Automated Video Analysis of the Specific Effects of the Size and Concentration of Polystyrene Microplastics on the Swimming Behavior of the Crucian Carp <italic>(Carassius carassius)</italic>

Abstract

The size and concentration of polystyrene microplastics (PS-MPs) may affect the behavior of fish, but sufficient evidence is lacking due to laborious, inefficient and fallible observation. In this study, an automated video analysis method based on convolutional neural network was developed to assess the behavior changes of crucian carp (Carassius carassius) exposed to different particle sizes (80 nm and $8~\mu $ m) and concentrations (0.3, 1.2, 4.8 and 19.2 mg/L) of PS-MPs for 20 h and then depurated for 20 h. The results showed that crucian carp were more sensitive to 80 nm PS-MPs, which manifested as increased swimming speed, shorter resting time, higher degree of dispersion, higher turnings, higher tail-beat frequency and amplitude. The stress response of fish was not obvious until exposed to medium and high concentrations of $8~\mu $ m PS-MPs, while there was no significant difference in behaviors at medium and high concentrations of 80 nm PS-MPs. Stress-induced changes to fish behavior were slightly ameliorated during the depuration period, although swimming speed had not decreased. PS-MPs with different sizes and concentrations had specific effects on the swimming behavior of C. Carassius. The results of this study will help to establish methods to assess stress-induced changes to the behavior of fish, and provide new perspective of behavioral changes in response to PS-MPs.

Keywords

• Microplastics
• Carassius carassius
• swimming behavior
• automated analysis