When behaviour is studied experimentally, individuals are often placed in a specific set-up that allows the variables of interest to be manipulated, controlled, or measured. On the one hand, this involves a change in environment, as subjects are moved from their initial environment (e.g. home tank) to the experimental apparatus. On the other hand, it typically requires interaction with the experimenter, who handles the subjects in order to transfer them.
These two experimental constraints—environmental change and handling—have been identified as potential sources of methodological bias within the STRANGE framework (Webster & Rutz, 2020). Specifically, they relate to the “Acclimation and Habituation” factor, which can influence behavioural and physiological responses. Based on definitions proposed by Nanninga et al. (2017), we consider acclimation to be the process by which an individual becomes familiar with a new stimulus or environment upon first exposure, whereas habituation refers to a learned decrease in response to repeated exposure.
It is now widely accepted that handling and environmental change can induce stress in a broad range of animal species (Balcombe et al., 2004; Webster & Rutz, 2020). However, the nature and intensity of these effects may vary depending on the type and purpose of handling or environmental transition. For instance, manipulation may occur when animals are collected from the wild and brought to the laboratory, during laboratory rearing, or when subjects are transferred from their rearing environment to the test apparatus.
Several studies have examined the impact of handling on animal stress, mostly in vertebrates and particularly in mammals. For example, Gessner et al. (2022) reported that habituating mice to handling improved performance in a novel object recognition task. Gouveia & Hurst (2019) showed that reducing the invasiveness of handling procedures lowered stress in mice. Shishis et al. (2023) found that different methods of transferring zebrafish from their home tank to an experimental device affected multiple behavioural variables, with more invasive handling associated with increased immobility. Philippe et al. (2023) similarly identified stress-related physiological responses (e.g. cortisol elevation) in zebrafish after handling, even though they did not observe significant differences across handling types; however, they noted both habituation and acclimation effects over time. Comparable results were reported by Wilson et al. (2021) in green shore crabs, where less invasive techniques were associated with lower stress indicators (oxygen consumption, haemolymph glucose, and L-lactate). Taken together, these findings underscore the importance of standardising handling procedures, as well as reporting them in detail to improve replicability and interpretation of experimental outcomes.
The transfer from one environment to another thus appears to be a consistent stressor, even when attempts are made to minimise its invasiveness. A commonly recommended mitigation strategy is to impose an acclimation period between handling and the onset of the experimental task. Acclimation is believed to promote a return to homeostasis following the stress induced by manipulation and/or environmental change (Arbon et al., 2023). According to our literature review, fewer studies have focused specifically on acclimation duration compared to the growing literature on handling effects. Most research on this topic appears to have been conducted in the context of fish behaviour (Burns, 2008; Makaras et al., 2021; O’Neill et al., 2018), showing that pre-test acclimation duration can influence both stress levels and behavioural responses, but that these effects are highly species- and context-dependent.
These authors argue that acclimation is essential to allow subjects to recover from transfer-related stress, which could otherwise interfere with spontaneous behaviour. In this sense, the lack of acclimation may affect both the psychometric properties of behavioural measures and the validity (particularly internal validity) of experimental protocols, thereby undermining the robustness and reproducibility of findings. Although this issue could theoretically affect a wide range of behaviours, it is likely to be especially problematic in studies investigating stress-related mechanisms.
In the present study, we focus on acclimation in terrestrial isopods, where experimental research on stress has both theoretical and applied relevance. These organisms are known bioindicators of environmental quality. Laboratory studies have shown that environmental perturbations—such as vibrations or chemical stressors—negatively affect their physiology and behaviour (Ďurajková et al., 2022; Roques et al., 2020). Their ecological role as decomposers has prompted investigations into the impact of climate-related variables (e.g. heat or moisture stress) on their biological functioning (Delhoumi et al., 2023; Depeux et al., 2023; Wu et al., 2021). Meanwhile, several studies have explored how different stressors influence behavioural responses in various woodlouse species (Anselme, 2018, 2019; Anselme et al., 2023), helping to bridge the gap in our understanding of invertebrate behaviour, which remains underrepresented in comparison to vertebrate models (Schnell & Clayton, 2021).
Here, we adapted the protocol developed by Broly & Deneubourg (2015) in their third experimental condition, with four key modifications to explore acclimation effects at the individual level. First, we manipulated the duration of acclimation, i.e. the waiting time imposed on the woodlouse before the task (referred to as “retention time” by Broly & Deneubourg, 2015) as a continuous variable. Second, we included three species of woodlice—Porcellio dilatatus dilatatus (Brandt, 1833), Oniscus asellus (Linnaeus, 1758), and Porcellio scaber (Latreille, 1804), thereby increasing interspecies variability. Third, this also enabled a larger overall sample size. Fourth, we used the time to traverse a straight corridor as our main dependent variable, whereas Broly & Deneubourg (2015) focused primarily on departure time (which we analysed as a secondary outcome). Consequently, in our design, acclimation duration served as the independent variable, while straight-line traversal time was the principal dependent variable.
Preregistered Hypotheses
We formulated a set of concurrent hypotheses concerning the effect of pre-task waiting time (i.e., acclimation duration) on woodlice behaviour. Pre-registration can be consulted online at osf.io/6cbnp. The null hypothesis (H0) posits that waiting time before the task has no impact on behavioural performance. In this case, regardless of how long a woodlouse waits before the beginning of the task, the speed at which it travels the straight line would remain unchanged.
The first alternative hypothesis (H1a) suggests that a longer waiting time before the task leads to a slower traversal of the straight line. This prediction is consistent with the findings of Broly & Deneubourg (2015), who observed that individuals subjected to longer initial retention times were slower after release. The underlying rationale is that handling and environmental change initially provoke stress, which induces a heightened state of arousal, often associated with increased locomotor activity. As acclimation proceeds, this stress is reduced, and the individual enters a calmer physiological state, resulting in slower movement. This interpretation aligns with Anselme (2019) dual-process model, which proposes that exposure to a novel and non-rewarding experimental environment triggers a moderate stress response that promotes horizontal escape behaviour (i.e., running). Reduced stress through acclimation would therefore attenuate this escape drive, decreasing speed.
Conversely, the second alternative hypothesis (H1b) predicts that a longer waiting time before the task may result in a faster traversal of the straight line. Although this hypothesis is less supported by existing literature, it accounts for the possibility that handling and environmental change induce a high-intensity stress response, sufficient to temporarily suppress behavioural activity or even result in immobility. In this view, a prolonged pre-task delay might allow individuals to recover from this intense stress, thereby restoring or even enhancing locomotor performance. This reasoning is supported by observations in other species showing behavioural inhibition following strong stressors (Anselme, 2019; Shishis et al., 2023).
Finally, the third hypothesis (H2) proposes a non-linear effect of waiting time. Specifically, it predicts that as acclimation duration increases, woodlice will initially move more slowly, but this trend may reverse if the waiting period becomes excessively long. Prolonged isolation without sensory stimulation or opportunities to explore (conditions typical of extended acclimation) could itself become a source of stress, particularly due to social separation or environmental deprivation (O’Neill et al., 2018). As a result, stress levels may first decrease with acclimation, but begin to rise again after a certain threshold, leading to a U-shaped relationship between acclimation duration and behavioural performance.
Methods
Animals
Three species of woodlice were involved: Porcellio dilatatus dilatatus (Brandt, 1833), Oniscus asellus (Linné, 1758) and Porcellio scaber (Latreille, 1804). These species were specifically chosen because of their reaction to disturbance. These three species do not conglobate. When disturbed, their fleeing or freezing behaviour can be observed without latency. They were derived from natural populations collected in France and Scotland (Saint-Martin de Connée, Mayenne, 2012 and Rom, Deux-Sèvres, 1988, for P. dilatatus dilatatus, Cosne-sur-Loire, Nièvre, 2007 and Edimbourg, south-east Scotland 2001 for O. asellus, Mazé, Maine-et-Loire 2012 and La-Molle-St-Héray, Deux-Sèvres 2009 for P. scaber). The animals were maintained in standard laboratory conditions (in boxes of moistened compost, at 20°C, and with the natural photoperiod of Poitiers, France 46.58°N, 0.34°E) and fed ad libitum fresh organic carrots and dried leaves of linden (Tilia sp.). None of the tested individuals were moulting during the course of the experiment. Gravid females were excluded from the experimental pool. French and European reglementations do not require formal ethical evaluation concerning research on arthropods. We followed the guidelines for ethical conduct in the care and use of nonhuman animals in research (American Psychological Association, 2022). The protocol was designed to avoid any unnecessary distress.
As our individuals come from populations bred in the laboratory for several generations, their dimensions are on average smaller than those of wild individuals. For each species, we measured the length and width of 20 randomly selected males and 20 randomly selected females, and calculated the mean dimensions for each species and their standard deviations (see Table 1). The individuals were collected from the rearing box at least 72 h before the experiment, separated by sex and placed in groups of 10 for P. dilatatus dilatatus and P. scaber and a maximum of 15 individuals for O. asellus, in plastic boxes measuring 16.9 cm × 14.2 cm × 5.9 cm (Figure 1). The bottom of each pre-experimental box was covered with moist potting soil and rehydrated dried leaves of linden. Water was added regularly.
638 trials were run after official data sharing (i.e. after the first pre-tests and pilots). Data was collected during the experiment by the experimenter and double-checked a posteriori by independent video coders. After data verification, the total number of woodlice for valid trials was 473 (see Figure 3). 39 woodlice performed the task twice. This did not significantly affect the main results of the experiment (see this document for more details). The minimum interval between the two sessions for the animals concerned was 15 days. Unpublished results from previous experiments conducted by the last author indicated that memory of the experiment and the apparatus appeared to be preserved for a maximum of one week in woodlice. The minimum interval between the two sessions for our current experiment was therefore at least twice the estimated limit of possible memory retention.
In order to approximate a correct estimate of the optimal waiting time before the task, we aimed to collect as much data as possible in April and May 2023. The goal was to test at least 480 woodlice in total, belonging to the three species. This minimum sample size was planned on the basis of a stability approach. According to the simulations, considering a pessimistic scenario where the correlation between waiting time and travel time exists but is weak (ρ = .1), we needed 470 individuals for the correlation to stabilise. In other words, we needed 470 individuals (critical point of stability) to ensure that the correlation does not drop below 0 and does not rise above 0.2, with 95% confidence (Schönbrodt & Perugini, 2013). Based on our pret-tests, anticipating potentially 10 invalid trials due to possible technical problems or unusual behaviour, we therefore planned a minimum sample size of 480 woodlice. This preregistered sample size is therefore respected.
Apparatus
We presented individuals with a task consisting in crossing a corridor linking a rectangle starting arena to a smaller rectangle finishing arena (Figure 1). The corridor and the arenas were built with LEGOⒸ blocks inspiered by McCreery et al. (2016) and Carlesso et al. (2023), allowing both modularity and easy reproducibility. Two apparatus were built. For the first one (experiments on P. dilatatus), the blocks were assembled to form a wall 2.9 cm high and 1.6 cm wide. This wall surrounded two arenas, a starting arena (5.5 cm × 4.9 cm), a finishing arena (3.9 cm × 2.4 cm), and a corridor (21.5 cm × 2.3 cm) connecting them. A second apparatus, in which the width of the corridor was reduced to 1.5 cm, was built in parallel. This second apparatus was used for P. scaber and O. asellus, which are smaller than P. dilatatus dilatatus (see Table 1). In the centre of the start area was a transparent plastic cylinder (diameter 2.4 cm, height 2.8 cm). This cylinder was used as a waiting area within the apparatus where the animals were placed at the start of the task. The apparatus was placed on a sheet of thick white paper which was changed regularly to avoid any potential chemical residual bias.
