Expanding the dynamic shift detector algorithm to monitor insect populations through time

Insect populations are naturally variable and can experience significant oscillations over time. Variations in intrinsic growth rate (r) and carrying capacity (k) cause fluctuations in population size. Identifying shifts in these parameters is imperative to understanding natural fluctuations in insect populations and has implications for identifying drivers of insect declines, identifying species in need of conservation action, and monitoring and understanding the dynamics of invasive insect species. To that end, the Dynamic Shift Detector (DSD) algorithm was built to interrogate time-series data to identify shifts in parameter values. This tool has proven effective in monitoring invasive ladybeetle populations, and identifying trends in Monarch butterfly populations. However, the DSD is limited by testing a single population model, the Ricker Model, which may not fit all real-world data. Here, we expanded the DSD to include the Beverton-Holt Model, Logistic Model, and the Ricker Model. In the expanded DSD, the user indicates which population model will detect time points when there are shifts in population trajectory, then with those time points the algorithm fits all the population models and provides an AIC value for each population model to indicate which is best supported. Here, we present simulations that test the performance of the expanded DSD under scenarios of different levels of environmental noise and scenarios with different start values and change rates for r and k. Further, to demonstrate the expanded DSD’s utility in real-world settings, we present a case study using a 20-year butterfly dataset from Cuyahoga Valley National Park.