Acoustic Counting of Adult Insects with Differing Rates and Intensities of Sound Production in Stored Wheat

Automated acoustic detection systems count the insects in a grain sample by analyzing the spatial and temporal distribution of sounds. The acoustic location fixing insect detector is an automated system developed originally to quantify hidden larval infestations in 1-kg samples of wheat. The detector analyzes input from an array of sensors embedded in the sample container walls. It identifies (scores) a specific pattern of input as an insect if the spatial and temporal distribution matches the criteria based on a calibration with 4 th-instar rice weevil, Sitophilus oryzae (L). However, expanded testing has revealed considerable differences in the spatial and temporal distributions of sounds made by insects of different species and sizes. These differences were examined in a series of tests with insects that range an order of magnitude above and below the 1.5-mg weight of the S. oryzae larvae. A particular focus was the detection order of the first 2 sensors registering each sound. Multiple sounds from an insect ten to cluster together into a small number of contiguous Ist:2nd sensor detection pairs, but the pattern for background noises is random. It was determined that the cluster size, the number of contiguous Ist:2nd detection pairs, was proportional to insect weight. The rate of sound detection was inversely proportional to weight. Thus, to reliably count insects with widely varying sound production patterns, the sound pattern identification algorithm needs to self correct, depending on the input received from the grain sample. Adults or larvae generating large numbers of loud sounds, typically weighing> 1 mg, can be scored in a few seconds, but those generating small numbers of weak sounds, typically 10 min. The possibility of using differences in cluster size to distinguish among species is also discussed.