HTBPPS: A high-throughput behavioral phenotyping platform for shrimp

In shrimp breeding, phenotypic measurement is crucial for identifying key traits. Meanwhile, behavioral traits typically exhibiting moderate to high heritability are gaining recognition as novel traits in aquatic species breeding. However, traditional shrimp behavioral phenotype measurement is unsuitable for large-scale evaluations as it relies primarily on time-consuming and error-prone manual observations. With advances in computer vision, various behavior recognition and tracking algorithms have effectively overcome these limitations. Yet, such algorithms often prove inadequate for long-duration behavioral video data. Inspired by high-throughput phenotyping platforms in plant research, this study designed specialized temporary housing and observation apparatus to track shrimp behavior and the shrimp.tracker software based on a boundary-constrained Kalman filter algorithm. Additionally, this study established an analysis system for shrimp behavioral phenotypic data and a framework for long-term high-throughput behavioral data acquisition. Ultimately, these hardware and software systems form a high-throughput behavioral phenotyping platform for shrimp. The platform's algorithm scored 92.4 in sMOTSA, 99.5 in MOTA, and 95.5 in MOTSP, with a detection accuracy of 99.79 %, surpassing the deep learning-based Track-RCNN algorithm. Case studies illustrated the platform's capability to track various shrimp species, multiple individuals over extended periods, and shrimp under high- temperature stress. The experimental results revealed significant behavioral phenotype differences among different types and sizes of shrimp. During long-term behavioral observations, shrimp behaviors exhibited 24-h cyclical rhythm changes. Different types of shrimp exhibited the same M-shaped trend in mobility during linear temperature increases, with the last peak of the M shape moderately correlating with shrimp heat tolerance. These findings emphasize the importance of behavioral phenotypes in shrimp breeding, serving as potential indicators for assessing heat tolerance in shrimp breeding. Thus, the proposed platform has significant potential for future applications in shrimp breeding. •Developed a high-throughput behavioral phenotyping platform for shrimp.•Achieved 99.79 % detection accuracy, outperforming Track-RCNN.•Enabled analysis of long-duration behavioral data for various shrimp species.•Identified behavioral phenotypes as indicators of heat tolerance in shrimp.