Access to this full-text is provided by The Royal Society.
Content available from Royal Society Open Science
This content is subject to copyright.
royalsocietypublishing.org/journal/rsos
Research
Cite this article: Mathot KJ, Arteaga-Torres JD,
Wijmenga JJ. 2022 Individual risk-taking behaviour
in black-capped chickadees (Poecile atricapillus)
does not predict annual survival. R. Soc. Open Sci. 9:
220299.
https://doi.org/10.1098/rsos.220299
Received: 9 March 2022
Accepted: 6 July 2022
Subject Category:
Organismal and evolutionary biology
Subject Areas:
behaviour/ecology
Keywords:
animal personality, risk-taking, survival,
foraging behaviour
Author for correspondence:
Kimberley J. Mathot
e-mail: mathot@ualberta.ca
Electronic supplementary material is available
online at https://doi.org/10.6084/m9.figshare.c.
6108476.
Individual risk-taking
behaviour in black-capped
chickadees (Poecile atricapillus)
does not predict annual
survival
Kimberley J. Mathot
1,2
, Josue D. Arteaga-Torres
1
and
Jan J. Wijmenga
1
1
Department of Biological Sciences, and
2
Canada Research Chair in Integrative Ecology,
University of Alberta, Edmonton, Alberta, Canada T6G 2E9
KJM, 0000-0003-2021-1369; JDA-T, 0000-0001-5025-6109;
JJW, 0000-0002-3817-2153
Within species, individuals often show repeatable differences
in behaviours, called ‘animal personality’. One behaviour
that has been widely studied is how quickly an individual
resumes feeding after a disturbance, referred to as boldness
or risk-taking. Depending on the mechanism(s) shaping risk-
taking behaviour, risk-taking could be positively, negatively,
or not associated with differences in overall survival. We
studied risk-taking and survival in a population of free-living
black-capped chickadees (Poecile atricapillus) in which we
previously showed repeatable among-individual differences
in risk-taking over the course of several months. We found
no evidence that variation in risk-taking is associated with
differences in annual survival. We suggest that variation
in risk-taking is likely shaped by multiple mechanisms
simultaneously, such that the net effect on survival is small
or null. For example, among-individual differences in energy
demand may favour greater risk-taking without imposing an
overall mortality cost if higher energy demand covaries with
escape flight performance. We propose directions for future
work, including using a multi-trait, multi-year approach to
study risk-taking, to allow for stronger inferences regarding
the mechanisms shaping these behavioural decisions.
1. Introduction
Within populations, individuals often exhibit repeatable differences
in behaviour, referred to as animal personality [1]. For example,
© 2022 The Authors. Published by the Royal Society under the terms of the Creative
Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits
unrestricted use, provided the original author and source are credited.
within populations, individuals vary consistently in how quickly they return to baseline levels of activities
following manipulations of perceived predation risk (e.g. [2–7]). This behaviour is often referred to as
‘boldness’or ‘risk-taking’. Understanding the causes and consequences of animal personality is a major
area of research in contemporary behavioural and evolutionary ecology and models of adaptive animal
personality variation often invoke state-dependent behavioural decisions and/or trade-offs to explain
the maintenance of such variation within populations [8–11].
Among-individual differences in risk-taking may arise due to trade-offs and/or state-dependent
payoffs. For example, responsiveness to predators outside of the breeding season may trade off
different components of mortality risk; risk of mortality due to starvation versus risk-of mortality due
to predation such that variation in risk-taking results in similar overall survival. However, the
resolution of these trade-offs may also be shaped by an individual’s state, in which case, differences in
risk-taking may be associated with differences in fitness. Individuals with intrinsically lower
vulnerability to predators may behave more boldly in response to predators and therefore increase
resource acquisition without incurring higher risks of mortality due to predation [12]. For example, in
prey with gape-limited predators, larger individuals may be intrinsically less vulnerable [12], while in
birds, individuals with larger pectoral muscles may be less vulnerable due to greater escape flight
performance [13]. In this case, higher risk-taking would be associated with higher survival. On the
other hand, individuals with intrinsically lower risk of starvation (e.g. low metabolic rates) may
exhibit relatively low boldness compared to individuals with intrinsically higher risk of starvation
(e.g. high metabolic rates) to achieve similar realized risk of starvation, while at the same time having
lower risk of mortality due to predation [6]. Given that the consequences of boldness for survival are
likely to depend not only on context (e.g. food and/or predator abundance), but also depend on
aspects of an individual’s state (e.g. body condition, ability to evade predators, etc.), it is perhaps not
surprising that studies that have attempted to evaluate survival consequences of among-individual
differences in boldness or risk-taking have yielded mixed results [14–16].
In an earlier study, we showed that response to manipulations of perceived predation risk in
free-living black-capped chickadees (Poecile atricapillus) vary markedly both within and among
individuals [2]. Within individuals, variation in risk-taking was shaped by ambient temperatures, with
lower temperature generally favouring shorter latencies to resume feeding [2]. This effect was
presumably because lower temperatures (below thermoneutrality) increase energy expenditure in
small birds [17], favouring faster resumption of feeding behaviour. Individuals also exhibited
repeatable differences in risk-taking. After experimental manipulations of perceived predation risk,
some individuals resumed feeding within minutes, while others resumed feeding after hours, and
these among-individual differences were repeatable for at least several months over the non-breeding
season [2]. Here, we evaluate whether among-individual differences in response to experimental
manipulations of perceived predation risk are associated with differences in feeding rates and annual
survival and evaluate support for the three non-exclusive mechanisms outlined above: (i) differences
in allocation to starvation versus predation avoidance, (ii) differences in intrinsic vulnerability to
predation and (iii) differences in intrinsic vulnerability to starvation. To do this, we quantified
foraging and risk-taking in 79 black-capped chickadees over the course of a winter, and subsequently
tracked their survival to the following winter. We include feeding rates prior to the experimental
manipulations of perceived predation risk because foraging can also be viewed as a form of risk-
taking in that it provides access to food resources while exposing individuals to greater risk [18].
Thus, we expected feeding rates and latency to resume feeding following a disturbance to be
negatively correlated (i.e. higher feeding rates associated with shorter latencies). Further, if these are
both expressions of risk-taking, then higher feeding rates and shorter latencies to resume feeding
should have similar survival consequences.
2. Methods
2.1. Study site and study population
This study was carried out in a marked population of black-capped chickadees at the University of
Alberta Botanic Garden (UABG) in Devon, Alberta, Canada (53 °2 402 700 N, 113 °4 504 100 W). The
study area is 97 hectares: 32 hectares of display gardens and 65 hectares of mixed forest. A marked
population of black-capped chickadees was established in the fall of 2017 and catching and banding
are conducted annually to maintain the marked population. Chickadees were caught using mist-nets
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220299
2
placed near feeders. Upon capture, any birds that had not previously been caught were fitted with
Canadian Wildlife Service aluminium bands for unique identification. Two short-behavioural assays
were carried out as part of another study totally less than 4 min; a field cage exploration test
(following the protocol outlined in [19]) and handling aggression tests (following the protocol
outlined in [20]). Following these two tests, standard morphometric data were collected (e.g. tarsus,
bill length, bill depth, wing length), body mass was recorded, and a small blood sample was collected
from the brachial vein to allow for molecular sexing [21]. As part of another study aimed at assessing
the effects of passive integrated transponder (PIT) tags and methods on chickadees, from 2017 to
2019, birds were randomly assigned to receive no PIT tag, a PIT tag attached a colour band, or a PIT
tag implanted subcutaneously [22]. Because we relied on radio frequency identification (RFID) to
detect birds at feeders (see below), the data presented here are only for chickadees that were fitted
with leg band PIT tags, as the implanted PIT tags were found to have unreliable detectability (see [2]
for further details). Comparisons of observations of 353 feeder visits by chickadees with leg band
embedded PIT tags from video recordings against RFID registrations confirmed that the leg band
embedded PIT tags were registered with 100% reliability [2].
2.2. Foraging and risk-taking behaviour
We analysed within- and among-individual differences in both foraging (i.e. visit rates to feeders prior to
a disturbance) and risk-taking (i.e. latency to resume feeding following a disturbance). Within-individual
changes (i.e. behavioural plasticity) refer to changes in behaviour expressed by the same individual in
different instances, while among-individual differences refer to differences in the average behaviour
expressed by different individuals. By analysing both feeding rates and latency to resume feeding
following a disturbance, we were able to assess whether associations between risky behaviours and
survival were generalizable, or context-dependent (i.e. whether they depend on the degree of
perceived risk).
Foraging behaviour and risk-taking were assessed in experiments that took place between November
2018 and March 2019 at eight feeder locations spread throughout the study site. Feeders were placed at
least 270 m apart based on previously reported winter territory sizes from other populations of black-
capped chickadees [23], with the aim of providing a single feeder per winter flock (see electronic
supplementary material, figure S1). Feeders were filled with black oil sunflower seeds and equipped
with an RFID antenna around the feeder opening that is connected to a circuit board with internal
clock and SD card for data storage. Details on the feeders and RFID system are provided elsewhere
[2], but briefly, when a bird with a leg band embedded PIT tag visits the feeder, its unique
transponder code, and the date and time of the visit are registered to the SD card. Throughout the
experiments, feeders were visited every 4 days to top up the seeds, to replace the batteries and to
collect the data that had been written to the SD cards. These visits to the feeder were always
conducted on non-experimental days so that they did not create disturbances that might interfere with
interpretation of treatment effects.
Experiments were conducted using a 2 × 2 factorial design using different combinations of acoustic
(yes/no) and visual (yes/no) cues of predation resulting in a total of four treatment combinations.
The acoustic cue was comprised of mobbing calls of chickadees recorded in another population
(located ca 40 km from the current study population) produced in response to merlin (Falco
columbarius) mounts. Eight unique 1 h files were created that were made up of alternating sequences
of mobbing calls (ranging from 5 to 20 s in length and comprised of the mobbing calls of between 1
and 4 chickadees, repeated over 1 min periods) and bouts of silence (ranging from 60 to 180 s). Each
of the eight unique files included the same range of flock sizes in the mobbing bouts (1–4). The
sequence files were played back using portable speakers (Shockwave, FoxPro, Lewistown, PA, USA)
that were placed on top of a pole 3 m in front of the feeder. The volume at which the calls were
broadcast could be heard up to distances of approximately 80 m (J.D.A.-T., 2018, personal observation).
Further details about the recordings are provided in Arteaga-Torres et al. [2]. We used six different
taxidermic mounts of juvenile merlin as our visual cue of predation risk. During presentation, the
mount was carried to the focal feeder in a plastic box and then removed from the box and placed
on a pole that was 3 m in front of the feeder (as with the speaker during acoustic treatments) with the
mount facing the feeder (see electronic supplementary material, figure S2). The height of the pole was
such that the mount lined up with the height of the feeder opening. These mounts were visible at
distances ranging from around 20 to 50 m (J.D.A.-T., 2018, personal observation). Merlin were selected
as the model predator because they specialize in small birds, including chickadees [24], and are
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220299
3
present in the study area throughout the winter (based on records in the eBird digital repository: https://
ebird.org/species/merlin). Our control treatments for both the acoustic and visual cues of predation risk
were designed to control for the non-biological components of our experimental treatments. As such, the
control for the acoustic playback consisted of the presence of the speaker (not broadcasting sound). The
control for the visual cue consisted of the presence of the pole placed near the feeder, but without a
taxidermic mount. All treatments were 1 h in duration. This treatment length was chosen to allow for
sufficient time for birds to experience the treatment, but was intended to be short enough to prevent
habituation. Previous analyses found no evidence of habituation by chickadees to the treatments [2].
We used a stratified random design to assign treatments to feeders such that (i) each experimental
day a maximum of one treatment was carried out at any given feeder and (ii) each experimental day,
each of the four treatment levels was carried out (i.e. one control, one acoustic, one visual and one
combined). A complete replicate consisted of each of the four treatments being carried out at each of
the eight feeders. Treatment start times were 09:30, 11:00, 12:30 and 14:00. Within each experimental
day, the order of the treatments was randomized. To minimize potential carry-over and/or
habituation effects of our treatments, we only conducted treatments (experimental days) every second
day during any given replicate, with at least 7 days break between successive replicates. This meant
that within a given replicate, three 1 h long predator presentations (one visual, one acoustic and one
visual + acoustic) plus 1 h control treatment occurred at a single feeder over the course of 15 days
(8 days to complete the replicate + 7 days between replicates). We carried out four complete replicates
of the experiment at each of the eight feeders.
Only birds that were present at feeders in the 1 h immediately preceding a treatment were included in
the analysis. This was done to increase the likelihood that birds detected at a feeder following a given
treatment had been in the vicinity of the feeder when the treatment was carried out, and therefore,
likely to have experienced the treatment. We used the count of feeder visits in the hour immediately
preceding the treatment as a measure of feeding rate (visits/h). We define risk-taking as the latency to
resume feeding following the presentation of a treatment at a feeder (i.e. the time in seconds from the
start of the treatment to the first visit by each individual to the feeder), with shorter latencies
corresponding to greater risk-taking.
2.3. Survival data
We monitored detections of the 79 birds (47 males and 32 females) for which we obtained feeding rate
and risk-taking measures at RFID equipped feeders over the subsequent 2 years to allow us to estimate
annual survival. Chickadees are non-migratory and form stable winter flocks [23]. Surviving birds
remain part of the same winter flock in subsequent years, and occupy the same core winter territory
[23]. As such, we used re-detections in the study area as a proxy for survival. Birds that were not
detected in the year after the collection of the behavioural data (N= 35) were also not detected in the
following year, suggesting that they were not simply transiently absent from the study area. We used
detection (yes/no) in the year following collection of behavioural data as our estimate of annual survival.
2.4. Data analysis
We analysed the within- and among-individual correlations between foraging rate and latency to resume
feeding after manipulations of perceived predation risk-taking using a bivariate mixed effects models
using the R package MCMCglmm [25] in the R Statistical Environment [26] using the R Studio
interface [27]. We constructed a two-trait model with feeding rate (continuous variable), and latency
to resume feeding (continuous variable) as response variables. Foraging rate and latency to resume
feeding were both log transformed prior to analysis so that model residuals met the assumption of
normality. We opted to transform these variables rather than construct models with Poisson error
distributions because the latter resulted in models that were overdispersed (results not shown). We
included sex as a fixed effect for both foraging rate and latency to resume feeding because in
chickadees males are dominant over females [23], which might logically be expected to affect both
foraging and risk-taking decisions. We additionally modelled the effect of temperature because it has
previously been shown to affect both foraging and risk-taking in our population [2], presumably
because temperatures below thermoneutrality increase energy demand. Daily average temperature,
obtained from a nearby weather station (Edmonton International Airport Weather Station, 10 KM SE
of field site, https://agriculture.alberta.ca/acis), was included as a fixed effect for hourly feeding
rates. We only modelled the treatment effect (categorical variable with four levels), and its interaction
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220299
4
with temperature, for the latency to resume feeding as the foraging data was collected prior to treatments
(which were assigned using a stratified random design) and therefore, logically could not have been
affected by treatment. The temperature was centred and standardized prior to analyses. We included
individual ID as a random effect, allowing us to quantify the pairwise among-individual relationships
between foraging rate, latency to resume feeding, and survival. We did not include feeder ID as a
random effect as earlier analyses showed it to be of minor importance [2]. To ensure model
convergence, we ran three chains with 103000 iterations with a burnin of 3000 and a thinning of 100
to generate 1000 estimates. Posterior density plots were inspected visually to ensure proper model
mixing and convergence (see electronic supplementary material, figure S3). Results presented in the
text below are from a representative chain using a parameter expanded prior. However, we verified
that results were robust to modest changes in the prior specifications (result not shown). Results are
presented as the mean and 95% credible interval (CrI) of the 1000 estimates from a single chain. The
point estimates and 95% CrIs were used to evaluate support for a given effect (see below). We
calculated adjusted repeatability following Nakagawa & Schielzeth [28] as: [among-individual
variance]/[among-individual variance + residual variance].
Although we initially attempted to include survival as a third trait in the multivariate model to estimate
the among-individual covariance between foraging (continuous trait with repeated measures), latency
(continuous trait with repeated measures) and survival (binary trait with a single measure per
individual), we were unable to achieve good model convergence. This was also true when constructing
bivariate models for behaviour (either foraging or latency) and survival. Thus, to quantify the among-
individual correlation between behaviour and survival, we instead ran univariate generalized linear
models (glms) of survival (yes/no) as a function of the best linear unbiased predictors (BLUPs) of
behaviour derived from the bivariate model described above. To account for the uncertainty in BLUPs,
we ran each glm of survival 1000 times using an estimate drawn from the distribution of BLUPs for
each behavioural trait [29]. The 1000 estimated effects sizes of behaviour on survival were used to
derive an estimated effect size and 95% CrI for the relationship between behaviour (foraging rate or
latency) on annual survival. We additionally investigated sex-related differences in survival using a
general linear model with survival (yes/no) as the response variable and sex as the predictor. All glms
with survival as a fixed effect were fitted with a binomial error distribution.
We describe estimates with CrIs that did not overlap zero as providing strong support for an effect,
while estimates that were centred on zero are described as providing no support for an effect, or strong
support for lack of an effect. For estimates that were not centred on zero, but whose CrIs overlapped zero,
we calculated Bayesian p-values based on the proportion of counts of estimates that were above or below
zero, depending on the direction of the estimated mean. Estimates with Bayesian p-values less than 0.15
are referred to as showing moderate support for an effect because this corresponds to more than five
times greater support for the interpretation of an effect compared to the interpretation of no effect.
3. Results
Of the 79 birds for which we quantified risk-taking behaviour in the winter of 2018/2019, 44 were
detected using the feeders the following winter (2019/2020). The 35 birds that were not detected
using the feeder the following winter were assumed to have died. This reflects an annual survival rate
of approximately 56%.
Feeding rate varied as a function of both temperature and sex (table 1). As temperature increased (i.e.
conditions became less energetically challenging), foraging rates decreased (temperature effect: β=−0.14,
95% CrI = −0.19, −0.08). Males also exhibited higher feeding rates compared to females (sex effect: β=
0.26, 95% CrI = 0.02, 0.47). Further, there was significant among-individual variation in feeding rates
(r= 0.22, 95% CrI = 0.14, 0.34). Analyses of the response to different cues of predation is presented in
detail elsewhere [2]; however, our multivariate models reproduce the key findings of no evidence for
sex-related differences in latency to resume feeding, and longer latencies to resume feeding with
increased ambient temperatures for most treatments (table 1). As previously reported [2], there was
also significant among-individual variation in latency to resume feeding (r= 0.22, 95% CrI = 0.15,
0.37). Feeding rate and latency to resume feeding after manipulations of perceived predation risk were
negatively correlated within-individuals (r=−0.18, 95% CrI = −0.24, −0.12), presumably due to the
common effect of ambient temperature. Foraging and latency to resume feeding were also strongly
negatively correlated at the among-individual (r=−0.50, 95% CrI = −0.76, −0.25, figure 1).
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220299
5
We found no support for an effect of either feeding rate (log odds ratio = −0.16, 95% CrI = −0.86, 0.38)
or latency to resume feeding (log odds ratio = −0.07, 95% CrI = −0.86, 0.38) on annual survival (figures 1
and 2). Although both point estimates were negative, the confidence intervals overlapped zero
substantially (foraging: Bayesian p-value = 0.33; latency: Bayesian p-value = 0.40), indicating no
support for an effect. There was moderate support for a sex effect on survival (Bayesian p-value =
0.10), with males (log odds ratio = 0.48, 95% CrI = −0.10, 1.08) tending to have higher annual survival
compared with females (log odds ratio = −0.13, 95% CrI = −0.83, 0.57).
4. Discussion
We previously showed repeatable among-individual differences in the latency to resume feeding after
exposure to predator cues (risk-taking) in the same population of chickadees [2]. Here, we asked
whether this variation in risk-taking was associated with variation in foraging rates and annual survival
to evaluate support for three potential non-exclusive mechanisms underlying variation in risk-taking:
differences in allocation to starvation versus predation avoidance, differences in vulnerability to
starvation, and/or differences in vulnerability to predation. Although latency to resume feeding and
foraging were strongly negatively correlated both within and among individuals, there was no support
for a correlation between either risk-taking or foraging on annual survival. Below, we discuss the
implications and limitations of these results, and provide directions for future work.
Both foraging and risk-taking allow individuals to acquire resources while exposing them to risk [18].
Therefore, we expected to observe a correlation between these two traits, which we did. Individuals that
had higher foraging rates on average in the observation period prior to our predator treatments also had
shorter latencies to resume feeding following predator treatments. We found no support for the
Table 1. Sources of variation in feeding rates (feeder visits per hour) prior to manipulations, and latency to resume feeding
following manipulation of perceived predation risk (seconds). Estimates derived from MCMCglmm model with log feeding rate
and log latency to resume feeding as response variables. See main text for full model details.
fixed effects
log feeding rate (visits per hour) log latency to resume feeding (seconds)
β(95% CI) β(95% CI)
intercept
a
−0.16 (−0.34, −0.01) 0.34 (0.15, 0.53)
sex
b
0.26 (−0.03, 0.48) −0.01 (−0.22, 0.20)
control treatment
c
n.a. −0.70 (−0.85, −0.56)
acoustic treatment
c
n.a. −0.56 (−0.69, −0.40)
visual treatment
c
n.a. 0.06 (−0.06, 0.21)
temperature
d
−0.14 (−0.19, −0.08) n.a.
temperature
d
: control n.a. 0.15 (0.04, 0.24)
temperature
d
: acoustic n.a. 0.11 (0.01, 0.22)
temperature
d
: visual n.a. −0.01 (−0.11, 0.09)
temperature
d
: acoustic + visual n.a. 0.11 (−0.01, 0.22)
random effects σ(95% CI) σ(95% CI)
individual ID 0.19 (0.12, 0.28) 0.17 (0.10, 0.25)
residual 0.80 (0.76, 0.87) 0.72 (0.66, 0.78)
repeatability r (95% CI) r (95% CI)
individual
e
0.22 (0.14, 0.34) 0.19 (0.13, 0.26)
a
Intercept estimated average temperature over study period and for female sex. For latency to resume feeding data, the intercept
was estimated during the highest risk treatment (i.e. acoustic + visual).
b
Sex effect (reference category = female): estimates difference between males and females.
c
Treatment effects relative to highest risk treatment (i.e. reference category = acoustic plus visual).
d
Temperature, centred and standardized.
e
Adjusted repeatability estimated after taking into account fixed effects.
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220299
6
interpretation that among-individual differences in either foraging or risk-taking were associated with
variation in annual survival. Studies that quantify the relationships between behaviour and survival at
the among-individual level, rather than at the phenotypic level, are relatively rare [30]. Nonetheless,
our result is in line with two recent meta-analyses that broadly find little support that among-
individual differences in foraging behaviour [14], or risk-taking [14,15], are associated with predictable
differences in survival.
One explanation for our finding that neither foraging or risk-taking are associated with differences in
overall survival is that variation in foraging and risk-taking reflects differences in allocation to avoidance
of starvation versus predation. For example, individuals that invest more in starvation avoidance have
higher feeding rates and shorter latencies to resume feeding on average, but this does not translate to
a net survival benefit because it comes at the cost of increased mortality due to predation. Proper
evaluation of this possibility would require data on the sources of mortality, which we do not have.
However, we suggest it is unlikely that this can fully account for our results. There is ample evidence
that among-individual differences in state variables shape foraging and risk-taking. For example,
across a range of taxa, higher energetic needs and/or lower nutritional status are associated with
increased foraging and risk-taking [31,32]. We show that within-individual variation in foraging and
risk-taking in our population are shaped by within-individual variation in energetic needs, as inferred
from ambient temperature. Given that among-individual differences in energetic needs are near
ubiquitous in animals [33], it seems likely that among-individual variation in energetic needs would
similarly shape among-individual differences in foraging and risk-taking in our population. Such
state-dependent behaviour often results in individuals achieving different fitness outcomes [11], yet
we found no effect of variation in either foraging or risk-taking on annual survival. How can this be?
We suggest that the lack of relationship between either foraging or risk-taking and annual survival
may arise because the expression of these behaviours is shaped by multiple state-dependent
–1
–1.0
–5.0
0
0.5
1.0
0
feeding rate (BLUP ± 95% CrI)
survive
0
1
latency to resume feeding (BLUP ± 95% CrI)
1
Figure 1. Among-individual correlations between feeding rate, latency to resume feeding after predator cues. Each point represents
an individual’s mean BLUP derived from the bivariate model. Whiskers denote 95% CrIs. Annual survival outcome is illustrated by
colour: survived = black circles, died = red triangles. There is a strong negative correlation between feeding rate and latency to
resume feeding, but no evidence that either trait predicts survival outcome.
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220299
7
mechanisms simultaneously. For example, if vulnerability to predation and vulnerability to starvation act
together to shape the expression of foraging and risk-taking behaviour, we may not expect a net effect of
either behaviour on overall survival. This is because under among-individual differences in vulnerability
to predation (e.g. due to differences in feather condition, flight muscle size, etc.), increased foraging
or risk-taking would be associated with lower risk of starvation, while under among-individual
differences in starvation (e.g. due to differences in metabolic rate, priority access to feeders, etc.)
it would be associated with increased risk-of mortality due to predation. This could occur if
vulnerability to predation and energy requirements were themselves linked, for example,
if individuals with higher energy requirements also have greater ability to evade predators. Such a
pattern would be predicted under the ‘performance’model of metabolism, whereby individuals with
higher energetic requirements are also able to exhibit higher expression of performance related traits
such as escape behaviour [34]. Although there is general support for performance models in various
animal taxa [31], the relationship has yet to be tested in this system. Nonetheless, at least one study
provides some suggestion that this may be the case. In another population of chickadees, individuals
that carried more fat also had larger pectoral muscles [35]. Fat stores are sometimes maintained as a
buffer against starvation risk associated with high energy requirements [36], and pectoral muscle is
important for powering rapid escape flights [13]. Therefore, the correlation between fat stores and
pectoral muscle mass reported previously is suggestive of a potential correlation between energy
demand and vulnerability to predation, though this is speculative, and requires further testing.
We also observed trait-specific sex effects. Males had significant higher feeding rates and higher
annual survival compared to females, but there was no evidence of sex-related differences in latency
to resume feeding following a manipulation of perceived risk. These results can be understood in
light of the fact that in black-capped chickadees, males are structurally larger and are dominant over
females [23,37]. Males are therefore expected to have higher energy requirements (due to their large
size) but are also expected to be able to maintain priority access to the best feeding sites. Higher
0
0
1
survival
10 20 30
foraging rate (visits/hour)
40 50
0
0
1
survival
latency to resume feeding (seconds)
30 000
20 00010 000
Figure 2. Distribution of foraging rates (a) and latency to resume feeding (b) as a function of annual survival (0 = no, 1 = yes).
Each dot represents an observation, with multiple observations per individual. The outer bounds of the boxes indicate the 25th and
75th interquartile range, and the centre line is the median value. Whiskers denote 10th and 90th percentiles.
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220299
8
dominance is associated with higher survival in small birds in winter in general [38], and previous
studies in black-capped chickadees have also reported higher survival in males [39,40]. If latency to
resume feeding after a manipulation of perceived risk was solely due to differences in energy
requirements, then we would predict males to return sooner than females. The fact that they did not
is consistent with previous work showing that subordinates take relatively more risk in resuming
feeding following encounters with predators to offset their generally lower access to food [36,41]. The
simultaneous effect of high energy demands in males and low priority access to food in females may
lead to a lack of overall sex-effect on latency to resume feeding.
Taken together, our results show that foraging behaviour and risk-taking share common drivers (e.g.
temperature), but can also vary independently (e.g. sex affected foraging but not risk-taking). Although
we found no support for the interpretation that either foraging rate or latency to resume feeding were
associated with differences in annual survival (after controlling for sex effects), we suggest that this
pattern may reflect the net effect of multiple mechanisms shaping the expression of foraging and risk-
taking, and their relationship to annual survival simultaneously. Many models for adaptive animal
personality are based on state-dependence and predict non-equal fitness outcomes for different
behavioural types [8,11]. We suggest that it will often be the case that the expression of a given
behaviour is simultaneously shaped by multiple state variables whose fitness consequences may
cancel out. Studies that simultaneously consider multiple traits are needed to tease apart the relative
contributions of different mechanisms. For example, future studies of among-individual differences in
risk-taking could aim to quantify individual state variables related to energy requirements and
vulnerability to predation, simultaneously. In birds, this could mean measuring metabolic rates and
escape flight performance, respectively. Additionally, extending studies over multiple years could
provide important insights if the relative contributions of starvation versus predation to overall
mortality rates fluctuates across years, which would be expected under variable food and/or predator
abundances. Using a multi-trait, multi-year approach to study risk-taking decisions will allow for a
more holistic understanding of the mechanisms shaping these behavioural decisions, which is
necessary to understand the consequences of variation in these behaviours on survival.
Ethics. This work was carried out under permits to K.J.M. and J.J.W. for catching and banding chickadees from the Bird
Banding Office in Canada (banding permits 10277 AK, 10277 AL, 10936 and 10936A), permits from the University of
Alberta Biosciences Animal Care and Use Committee (ACUC) (permits AUP00002542 and AUP00002210), and
Environment Canada Canadian Wildlife Service Scientific permit (no. 13-ABSC004), and Alberta Fish and Wildlife
Capture and Research permits (nos. 56066, 56065, 19-056).
Data accessibility. All data and code required to reproduce the results and figure in this manuscript are available on the
Open Science Framework digital repository: https://osf.io/p3hz4/. Electronic supplemental material is available
online [42].
Authors’contributions. K.J.M.: conceptualization, data curation, formal analysis, funding acquisition, investigation,
supervision, visualization, writing—original draft, writing—review and editing; J.D.A.-T.: conceptualization, data
curation, funding acquisition, investigation, writing—review and editing; J.J.W.: conceptualization, investigation,
methodology, writing—review and editing.
All authors gave final approval for publication and agreed to be held accountable for the work performed therein.
Conflict of interest declaration. At the time of writing, K.J.M. was a Board Member of Royal Society Open Science, but had no
involvement in the review or assessment of the paper.
Funding. This field work was supported by an ACA Biodiversity Grant to JDAT and University of Alberta Startup funds,
an NSERC Discovery grant no. (RGPIN-2018-04358), and CRC Research grant to K.J.M.
Acknowledgements. Thank you to Sheeraja Sridharan for molecular sexing of the birds, and Elene Haave Audet for help
with field work. We thank the Wildbird General Store in Edmonton for providing the sunflower seeds. Four
anonymous referees provided helpful feedback on an earlier version of the manuscript. We would especially like to
thank referee 4 for suggesting figure 2 and providing the r-script to generate it.
References
1. Bell AM, Hankison SJ, Laskowski KL. 2009 The
repeatability of behaviour: a meta-analysis.
Anim. Behav. 77, 771–783. (doi:10.1016/j.
anbehav.2008.12.022)
2. Arteaga-Torres JD, Wijmenga JJ, Mathot KJ.
2020 Visual cues of predation risk outweigh
acoustic cues: a field experiment in black-
capped chickadees. Proc. R. Soc. B 287,
20202002. (doi:10.1098/rspb.2020.2002)
3. Brodie III ED, Russell NH. 1999 The consistency
of individual differences in behaviour:
temperature effects on antipredator behaviour
in garter snakes. Anim. Behav. 57, 445–451.
(doi:10.1006/anbe.1998.0990)
4. Johnson JC, Sih A. 2007 Fear, food, sex and
parental care: a syndrome of boldness in the
fishing spider, Dolomedes triton. Anim. Behav.
74, 1131–1138. (doi:10.1016/j.anbehav.2007.
02.006)
5. Kralj-Fišer S, Scheiber IBR, Blejec A, Moestl E,
Kotrschal K. 2007 Individualities in a flock of
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220299
9
free-roaming greylag geese: behavioural and
physiological consistency over time and across
situations. Horm. Behav. 51, 239–248. (doi:10.
1016/j.yhbeh.2006.10.006)
6. Mathot KJ, Nicolaus M, Araya-Ajoy YG,
Dingemanse NJ, Kempenaers B. 2015 Does
metabolic rate predict risk-taking behaviour?
A field experiment in a wild passerine bird.
Funct. Ecol. 29, 239–249. (doi:10.1111/1365-
2435.12318)
7. van Oers K, Drent PJ, De Goede P, van
Noordwijk AJ. 2004 Realized heritability and
repeatability of risk-taking behaviour in relation
to avian personalities. Proc. R. Soc. Lond. B 271,
65–73. (doi:10.1098/rspb.2003.2518)
8. Dingemanse NJ, Wolf M. 2010 Recent models
for adaptive personality differences: a review.
Phil. Trans. R. Soc. B 365, 3947–3958. (doi:10.
1098/rstb.2010.0221)
9. Wolf M, Kraus J. 2014 Why personality
differences matter for social functioning and
social structure. Trends Ecol. Evol. 29, 306–308.
(doi:10.1016/j.tree.2014.03.008)
10. Wolf M, McNamara JM. 2012 On the evolution
of personalities via frequency-dependent
selection. Am. Nat. 179, 679–692. (doi:10.1086/
665656)
11. Wolf M, Weissing FJ. 2010 An explanatory
framework for adaptive personality differences.
Phil. Trans. R. Soc. B 365, 3959–3968. (doi:10.
1098/rstb.2010.0215)
12. Luttbeg B, Sih A. 2010 Risk, resources and
state-dependent adaptive behavioural
syndromes. Phil. Trans. R. Soc. B 365,
3977–3990. (doi:10.1098/rstb.2010.0207)
13. van den Hout PJ, Piersma T, Dekinga A, Lubbe
SK, Visser GH. 2006 Ruddy turnstones Arenaria
interpres rapidly build pectoral muscle after
raptor scares. J. Avian Biol. 37, 425–430.
(doi:10.1111/j.0908-8857.2006.03887.x)
14. Haave Audet EG, Besson AA, Nakagawa S,
Mathot KJ. 2021 Differences in resource
acquisition, not allocation, mediate the
relationship between behaviour and fitness:
a systematic review and meta-analysis. Biol. Rev.
97, 708–731. (doi:10.1111/brv.12819)
15. Moiron M, Laskowski KL, Niemelä PT. 2020
Individual differences in behaviour explain
variation in survival: a meta-analysis. Ecol. Lett.
23, 399–408. (doi:10.1111/ele.13438)
16. Smith BR, Blumstein DT. 2008 Fitness
consequences of personality: a meta-analysis.
Behav. Ecol. 19, 448–455. (doi:10.1093/beheco/
arm144)
17. Broggi J, Hohtola E, Koivula K, Orell M,
Thomson RL, Nilsson J-Å. 2007 Sources of
variation in winter basal metabolic rate in the
great tit. Funct. Ecol. 21, 528–533. (doi:10.
1111/j.1365-2435.2007.01255.x)
18. Houston AI, McNamara JM, Hutchinson JMC.
1993 General results concerning the trade-off
between gaining energy and avoiding
predation. Phil. Trans. R. Soc. B 341, 375–397.
(doi:10.1098/rstb.1993.0123)
19. Stuber EF, Araya-Ajoy YG, Mathot KJ, Mutzel A,
Nicolaus M, Wijmenga JJ, Mueller JC,
Dingemanse NJ. 2013 Slow explorers take less
risk: a problem of sampling bias in ecological
studies. Behav. Ecol. 24, 1092–1098. (doi:10.
1093/beheco/art035)
20. Dubuc-Messier G, Réale D, Perret P, Charmantier
A. 2017 Environmental heterogeneity and
population differences in blue tits personality
traits. Behav. Ecol. 28, 448–459. (doi:10.1093/
beheco/arw148)
21. Griffiths R, Double MC, Orr K, Dawson RJG. 1998
A DNA test to sex most birds. Mol. Ecol. 7,
1071–1075. (doi:10.1046/j.1365-294x.1998.
00389.x)
22. Farr JJ, Haave-Audet E, Thompson PR, Mathot KJ.
2021 No effect of passive integrated transponder
tagging method on survival or body condition in
a northern population of black-capped chickadees
(Poecile atricapillus). Ecol. Evol. 14,9610–9620.
(doi:10.1002/ece3.7783)
23. Smith SM. 1992 The black-capped chickadee:
behavioral ecology and natural history. Ithaca,
NY: Cornell University Press.
24. Templeton CN, Greene E, Davis K. 2005
Allometry of alarm calls: black-capped chicadees
encode information about predator size. Science
308, 1934–1937. (doi:10.1126/science.1108841)
25. Hadfield J. 2010 MCMC Methods for multi-
response generalized linear mixed models: the
MCMCglmm R Package. J. Stat. Softw. 33,1–22.
(doi:10.18637/jss.v033.i02)
26. R Development Core Team. 2020 R Development
Core Team. R: A language and environment for
statistical computing (version 4.0.2). Vienna,
Austria: R Foundation for Statistical Computing.
See http://www.R-project.org.
27. R Studio Team. 2020 RStudio: integrated
development for R. Boston, MA: RStudio, PBC.
Retrieved from http://www.rstudio.com/
28. Nakagawa S, Schielzeth H. 2010 Repeatability
for Gaussian and non-Gaussian data: a practical
guide for biologists. Biol. Rev. 85, 935–956.
(doi:10.1111/j.1469-185X.2010.00141.x)
29. Hadfield JD, Wilson AJ, Garant D, Sheldon BC,
Kruuk LEB. 2010 The misuse of BLUP in ecology
and evolution. Am. Nat. 175, 116–125. (doi:10.
1086/648604)
30. Dingemanse NJ, Araya-Ajoy YG, Westneat DF.
2021 Most published selection gradients are
underestimated: why this is and how to
fix it. Evolution 75, 806–818. (doi:10.1111/
evo.14198)
31. Mathot KJ, Dingemanse NJ, Nakagawa S.
2019 The covariance between metabolic
rate and behaviour varies across behaviours
and thermal types: meta-analytic insights.
Biol. Rev. 94, 1056–1074. (doi:10.1111/
brv.12491)
32. Moran NP, Sanchez-Tojar A, Schielzeth H,
Reinhold K. 2021 Poor nutritional condition
promotes high-risk behaviours: a systematic
review and meta-analysis. [Review]. Biol. Rev.
96, 269–288. (doi:10.1111/brv.12655)
33. Nespolo RF, Franco M. 2007 Whole-animal
metabolic rate is a repeatable trait: a meta-
analysis. J. Exp. Biol. 210, 2000–2005. (doi:10.
1242/jeb.02780)
34. Careau V, Thomas DK, Humphries MM, Réale D.
2008 Energy metabolism and animal
personality. Oikos 117, 641–653. (doi:10.1111/j.
0030-1299.2008.16513.x)
35. Lewden A, Petit M, Vézina F. 2012 Dominant
black-capped chickadees pay no maintenance
energy costs for their wintering status and are
not better at enduring cold than subordinate
individuals. J. Comp. Physiol. B 182, 381–392.
(doi:10.1007/s00360-011-0625-8)
36. Dall SRX. 2010 Managing risk: the perils of
uncertainty. In Evolutionary behavioral ecology
(eds DF Westneat, CW Fox), pp. 194–206.
Oxford, UK: Oxford University Press.
37. Desrochers A. 1989 Sex, dominance, and
microhabitat use in wintering black-capped
chickadees: a field experiment. Ecology 70,
636–645. (doi:10.2307/1940215)
38. Lahti K. 1998 Social dominance and survival in
flocking passerine birds: a review with an
emphasis on the Willow Tit Parus montanus.
[Review]. Ornis Fennica 75,1–17.
39. Desrochers A, Hannon SJ, Kelly EN. 1988 Winter
survival and territory acquisition in a northern
population of black-capped chickadees. The Auk
105, 727–736. (doi:10.1093/auk/105.4.727)
40. Smith SM. 1995 Age-specific survival in
breeding black-capped chickadees (Parus
atricapillus). [Article]. Auk 112, 840–846.
(doi:10.2307/4089016)
41. Ekman J. 1987 Exposure and time use in willow
tit flocks: the cost of subordination. Anim.
Behav. 35, 445–452. (doi:10.1016/S0003-
3472(87)80269-5)
42. Mathot KJ, Arteaga-Torres JD, Wijmenga JJ.
2022 Individual risk-taking behaviour in black-
capped chickadees (Poecile atricapillus) does not
predict annual survival. Figshare. (doi:10.6084/
m9.figshare.c.6108476)
royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220299
10
Available via license: CC BY 4.0
Content may be subject to copyright.
