage caused by invasive insect species is a crucial lever to achieve this objective. In support of these challenges, and in line with the principles of precision agriculture and Integrated Pest Management (IPM), an innovative simulation framework is presented, aiming to become the digital twin of a pest invasion. Through a flexible rule-based approach of the Agent-Based Modeling (ABM) paradigm, the framework supports the fine-tuning of the main ecological interactions of the pest with its crop host and the environment. Forecasting insect infestation in realistic scenarios, considering both spatial and temporal dimensions, is made possible by integrating heterogeneous data sources: pest biodata collected in the laboratory, environmental data from weather stations, and GIS data of a real crop field. In this study, an application to the global pest of soft fruit, the invasive fruit fly Drosophila suzukii, also known as Spotted Wing Drosophila (SWD), is presented.

Figures (5)

• Figure 1: Schema of the ABM implemented in PesTwin. In this example, three main agents are modeled: the Male and Female SWD pest, and the Berry host fruit. Agents’ lifecycle is decomposed into states: for SWD agents, the Juvenile and Adult states; for the Berry agent, the Mature and Infested states. Agents’ behavior, such as agent-to-agent interactions as well as interactions with the environment, is modeled via processes, represented as colored circles. Each process is a reusable and configurable software element, so that complex ecological relations can easily be modeled by assembling processes from a predefined library.
• Figure 2: A) Aerial view showing the collection areas in the municipality of Ardenno, in the province of Sondrio; B) Close-up of the trap locations: the forest field (orange marker, FF) and the blueberry field trap (blue marker, BF) (Maps Data: Google Earth ©2023, V 10.55.01, Google LLC); C-E) Images of the biological trap used for SWD collections.
• Figure 3: Dynamics of the growth of the SWD population in the cage experiment. Three cage replicas were set up with a starting population of three mated SWD females. (a) The number of hatched eggs was recorded at each generation, for the three subsequent generations (G0 to G3). 50 stochastic simulations were run (grey lines) using parameters collected via biology assays. (b) The pest adult population mortality rate was recorded daily, producing the time series of the pest population with overlapping generations.
• Figure 4: Trapping data of the adult SWD population collected in the Fondazione Fojanini field in Ardenno during 2024. Ten stochastic simulations were run, and their mean is shown (grey line) for comparison against trap data (colored dots). (a) Female adult SWD population. (b) Male adult SWD population. (c, d, e) Details of the spatial dispersal of the SWD total population in one simulation at three different times of the pest invasion: (c) early August, (d) mid-August, and (e) early September.
• Figure 5: Simulation of the dynamics of pest population as the target of an SIT control strategy. 10 stochastic simulations were run, and their mean is shown for comparison. (a) Female adult SWD population dynamics in a no-control scenario (red) and when a Sterile Insect Technique is released weekly (green) during the blueberries’ ripening season. (b) Population of sterile SWD males released during the SIT control strategy. In the inset, the location of the two simulated release sites inside the crop field is shown.