An investigation of novel population models to control aphid outbreaks via a data driven “adaptive” framework

The soybean aphid ( Aphis glycines ) is an invasive pest that causes severe yield loss to soybeans in the North Central United States. Its populations frequently exhibit multiple, unequal-amplitude peaks within a single growing season, a pattern that classical boom-bust models cannot reproduce. We investigate and develop a sequence of population models aimed at capturing this multi-peak behaviour while remaining biologically realistic and useful for pest management. We first show that the variable carrying capacity model (VCM), in which seasonally fluctuating host-plant quality drives a time-dependent carrying capacity, can generate multiple peaks but also admits finite-time blow-up, or population explosion, rendering it unrealistic for high colonisations. We propose an adaptive behavioural model with a density-dependent switch that preserves the multi-peak dynamics while guaranteeing bounded populations. Guided by empirical evidence that abiotic stressors such as flooding alter host-plant fitness, we further devise a novel model with time-dependent, data-fitted periodic plant fitness and aphid growth rate; we prove that it produces transient multi-peak oscillations, with no sustained periodic state. We assess all models against the Economic Threshold (ET) and Economic Injury Level (EIL), and a global Sobol sensitivity analysis identifies the parameters governing outbreak frequency. Across all models, simulated aphid abundances exceed the ET and EIL, indicating that control is required. Lastly, applications of these models to improving current pest management strategies are discussed. Our work provides a stable, data-informed, and biologically grounded framework for predicting aphid outbreaks and guiding management strategies.