Motion-detection computational approach to time-to-knockdown bioassay scoring

The study of thermal tolerance physiology requires large-scale screenings of animals subjected to a variety of thermal performance assays. Recovery or termination of neuromotor activity is used as a proxy to measure thermal tolerance in small ectotherms, and experiments have been primarily scored visually since their original description. Specialized video tracking software has been used to score video-recorded assays to a limited extent, but in some cases, there are confounding results influenced by the detection thresholds used in particle detection algorithms. Here, we apply a different algorithm for detecting recovery or cessation of biological activity based on movement detection. We compare computational approximations to visually scored estimates of critical thermal limits in Drosophila melanogaster using three popular thermal performance bioassays: static heat knockdown time, chill coma recovery and critical thermal minima with ramping temperature. We compare computational definitions of motor activity onset- or endpoints and describe their relationship to male and female fertility decay. The statistical properties of the computationally derived estimates equal or exceed those made by direct visual inspection, but underestimate the magnitude. We discuss the biological interpretation and applicability of computational scoring of thermal limit estimates in high throughput methodologies.