On the use of rapid assays in personality research: a response to Edwards et al.

On the use of rapid assays in personality research: a response to Edwards et al.

Animal Behaviour 86 (2013) e1ee3 Contents lists available at SciVerse ScienceDirect Animal Behaviour journal homepage: www.elsevier.com/locate/anbeh...

201KB Sizes 0 Downloads 22 Views

Animal Behaviour 86 (2013) e1ee3

Contents lists available at SciVerse ScienceDirect

Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav

Forum

On the use of rapid assays in personality research: a response to Edwards et al. Peter A. Biro* Centre for Integrative Ecology, School of Life and Environmental Sciences, Deakin University, Geelong, VIC, Australia

a r t i c l e i n f o Article history: Received 21 February 2013 Initial acceptance 15 March 2013 Final acceptance 8 April 2013 Available online 31 May 2013 MS. number: 13-00165 Keywords: acclimation habituation personality random regression

When we assay behaviour in the laboratory, we assume that it should be informative of behaviour that would also be expressed in nature. However, our assays usually (forcibly) remove individuals from the field, or from their holding cages/tanks, and then confine them in some novel testing apparatus where they may or may not be briefly acclimated for some minutes before quantifying their behaviour. Then, animals are returned to the field or holding cages, and the assay later repeated one or more times to see whether behaviour is consistent within individuals, and to determine some mean value for each individual (its behavioural type, or BT). Of course, this is highly artificial, but we assume that assays under this level of unnatural stress and novelty tell us something about behaviour under natural conditions, where the animals’ behavioural repertoire evolved. In other words, we assume individual rank order is maintained across these contexts, and thus is correlated across contexts. As a first step towards testing this assumption (Biro 2012), I repeatedly assayed fish behaviour over time, beginning shortly after introduction to their individual home tanks and then twice per day for 1 week. My intent was to test ‘whether or not observations made under forced novel conditions predict behavioural traits under familiar and (presumably) less stressful conditions, conditions that are likely to be most similar to what animals in nature * Correspondence: P. Biro, Centre for Integrative Ecology, School of Life and Environmental Sciences, Deakin University, Geelong, VIC 3216, Australia. E-mail address: [email protected].

experience, most of the time’ (Biro 2012, page 1296). In essence, I observed that behaviour was highly variable within individuals during the first 2 days in captivity, but was repeatable thereafter. In addition, there was no indication that initial behaviour predicted BTs that were evident from subsequent assays. On this basis, I suggested that we should not assume that a rapid series of one, two or three assays as commonly employed in personality studies is sufficient to characterize individual behaviour types. Why? Because it seems that individuals can markedly and systematically differ in the rate at which they habituate and/or acclimate with repeated behavioural assays (Biro 2012), and in intraindividual variability (see below), meaning that rank order of individual behaviour is not maintained over the time during which habituation/acclimation is occurring. That individuals might differ in patterns of habituation/ acclimation is not new (see below), but the great majority of personality studies conduct only one, two or three assays of behaviour and thus cannot address this issue head on in the face of highly variable behavioural data. I then speculated that the observed patterns of relatively unpredictable followed by predictable behaviour may result from the highly stressful and artificial assays we tend to use in our research, something with which animals have not had any evolutionary experience. It was my opinion that characterizing an individual’s BT using data solely from the brief interval when animals seem to be really stressed and (apparently) behaving erratically and unpredictably may not be a good idea, because animals in nature would never experience such a forced and unnatural novel experience. Of course, animals in nature

0003-3472/$38.00 Ó 2013 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.anbehav.2013.04.020

e2

P. A. Biro / Animal Behaviour 86 (2013) e1ee3

experience novelty, but in nature animals are usually choosing to take risks and explore novelty, not being suddenly removed and confined in a net or box, then forced into some apparatus. Edwards et al. (2013) recently commented on my note, and raised three main concerns with my study and interpretations from those data: (1) my assay of activity is conflated with exploration, (2) many repeated assays habituate individuals and so my results cannot be extrapolated to assaying in a novel environment and (3) they question my suggestion to gather many repeats per individual, when studying wild populations. Regarding the first concern, it is indeed possible that what I called ‘activity’ was actually ‘exploration’ during the earliest observations, something that I had not previously considered. But how do we distinguish the motivation behind the patterns of spontaneous activity I observed? One could argue that the highly variable initial activity reflected exploration, because the environment was novel (e.g. Reale et al. 2007). But do we expect exploration to possess inherently greater within-individual variation than compared to some other behaviour, like activity? Conversely, one could argue it represented boldness, because the animals were expressing activity in the presence of high stress and/or fear. I do not have answers to these questions, but altogether these questions and concerns indicate that we should strive to take control over the context of our assays and/or gather sufficient samples over time to help us identify whether context has changed. Edward et al.’s second concern, about the effect of habituation, seems to have two parts. First, they suggest my ‘boldness scores, after repeated training and sampling, might not reflect boldness, because of habituation to the stimulus’ (page e2). Second, they state that ‘he did not consider the likelihood of habituation or that individuals can differ in their rate of habituation and learning (Glowa & Hansen 1994; Gallistel et al. 2004). We therefore consider it vital to consider context, habituation and individual plasticity in habituation responses when designing and drawing conclusions from behavioural assays’ (page e2). Together, these two related concerns seem somewhat misplaced. First, only a few individuals required training before they would use the artificial shelter I provided them, and at the end of the experiment individuals still consistently and markedly differed in their latency to emerge from shelter. Second, I explicitly tested for and revealed complex individual differences in patterns of habituation (i.e. the individual ‘plasticity’ they refer to; see my predictions in Fig. 1 of Biro 2012). It is precisely these sorts of individual-specific responses that can make rapid (one, two or three) assays not informative on their own. To characterize an individual’s BT, we must repeatedly sample them over time, and so habituation/acclimation seems an unavoidable possibility in any assay. Habituation does not mean that my assay does not reflect boldness; rather, I would consider it an integral part of boldness (for a nice discussion of the importance of habituation, see Bell & Peeke 2012). Habituation is an old concept, but the idea that individuals might differ in the rate or pattern of habituation is relatively recent (reviewed in Bell & Peeke 2012). Whereas I suggest we tackle this issue head on by trying to quantify it, and determine appropriate acclimation intervals, Edwards et al. suggest that we should try to minimize the potential for habituation, namely by increasing intervals between measures. Indeed, increasing intervals might change the rate of habituation, but I would suggest that individuals will probably still differ in the rate and/or temporal pattern of habituation. Individual differences in habituation/acclimation are not uncommon, occurring for a variety of different behaviours at short (Montiglio et al. 2010; Bell & Peeke 2012), intermediate (Rodríguez-Prieto et al. 2011; Biro 2012; Stamps et al. 2012) and long interobservation intervals (Dingemanse et al. 2012). I fully agree with Edwards et al. that habituation/acclimation reflects changing contexts (levels of novelty/stress change across

successive observations), but I do not see how we can avoid such effects when we perform repeated sampling. Rather, it seems we need to consider multiple parameters to characterize individual BTs (individual intercepts, slopes and covariances), rather than a single mean value. Importantly, depending upon the stimulus, or strength of stimulus, individuals may habituate too quickly over time to the extent that we no longer see any meaningful individual differences. In that scenario, of course, gathering many samples will not help us discern whether individuals consistently vary. Another possibility is that if a stimulus is too strong, this can cause individuals to have a maximal or near maximal response, again resulting in the variance across individuals approaching zero. Edward et al.’s third concern was with my suggestion for obtaining 10þ assays per animal, saying this is ‘often not logistically feasible in wild populations’ (page e1). While I fully agree that sampling is definitely a lot more difficult in the field than my conveniently housed fish in the laboratory, I think that gaining multiple samples (more than most studies presently employ) can be feasible in the field. For example, flight initiation data are easily obtained on marked (or identifiable) individuals in the field, and one study on lizards had on average 10 samples, and up to 15 repeats per animal (Carter et al. 2012). It is worth noting that this relatively large sample size not only permitted a test of individual differences in habituation, but also permitted powerful tests for the effects of several covariates, such as temperature and season. Perhaps the trick to getting more samples in field-based studies is to design assays that do not require capturing the animals prior to each assay. In doing so, stress associated with capture and handling would be eliminated, reducing the possibility of complex and variable responses that introduce unwanted variance in our data, or conflating the experimental context under study as suggested. For instance, the presence of high stress when it goes unnoticed could mean that when we measure activity we might actually be measuring boldness (Carter et al. 2013). In any case, most studies of personality and syndromes observe each individual just a few times, or less (Bell et al. 2009; Garamszegi et al. 2012), and I think that we need to sample individuals far more than this to have power to characterize individual BTs rigorously in any study, particularly when in the presence of confounding variables. Given that within-individual variation appears to be the largest component of behavioural variation (Bell et al. 2009), and given that it may also be common for individuals to differ in their responses to all sorts of biotic and abiotic factors, including habituation/acclimation (Mathot et al. 2012), then correctly partitioning this complex behavioural variation will require hundreds of samples, and possibly many more than implied by recent simulations (Adolph & Hardin 2007; Martin et al. 2011; van de Pol 2012; Wolak et al. 2012). In conclusion, I have welcomed this debate and appreciate the constructive criticisms from Edwards et al, and from the reviewers of this response. I suspect that topics such as these, and related ones (Carter et al. 2013), will continue to be debated for some time, because at present there does not seem to be much agreement in the literature about exactly how we should sample, design experiments and assay behaviour (but see Reale et al. 2007). The way forward, then, should be to challenge our approaches and definitions, by evaluating our methodology and testing assumptions, and not simply perpetuating established ways of conducting personality research. Thanks to the editor and two referees for their comments, and to the ARC for funding. References Adolph, S. & Hardin, J. 2007. Estimating phenotypic correlations: correcting for bias due to intraindividual variability. Functional Ecology, 21, 178e184.

P. A. Biro / Animal Behaviour 86 (2013) e1ee3 Bell, A. M. & Peeke, H. V. S. 2012. Individual variation in habituation: behaviour over time toward different stimuli in threespine sticklebacks (Gasterosteus aculeatus). Behaviour, 149, 1339e1365. Bell, A. M., Hankison, S. J. & Laskowski, K. L. 2009. The repeatability of behaviour: a meta-analysis. Animal Behaviour, 77, 771e783. Biro, P. A. 2012. Do rapid assays predict repeatability in labile (behavioural) traits? Animal Behaviour, 83, 1295e1300. Carter, A. J., Heinsohn, R., Goldizen, A. W. & Biro, P. A. 2012. Boldness, trappability and sampling bias in wild lizards. Animal Behaviour, 83, 1051e1058. Carter, A. J., Feeny, W., Marshall, H., Cowlishaw, G. & Heinsohn, R. 2013. Animal personality: what are behavioural ecologists measuring? Biological Reviews, 88, 465e475. Dingemanse, N. J., Bouwman, K. M., van de Pol, M., van Overveld, T., Patrick, S. C., Matthysen, E. & Quinn, J. L. 2012. Variation in personality and behavioural plasticity across four populations of the great tit Parus major. Journal of Animal Ecology, 81, 116e126. Edwards, H. A., Winney, I. S., Schroeder, J. & Dugdale, H. L. 2013. Do rapid assays predict repeatability in labile (behavioural) traits? A reply to Biro. Animal Behaviour, 85, e1ee3. Garamszegi, L., Markó, G. & Herczeg, G. 2012. A meta-analysis of correlated behaviours with implications for behavioural syndromes: mean effect size, publication bias, phylogenetic effects and the role of mediator variables. Evolutionary Ecology, 26, 1213e1235.

e3

Martin, J. G. A., Nussey, D. H., Wilson, A. J. & Réale, D. 2011. Measuring individual differences in reaction norms in field and experimental studies: a power analysis of random regression models. Methods in Ecology and Evolution, 4, 362e374. Mathot, K. J., Wright, J., Kempenaers, B. & Dingemanse, N. J. 2012. Adaptive strategies for managing uncertainty may explain personality-related differences in behavioural plasticity. Oikos, 121, 1009e1020. Montiglio, P.-O., Garant, D., Thomas, D. & Réale, D. 2010. Individual variation in temporal activity patterns in open-field tests. Animal Behaviour, 80, 905e912. van de Pol, M. 2012. Quantifying individual variation in reaction norms: how study design affects the accuracy, precision and power of random regression models. Methods in Ecology and Evolution, 3, 268e280. Reale, D., Reader, S. M., Sol, D., McDougall, P. T. & Dingemanse, N. J. 2007. Integrating animal temperament within ecology and evolution. Biological Reviews, 82, 291e318. Rodríguez-Prieto, I., Martín, J. & Fernández-Juricic, E. 2011. Individual variation in behavioural plasticity: direct and indirect effects of boldness, exploration and sociability on habituation to predators in lizards. Proceedings of the Royal Society B, 278, 266e273. Stamps, J. A., Briffa, M. & Biro, P. A. 2012. Unpredictable animals: individual differences in intraindividual variability (IIV). Animal Behaviour, 83, 1325e1334. Wolak, M. E., Fairbairn, D. J. & Paulsen, Y. R. 2012. Guidelines for estimating repeatability. Methods in Ecology and Evolution, 3, 129e137.