If bold animals are more likely to be trapped than shy animals, we take a biased sample of personalities—a problem for behavioural research. Such a bias is problematic, also, for population estimation using mark-recapture models that assume homogeneity in detection probabilities. In this study, we investigated whether differences in boldness result in differences in detection probability in a native Australian rodent, the grassland melomys (Melomys burtoni). During a mark-recapture study of this species, we used modified open field tests to assess the boldness (via emergence, and interaction with a novel object) of melomys trapped on the last night of four trapping nights in each of two trapping sessions. Despite melomys showing repeatable variation in these behavioural traits, neither boldness nor emergence latency had an effect on detection probability, and we found no evidence that detection probability varied between individuals. This result suggests that any neophobia is experienced and resolved in individuals of this species on a scale of minutes, rather than the hours across which traps are made available each night. Our work demonstrates that personality-caused sampling bias may not be inevitable, even in situations where animals are required to respond to novelty to be detected, such as in baited traps. Heterogeneity in personality does not inevitably lead to heterogeneity in detection probability. Historically, passive traps were assumed a non-biased means of sampling animal populations. Increasingly behavioural ecologists suggest that personality traits, particularly individual boldness, may influence behaviour and, as a consequence, could result in sampling bias. Here, we present a comprehensive example of when animal personality has no effect on detection probability. Despite having distinct personalities, detection probabilities of a native Australian rodent, grassland melomys Melomys burtoni, were not influenced by whether they were ‘shy’ or ‘bold’. We provide evidence that heterogeneity in personality does not inevitably lead to heterogeneity in detection probability. Given that population estimation models typically assume homogeneity in detection probability between individuals, if this is a broad phenomenon, consistently similar results may improve our confidence in this assumption.
Read full abstract