How do we characterize temperament? Broad testing of temperament across time and contexts in low-variable conditions

In many species, individual animals often maintain relatively consistent individual differences in behaviour, indicative of temperament or personality. These stable individual differences are important for life history, survival and health consequences, and the field has seen exponential growth in temperament (e.g. boldness, exploration, sociability, aggression and activity) research. Quantification of animal temperament can be difficult because it requires (1) repeated behavioural measures from the same individual (to gauge within-individual consistency across time and contexts) and (2) identification of the most appropriate behaviour(s) to characterize temperament. In the current study, we took an atheoretical approach and conducted repeat, diverse behavioural tests in a low-variable environment to determine behaviours that were individually consistent over time and across conditions. To control environmental effects on behaviour, we conducted this work with individually housed outbred male Sprague–Dawley rats in a laboratory. To determine individual consistency, we measured each rat's behavioural response to five different arenas (i.e. different contexts) and repeated the five tests at three time points. To identify potential composite, multibehaviour estimates of temperament, we used factor analyses to determine behaviours that covaried on each test arena, and calculated factor scores. To determine individual consistency of behaviours and factor scores across time and contexts, we used reliability and correlation analyses. Several behaviours consistently covaried on each test, and with a few notable exceptions, the resulting factors tended to align with five commonly studied temperament categories. Certain behaviours/factors were more consistent across time and contexts than others; distance travelled/activity were highly consistent, as were sociality- and boldness-related behaviours, whereas aggression and exploration were less consistent. We did not find strong evidence that multibehaviour estimates of temperament (factor scores) were more consistent over time or across conditions than single-behaviour metrics. Where possible, similar atheoretical approaches to cross-time and cross-context behaviour testing will be beneficial for initial temperament characterization.