Estimation of population density of stored grain pests via bioacoustic detection

Abstract

The potential of bioacoustics in estimating the population density of insect pests inside the stored grain mass was evaluated in the laboratory. We used a piezoelectric sensor and a portable acoustic emission amplifier connected to a computer for recording acoustic emissions of insects. The software analyses the vibration recordings of the piezoelectric sensor, performs signal parameterization and eventually classification of the infestation severity inside the grain mass in four classes, namely: Class A (densities ≤1 adult/kgr), Class B (densities 1–2 adults/kgr), Class C (densities 2–10 adults/kgr) and Class D (densities >10 adults/kgr). Adults of the most important beetle pests of stored cereals and pulses, in various population densities (1, 2, 10, 20, 50, 100, 200 & 500 beetle adults/kgr grain) were used during the present study. The linear model was very effective in describing the relationship between population density and number of sounds. Multiple classifiers were used to evaluate the accuracy of bioacoustics on predicting the pest density given per minute counts of vibration pulses. Based on our results, our system's performance was very satisfactory in most cases (∼68%) given that probabilities for successful prediction typically exceeding 70%. Our study suggests that automatic monitoring of infestations in bulk grain is feasible in small containers. This kind of service can assist with reliable decision making if it can be transferred to larger storage establishments (e.g. silos). Our results are discussed on the basis of enhancing the use of acoustic sensors as a decision support system in stored product IPM.