Individual versus pseudo-repeatability in behaviour: Lessons from translocation experiments in a wild insect.

Repeatability represents a key parameter in ecological and evolutionary research. Repeatability is underpinned by developmental plasticity and genetic variation but may become biased upwards by repeatable differences in environments to which individuals respond plastically. The extent of upward bias caused by the latter mechanism (causing "pseudo-repeatability") is important yet rarely investigated in ecological research. We repeatedly assayed a key behaviour (flight initiation distance) affecting longevity in a wild cricket population (Gryllus campestris). We used naturally moving, translocated and forced-stationary individuals to study bias in repeatability caused by spatial variability in environmental conditions. Our experiments acknowledged that translocations might themselves bias repeatability estimates if animals respond to handling procedures (a necessary component of translocations). Individuals were, therefore, either (i) repeatedly translocated and assayed or (ii) assayed at multiple burrows as part of natural movements. This enabled estimation of behavioural variance attributable to individual, burrow and residual components within each treatment; comparison across treatments addressed whether translocations caused bias. We also calculated repeatability for individuals that were forced to be stationary to investigate whether this led to upward bias of repeatability. For adult crickets, individual explained 17.8% versus 17.2%, and burrow 8.7% versus 10.3%, of the behavioural variance in translocated versus natural-movement treatments. Repeatability for forced-stationary adults was 31.1%, thereby demonstrating experimentally that certain study designs bias repeatability upwards. For translocated juveniles, individual explained 10.0% and burrow 6.0% of the variance, while in the natural-movement treatment, those components could not be separated as juveniles do not switch burrows. Translocations did not lead to detectable biases in behavioural mean or variance. Repeatability was not biased for adults subjected to the natural-movement treatment because individuals were assayed under many different environments, facilitating the separation of individual from burrow effects. Upward bias would have occurred with less optimal sampling schemes: if individuals had been assayed repeatedly at the same burrow. We, therefore, recommend that translocation experiments are more commonly applied, particularly in stationary species, to ensure the unbiased estimation of repeatability.