Do animal personality components independently evolve and develop in response to environmental complexity?
Abstract
Widespread existence of consistent differences in behaviour among individuals even within species and populations, i.e., animal personality, has been established since the last decades in a wide array of taxa. However, little is known about personality traits' ontogeny and evolution. This study aimed at exploring eco-evolutionary mechanisms driving the emergence and development of animal personality. Focusing on boldness as a personality trait, we assessed how personality components (mean, among- and within-individual variabilities, repeatability, and plasticity in response to environmental complexity) develop at early age. Investigating developmental trajectories of all personality components, rather than averages only, offers a more exhaustive and comprehensive picture of how personality emerges, develops, and evolves. We compared personality components between juveniles of five morphs of Arctic charr (Salvelinus alpinus) ranging along gradients of ecological and genetic divergence from a common ancestor, raised from hatching in plain vs. structurally complex treatments. On the one hand, we show that some of these personality components evolve and develop independently from the others: mean boldness, which increases with divergence from the ancestor, was predominantly genotype-dependent and suspectedly a highly heritable trait with strong and stable selective pressures acting on it in the wild, while boldness repeatability might rather depend on the ecology of each morph. These two components were not affected by environmental complexity. On the other hand, variability-related components of personality, including their plasticity in response to environmental complexity, were rather dependent on genotype-by-environment effects and seemingly evolve and develop jointly. Boldness tended to be more consistent within the treatment mimicking the structural complexity of a given morph’s natural habitat, hinting that personality emergence might be favoured for individuals experiencing conditions to which they have been adapted. These findings suggest mechanisms by which personality components could be implicated in adaptability to environmental changes or even sympatric diversification and biodiversity.
README: Do animal personality components independently evolve and develop in response to environmental complexity?
https://doi.org/10.5061/dryad.xgxd254q2
Research questions
In the present study, we examined the eco-evolutionary mechanisms shaping the emergence and development of animal personality profiles. For that, we tested the influence of a key environmental factor in behavioural and cognitive development – structural complexity – on each personality trait's components (mean, among- and within-individual variabilities, repeatability and plasticity). We focused on the most studied personality trait, boldness – the individual propensity to take risks – that has recently been reported in the Arctic charr. We compared the developmental trajectories of all boldness components in juveniles of five wild Arctic charr morphs (Salvelinus alpinus), ranging along gradients of ecological and genetic divergence, exposed in common garden to plain vs. structurally complex treatments.
We hypothesized that:
(H1) environmental complexity experienced at young age influences the developmental trajectory of personality components (developmental plasticity of personality), which would translate into boldness components varying between treatments, either overall (environment effect) or within morphs (genotype-by-environment effect).
(i) The mean component: if environmental complexity has a uniform effect at the group level on personality development, we expected differences in mean boldness between treatments.
(ii) The among-individual variability component: if a complex environment provides a variety of microenvironments experienced differently and eliciting different personality responses for each individual, we would expect a higher boldness variability among individuals raised in the enriched treatment when compared to the plain one.
(iii) The within-individual variability and repeatability components: if behaviours are more consistently expressed when an individual’s current environmental conditions match those to which it has been adapted, this would translate into boldness being more repeatable, and/or less variable within individuals in the enriched treatment than in the plain one in every morph. We might also expect a morph-by-treatment interaction effect, resulting in boldness being more repeatable, and/or less variable within individuals, in the enriched treatment for morphs whose natural habitats are structurally more complex, as opposed to morphs living in lower-complexity environments that would display more consistent boldness in the plain treatment.
(H2) The evolutionary history of individuals can impact the development of personality profiles and their response to environmental complexity. Instead of evolving jointly as a block, personality components could rather evolve independently from each other. If this is the case, we expected morph-specific personality profiles, of which personality components do not show gradient-like values correlated to the divergence degree across morphs.
Data acquisition
All fish were submitted twice to an Open Field Test (OFT) with shelter, with a 7-day interval between each Replicate (video-recorded at 30 frames per second). At the beginning of the test, the focal fish was placed in the closed shelter of the OFT arena. After 5 minutes acclimation, the shelter door was opened and the fish was free to explore the arena for 20 minutes. The fish were tracked from the obtained videos with Ethovision XT version 15 (Noldus Information Technology). The arena was divided into four virtual zones: the shelter zone covering the shelter area, overlapping with an entry zone; the border zone along the edges of the arena; the centre zone being the remaining part of the arena. The fish barycentre was used to calculate behavioural variables presented in the Metadata.
Metadata
- Date = the day at which the OFT test took place for the focal fish (DD/MM/YYYY)
- Round = the order (1 to 13) in which the OFT test was carried out for the focal fish that day
- Twosome = the team of manipulators that set up the focal OFT trial
- Trial = the number of the OFT trial, i.e., the order in which fish were tested, within a morph. NB: one trial consisted of 4 arenas tested simultaneously, hence 4 fish having the same trial.no
- Tank = the tank replicate to which the focal fish belongs
- Treatment = the environmental conditions (E - Enriched or P - Plain) under which the focal fish was raised
- Morph = the morph (AN, VS, VB, PL, LB) to which the focal fish belongs
- Replicate = the Replication (either A or B) as OFT tests were repeated twice for each fish
- Arena = the number of the arena in which the focal fish was tested during the trial
- Indiv = the PIT tag ID number of the focal fish
- W1 = weight of the fish (g) at tagging event
- FL1 = fork length of the focal fish at tagging event, measured from the tip of the snout to the fork of the caudal fin (cm)
- TL1 = total length of the focal fish at tagging event, measured from the tip of the snout to the tip of the caudal fin lobes (cm)
- W = weight of the fish (g) after the focal OFT trial
- FL = fork length of the focal fish after the focal OFT trial, measured from the tip of the snout to the fork of the caudal fin (cm)
- TL = total length of the focal fish after the focal OFT trial, measured from the tip of the snout to the tip of the caudal fin lobes (cm)
- T1 = temperature in the OFT arena at the beginning of the OFT trial (°C)
- T2 = temperature in the OFT arena at the end of the OFT trial (°C)
- T.mean = the average temperature over the OFT trial, i.e., (T1 + T2)/2 (°C)
- SGR = Specific Growth Rate, calculating the weight gain between the tagging event and the focal OFT trial, i.e., (ln(W)-ln(W1))*100/nb of days
Behavioural variables recorded for the focal fish in the OFT arena during the 20 minutes free phase:
- Dtot = the total distance travelled (cm)
- *Velocity.initial *= the mean swimming velocity (cm.s^-1)
- Exit.time = the latency to exit the shelter for the first time (head and trunk visible, s)
- *XXX.freq *= the frequency of visits to the XXX zone
- XXX.duration = the amount of time spent in the XXX zone (s)
- Ang.vel = absolute angular velocity (deg.s^-1)
- Velocity = the mean swimming velocity divided by the total length of the focal fish (i.e., Velocity.initial/TL, s^-1)
Description of the data and file structure
The data is presented in the form of an excel table, where each variable described above represents a column, and each line represents the information for one specific fish, in one specific OFT trial within one specific OFT replication. Missing data are reported as "NA". The associated dataset can be found under the name "Data.xlsx".
Code/Software
The code used to obtain boldness scores and analyse the data is available here under the name "Accompanying_R_Script_2025.123077.Rmd". Supplementary figures and tables associated to the main text can be found in the electronic supplementary material of the related article (https://doi.org/10.1016/j.anbehav.2025.123077). All statistical analyses, following those of Dellinger, M., Steele, S. E., Sprockel, E., Philip, J., Pálsson, A., & Benhaïm, D. (2023). Variation in personality shaped by evolutionary history, genotype and developmental plasticity in response to feeding modalities in the Arctic charr. Proceedings of the Royal Society B, 290, 20232302. https://doi.org/10.1098/rspb.2023.2302, were performed with R v.4.3.0 software (R Core Team. (2018). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.r-project.org/ ) We used linear mixed models fitted in Stan Bayesian language (Carpenter, B., Gelman, A., Hoffman, M. D., Lee, D., Goodrich, B., Betancourt, M., Brubaker, M. A., Guo, J., Li, P., & Riddell, A. (2017). Stan: A probabilistic programming language. Journal of Statistical Software, 76(1), 11–32. https://doi.org/10.18637/jss.v076.i01) using the brms package v.2.19.0 (Bürkner, P. C. (2017). brms: An R package for Bayesian multilevel models using Stan. Journal of Statistical Software, 80(1), 1–28. https://doi.org/10.18637/jss.v080.i01).