A New Method of Aquatic Animal Personality Analysis Based on Machine Learning (PAML): Taking Swimming Crab Portunus trituberculatus as an Example

• Main Authors: Chao Yang, Xianpeng Su, Dapeng Liu, Zhongwen Guo, Fang Wang, Yunliang Lu
• Format: Article
• Language: English
• Published: Frontiers Media S.A. 2020-02-01
• Series: Frontiers in Marine Science
• Subjects: Portunus trituberculatus; machine learning; clustering analysis; animal personality; dynamic change
• Online Access: https://www.frontiersin.org/article/10.3389/fmars.2020.00032/full

Personality differences are an important part of animal behavior research and a potential factor affecting the physiochemical response of animals to stimuli. Therefore, accurate classification of personality has an impact on animal experiment research that cannot be ignored. However, no completely objective methods have been noticed hitherto to be employed in aquatic animal personality studies. To address this issue, we established a personality analysis method based on machine learning (PAML) to avoid the influence of subjective factors. Then we analyzed the personality changes of swimming crab Portunus trituberculatus exposed to external stimuli i.e., food deprivation. Results showed that the high-accuracy PAML method could provide a comprehensive understanding of crab personality classification. Based on PAML, the personality of P. trituberculatus was detected to change dynamically under stress. After analyzing three typical personality characteristics: boldness, activity, and hesitancy, we found that the stability of boldness was the strongest, which is the best single candidate for personality evaluation, while the activity changed greatly under stress. This study confirms that animal personality changes dynamically with external factors. Based on PAML, we can not only avoid subjective factors and improve classification accuracy, but also quantitatively trace the dynamic influence of external factors on aquatic animal personality.