Résumé
Within species, individuals vary in their preference to interact with conspecifics, shaping ecological and evolutionary processes. Measures of individual sociability are essential in behavioural and physiological ecology, yet the impact of experimental design, particularly arena size, on sociability estimates has been largely overlooked. Arena size is known to influence other fish behaviours, but its effect on accurately measuring sociability remains understudied. Here, measuring individuals twice at each of four different arena sizes in a binary choice assay, we show that the size of the experimental arena strongly affects absolute and relative measures of sociability in wild-caught threespine sticklebacks Gasterosteus aculeatus. Individual fish were less social with increasing arena size (arena size range: 10-40 cm length x 30 cm width), and sociability also decreased as the ratio of arena size to fish body length decreased. In addition, smaller individuals were more social than larger conspecifics at all arena sizes. Importantly, while absolute sociability estimates varied with arena size, the effect of body size remained relatively consistent. However, individual repeatability of sociability varied across arenas sizes, with the highest observed in the 30 cm arena, suggesting this may represent a practical minimum size for detecting consistent behavioural differences in G. aculeatus or similarly sized species. Together, these results provide the first empirical evidence that experimental arena size can bias both absolute estimates of individual sociability and their repeatability in fishes. This effect may have implications for the comparability and transferability of behavioural data, possibly limiting comparisons among studies and species. However, the results here represent a step toward understanding how methodological details can bias measurements in standardised assays in behavioural ecology.