Allometry of stridulatory signals in the Japanese rhinoceros beetle

Vibroacoustic signals, one of the major types of insect communication, occur in diverse behavioural contexts, including courtship and combat. Stridulation is one of the mechanisms that generates vibroacoustic signals. Because sexual selection can influence the relationship between overall body size and changes in the size or shape of body parts, understanding functional consequences of allometry in acoustic communication can provide broad insights on evolution. However, this phenomenon has not been thoroughly investigated across insect groups. In this study, we revealed the allometric patterns in acoustic features of stridulation in the Japanese rhinoceros beetle, Trypoxylus dichotomus. Over 2000 stridulatory syllables were recorded and analysed to examine allometry. We trained a two-dimensional convolutional neural network (2D-CNN) to classify signals into three types and employed unsupervised clustering methods to detect underlying subtypes. Multivariate analysis of variance and standardised major axis regression revealed that one signal type exhibits a significant allometric relationship with elytral area. These findings indicate that the stridulatory signals of T. dichotomus can be reliably classified using 2D-CNN, and one signal type correlates with body size. This study establishes a foundation for future research on the functions of acoustic allometry in insect communication.