DaphTrack: A deep learning-based multidimensional behavior analysis system for neonatal Daphnia magna

Pollutant monitoring is essential for evaluating ecosystem health. As a sensitive aquatic indicator species, Daphnia magna (D. magna) is widely employed in environmental toxicology. However, existing methodologies exhibit critical limitations in detecting and tracking neonatal D. magna, impeding accurate quantification of their behavioral responses to contaminants. In this study, we present DaphTrack, a deep learning-enabled system for automated detection, counting, and multidimensional behavioral analysis of neonatal D. magna. The system incorporates the YOLO11n object detection framework and an optimized ByteTrack++ tracking algorithm, achieving high identification accuracy (96.5 %) and robust multi-object tracking performance (IDF1 = 88.8 %). A behavioral parameter matrix was established, comprising mean speed, mean scalar acceleration, maximum speed, maximum acceleration, trajectory density, movement distance, activity range, and inter-individual distance, to comprehensively quantify locomotor characteristics. The acute toxic effects of different concentrations of the marine antifouling biocide 4,5-Dichloro-2-n-octyl-4-isothiazolin-3-one (DCOIT) on the behavior of neonatal D. magna were assessed to validate the accuracy of the system. In summary, DaphTrack significantly enhances the accuracy and efficiency of neonatal D. magna identification and behavioral analysis, offering a novel tool and perspective for elucidating the toxicological mechanisms of contaminants such as DCOIT in aquatic organisms.