© 2024. Published by The Company of Biologists Ltd | Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.

244662

RESEARCH ARTICLE

Differential metabolic responses in bold and shy sea anemones

during a simulated heatwave
Daniel K. Maskrey1, Shaun S. Killen2, Lynne U. Sneddon3, Kathryn E. Arnold4, David C. C. Wolfenden3 and
Jack S. Thomson1, *

ABSTRACT INTRODUCTION
As climate change-induced heatwaves become more common, The pace-of-life syndrome hypothesis (POLS; Réale et al., 2010)
phenotypic plasticity at multiple levels is a key mitigation strategy by posits that organisms must trade-off long-term survival against
which organisms can optimise selective outcomes. In ectotherms, short-term reproductive success, leading to predictable correlations
changes to both metabolism and behaviour can help alleviate between risk-related or metabolically costly behaviour and
thermal stress. Nonetheless, no study in any ectotherm has yet energetic physiology. POLS predicts that individuals with
empirically investigated how changing temperatures affect more risk-prone, ‘bolder’ personalities (consistent behavioural
among-individual differences in the associations between these differences among individuals; Sih et al., 2004), should show
traits. Using the beadlet anemone (Actinia equina), an intertidal ‘faster’ physiological characteristics, such as a high metabolic rate,
species from a thermally heterogeneous environment, we that are conducive to higher early fecundity and growth, but
investigated how individual metabolic rates, linked to morphotypic not to long-term survival (Montiglio et al., 2018; Réale et al., 2010).
differences in A. equina, and boldness were related across It is well established that life-history trade-offs can drive the
changing temperatures. A crossed-over design and a temporal maintenance of personality types (Wolf and McNamara, 2012; Wolf
control were used to test the same individuals at a non-stressful and Weissing, 2010); bold individuals often place themselves at
temperature, 13°C, and under a simulated heatwave at 21°C. At greater risk of mortality than shy conspecifics, but benefit from
each temperature, short-term repeated measurements of routine riskier lifestyles via increased foraging opportunities and more
metabolic rate (RMR) and a single measurement of a repeatable energetic scope for reproduction (Smith and Blumstein, 2008).
boldness-related behaviour, immersion response time (IRT), were POLS simply extends these trade-offs by predicting that the
made. Individual differences, but not morphotypic differences, were physiology of bolder individuals should be primed to maintain
highly predictive of metabolic plasticity, and the plasticity of RMR high levels of risky activity, exploration and foraging, even under
was associated with IRT. At 13°C, shy animals had the highest periods of heightened environmental stress (Montiglio et al., 2018;
metabolic rates, while at 21°C, this relationship was reversed. Réale et al., 2010). As such, according to POLS, bolder animals
Individuals that were bold at 13°C also exhibited the highest should exhibit reduced physiological stress reactivity, faster growth
metabolic rates at 21°C. Additional metabolic challenges during and higher metabolic rates than shyer conspecifics (Biro and
heatwaves could be detrimental to fitness in bold individuals. Stamps, 2010). The relationships proposed by POLS could be of
Equally, lower metabolic rates at non-stressful temperatures could great value to conservationists, indicating the life-history strategies
be necessary for optimal survival as heatwaves become more present in a given population and informing whether some
common. These results provide novel insight into the relationship behavioural phenotypes might be more susceptible to novel
between metabolic and behavioural plasticity, and its adaptive selective pressures than others.
implications in a changing climate. Evidence for POLS is mixed, with studies finding positive
(Polverino et al., 2018), negative (Le Galliard et al., 2013) and
KEY WORDS: Climate change, Boldness, Metabolism, Pace of life, inconclusive associations (Killen et al., 2012) between metabolic
Marine invertebrate rate and boldness. As such, a recent meta-analysis concluded that
empirical evidence for POLS is weak (Royauté et al., 2018).

Journal of Experimental Biology

However, no studies have yet investigated how individual variation
in the scope and nature of phenotypic plasticity to environmental
1
Department of Earth, Ocean and Ecological Sciences, School of Environmental
fluctuation might affect relationships between bioenergetics and
Sciences, Nicholson Building, University of Liverpool, Liverpool L69 3GP, UK. boldness (Killen et al., 2013; Montiglio et al., 2018). ‘Activational’
2
Institute of Biodiversity, Animal Health & Comparative Medicine, Graham Kerr phenotypic plasticity (Snell-Rood, 2013), an animal’s ability to
Building, University of Glasgow, Glasgow G12 8QQ, UK. 3Department of Biological
& Environmental Sciences, University of Gothenburg, Box 463, SE-405 30
rapidly alter aspects of its phenotype in response to acute
Gothenburg, Sweden. 4Department of Environment and Geography, Wentworth environmental change (Seebacher et al., 2015; Stamps, 2016),
Way, University of York, Heslington, York YO10 5NG, UK. could be crucial to understanding these relationships under a rapidly
*Author for correspondence ( [email protected]) changing climate, where extreme weather events are becoming
increasingly common (IPCC, 2013, 2018). Activational plasticity
D.K.M., 0000-0002-1514-1565; L.U.S., 0000-0001-9787-3948; J.S.T., 0000- can have selective advantages, helping many species remain robust
0003-2822-5589
to fluctuating environments (Snell-Rood, 2013), and disadvantages,
This is an Open Access article distributed under the terms of the Creative Commons Attribution as maintaining scope for short-term plasticity is energetically costly
License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use,
distribution and reproduction in any medium provided that the original work is properly attributed. (Dall et al., 2004). This leads to variation within populations in the
ability of different individuals to deal with acute environmental
Received 15 June 2022; Accepted 19 December 2023 changes (Dingemanse and Wolf, 2013) and could thus influence

1
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

relationships between metabolism and boldness, and their selective differences in boldness-related behaviours. Of particular relevance
implications under different environmental scenarios. could be immersion response time (IRT), which not only displays
Although activational plasticity does not itself form part of POLS among-individual variation within and across environments but also
(Montiglio et al., 2018), the relationship between metabolism and may be closely associated with an anemone’s metabolic rate
behaviour is likely to be both species and context dependent (Killen (Maskrey et al., 2020, 2021). IRT is a risk-related behaviour, which
et al., 2012). In mosquitofish, for example, lineages from more measures how long after a simulated tide cycle it takes an anemone
stochastic environments grow faster, but also exhibit lower to re-extend its feeding tentacles and resume foraging. Because
metabolic rates and are shyer than those from stable environments A. equina is unable to feed when the tide is out, more protracted
(Polverino et al., 2018). Among-individual variation in plasticity tentacle retraction upon immersion will lead to an animal having
may be an especially important driver of relationships between less time to forage and thus less energy for metabolic processes.
behaviour and metabolism in heterogeneous environments, where Further, tentacle extension in A. equina is inherently linked to
the coexistence of different plastic strategies can be driven by spatial respiratory efficiency and gas exchange (although anemones do
and temporal variation both in normal environmental conditions and continue to respire with their tentacles retracted or when exposed;
in how much those conditions fluctuate (Dingemanse and Wolf, Griffiths, 1977; Navarro et al., 1981), so spending longer with
2013; Wolf and McNamara, 2012). Despite the clear need to tentacles retracted can be highly costly in the face of metabolic
incorporate activational plasticity into investigations of POLS and challenges (Griffiths, 1977). Conversely, while its primary function
its selective implications, we currently do not have any empirical is to support essential metabolic processes and gas exchange,
information on how short-term environmental changes, and the tentacle extension is also risky, placing anemones under increased
overall variability of an animal’s environment, might affect the threat of predation (Edmunds et al., 1974, 1976). Given that
relationships between metabolism and behaviour. anemones are still able to respire with their tentacles retracted
When investigating how acute environmental changes affect (Griffiths, 1977; Navarro et al., 1981), and thus tentacle extension is
these relationships, temperature is a particularly important not perfectly correlated with routine metabolic rate (RMR, the
environmental variable to consider. As climate change leads to metabolic rate of an animal undergoing normal activity), tentacle
higher average temperatures and increased frequency of heatwaves extension behaviour should represent a trade-off. If an individual is
(IPCC, 2013, 2018), animal populations are being placed under more cautious to predation risk, it is likely to face the costs of
substantial, novel selective pressures (Parmesan, 2006). Thermal reduced efficiency of gas exchange and reduced foraging
physiology and thermal preferences are especially important drivers opportunities but may survive longer in the face of predators. As
of associations between metabolism and behaviour in ectotherms such, general tentacle extension behaviour, and in turn IRT, should
(Abram et al., 2017), where many physiological processes, act as a useful gauge of boldness. This trade-off should be
including metabolic rate, are intrinsically linked to environmental particularly pronounced at high temperatures, which drive increases
temperatures (Seebacher et al., 2015). In ants, for example, colonies in metabolic demand (Abram et al., 2017) but are also likely to
from warmer environments are more likely to show the positive place anemones under greater risk of predation from other
associations among energetically costly or risk-related behaviours ectotherms (e.g. Twardochleb et al., 2020; Yamane and Gilman,
that are predicted by POLS (Segev et al., 2017). Adult parasitic 2009). Relationships between IRT and metabolic rate at different
wasps with a preference for low temperatures are another example, temperatures should thus provide clear indications of whether
exhibiting high metabolic rates, low foraging efficiency and short A. equina follows POLS and enable prediction of the metabolic
lifespans when exposed to high temperatures (Le Lann et al., 2011). responses of different individuals to thermal extremes.
As an acute thermal stressor to which they are unable to immediately In this study, we investigated for the first time how relationships
physiologically adjust, climate change-induced heatwaves present between behavioural phenotypes and metabolic rates were
ectotherms with a different set of ecological challenges compared influenced by different temperatures, both within and among
with chronic temperature change (Vajedsamiei et al., 2021), and individuals. We measured the associations between RMR (Metcalfe
many species rely heavily on behavioural plasticity to address these et al., 2016; Velasque and Briffa, 2016) of two morphotypes of
challenges (Abram et al., 2017). Nonetheless, the effect of acute A. equina and their IRTs at two temperatures using a crossed-over
temperature rises on the relationship between ectothermic temperature design, where one temperature was non-stressful and
physiology and behaviour has yet to be investigated. the other simulated a heatwave. We also incorporated a temporal
The beadlet anemone, Actinia equina, lives across a gradient of control where the temperature remained non-stressful throughout.

Journal of Experimental Biology

shore heights in the spatially and temporally heterogeneous Importantly, all experimental individuals had their IRTs and RMRs
intertidal zone (Allcock et al., 1998). Although A. equina is measured at both temperatures, allowing us to use Bayesian
particularly robust to temperature stress (Griffiths, 1977), making it multivariate methodologies to investigate associations between
an ideal organism with which to investigate the effects of thermal IRT and RMR both within and across environments. We expected
perturbation without risk of mortality, the heterogeneous nature of that RMR would show a general increase across all individuals at
its environment leads it to show ranges of thermal preferences high temperatures and that bolder animals, exhibiting shorter IRTs,
within single populations (Navarro et al., 1981). Anemones living would have higher RMRs at both temperatures, as predicted by
further from the low-tide line (higher up the shore), experience POLS. As both metabolic demand and predation risk should be
increased exposure to stressful temperatures (Brahim et al., 2019) exacerbated at high temperatures, which should theoretically
and greater thermal variation than those living lower down intensify any trade-off between the efficiency of gas exchange
the shore (Chapperon et al., 2016). Multiple genetically distinct and foraging, and risk taking, we further predicted that this
morphotypes have developed which favour different shore heights; relationship would be stronger at high temperatures. We also
the red morphotype is associated with the high shore, and is predicted that the degree of change in RMR between temperatures
thus predicted to deal more effectively with temperature extremes would vary at individual and morphotypic levels, with discrepancies
than the green morphotype, which is associated with the low shore being influenced by the red morphotype’s tolerance for higher, more
(Allcock et al., 1998). Morphotypes further display consistent variable temperatures than the green.

2
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

MATERIALS AND METHODS period of 24 h without food before metabolic experimentation

Experimental schedule commenced. After the full 48 h, each group of five anemones was
Data collection was carried out between January and March 2020 first subject to an IRT measurement at ∼17:00 h, using the
over the course of four, 9 day data collection blocks. Each block anemones’ natural response to changes in tides to measure
began with 3 days of anemone, Actinia equina (Linnaeus), boldness. IRTs have previously been shown to be repeatable
collection from rocky sea defences at the top of the beach at New within contexts (r=0.26–0.4), and show significant among-
Brighton, UK (latitude: 53.4400, longitude: −3.0565). Five animals individual variation in how they change across temperatures
were collected on each day, split between the red and green (Maskrey et al., 2020, 2021). Similarly, risk-related tentacle
morphotypes, such that three of one morphotype and two of the extension behaviour has been shown to exhibit both repeatability
other were collected. This uneven split, necessary because of 15 and among-individual variation in plasticity in A. equina and other
being the maximum number of individuals that could be anemone species (Briffa and Greenaway, 2011; Hensley et al., 2012;
investigated per block in our respirometry apparatus, was Rudin and Briffa, 2012). As such, although not taking repeated
randomised and evened out over the course of the study. Overall, measurements of a trait can give rise to statistical concerns in some
30 anemones of each morphotype were used. Collection and cases (Brommer, 2013), a single measurement of IRT was taken
identification of morphotypes were carried out using previously before the commencement of metabolic testing, both because of the
described methods (Maskrey et al., 2020). Anemones were size weight of previous evidence indicating that variation in IRT is
matched at the point of collection as far as was possible by associated with among-individual differences and because of
measuring pedal disc diameter (PDD). Because the raw RMR logistical constraints. Cups containing anemones were removed
(Metcalfe et al., 2016; Velasque and Briffa, 2016) of larger from holding tanks and drained. Anemones were left emersed for
individuals of all species is higher than that of smaller individuals half an hour, before being re-immersed in their holding tank. After
(Glazier, 2005), and thus easier to accurately detect, only anemones re-immersion, IRT, defined as the length of time to re-extend
with a PDD of over 20 mm, the largest individuals in the population, feeding tentacles fully (Maskrey et al., 2020), was recorded by two
were selected. Further, because pilot work found the measurement GoPro Hero 4 (GoPro Inc., San Mateo, CA, USA) cameras mounted
of wet mass to be highly invasive and cause significant damage to directly above the tank, taking time-lapse photographs every 30 s;
anemones, and because the volume of A. equina is not static 50 min of footage was recorded, of which the 45 min immediately
(Griffiths, 1977) so cannot be reliably measured, this size matching after each individual’s immersion was used to determine IRT. The
was also used to minimise any effect of anemone volume on number of photos an anemone took to re-extend its tentacles was
metabolic results. recorded by D.K.M. (blinded to treatment) and converted into
Each group of five anemones was returned to the laboratory, seconds. Anemones which showed no complete response within
again using previously described methods (Maskrey et al., 2020), 45 min were given a maximum value of 2700 s (Maskrey et al.,
and left to acclimate to their surroundings for 48 h. Anemones were 2020).
housed in a 13±1°C temperature-controlled room on a 12 h:12 h After their evening IRT measurement, anemones were gently
day:night cycle. Each anemone was placed in a 7×15 cm plastic cup separated from their pebbles and transferred to one of six freshly
containing drainage holes and a rock to which the anemone could sterilised (with bleach-water and rinsed with fresh water) 14×6 cm,
adhere. Cups were situated in one of two 80×45×40 cm tanks filled 425 ml glass intermittent-flow respirometry chambers sealed
to a depth of ∼12 cm with artificial seawater, at a salinity of with polypropylene lids (IKEA, Älmhult, Sweden) and each
34±1 ppt (RO water with Tropic Marin Pro Reef Salt, Germany). containing a 2.5 g magnetic stir bar with which anemones could
One of these tanks, which housed 10 individuals, was maintained not interact. One of these chambers was designated as the blank, and
at the ambient temperature of 13°C (±1°C), while the other, thus remained empty. For identification purposes, anemones of the
which housed 5 individuals, was maintained at 21°C (±1°C): 13°C same morphotype were grouped in adjacent chambers, but where
was chosen as a suitable non-stressful temperature because it is the morphotypes were, and which chamber was designated as the
well within the normal range of ambient temperatures A. equina blank, was randomised. Chambers were situated in a temperature-
would experience on the shore at New Brighton (Maskrey et al., controlled respirometry tank (67×46×38 cm) filled to a depth
2021); 21°C, meanwhile, falls near the thermal maximum for of 20 cm with freshly made artificial seawater (salinity 34±1 ppt).
this species and is well above the normal temperature range The seawater was heated or chilled to the appropriate temperature
experienced in this region (Maskrey et al., 2020). As such, exposure (±0.3°C) prior to anemone introduction. Chambers were also

Journal of Experimental Biology

to three consecutive days at this temperature was deemed an subject to blank measurements prior to the introduction of
appropriate method by which to simulate the effects of an anemones to ensure that only negligible background respiration
atmospheric heatwave on the intertidal zone (Perkins and was present; where background was detected, chambers were re-
Alexander, 2013). Tanks were situated in two flow-through sterilised to ensure minimal microbial activity. After their
systems, which each contained three tanks. Every tank contained introduction, anemones were left for 12 h overnight to acclimate
anemones at least once and the order in which tanks were used was and attach to their chamber. We used automated intermittent-
randomised across blocks. Across the four blocks, each system was flow respirometry (Svendsen et al., 2016) to measure aquatic RMR,
maintained at each temperature twice, and the order of this was so this 12 h period also served to acclimate anemones to the
randomised. Tanks were retained at a single temperature within intermittent-flow cycle. Measurements could not be conducted
blocks. Water quality (salinity, pH, nitrate, nitrite, ammonia) and throughout the night because of difficulties with having the
water temperature in holding tanks were regularly recorded and full magnetic stir bars move continuously for 12 h. The intermittent-
water changes in each system were carried out before each block. flow cycle consisted of a 3 min water flush period and a 38 min
After 24 h of the 48 h acclimation period, anemones were fed ad closed period where oxygen consumption was measured. This
libitum on defrosted Artemia (Monkfield Nutrition, Ely, UK). This measurement period was shorter than those used in previous
gave anemones time to bind to the substrate and limited variation A. equina respirometry studies (e.g. Navarro et al., 1981), but with
among individuals in levels of satiation, whilst also allowing a this volume of chamber, our pilot investigations found it

3
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

produced sufficient data with which to measure oxygen utilising free software or statistical packages lack flexibility (e.g. lme4:
consumption, while allowing more repeated metabolic Bates et al., 2015), and relatively more flexible frequentist packages
measurements to be taken each day. After this 12 h overnight require paying a subscription (e.g. ASReml-r v.4.2: https://asreml.kb.
acclimation period, experimentation was carried out for 4 h in the vsni.co.uk/knowledge-base/asreml_r_documentation/). By contrast,
morning and 4 h in the afternoon, with a brief (∼10 min) break multiple free, flexible Bayesian packages and software exist for fitting
period between the two where any small bubbles that had formed mixed effects models across multiple response variables or across
were removed from chambers. During trials, aquatic oxygen multiple levels of variance (e.g. MCMCglmm: https://CRAN.R-
concentration within chambers never fell below 80% saturation. project.org/package=MCMCglmm; JAGS: https://mcmc-jags.
Four metabolic slopes of oxygen consumption were produced sourceforge.io/; Stan: Carpenter et al., 2017). By fitting Bayesian
for each anemone at each time of day (for full descriptions models over uninformative priors as we do in this study, study data
and checklist, see Supplementary Materials and Methods, ‘1.1 drive modelled posterior distributions, meaning that the information
Metabolic apparatus and additional measurement detail’ and and estimates contained in those distributions are functionally the
Table S1; Killen et al., 2021). same as those contained in frequentist models (Cleasby et al., 2015;
At the end of their first day of testing, groups were transferred Houslay and Wilson, 2017). Moreover, Bayesian methodologies are
back to one of the two holding tanks and the whole cycle was more robust to small sample sizes, or violations of the assumptions of
repeated, beginning with another 48 h acclimation period. To frequentist modelling approaches (McNeish, 2016). Bayesian
account for temporal variation and treatment order effects, we modelling has become standard practice when investigating
utilised a crossed-over design (Briffa et al., 2013) and a temporal associations between multiple phenotypic traits within or across
control (White and Briffa, 2017). Within blocks, each group of five environments (e.g. Briffa et al., 2013; Jolles et al., 2019; Maskrey
anemones was designated to one of three treatment groups, which et al., 2021; Mitchell et al., 2020), and current advancements in these
were subjected to repeated RMR measurements and a single IRT approaches use predominantly Bayesian frameworks (Mitchell et al.,
measurement at each temperature. Each treatment had an overall 2016; Mitchell and Houslay, 2021; Prentice et al., 2020). We utilised
sample size of 10 red and 10 green individuals: (1) The temporal these methods, exploring associations between traits at multiple
control (L–L) group was housed and tested at 13°C throughout; phenotypic levels (i.e. behavioural, as IRT, and physiological, as
(2) the low temperature to high temperature group (L–H) was first RMR), here.
housed and tested at 13°C, then at 21°C; (3) the high temperature to
low temperature group (H–L) was first housed and tested at 21°C, Calculation of RMR
then at 13°C. Metabolic data were analysed using the respR package (Harianto
After the second round of metabolic testing, anemones were et al., 2019) in R version 3.6.2 or later (http://www.R-project.org/).
placed directly into a −20°C freezer in individually labelled plastic A minimum of two slopes were extracted for every chamber for both
bags, to be stored for later drying and weighing. This method morning and afternoon sets of readings, providing up to six
of euthanasia added an extra step compared with previous studies of measurements of oxygen consumption per chamber, per day.
A. equina, where anemones were dried from live (e.g. Navarro et al., The first 41 min cycles in both the morning and afternoon were
1981; Rudin and Briffa, 2012). Fig. 1 provides a visualisation of this always excluded, as preliminary data indicated that RMR settled
schedule. during the first cycle, such that values were different from
subsequent slopes in a sampling repeat. After a wait period before
Morphological measurement recording of at least 1 min to ensure only the linear portion of slopes
To measure the dry mass of each individual, frozen anemones were were measured (Svendsen et al., 2016), the r 2 of recorded oxygen
placed in a Carbolite (Hope Valley, UK) CWF100 muffle furnace, consumption slopes for all anemones fell above a threshold of
maintained at 110°C for 90 h. Dry anemones were then weighed 0.9. For each repeated measure, slopes from blank chambers were
using a Sartorius (Stonehouse, UK) R2000 balance scale, accurate used to calculate background-adjusted whole-organism oxygen
to the nearest hundredth of a milligram. consumption (RMR) by subtracting blank oxygen consumption
values from the values of each experimental chamber. Raw blank-
Statistical analysis adjusted measurements were converted to provide final whole-
The question of how behavioural and other phenotypic traits covary organism short-term RMR estimates in milligrams of oxygen per
with one another within and across environments is multivariate by hour (mg O2 h−1; Supplementary Materials and Methods, ‘1.2

Journal of Experimental Biology

its nature. Multiple traits can be influenced by the same set of fixed Example slopes and conversion’). Each of these slopes was treated
effects, including environmental variation, while simultaneously as a repeated measure in subsequent analyses.
covarying with one another. While univariate modelling approaches
facilitate investigating single traits (Nakagawa and Schielzeth, Bayesian analyses
2010), they are not appropriate for investigating the associations All individual-level models were run within a Bayesian
between different traits (Houslay and Wilson, 2017), giving rise to Markov chain Monte Carlo framework using the R package
anti-conservative correlation estimates and spurious conclusions. MCMCglmm (https://CRAN.R-project.org/package=MCMCglmm).
Simple correlational analyses using raw, rather than modelled, data See Supplementary Materials and Methods, ‘1.3 Statistical detail’, for
are also of limited value when investigating relationships between details of general model specification and convergence checks,
phenotypic traits as they do not account for the complexity of the additional analytical details, and discussion of a confirmatory
relationships in question (Nakagawa and Schielzeth, 2010). Many analysis into the relationship between RMR and dry mass.
researchers in behavioural ecology and quantitative genetics have
thus turned to multivariate mixed effects modelling to better Repeatability and plasticity of RMR
describe these multivariate questions (Cleasby et al., 2015; Houslay Before investigating relationships between RMR and IRT across
and Wilson, 2017; Mitchell et al., 2016). While it is possible to temperatures, we first sought to confirm the presence of among-
fit multivariate models using a frequentist framework, methods individual variation in RMR and in its plasticity, a pattern previous

4
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

A Testing
13°C Testing 13°C and
freezing
Testing
and
21°C Testing 13°C
freezing
Testing
13°C Testing 21°C and
freezing

Day 1 2 3 4 5 6 7 8 9
Collection of 5 Collection Collection
anemones (am)

B 48 h acclimation 48 h acclimation
(13) (21)
Respirometry Respirometry and freezing
Collection

Day 1 2 3 4 5 6 7
Feeding and PDD IRT (pm) Feeding and PDD IRT (pm)

C Respirometers running, 38 min per measurement, 2 min flush Bleach, water change,
Into respirometer and acclimation overnight (160 min for 4 measurements) blank run Next anemones in

Experimental tank pumps and temp. control

running

Time 0 0 min 3 h am, 3 h pm 2h

Fig. 1. Schematic diagram of the experimental design. (A) A full example block schedule (where grey bars indicate acclimation periods), (B) the schedule
for a low to high temperature (L–H) group (kept initially at 13°C and changed to 21°C on day 4) and (C) an example respirometry schedule. IRT, immersion
response time; PDD, pedal disc diameter.

research has already shown in IRT (Maskrey et al., 2020, 2021). We To investigate whether individual differences in metabolic rate
also sought to investigate the influence of morphotypic differences were consistent across the two temperatures, adjusted cross-context
on RMR. To achieve this, a random slopes model was run (see repeatability was calculated. To test the explanatory importance of
Supplementary Materials and Methods, ‘1.3 Statistical detail’, for individual variation in determining metabolic plasticity to
details of comparison with the equivalent random intercepts model). temperature, the temperature-driven I×E, or random slope effect
RMR (z-transformed) was the response variable and morphotype, estimate, was extracted. Finally, to explore how an individual’s
temperature, treatment, dry mass (z-transformed), sampling metabolic rate at 13°C related to their metabolic plasticity to
occasion, sampling day (i.e. whether it was an individual’s first temperature, the correlation between individual intercepts at 13°C
or second day of metabolic testing), and data collection block and their random slopes between 13°C and 21°C was calculated.
(z-transformed) were all included as fixed effects. Because data Statistical significance was inferred for fixed effects and covariance
collection block was temporally directional, and to improve model terms where 95% credible intervals did not cross zero. For
fits and simplify outputs, it was included in this and subsequent repeatability and random slope effect estimates, significance was
models as a continuous variable. This was denoted by the number of inferred where 95% credible intervals were not close to zero and
days between the start of the first block and the start of the block in histograms of the term’s posterior estimates were not pushed up
question (0, 12, 28 and 41, respectively). z-Transformation of against zero (Maskrey et al., 2020).
continuous variables, which centres data on zero based on how

Journal of Experimental Biology

many standard deviations away from the mean each data point falls, The relationship between RMR and IRT
was carried out in this and subsequent models to improve model We explored several approaches to investigate the relationship
convergence by placing continuous variables on the same scale between RMR and IRT, how it might relate to POLS at different
(Houslay and Wilson, 2017). Temperature was allowed to interact, temperatures, and whether boldness at 13°C or 21°C could be used
separately, with morphotype and treatment order. Previous to predict RMR at 21°C, whilst accounting for censored IRT data.
iterations of univariate random regressions showed no significant Initially, we used a bivariate analysis with the brms package in Stan,
interaction between temperature and dry mass in relation to RMR using a Gaussian distribution for rate measures and a Weibull
(P=0.111). This interaction was thus removed from the models as it distribution for the censored IRT data; however, the software was
was not of direct interest, unlike those between temperature and unable to fit the appropriate model. We were also unable to develop
morphotype or temperature and treatment order. A random slope a model in MCMCglmm that allowed for both a multi-variable
effect, accounting for individual variation and the interaction response and censored observations. As such, we used a Gaussian
between individual identity and metabolic plasticity to temperature distribution for IRT without censoring; whilst there are problems
(individual×environment, I×E), was specified as the random effect. associated with this approach, described below, current approaches
The inclusion of temporal control data in the model, alongside struggle to deal with these data (as explored in Stamps et al., 2012),
including sampling occasion as a fixed effect, allowed robust and the approach used had been previously justified (see Maskrey
control for both treatment order- and time-related effects. et al., 2021).

5
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

Two separate datasets were used. The ‘13°C dataset’ included all 4
values measured at 13°C, including 13°C values from both the
control and crossed-over temperature treatments. The ‘crossover
dataset’ included all values measured in the crossed-over
temperature treatments at either temperature. A bivariate mixed
effects model was run on the 13°C dataset to determine the
covariance between the two traits at 13°C, with RMR and IRT (both

Scaled RMR (mg O2 h−1)

z-transformed) set as response variables. For the crossover dataset, a 2
multivariate model was run where the four response variables were
z-transformed RMR and IRT at each temperature. For both models,
Morphotype
morphotype, dry mass (z-transformed), treatment, sampling
Green
occasion and data collection block were all fixed effects, with Red
individual identity included as a random effect. Residual (within-
individual) covariance was not identifiable, both because IRT and
0
RMR were not measured at the same time in any treatment, and
because only one IRT measurement per individual was modelled
within each temperature for crossed-over treatments. As such,
values within the residual variance–covariance matrices of each
model were therefore fixed at 1. After the appropriate model checks
had been carried out, among-individual correlations between IRT
and RMR within each temperature and between temperatures, −2
alongside associated 95% credible intervals (CIs), were extracted.
13 21
Statistical significance was inferred where 95% CIs did not cross
Temperature (°C)
zero. This analysis was split into a bivariate and a multivariate
component to allow the incorporation of the control data into the Fig. 2. Variation in predicted mean routine metabolic rate (RMR)
13°C correlation estimate. Including control data in the bivariate between red and green morphotypes of the beadlet anemone (Actinia
model was especially important to account for temporal effects, as equina) at 13°C and 21°C. Data are scaled to standard deviation units and
centred on the mean at zero for all individuals (n=60), derived from a
13°C experimental data only covered either the first or second
Bayesian random slopes analysis. Boxes denote the median value, with the
sampling day, depending on treatment. Conversely, while control first and third quartiles forming the box limits; dashed lines denote the mean;
data were not included in the multivariate model, the nature of the whiskers extend to encompass all data or 1.5 times the interquartile range.
crossover dataset, incorporating both sampling days for both Any point falling outside the whiskers can be deemed to be an outlier.
treatments, means that the multivariate model should still have
been robust to both treatment order- and time-related effects.
At the individual level, repeatability estimates indicated that
Ethical note RMR varied consistently among individuals and that some of this
No licences or ethical approval was required to carry out this study variation was retained across temperatures, such that the rank order
as this species is not currently protected under UK legislation. of individual RMRs at 21°C remained partially similar to the rank
Nonetheless, anemones were treated humanely, temperatures to order at 13°C. As such, adjusted cross-context repeatability was low,
which anemones were exposed fell within their tolerance ranges but still significant (Radj=0.18, 95% CI=0.11, 0.25). This result
(Maskrey et al., 2020), and no mortality occurred during the should be taken with some caution, given that while individuals did
experimental periods. show variation in the magnitude of their metabolic slopes within
temperatures, within-temperature measurements may have been
RESULTS subject to some level of temporal autocorrelation (Mitchell et al.,
Repeatability and plasticity of RMR 2020), leading to a lower cross-context repeatability estimate.
There was a general trend across almost all animals for increased Nonetheless, in these data, much of the among-individual variation
RMR at 21°C such that sample mean RMR increased by over 50% in repeated RMR measurements was explained by I×E, translating

Journal of Experimental Biology

from 0.12±0.04 mg O2 h−1 (mean±s.d.) at 13°C to 0.19±0.06 mg into a highly significant, strong random slope effect (estimate=0.70,
O2 h−1 at 21°C. This translated into a significant modelled 95% CI=0.60, 0.80). Fig. 3 shows that individuals differed greatly
temperature effect on RMR, but the size of this effect was in the degree to which their short-term RMR increased as the
estimated with a large degree of uncertainty (estimate=0.98, 95% temperature was raised: 2/40 exhibited lower RMRs at 21°C than at
CI=0.05, 1.95). Both morphotype and dry mass were also significant 13°C and 2/40 showed minimal change between the two
predictors of RMR, and neither of these relationships was affected by temperatures. How the RMR of different individuals changed
temperature. The green morphotype exhibited significantly higher between temperatures was associated with their RMR at 13°C, with
mean RMR than the red morphotype at both temperatures (25% individuals that exhibited lower RMRs at 13°C showing larger
higher at 13°C and 21% higher at 21°C; estimate=−0.34, 95% increases at 21°C than those that exhibited higher RMRs at 13°C. This
CI=−0.59, −0.10), although Fig. 2 does show a wider range of RMR translated into a significant negative correlation between individual
values for both morphotypes at 21°C. Heavier individuals exhibited RMRs at 13°C and the gradient of individual random slopes between
significantly higher RMRs than lighter individuals (estimate=0.41, the two temperatures (r=−0.53, 95% CI=−0.70, −0.21).
95% CI=0.29, 0.53). The relationship between data collection block
and RMR was also on the verge of significance (estimate=−0.12, The relationship between RMR and IRT
95% CI=−0.24, −0.01), such that animals in later blocks had lower RMR was correlated with IRT, and the nature of this correlation was
RMRs than those in earlier blocks. related to environmental temperature such that it did not always

6
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

DISCUSSION
As the climate warms and marine invertebrates experience extreme
2 temperatures with increasing regularity (IPCC, 2013), the survival
of individuals under the greatest metabolic demand at those
Scaled RMR (mg O2 h−1)

temperatures may be placed in jeopardy (Montiglio et al., 2018).

1 The POLS hypothesis (Réale et al., 2010), which predicts positive
RMR quartile
correlations between boldness and metabolic rate, could be used to
High help indicate those individuals most vulnerable to warming, but how
0
Low
Moderately high
these correlations change when individuals are exposed to
Moderately low ecologically relevant heat stress has not been tested. In this study,
we found evidence in beadlet sea anemones for a complex
−1 association between RMR and IRT across temperatures, such that
the correlations between the two did not follow the assumptions of
POLS at a non-stressful temperature of 13°C but did during a
simulated heatwave at 21°C. We further showed that individual
−2
differences explained most of the variance in how RMR changed
13 21 between temperatures. If individuals of certain personality types are
Temperature (°C) behaviourally or physiologically more sensitive to temperature
stress, using more energy or experiencing reduced foraging
Fig. 3. The change in individual-level posterior mean predictions for
RMR between 13°C and 21°C for experimental individuals. Data are opportunities or increased predation risk, then more regular
scaled to standard deviation units and centred on the mean at zero. Each line heatwaves could put those individuals at a selective disadvantage.
corresponds to a single individual’s (n=40) predicted RMR at each At a non-stressful temperature, shyer individuals exhibited higher
temperature, coloured based on their RMR quartiles, compared with other RMRs, a result entirely at odds with POLS. One intuitive
experimental individuals, at 13°C. Predictions are derived from a random explanation for this could be that, at lower temperatures, shyer
regression model across 13°C and 21°C run using the full dataset
individuals were investing more than bolder individuals in
incorporating control individuals (n=60). For visualisation purposes, one
individual that exhibited especially high scaled estimates at both temperatures
metabolically costly maintenance processes such as growth, as
(13°C: 1.54 mg O2 h−1, 21°C: 3.76 mg O2 h−1), was excluded from the plot. suggested by the allocation model of resource management (Biro
et al., 2020; Careau et al., 2019). Although population-wide studies
tend to find faster growth at warmer temperatures in marine
follow the assumptions of POLS. At 13°C, shyer individuals invertebrates (reviewed in Angilletta et al., 2004), this rule is not
(exhibiting longer IRTs) showed higher RMRs than bolder universal (Angilletta and Dunham, 2003; Biro et al., 2020; Careau
individuals, translating into a moderate, significant positive et al., 2019). In A. equina, even very regular feeding regimens are
correlation between IRT and RMR at that temperature (Table 1, not enough to counteract a loss of body mass at high temperatures
Fig. 4A). At 21°C, the pattern of this correlation swapped, such that in laboratory environments (Chomsky et al., 2004b). At low
bolder individuals (with shorter IRTs) at either temperature temperatures, meanwhile, A. equina is able to grow even when fed
exhibited higher RMRs at 21°C than individuals that were shyer only once or twice a week (Chomsky et al., 2004a). As such, while
at either temperature (Figs 4B,D). This correlation, while still of 24 h of fasting should have eliminated the cost of digestion (Navarro
interest, was not significant between IRT and RMR both measured et al., 1981), individuals probably still had flexibility in how they
at 21°C (Table 1, Fig. 4B). It was, however, at the bound of expended their energy at 13°C. This flexibility could have provided
significance between IRT at 13°C and RMR at 21°C, as repeated the scope for different personality types to employ different resource
runs of the model found the lower bound of the 95% CI to hover allocation strategies and for RMR to be positively correlated with
within 0.01 units either side of 0 (Table 1, which shows the results growth (Killen, 2014). Shy individuals may thus have prioritised
of one such run; Fig. 4D). This indicates that an individual’s IRT at growth to a greater degree than bold individuals at low temperatures,
13°C was moderately predictive of its RMR at 21°C. It is worth both because of the greater flexibility in resource allocation that
noting that this significance may have been clearer but for the those temperatures afforded, and because low risk of predation from
smaller effective sample size of the crossover dataset (n=40). There ectothermic predators at those temperatures (e.g. Twardochleb et al.,

Journal of Experimental Biology

was no relationship between RMR at 13°C and IRT at 21°C 2020; Yamane and Gilman, 2009) should have allowed them to
(Table 1, Fig. 4C). forage more freely. Alternatively, the move from holding tanks to
respirometry chambers, or the slight disturbance in the chambers
caused by the intermittent-flow cycle itself, might have caused
disproportionate stress responses in shyer animals as compared
Table 1. Bivariate relationships between immersion response time (IRT) with their bolder conspecifics, driving their higher RMRs at 13°C
and routine metabolic rate (RMR) across different temperatures (e.g. Martins et al., 2011). In turn, shy individuals may already have
Correlation Lower Upper been maximally stressed at that temperature and thus had little
Relationship estimate (r) 95% CI 95% CI capacity to further increase their RMR at higher temperatures.
Future studies could investigate the relationships between growth
IRT 13°C and RMR 13°C 0.34 0.02 0.63
IRT 13°C and RMR 21°C −0.35 −0.68 −0.01 rate, dry mass and IRT to see whether shy individuals do grow faster
IRT 21°C and RMR 13°C 0.04 −0.33 0.45 than bold individuals at non-stressful temperatures, and what
IRT 21°C and RMR 21°C −0.30 −0.65 0.06 implications this has for selection.
CI, credible interval. Significant relationships, and those falling on the bound of In contrast to 13°C, individuals that were bolder at either
significance, are denoted by bold text. Relationships were extracted from one temperature exhibited higher RMRs when under thermal stress than
bivariate and one multivariate Bayesian model. those that were shyer. One possible mechanistic underpinning this

7
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

A B
3 3
***
Scaled RMR at 13°C (mg O2 h−1)

Scaled RMR at 21°C (mg O2 h−1)

2 2

1 1

0 0

−1 −1

−2 −2

−2 −1 0 1 2 −2 −1 0 1 2
Scaled IRT at 13°C (s) Scaled IRT at 21°C (s)

C D
3 3
**
Scaled RMR at 13°C (mg O2 h−1)

Scaled RMR at 21°C (mg O2 h−1)

2 2

1 1

0 0

−1 −1

−2 −2

−2 −1 0 1 2 −2 −1 0 1 2
Scaled IRT at 21°C (s) Scaled IRT at 13°C (s)
Fig. 4. The relationship between individual-level posterior mode estimates (Bayesian BLUPs) for RMR and immersion response time (IRT) at 13°C
and 21°C. Data are scaled to their respective standard deviation units and centred on their respective means at zero, at (A) 13°C for both traits, (B) 21°C for

Journal of Experimental Biology

both traits, (C) 13°C for RMR and 21°C for IRT and (D) 21°C for RMR and 13°C for IRT. ***Statistically significant relationship between the two traits.
**Relationship falling at the bound of statistical significance. Estimates are derived from bivariate (A) and multivariate (B–D) Bayesian mixed effects models.
Error bars denote 95% credible intervals around each of the traits at each temperature.

relationship could have involved shy individuals suppressing their but, as heatwaves become more frequent, suppression strategies
RMRs to a greater degree than bold individuals at stressful high could place individuals at a selective disadvantage by creating
temperatures, in preparation for cooler ‘recovery periods’ in the energetic deficits which they are unable to fulfil during
future (Schulte et al., 2011). Temperature-driven adaptive metabolic progressively shorter recovery periods (Pörtner, 2012). Although
suppression (either limiting the increase or even decreasing they were of varying personality types, those individuals
metabolic rates rapidly by curbing processes with high energy suppressing their RMRs most in our sample could have been the
requirements; Schulte et al., 2011) appears to be specific to intertidal four that failed to increase their RMRs at all at 21°C (e.g.
and shallow-water invertebrates, having been so far documented in Vajedsamiei et al., 2021). It is also possible that these individuals
gastropods (McMahon et al., 1995), molluscs (Vajedsamiei et al., had surpassed their metabolic thermal maximum (Schulte et al.,
2021) and freshwater crustaceans (Glazier et al., 2020). Animals 2011). Although the high RMRs of all four of these individuals at
employing this strategy may be equipped during heatwaves to avoid 13°C might suggest this is plausible, it would seem unlikely that any
their energetic demands exceeding their available energy supply of them had surpassed this threshold given that lethal temperatures

8
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

for other UK populations of A. equina are at least 2°C higher than directly, and incorporate life-history characteristics such as survival
the 21°C used in this experiment (Maskrey et al., 2020). Future or fecundity. This could provide clearer information on not only the
studies could investigate molecular changes associated with vulnerability of different A. equina individuals and genotypes to
metabolic suppression (Richier et al., 2008; Tomanek and heatwaves but also that of other species, as the heterogeneous nature
Somero, 1999) to address these two possibilities directly. of the seashore leads many intertidal ectotherms to display
The effects of individual and morphotypic variation on the intraspecific variation in behavioural and physiological responses
plasticity of RMR could shed further light on its relationship with to stressful high temperatures (Briffa et al., 2013; Chapperon et al.,
IRT. There was a general trend towards higher RMRs at 21°C, 2016; Dias et al., 2018; Maskrey et al., 2021).
which remained unexpectedly consistent between the two Our measures of IRT resulted in censored data, given that there
morphotypes. Importantly, however, as predicted, this response was a maximum cut-off time for shy individuals. Such data are
was not ubiquitous, possibly indicating variation among individuals common among behavioural studies (e.g. Stamps et al., 2012), but
in strategies for dealing with extreme high temperatures. One nevertheless they complicated our analysis of the relationships
potential explanation is that some individuals may compensate for between RMR and IRT. We attempted a variety of approaches
increased RMR induced by higher temperatures by reducing but the ability of current statistical packages to deal with these sorts
activity. This could have led to the observed negative correlation of data were limited (although future developments may provide
between IRT and RMR at 21°C. After considering the available ways forward). Though our final statistical design was able to
evidence, however, we believe that such a compensatory reduction produce appropriate models, the likes of which are utilised in
in activity may be unlikely. Foraging tentacle expansion is the main behavioural studies across the literature, there are concerns over its
energetically costly activity these anemones undertake (Griffiths, ability to deal with censored values (Stamps et al., 2012). As such,
1977) and IRT is thus not only an excellent proxy for boldness but these results may need to be considered with care but we are
also should act as a proxy for activity. As such, if a change in activity confident that they are reflective of the biology of these organisms.
was driving the change in RMR, IRT and RMR at 21°C should have This study shows that associations between RMR and boldness in
been closely correlated, a pattern that is not borne out by the data A. equina are highly temperature dependent, and that variation in
presented here. Indeed, RMR at 21°C was more closely correlated trait plasticity could be part of the reason why evidence for POLS is
with IRT at 13°C than at 21°C, which does not suggest a close so inconsistent. Individual differences were key drivers of how
association between an individual’s current level of activity and RMR changed between temperatures in this study, while morphotypic
its RMR. Previous research also shows differences in how differences remained consistent, suggesting changing associations
morphotypes change their IRTs, and thus their activity, across between RMR and boldness were being driven by individual-level
temperatures (Maskrey et al., 2020). These discrepancies were, plasticity. By measuring the boldness and metabolism of animals
unexpectedly, not reflected in morphotypic patterns of RMR in living in heterogeneous environments under different contexts, and
these data, further indicating that changing activity was not driving adding a life-history component, future work could draw a clearer
changing RMR. Instead, it seems more likely that relationships picture of POLS, gain insights into how different populations might
between IRT and RMR were at least partly driven by boldness- respond to more regular climate change-induced heatwaves, and
associated physiological differences, either via varied growth facilitate a more comprehensive understanding of the adaptive value
strategies at non-stressful temperatures and RMR suppression at of different personality types under climate change.
high temperatures as we speculate, or by other mechanisms.
The patterns elucidated here may have important selective Acknowledgements
implications for A. equina as the frequency of heatwaves Thanks to Leslie Connor for technical support with the muffle furnace, to Guillermo
Garcia-Gomez, Ruth Dunn and Curtis Horne, who assisted with anemone collection,
continues to increase (IPCC, 2013, 2018). At the broader and to Dr Matthew Spencer for invaluable assistance setting up the 21°C
genotypic scale, the red morphotype, associated with the high temperature control. We would also like to thank the reviewers for their valuable
shore, may be at a selective advantage as predicted. Not only do comments, and Dr Simone Blomberg, University of Queensland, for her advice on
individuals of the red morphotype, in contrast to their green the statistical analysis. Portions of this paper are reproduced from the PhD thesis of
counterparts, lower their IRTs (Maskrey et al., 2020, 2021) at high D.K.M. (Maskrey, 2022).

temperatures, which should provide better mitigation against

Competing interests
metabolic challenges under higher energetic demand (Griffiths,
The authors declare no competing or financial interests.
1977; Navarro et al., 1981), but these data show that they also

Journal of Experimental Biology

exhibit lower overall RMRs than green individuals, potentially Author contributions
further accentuating this increased efficiency. At the level of Conceptualization: D.K.M., S.S.K., L.U.S., K.E.A., J.S.T.; Methodology: D.K.M.,
individuals, the directionality of any selective implications is less S.S.K., L.U.S., K.E.A., D.C.C.W., J.S.T.; Software: D.K.M.; Validation: D.K.M.,
clear. It is possible that shy animals exhibit a more robust strategy S.S.K., L.U.S., K.E.A., J.S.T.; Formal analysis: D.K.M.; Investigation: D.K.M.;
Resources: D.K.M., S.S.K., L.U.S., D.C.C.W., J.S.T.; Data curation: D.K.M.; Writing -
than bold ones under current environmental conditions, up-
original draft: D.K.M.; Writing - review & editing: D.K.M., S.S.K., L.U.S., K.E.A.,
regulating their metabolism when not under heat stress to invest in D.C.C.W., J.S.T.; Visualization: D.K.M.; Supervision: S.S.K., L.U.S., K.E.A.,
increased body mass, and conserving energy when they are. Bold D.C.C.W., J.S.T.; Project administration: D.K.M., L.U.S., J.S.T.; Funding acquisition:
individuals, meanwhile, may be at greater risk of perishing during L.U.S., K.E.A., D.C.C.W., J.S.T.
heatwaves, by investing less in growth at lower temperatures and
exposing themselves to greater metabolic requirements and Funding
increased risks of predation when under heat stress. Equally, shy This project was funded by a Natural Environment Research Council ACCE PhD
studentship (ref: 1950009) and by the Blue Planet Aquarium. Open access funding
animals could be at a selective disadvantage if they are unable to provided by University of Liverpool. Deposited in PMC for immediate release.
recoup any energetic deficit incurred during metabolically
supressed periods, which could become increasingly likely as the Data availability
frequency of heatwaves increases and recovery periods become Data and analytical code are available from figshare: https://doi.org/10.6084/m9.
more infrequent. Future work should investigate these possibilities figshare.25071203.v1

9
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

References Griffiths, R. J. (1977). Thermal stress and the biology of Actinia equina
Abram, P. K., Boivin, G., Moiroux, J. and Brodeur, J. (2017). Behavioural effects L. (Anthozoa). J. Exp. Mar. Biol. Ecol. 27, 141-154. doi:10.1016/0022-
of temperature on ectothermic animals: unifying thermal physiology and 0981(77)90134-4
behavioural plasticity. Biol. Rev. 92, 1859-1876. doi:10.1111/brv.12312 Harianto, J., Carey, N. and Byrne, M. (2019). respR – an R package for the
Allcock, A. L., Watts, P. C. and Thorpe, J. P. (1998). Divergence of nematocysts in manipulation and analysis of respirometry data. Methods Ecol. Evol. 10, 912-920.
two colour morphs of the intertidal beadlet anemone Actinia equina. J. Mar. Biol. doi:10.1111/2041-210X.13162
Assoc. U. K. 78, 821-828. doi:10.1017/s0025315400044805 Hensley, N. M., Cook, T. C., Lang, M., Petelle, M. B. and Blumstein, D. T. (2012).
Angilletta, M. J. and Dunham, A. E. (2003). The temperature-size rule in Personality and habitat segregation in giant sea anemones (Condylactis
ectotherms: simple evolutionary explanations may not be general. Am. Nat. 162, gigantea). J. Exp. Mar. Biol. Ecol. 426-427, 1-4. doi:10.1016/j.jembe.2012.05.011
332-342. doi:10.1086/377187 Houslay, T. M. and Wilson, A. J. (2017). Avoiding the misuse of BLUP in
Angilletta, M. J., Steury, T. D. and Sears, M. W. (2004). Temperature, growth rate, behavioural ecology. Behav. Ecol. 28, 948-952. doi:10.1093/beheco/arx023
and body size in ectotherms: fitting pieces of a life-history puzzle. Integr. Comp. IPCC (2013). Summary for policymakers. Climate Change 2013 - The Physical
Biol. 44, 498-509. doi:10.1093/icb/44.6.498 Science Basis. Contribution of Working Group I to the Fifth Assessment Report
Bates, D., Mä chler, M., Bolker, B. M. and Walker, S. C. (2015). Fitting linear mixed- of the Intergovernmental Panel on Climate Change (ed. Intergovernmental
effects models using lme4. J. Stat. Softw. 67, 1-48. doi:10.18637/jss.v067.i01 Panel on Climate Change). Cambridge University Press. doi:10.1017/
Biro, P. A. and Stamps, J. A. (2010). Do consistent individual differences in CBO9781107415324
metabolic rate promote consistent individual differences in behavior? Trends Ecol. IPCC (2018). Summary for Policymakers. Global Warming of 1.5°C. An IPCC
Evol. 25, 653-659. doi:10.1016/j.tree.2010.08.003 Special Report on the impacts of global warming of 1.5°C above pre-industrial
Biro, P. A., Thomas, F., Ujvari, B., Adriaenssens, B. and Beckmann, C. (2020). levels and related global greenhouse gas emission pathways, in the context of
Spontaneous activity rates and resting metabolism: support for the allocation strengthening the global response to the threat of climate change, sustainable
model of energy management at the among-individual level. Ethology 126, 32-39. development, and efforts to eradicate poverty (ed. Intergovernmental Panel on
doi:10.1111/eth.12957 Climate Change). World Meteorological Organization, Geneva, Switzerland,
Brahim, A., Mustapha, N. and Marshall, D. J. (2019). Non-reversible and 32pp. https://www.ipcc.ch/sr15/download/.
reversible heat tolerance plasticity in tropical intertidal animals: responding to Jolles, J. W., Briggs, H. D., Araya-Ajoy, Y. G. and Boogert, N. J. (2019).
habitat temperature heterogeneity. Front. Physiol. 9, 1909. doi:10.3389/fphys. Personality, plasticity and predictability in sticklebacks: bold fish are less plastic
2018.01909 and more predictable than shy fish. Anim. Behav. 154, 193-202. doi:10.1016/j.
Briffa, M. and Greenaway, J. (2011). High in situ repeatability of behaviour indicates anbehav.2019.06.022
animal personality in the beadlet anemone Actinia equina (Cnidaria). PLoS One 6, Killen, S. S. (2014). Growth trajectory influences temperature preference in fish
e21963. doi:10.1371/journal.pone.0021963 through an effect on metabolic rate. J. Anim. Ecol. 83, 1513-1522. doi:10.1111/
Briffa, M., Bridger, D. and Biro, P. A. (2013). How does temperature affect 1365-2656.12244
behaviour? Multilevel analysis of plasticity, personality and predictability in hermit Killen, S. S., Marras, S., Ryan, M. R., Domenici, P. and McKenzie, D. J. (2012).
crabs. Anim. Behav. 86, 47-54. doi:10.1016/j.anbehav.2013.04.009 A relationship between metabolic rate and risk-taking behaviour is revealed
Brommer, J. E. (2013). On between-individual and residual (co)variances in the during hypoxia in juvenile European sea bass. Funct. Ecol. 26, 134-143. doi:10.
study of animal personality: are you willing to take the “individual gambit”? Behav. 1111/j.1365-2435.2011.01920.x
Ecol. Sociobiol. 67, 1027-1032. doi:10.1007/s00265-013-1527-4 Killen, S. S., Marras, S., Metcalfe, N. B., McKenzie, D. J. and Domenici, P.
Careau, V., Beauchamp, P. P., Bouchard, S. and Morand-Ferron, J. (2019). (2013). Environmental stressors alter relationships between physiology and
Energy metabolism and personality in wild-caught fall field crickets. Physiol. behaviour. Trends Ecol. Evol. 28, 651-658. doi:10.1016/j.tree.2013.05.005
Behav. 199, 173-181. doi:10.1016/j.physbeh.2018.11.023 Killen, S. S., Christensen, E. A. F., Cortese, D., Zá vorka, L., Norin, T., Cotgrove,
Carpenter, B., Gelman, A., Hoffman, M. D., Lee, D., Goodrich, B., Betancourt, L., Crespel, A., Munson, A., Nati, J. J. H., Papatheodoulou, M. et al. (2021).
M., Brubaker, M. A., Guo, J., Li, P. and Riddell, A. (2017). Stan: a probabilistic Guidelines for reporting methods to estimate metabolic rates by aquatic
programming language. J. Stat. Softw. 76, 1. doi:10.18637/jss.v076.i01 intermittent-flow respirometry. J. Exp. Biol. 224, jeb242522. doi:10.1242/jeb.
Chapperon, C., Volkenborn, N., Clavier, J., Sé ité , S., Seabra, R. and Lima, F. P. 242522
(2016). Exposure to solar radiation drives organismal vulnerability to climate: Le Galliard, J. F., Paquet, M., Cisel, M. and Montes-Poloni, L. (2013). Personality
evidence from an intertidal limpet. J. Therm. Biol. 57, 92-100. doi:10.1016/j. and the pace-of-life syndrome: variation and selection on exploration, metabolism
jtherbio.2016.03.002 and locomotor performances. Funct. Ecol. 27, 136-144. doi:10.1111/1365-2435.
Chomsky, O., Kamenir, Y., Hyams, M., Dubinsky, Z. and Chadwick-Furman, 12017
N. E. (2004a). Effects of feeding regime on growth rate in the Mediterranean Sea Le Lann, C., Wardziak, T., van Baaren, J. and van Alphen, J. J. M. (2011).
anemone Actinia equina (Linnaeus). J. Exp. Mar. Biol. Ecol. 299, 217-229. doi:10. Thermal plasticity of metabolic rates linked to life-history traits and foraging
1016/j.jembe.2003.09.009 behaviour in a parasitic wasp. Funct. Ecol. 25, 641-651. doi:10.1111/j.1365-2435.
Chomsky, O., Kamenir, Y., Hyams, M., Dubinsky, Z. and Chadwick-Furman, 2010.01813.x
N. E. (2004b). Effects of temperature on growth rate and body size in the Martins, C. I. M., Castanheira, M. F., Engrola, S., Costas, B. and Conceiçao, ̃
Mediterranean Sea anemone Actinia equina. J. Exp. Mar. Biol. Ecol. 313, 63-73. L. E. C. (2011). Individual differences in metabolism predict coping styles in fish.
doi:10.1016/j.jembe.2004.07.017 Appl. Anim. Behav. Sci. 130, 135-143. doi:10.1016/j.applanim.2010.12.007
Cleasby, I. R., Nakagawa, S. and Schielzeth, H. (2015). Quantifying the Maskrey, D. (2022). The behavioural, metabolic and proteomic effects of
predictability of behaviour: statistical approaches for the study of between- temperature stress on bold and shy beadlet anemones (Actinia equina). PhD
individual variation in the within-individual variance. Methods Ecol. Evol. 6, 27-37. thesis, University of Liverpool.
doi:10.1111/2041-210X.12281 Maskrey, D. K., Sneddon, L. U., Arnold, K. E., Wolfenden, D. C. C. and
Dall, S. R. X., Houston, A. I. and McNamara, J. M. (2004). The behavioural ecology Thomson, J. S. (2020). The impact of personality, morphotype and shore height
of personality: consistent individual differences from an adaptive perspective. on temperature-mediated behavioural responses in the beadlet anemone Actinia

Journal of Experimental Biology

Ecol. Lett. 7, 734-739. doi:10.1111/j.1461-0248.2004.00618.x equina. J. Anim. Ecol. 89, 2311-2324. doi:10.1111/1365-2656.13301
Dias, G. M., Christofoletti, R. A., Kitazawa, K. and Jenkins, S. R. (2018). Maskrey, D. K., Sneddon, L. U., Arnold, K. E., Wolfenden, D. C. C. and
Environmental heterogeneity at small spatial scales affects population and Thomson, J. S. (2021). Temperature-driven changes in behavioural
community dynamics on intertidal rocky shores of a threatened bay system. unpredictability and personality in the beadlet sea anemone, Actinia equina.
Ocean Coast. Manag. 164, 52-59. doi:10.1016/j.ocecoaman.2017.12.001 Anim. Behav. 181, 13-27. doi:10.1016/j.anbehav.2021.08.022
Dingemanse, N. J. and Wolf, M. (2013). Between-individual differences in McMahon, R. F., Russell-Hunter, W. D. and Aldridge, D. W. (1995). Lack of
behavioural plasticity within populations: causes and consequences. Anim. metabolic temperature compensation in the intertidal gastropods, Littorina
Behav. 85, 1031-1039. doi:10.1016/j.anbehav.2012.12.032 saxatilis (Olivi) and L. obtusata (L.). Hydrobiologia 309, 89-100. doi:10.1007/
Edmunds, M., Potts, G. W., Swinfen, R. C. and Waters, V. L. (1974). The feeding 978-94-011-0435-7_9
preferences of Aeolidia papillosa (L.) (Mollusca, Nudibranchia). J. Mar. Biol. McNeish, D. (2016). On using Bayesian methods to address small sample
Assoc. U. K. 54, 939-947. doi:10.1017/S0025315400057660 problems. Struct. Equ. Modeling 23, 750-773. doi:10.1080/10705511.2016.
Edmunds, M., Potts, G. W., Swinfen, R. C. and Waters, V. L. (1976). Defensive 1186549
behaviour of sea anemones in response to predation by the opisthobranch Metcalfe, N. B., Van Leeuwen, T. E. and Killen, S. S. (2016). Does individual
mollusc Aeolidia papillosa (L.). J. Mar. Biol. Assoc. U. K. 56, 65-83. doi:10.1017/ variation in metabolic phenotype predict fish behaviour and performance? J. Fish
s0025315400020440 Biol. 88, 298-321. doi:10.1111/jfb.12699
Glazier, D. S. (2005). Beyond the “3/4-power law”: variation in the intra- and Mitchell, D. J. and Houslay, T. M. (2021). Context-dependent trait covariances:
interspecific scaling of metabolic rate in animals. Biol. Rev. Camb. Philos. Soc. 80, how plasticity shapes behavioral syndromes. Behav. Ecol. 32, 25-29. doi:10.1093/
611-662. doi:10.1017/S1464793105006834 beheco/araa115
Glazier, D. S., Borrelli, J. J. and Hoffman, C. L. (2020). Effects of fish predators on Mitchell, D. J., Fanson, B. G., Beckmann, C. and Biro, P. A. (2016). Towards
the mass-related energetics of a keystone freshwater crustacean. Biology 9, 40. powerful experimental and statistical approaches to study intraindividual variability
doi:10.3390/biology9030040 in labile traits. R. Soc. Open Sci. 3, 160352. doi:10.1098/rsos.160352

10
RESEARCH ARTICLE Journal of Experimental Biology (2024) 227, jeb244662. doi:10.1242/jeb.244662

Mitchell, D. J., Dujon, A. M., Beckmann, C. and Biro, P. A. (2020). Temporal Seebacher, F., White, C. R. and Franklin, C. E. (2015). Physiological plasticity
autocorrelation: a neglected factor in the study of behavioral repeatability and increases resilience of ectothermic animals to climate change. Nat. Clim. Change
plasticity. Behav. Ecol. 31, 222-231. doi:10.1093/beheco/arz180 5, 61-66. doi:10.1038/nclimate2457
Montiglio, P. O., Dammhahn, M., Dubuc Messier, G. and Ré ale, D. (2018). The Segev, U., Burkert, L., Feldmeyer, B. and Foitzik, S. (2017). Pace-of-life in a social
pace-of-life syndrome revisited: the role of ecological conditions and natural insect: behavioral syndromes in ants shift along a climatic gradient. Behav. Ecol.
history on the slow-fast continuum. Behav. Ecol. Sociobiol. 72, 1-9. doi:10.1007/ 28, 1149-1159. doi:10.1093/beheco/arx079
s00265-018-2526-2 Sih, A., Bell, A. and Johnson, J. C. (2004). Behavioral syndromes: an ecological
Nakagawa, S. and Schielzeth, H. (2010). Repeatability for Gaussian and non- and evolutionary overview. Trends Ecol. Evol. 19, 372-378. doi:10.1016/j.tree.
Gaussian data: a practical guide for biologists. Biol. Rev. 85, 935-956. doi:10. 2004.04.009
Smith, B. R. and Blumstein, D. T. (2008). Fitness consequences of personality:
1111/j.1469-185X.2010.00141.x
a meta-analysis. Behav. Ecol. 19, 448-455. doi:10.1093/beheco/arm144
Navarro, E., Ortega, M. M. and Madariaga, J. M. (1981). Effect of body
Snell-Rood, E. C. (2013). An overview of the evolutionary causes and
size, temperature and shore level on aquatic and aerial respiration of Actinia
consequences of behavioural plasticity. Anim. Behav. 85, 1004-1011. doi:10.
equina (L.) (Anthozoa). J. Exp. Mar. Biol. Ecol. 53, 153-162. doi:10.1016/0022-
1016/j.anbehav.2012.12.031
0981(81)90016-2 Stamps, J. A. (2016). Individual differences in behavioural plasticities. Biol. Rev. 91,
Parmesan, C. (2006). Ecological and evolutionary responses to recent climate 534-567. doi:10.1111/brv.12186
change. Ann. Rev. Ecol. Evol. Syst. 37, 637-669. doi:10.1146/annurev.ecolsys. Stamps, J. A., Briffa, M. and Biro, P. A. (2012). Unpredictable animals: individual
37.091305.110100 differences in intraindividual variability (IIV). Anim. Behav. 83, 1325-1334. doi:10.
Perkins, S. E. and Alexander, L. V. (2013). On the measurement of heat waves. 1016/j.anbehav.2012.02.017
J. Clim. 26, 4500-4517. doi:10.1175/JCLI-D-12-00383.1 Svendsen, M. B. S., Bushnell, P. G., and Steffensen, J. F. (2016). Design and
Polverino, G., Santostefano, F., Dı́az-Gil, C. and Mehner, T. (2018). Ecological setup of intermittent-flow respirometry system for aquatic organisms. J. Fish Biol.
conditions drive pace-of-life syndromes by shaping relationships between life 88, 26-50. doi:10.1111/jfb.12797
history, physiology and behaviour in two populations of Eastern mosquitofish. Sci. Tomanek, L. and Somero, G. N. (1999). Evolutionary and acclimation-induced
Rep. 8, 14673. doi:10.1038/s41598-018-33047-0 variation in the heat-shock responses of congeneric marine snails (genus Tegula)
Pö rtner, H. O. (2012). Integrating climate-related stressor effects on marine from different thermal habitats: implications for limits of thermotolerance and
organisms: unifying principles linking molecule to ecosystem-level changes. Mar. biogeography. J. Exp. Biol. 202, 2925-2936. doi:10.1242/jeb.202.21.2925
Ecol. Prog. Ser. 470, 273-290. doi:10.3354/meps10123 Twardochleb, L. A., Treakle, T. C. and Zarnetske, P. L. (2020). Foraging strategy
Prentice, P. M., Houslay, T. M., Martin, J. G. A. and Wilson, A. J. (2020). Genetic mediates ectotherm predator–prey responses to climate warming. Ecology 101,
variance for behavioural ‘predictability’ of stress response. J. Evol. Biol. 33, 1-11. doi:10.1002/ecy.3146
642-652. doi:10.1111/jeb.13601 Vajedsamiei, J., Wahl, M., Schmidt, A. L., Yazdanpanahan, M. and Pansch, C.
Ré ale, D., Garant, D., Humphries, M. M., Bergeron, P., Careau, V. and Montiglio, (2021). The higher the needs, the lower the tolerance: extreme events may select
P. O. (2010). Personality and the emergence of the pace-of-life syndrome concept ectotherm recruits with lower metabolic demand and heat sensitivity. Front. Mar.
at the population level. Philos. Trans. R. Soc. B: Biol. Sci. 365, 4051-4063. doi:10. Sci. 8, 264. doi:10.3389/fmars.2021.660427
Velasque, M. and Briffa, M. (2016). The opposite effects of routine metabolic rate
1098/rstb.2010.0208
and metabolic rate during startle responses on variation in the predictability of
Richier, S., Rodriguez-Lanetty, M., Schnitzler, C. E. and Weis, V. M. (2008).
behaviour in hermit crabs. Behaviour 153, 1545-1566. doi:10.1163/1568539X-
Response of the symbiotic cnidarian Anthopleura elegantissima transcriptome to
00003371
temperature and UV increase. Comp. Biochem. Physiol. Part D: Genomics
White, S. J. and Briffa, M. (2017). How do anthropogenic contaminants (ACs) affect
Proteomics 3, 283-289. doi:10.1016/j.cbd.2008.08.001 behaviour? Multi-level analysis of the effects of copper on boldness in hermit
Royauté , R., Berdal, M. A., Garrison, C. R. and Dochtermann, N. A. (2018). crabs. Oecologia 183, 391-400. doi:10.1007/s00442-016-3777-0
Paceless life? A meta-analysis of the pace-of-life syndrome hypothesis. Behav. Wolf, M. and McNamara, J. M. (2012). On the evolution of personalities via
Ecol. Sociobiol. 72, 1-10. doi:10.1007/s00265-018-2472-z frequency-dependent selection. Am. Nat. 179, 679-692. doi:10.1086/665656
Rudin, F. S. and Briffa, M. (2012). Is boldness a resource-holding potential trait? Wolf, M. and Weissing, F. J. (2010). An explanatory framework for adaptive
Fighting prowess and changes in startle response in the sea anemone, Actinia personality differences. Philos. Trans. R. Soc. B: Biol. Sci. 365, 3959-3968.
equina. Proc. R. Soc. B 279, 1904-1910. doi:10.1098/rspb.2011.2418 doi:10.1098/rstb.2010.0215
Schulte, P. M., Healy, T. M. and Fangue, N. A. (2011). Thermal performance Yamane, L. and Gilman, S. E. (2009). Opposite responses by an intertidal predator
curves, phenotypic plasticity, and the time scales of temperature exposure. Integr. to increasing aquatic and aerial temperatures. Mar. Ecol. Prog. Ser. 393, 27-36.
Comp. Biol. 51, 691-702. doi:10.1093/icb/icr097 doi:10.3354/meps08276

Journal of Experimental Biology

11