Abstract Using social information can benefit individuals in many ways. Responding to alarm signals can, for instance, maximize survival under predation risk. However, foraging individuals should consider the reliability of such risk-based information to balance antipredator behavior and resource acquisition. Receiver decisions could depend on personality effects, as individual variation in risk-taking tendencies (i.e., boldness) could not only affect receiver perception of the signaled threat but also signaler reliability. Recent theoretical models support the possibility of coevolution between personality and communication strategies. Using a playback experiment, we show that wild eastern chipmunks (Tamias striatus) respond to alarm calls according to their own boldness level (measured as consistent individual differences in basal vigilance) and that they increase their vigilance response to bolder callers potentially considered as more reliable. Further, receivers respond to the callers’ boldness regardless of their own boldness and independently of their familiarity level with callers, therefore decoding this information from vocalizations. Such effects of individual behavioral variation on the perception and interpretation of social information could apply to signals used in a variety of ecological contexts. INTRODUCTION Individuals can enhance decision-making by using social information (Danchin et al. 2004; Dall et al. 2005). In various ecological contexts such as mating, foraging, and predator avoidance, individuals benefit from using information acquired from others to assess and respond to their environment and thus maximize their fitness (Giraldeau et al. 2002; Dall et al. 2005). Assessing sexual signals can for instance improve reproductive success in the context of mate choice (Andersson 1994). In a foraging context, individuals can also use social information to enhance their foraging efficiency and thus their long-term survival (by observing other individuals’ discovery of resources or through information exchange; Ward and Zahavi 1973; Danchin et al. 2004; Dall et al. 2005), but they also need to constantly trade off resource acquisition against safety. Many animals therefore engage in antipredator behaviors that can maximize short-term survival under predation risk (Lima and Dill 1990). Responding to alarm signals from conspecifics can therefore be highly beneficial for immediate survival (Blumstein 2007). However, although paying attention to trustworthy signals is worthwhile (Searcy and Nowicki 2005), using socially acquired information can become costly if the reliability of the signal is not ensured (Giraldeau et al. 2002). Responsive behaviors therefore remain adaptive only if they are adjusted to the accuracy of the signals. Devoting less importance to social information that comes from unreliable signalers becomes critical to optimize the trade-off between antipredator behavior and resource acquisition (Pollard 2011). This is highly relevant for alarm communication systems in which the accuracy of the information transferred regarding risk depends greatly on the signalers’ capacity to accurately evaluate the presence of a danger (Pollard 2011). The discrimination of unreliable signalers has been reported in a few alarm-call systems of social mammals, by testing differences in response to callers that differ in their reliability. Reliability was classically considered in relation to familiarity (through social relationship and spatial proximity to the receiver, Hare 1998; O’Connell-Rodwell et al. 2007; Hollén and Radford 2009) and age (less reliable juveniles are expected to be ignored, Blumstein and Daniel 2004; Hollén and Radford 2009), or even modified through artificial manipulation (Cheney and Seyfarth 1988; Hare and Atkins 2001; Blumstein et al. 2004). The reliability of public information can also be assessed using the number of signalers asserting that a threat is present (Sloan and Hare 2008). Although receivers would ideally adjust their response to signalers of varying reliability (Pollard 2011), the origin of variation in signaler reliability in natural alarm communication systems remains largely unclear (Hare and Warkentin 2012). This is central, however, to our understanding of why, how and when individuals use social information and how they interpret it. It is thus worth considering that an individual’s propensity to emit alarm calls could depend not only on the presence of kin (Maynard-Smith 1965; Blumstein et al. 1997) or on its social status (Fuong et al. 2015), but also on its intrinsic stress response and risk-sensitivity. Low thresholds of emission are linked to high production of stress hormones (high stress reaction along the HPA axis) (Blumstein et al. 2006), and alarm signals emitted by overly risk-sensitive individuals should particularly arouse suspicion in receivers as they could be false alarms produced in reaction to unthreatening environmental stimuli, due to individuals’ “skittishness” (Pollard 2011), “jumpiness” (Searcy and Nowicki 2005) or “low threshold of excitation” (Hare and Atkins 2001). Receivers could thus truly adjust their responses to the so-called personality of the signalers, as inherent individual risk-taking profiles are known to vary consistently within a population, along a shyness-boldness continuum (Sloan Wilson et al. 1994, Réale et al. 2007). The concept of animal personality (Réale et al. 2007) has recently shed light on many aspects of animal evolutionary ecology (Wolf and Weissing 2012) through its individual-based approach, and consistent individual differences in behavior could be closely linked to intrinsic physiological constraints (POLS concept, Réale et al. 2010). In a stressful risky situation, less proactive shy individuals would respond with high concentration of stress hormones (Clary et al. 2014), and these high risk-sensitive individuals could be expected to have low thresholds of reaction for emission of alarm signals and for response to the alarms. Personality could thus be one of the critical drivers of variation in responses to signals, an idea strengthened by recent theoretical models which highlight how personality and communication can coevolve (McNamara et al. 2009; Botero et al. 2010; Wolf et al. 2011). Polymorphism in communication strategies is predicted to naturally emerge for both signalers and receivers, and the existence of individual variation in the production and interpretation of signals could maintain consistent individual differences in behavior (Botero et al. 2010). Behavioral consistency could also be favored by individual differences in responsiveness and reliability (McNamara et al. 2009; Wolf et al. 2011). Boldness, one of the most commonly studied personality traits, could thus play a significant role in differential response to alarm signals. Receivers’ reaction could be affected by their own boldness, but also be adjusted to the boldness of the alarm signalers, as shy signalers may appear less reliable. Personality could therefore affect individuals’ decisions not only through their tendency to use the available social information (Amy et al. 2010; Kurvers et al. 2010) but also through their interpretation of the signals, which could depend on both their own personality and that of the signaler. In this study, we examine variation in receivers’ reaction to conspecific alarm calls beyond any expected effect of familiarity, to determine whether the boldness of the receiver and that of the alarm caller affect these reactions. We also investigate if individuals’ boldness influences the degree to which individuals plastically use social information on caller boldness, while assessing if the estimation of such information depends on the receiver’s familiarity with the caller. To test these hypotheses, we conducted a field playback experiment with a free-ranging population of individually-marked eastern chipmunks (Tamias striatus). Our playback experiment integrated multiple levels of natural individual variation, as we repeatedly measured the reaction of several receivers to alarm calls from different callers varying in boldness and in their level of familiarity to the receiver. This design offers the opportunity to disentangle the complexity of social information use and interpretation in a natural alarm communication system. MATERIAL AND METHODS Study system Eastern chipmunks are ground-dwelling Sciurids (Elliot 1978) generally considered as solitary, but they produce different types of alarm calls (Elliot 1978; Burke da Silva et al. 1994) that elicit antipredation vigilance in conspecifics (Weary and Kramer 1995). The alarm communication system in this species may have evolved through kin selection (Maynard-Smith 1965; Burke da Silva et al. 2002), as female kin show spatial clustering (Elliot 1978; Messier et al. 2012). Chipmunks also show consistent individual differences in behaviors (e.g., exploration, docility), which are linked to many facets of their life-history strategies (Montiglio et al. 2012; Dammhahn et al. 2017). The study was conducted on 2 sites, separated by approximately 1 km, in a deciduous forest in Southern Québec, Canada (45°06.785’N; 72°26.078’W). Free-ranging chipmunks were identified and sexed through trapping, and we attached colored flags on their ear tags for recognition from a distance. To facilitate observations, individuals were attracted at temporary feeding patches (supplied with a handful of sunflower seeds mixed with leaves on the ground) established 10 m away from their burrows (previously found using visual observation and radio-telemetry), while observers were placed 10 m away, in the direction opposite to the burrow. Field recording and selection of alarm calls We based our playback experiment on “chucking” calls, which are typically emitted in response to aerial threats and consist of low frequency notes repeated within a calling bout (see Figure 1b and Burke da Silva et al. 1994). These vocalizations, along with occasional flight-associated multi-note calls (“trills”) (Burke da Silva et al. 1994), were primarily elicited under simulated aerial predation risk by experimentally throwing a hat above a foraging individual (see Figure 1a and Hare 1998; Hare and Atkins 2001), and recorded using a directional microphone (Sennheiser ME67/K6, protected by a Rycote Windshield kit 7) connected to a portable digital recorder (Marantz PMD661, sampling frequency of 96 kHz, with Sennheiser HD 280 Pro headphones to monitor the audio recording). From these original recordings, we first selected 20 calls from 15 callers (3 calls were recorded in 2013 and 17 in 2014) through a screening process (using Avisoft-SASLab Pro version 5.0) maximizing 1) the playback experiment’s crossed design (Figure 1c, to consider only calls from individuals present as potential receivers in 2015), 2) the number of callers with 2 calls available (ensuring that differential responses to calls would relate to callers’ characteristics and not to the context of recording of the calls), and 3) calls with high signal-to-noise ratios (for a broadcast with the most natural effect). We further supplemented this sample by retaining 3 calls recorded (in 2014) with a digital voice recorder (Olympus VN-702PC, sampling frequency of 48 kHz), for a total of 23 calls recorded during aerial-predator simulations (at a distance of 7.89 ± 1.95 [mean ± SD] meters, range: 5–11). Calls naturally emitted were also derived from 10 recordings made with the directional microphone during opportunistic encounters with calling individuals (2 calls were recorded in 2013 and 8 in 2014, at a distance of 9.88 ± 5.64 (mean ± SD) meters, range: 4–20), and from 3 recordings made in 2014 from video cameras positioned at 3.33 ± 1.53 (mean ± SD) meters (range: 2–5) from the feeding patches (Sony HDR CX220 and CX240, sampling frequency of 48 kHz). Thirty-six calls from 20 callers were thus available for playback (including 16 callers with 2 calls and 4 with 1 call; using each recording technique for callers of varying personalities). Figure 1 View largeDownload slide Main steps used to implement the field playback experiment. (a) Simulation of aerial predator attack for recording of elicited chucking alarm calls. (b) Acoustical preparation of 30-s calling sequences for playback. (c) Playback design for broadcast of multiple callers’ calls randomly assigned to multiple receivers. Callers and receivers varied in their boldness (color code is similar to the one used in Figure 2) and in their level of familiarity. Figure 1 View largeDownload slide Main steps used to implement the field playback experiment. (a) Simulation of aerial predator attack for recording of elicited chucking alarm calls. (b) Acoustical preparation of 30-s calling sequences for playback. (c) Playback design for broadcast of multiple callers’ calls randomly assigned to multiple receivers. Callers and receivers varied in their boldness (color code is similar to the one used in Figure 2) and in their level of familiarity. Acoustical preparation of calling sequences We prepared the 36 calls for playback as chucking sequences lasting 30 s (Figure 1b), following Weary and Kramer (1995) and to allow consistent comparison of receivers response (natural chucking bouts emitted during 129 hat tests for instance lasted on average (± SD) 186.34 ± 215.49 s, median = 98, range: 3–600, and 100 contained on average (± SD) 2.65 ± 2.68 trills, range: 1–16). To retain as much individual variation as possible, we kept the trills naturally contained within the chucking bouts (21/36 calls contained on average (± SD) 3.67 ± 3.12 trills, range: 1–13, which accounted for 0.81% ± 0.76 (mean ± SD) of the 30 s bout, range: 0.10–2.91%). For 9 calls that originally lasted less than 30 s (19.98 ± 5.29 [mean ± SD] seconds, range: 14–28, which belonged to individuals of varying personalities), we selected a chucking portion (of 7.27 ± 4.83 [mean ± SD] seconds, range: 2–13.5) and repeated it on average (± SD) 1.78 ± 1.30 times (range: 1–4) right after the cut, insuring smooth and inaudible transitions during broadcast, and keeping the chucking bout’s first and last 4 s natural. This type of procedure is commonly implemented for playback experiments (Weary and Kramer 1995; Hare and Atkins 2001; Blumstein and Daniel 2004) and should not have damaged any individuality encoded in the vocalizations as chipmunks tend to emit similar notes in a calling bout (Burke da Silva et al. 1994), and internote intervals were consistently preserved (Frommolt et al. 2003). We also improved the quality of the sequences with a gentle High Pass Filter (set at 0.63 ± 0.03 [mean ± SD] kHz, range: 0.6–0.7) which only attenuated low frequencies (through a smooth transition), allowing to get rid of background noise while ensuring that chucking notes at lowest frequencies were not cut off (we confirmed that chucking frequencies as low as 300 Hz were still preserved, both at the hearing and through acoustic measurements applied in Avisoft). For convenience of playback control we made the calling sequences start and end with 2 s and 1 min of silence respectively, and they were further down-sampled to 48 kHz for broadcast compatibility through the digital caller (Foxpro Scorpion X1B). Based on previous work (Weary and Kramer 1995; Emmering and Schmidt 2011) and on the experience of observers, the volume of the calls was calibrated to reach a natural loudness of 50 dB SPLA (mean ± SD = 49.51 ± 3.58, range: 40–52.5) at 5 m from the source, using a digital sound level meter (Nady DSM-1, calibrated by the manufacturer). Playback experiment procedure We ran the playback experiment from 18 June to 10 July 2015, over 20 fieldwork days during which we performed 9.65 ± 3.59 (mean ± SD) trials per day (range: 3–16), alternating sites every other day to limit potential habituation to playbacks. A period of at least 5 h separated trials performed on a same individual, and we also maximized the distance between 2 successive trial locations (to avoid testing individuals that may have heard calls from the previous playback). Twenty-five receivers (12 and 13 individuals on each site, 16 females and 9 males) received a total of 193 playback trials, being each tested 7.72 ± 1.77 (mean ± SD) times (range: 3–9), every 2.3 ± 1.7 (mean ± SD) days (range: 0.2–7). Seventeen individuals played both roles of receivers and callers, as 3 individuals among the 20 selected alarm callers (10 individuals from each site, 12 females and 8 males) disappeared (either dispersed or died) once we started the experiment. Eight other individuals, from which we had acquired behavioral data in 2014, were added as receivers. Only adults were considered in the experiment, and individuals’ sex was included in the analyses. Receivers were exposed to calls from both familiar and unfamiliar callers (precluding their own calls; Figure 1c), allowing to test if they would respond to callers’ personality: 1) only for familiar callers, thus potentially assessing this information through acoustic individual recognition associated with social assessment of callers’ behaviors; or 2) for both familiar and unfamiliar callers, thus decoding this information directly from the calls. We considered individuals as familiar when they lived on the same site and most assuredly knew each other through natural encounters (C. Couchoux unpublished field observations from 2014, wherein individuals: 1) engaged in agonistic interactions, or 2) used the same feeding patch, or 3) lived at a close range [~50 m]). Unfamiliar individuals lived on our 2 different sites and therefore did not know each other (chipmunks have dispersal distance smaller than 1 km; Messier et al. 2012). Each receiver was intended to receive 8 calls that were randomly selected beforehand, 4 calls from familiar callers (random assignment performed for 9 receivers having more than 4 neighbors, the average [± SD] being at 4.44 ± 0.82, range: 3–7), and 4 calls from unfamiliar callers (Figure 1c). To increase the power to detect any possible personality effect for unfamiliar callers, we selected them among the 6 (out of 10) callers from the other site with the most extreme calling behaviors during aerial-predator simulations (based on the 129 hat tests mentioned above, Figure 1a). For each receiver, we thus randomly selected 2 (out of 3) callers that tended to call for short duration and 2 (out of 3) callers that called for longer duration. As most callers had 2 calling sequences (16/20), one of them was randomly selected for each playback trial. For each receiver, we then randomly assigned the calls’ playback order and reassessed it every evening before the next fieldwork day, among the remaining possibilities. The familiarity treatment was reassigned if chosen 3 times in a row or more than 3 times during the first 4 trials, to ensure that if individuals were to disappear they would have received at least 1 call from each treatment by their third trial and 2 of each by their fourth. While 4 individuals disappeared prematurely, with 2 individuals that received 5 trials and 2 that received 3 trials, 21 individuals received their 8 planned calls, among which 9 even got a ninth one (preferentially selected among familiar callers; total of 95 familiar and 98 unfamiliar treatments). A playback trial consisted in measuring the vigilance reaction of a foraging individual (as the amount of time spent in alert posture, with the head raised) in response: 1) to a 30-s control “soundscape” playback, and (upon subsequent return of the individual), 2) to a 30-s alarm call playback (Figure 1, Weary and Kramer 1995; Baack and Switzer 2000). The control playbacks allowed us not only to confirm that responses were elicited by alarm calls (and not just by the playback apparatus or by the presence of observers nearby), but also to obtain repeated measures of individual basal vigilance that could reflect differential individual risk-taking tendencies (i.e., boldness). We also opportunistically performed 35 separate soundscape tests that were not followed by an alarm call playback (or in 5 cases by one that was aborted due to the arrival of a conspecific), both to prevent receivers from associating our presence with the broadcast of calls and to gather supplementary observations to estimate the consistency of individuals’ basal vigilance (17/25 individuals received on average [± SD] 2.06 ± 1.09 of these tests, range: 1–4). Control playbacks consisted of 4 soundscape sequences of sufficient quality that were derived from recordings of neutral surrounding forest sounds made on our study sites using the directional microphone. They were also subjected to a higher High Pass Filter (set at 1.6 kHz, as it was possible here to remove as much background noise as possible without risking damage to meaningful low-frequency sounds) and to an addition of identical silence sections, and were further down-sampled to 48 kHz. We calibrated their broadcasting volume to reach a natural effect (preliminary setting it to match the volume of the surrounding sounds heard simultaneously in the forest) and we randomly assigned one sequence to each playback trial. To implement a playback trial, we first filled the feeding patch before the targeted individual’s arrival, and perpendicularly installed the digital caller (Foxpro Scorpion X1B) 5 m away from the middle point between feeding and observation spots (randomly choosing either left or right side, so calls would not always come from the same direction). We then started the trial once the foraging individual had made at least one return to the burrow with loaded cheek pouches (Weary and Kramer 1995), and after ascertaining the absence of potentially disruptive conspecifics (within a ~10 m visibility radius). The trial was performed in blind conditions: one experimenter (J.C.) was responsible for planning and playing back the sequences, which were renamed with neutral codes within the remote control (Foxpro TX-9) and on the playbacks’ field plan, while the other experimenter (C.C.) was in charge of measuring (with a stopwatch) individual’s vigilance during the playbacks and for setting up a backup video camera at the observation spot (useful to later confirm manual measurements if visibility conditions were not optimal). During each playback, we also accounted for confounding variables (habitat characteristics, wind, social environment, see details in Clermont et al. 2017). The 20 callers had their calls broadcast 9.65 ± 4.45 (mean ± SD) times (range: 3–15), and each of the 36 calling sequences was played back 5.36 ± 0.40 (mean ± SD) times (range: 1–11) (the 12 callers’ sequences played back on both sites being broadcast more often). On the other hand, the 4 soundscape sequences were each played back 57.00 ± 3.83 (mean ± SD) times (range: 52–60) during 228 tests. Statistical analyses To analyze repeated individual measurements of vigilance, we used linear mixed models in R. 3.1.3. (R Development Core Team 2013) with the REML method in package “lme4” (Bates et al. 2015). To most accurately model vigilance variation, we included environmental fixed effects as confounding variables (see Clermont et al. 2017) and we ran preliminary models investigating the effects of individual random slopes along with potentially meaningful interactions (random slope models allow to reduce type I errors and to obtain the most appropriate estimates of variance, Schielzeth and Forstmeier 2009). To reach a minimal final model, we sequentially dropped each nonsignificant effect by comparing models lacking the different terms using likelihood ratio tests (LRT). The LRT P values (α = 0.05) were also corroborated with 95% bootstrapped confidence intervals calculated for the fixed effects’ estimates (Bates et al. 2015). We removed nonsignificant random slopes from the model, but kept nonsignificant random effects to avoid potential statistical issues related to the nonindependence of data with repeated measures from the same individual. The models were checked and validated for normality, homogeneity, and nonindependence of their residuals through graphical visualizations. We modeled the variation of individuals’ basal vigilance (converted to 60 s) during the 228 control playbacks (for the 25 receivers) and from 28 focal observations performed in 2014 on the 3 individuals that were only considered as callers. The 256 focal observations lasted on average (± SD) 59.15 ± 7.33 s, (range: 20–90), and the 28 individuals were each observed on average (± SD) 9.14 ± 2.59 times (range: 3–14), every 2.2 ± 2.3 (mean ± SD) days (range: 1 min–13 days). Random effects included date and individual identity. The starting model included Z-scores for tree cover and number of neighbors as confounding environmental fixed effects (see Clermont et al. 2017), as well as a categorical measure of disturbance (yes or no) resulting from the occurrence of wind or from the presence of conspecifics. We also controlled for a potential habituation effect due to our presence by including the observation’s number, which was transformed by subtracting 1 and divided by its standard deviation to set the reference for the first observation. Preliminary models showed nonsignificant individual random slopes for the observation’s number (LRT: χ22 = 3.08, P = 0.21). After estimating a significant repeatability for the individual random effect (through a LRT comparing the model with and without it, Nakagawa and Schielzeth 2010), we extracted individual best linear unbiased predictors (BLUPs, Pinheiro and Bates 2000) as measures of consistent individuals’ basal vigilance level. We further used the reverse of these values (-BLUPs) as scores of individual boldness (highest values are thus for bold individuals, which have low values of basal vigilance). We analyzed the variation of individuals’ vigilance response to conspecifics’ alarms calls during the 193 30-s alarm call playbacks, for the 25 receivers (considering the focal chipmunk as still being vigilant when it fled from the feeding patch, see Clermont et al. 2017). Random effects included date and individual identity of both receiver and alarm caller. As in the model for basal vigilance, Z-scores for tree cover, number of neighbors and trial’s number were included as confounding fixed effects. Categorical wind intensity (none, low, high, see Clermont et al. 2017) and level of disturbance (none, low, or high number of disturbance events from neighbors) were however kept separated for this model (as we specifically made disturbance events rare by trapping intruders). To test our effects of interest, we included: receiver-callers’ familiarity, included in interaction with the pair’s sex combination; receiver’s and caller’s personality (using scores of boldness extracted from the model of basal vigilance); an interaction between the boldness of both receiver and caller; and an interaction between the caller’s boldness and its familiarity with the receiver. We preliminary tested and removed receivers’ random slopes for the trial’s number (LRT: χ22 = 0.45, P = 0.80), along with the interaction between the trial’s number and receiver’s boldness (LRT: χ21 = 0.54, P = 0.46), as well as callers’ random slopes for the trial’s number (LRT: χ22 = 0.67, P = 0.71), along with the interaction between the trial’s number and the caller’s boldness (LRT: χ21 = 0.54, P = 0.46). Nonsignificant receivers’ random slopes for callers’ boldness were also removed from subsequent analyses (LRT: χ22 = 0.68, P = 0.71). Ethics statement All captures and manipulations were carried out in accordance with the guidelines of the Canadian Council on Animal Care and the protocols were approved through the Université du Québec à Montréal (CIPA permit #760) and the Québec Ministry of Sustainable Development, Environment, Wildlife and Parks (SEG permits #2014-04-22-103-05-S-F and #2015-05-12-106-05-S-F). RESULTS Repeatable basal vigilance as a measure of boldness Basal vigilance levels, primarily measured during multiple control soundscape playbacks, consistently differed among individuals (repeatability estimated at 22.34% with mixed models, LRT: χ21 = 24.38, P < 0.001, Table 1). We therefore extracted and further used -BLUPs values as individual boldness scores (Figure 2a), with individuals with the lowest basal vigilance levels being considered as taking greater risks and classified as the bolder ones. Table 1 Results from linear mixed-effect models on vigilance in eastern chipmunks during control and alarm call playbacks Basal vigilance model Fixed effectsa Estimate 95% CI χ2 df P Random effects Variance Intercept 18.88 15.87: 21.93 Individual 16.89 Disturbance 1.90 -0.28: 4.00 3.01 1 0.083 Date 4.22 Observation number -3.22 -4.41: -2.02 19.27 1 <0.001 Residual 54.49 Vigilance response model Fixed effectsb Estimate 95% CI χ2 df P Random effects Variance Intercept 17.01 14.97: 18.97 Receiver ID 9.05 Wind 1.90 0.77: 3.14 9.52 1 0.002 Caller ID 7.49 Number of neighbors 1.80 0.32: 3.24 5.41 1 0.020 Date 1.23 Trial number -3.81 -4.98: -2.58 24.15 1 <0.001 Residual 53.27 Boldness receiver -0.62 -1.08: -0.14 6.83 1 0.009 Boldness caller 0.51 0.04: 0.99 4.19 1 0.041 Basal vigilance model Fixed effectsa Estimate 95% CI χ2 df P Random effects Variance Intercept 18.88 15.87: 21.93 Individual 16.89 Disturbance 1.90 -0.28: 4.00 3.01 1 0.083 Date 4.22 Observation number -3.22 -4.41: -2.02 19.27 1 <0.001 Residual 54.49 Vigilance response model Fixed effectsb Estimate 95% CI χ2 df P Random effects Variance Intercept 17.01 14.97: 18.97 Receiver ID 9.05 Wind 1.90 0.77: 3.14 9.52 1 0.002 Caller ID 7.49 Number of neighbors 1.80 0.32: 3.24 5.41 1 0.020 Date 1.23 Trial number -3.81 -4.98: -2.58 24.15 1 <0.001 Residual 53.27 Boldness receiver -0.62 -1.08: -0.14 6.83 1 0.009 Boldness caller 0.51 0.04: 0.99 4.19 1 0.041 Effects of interest are in bold. aRemoved from the model after LRT: Number of neighbors (χ21 = 1.69, P = 0.194) and Tree cover (χ21 = 2.38, P = 0.123). The effect of interest is in bold. bRemoved from the model after LRT: Tree cover (χ21 = 0.02, P = 0.877), Disturbance (χ21 = 0.10, P = 0.753), Boldness receiver × Boldness caller (χ21 = 0.18, P = 0.667), Sex combination × Familiarity (χ23 = 1.62, P = 0.655), Sex combination (χ23 = 3.06, P = 0.382), Familiarity × Boldness caller (χ21 = 1.21, P = 0.272) and Familiarity (χ21 = 0.001, P = 0.972). Effects of interest are in bold. View Large Figure 2 View largeDownload slide Boldness as a driver of risk-taking reaction in chipmunks. (a) Boxplot representing individuals’ consistent differences in basal vigilance (n = 256). Individuals (n = 28) are ordered according to their boldness score (-BLUPs), with an increase in blue darkness from shyer to bolder individuals (Interm. = Intermediate). Boxplots show medians, first and third quartiles (middle, upper and lower hinge respectively). (b) Scatterplot illustrating variation in vigilance response to 30-s alarm call playbacks (n = 193). Receivers (n = 25) have their boldness represented using the same colors as in (a), and their intercepts were extracted from the final model not including receiver’s boldness (which otherwise encompasses a part of individual variation). Individual slopes of vigilance response used the slope estimate for the population, as there was no significant interaction between receiver’s and caller’s boldness. Black and grey points represent response to familiar (n = 95) and unfamiliar (n = 98) callers, respectively. Figure 2 View largeDownload slide Boldness as a driver of risk-taking reaction in chipmunks. (a) Boxplot representing individuals’ consistent differences in basal vigilance (n = 256). Individuals (n = 28) are ordered according to their boldness score (-BLUPs), with an increase in blue darkness from shyer to bolder individuals (Interm. = Intermediate). Boxplots show medians, first and third quartiles (middle, upper and lower hinge respectively). (b) Scatterplot illustrating variation in vigilance response to 30-s alarm call playbacks (n = 193). Receivers (n = 25) have their boldness represented using the same colors as in (a), and their intercepts were extracted from the final model not including receiver’s boldness (which otherwise encompasses a part of individual variation). Individual slopes of vigilance response used the slope estimate for the population, as there was no significant interaction between receiver’s and caller’s boldness. Black and grey points represent response to familiar (n = 95) and unfamiliar (n = 98) callers, respectively. No familiarity effect on response to alarm calls The playback experiment revealed that vigilance during the 30-s alarm call playbacks was not significantly affected by the receiver’s familiarity with the alarm caller (Table 1). Moreover, calls received from familiar females (potentially related in our system) did not elicit higher vigilance levels (no interaction between the receiver-caller’s sex combination and their familiarity, Table 1). Overall effects of boldness on the interpretation of alarm calls Receivers’ vigilance response to alarm calls was significantly negatively affected by their own boldness score: the shyest individuals (with high levels of basal vigilance/low values of -BLUPs, represented in light blue in Figure 2a) reacted the most strongly to alarm calls (Table 1, with high levels of vigilance response, as reflected in Figure 2b by intercepts of lighter blue lines being higher than intercepts of darker blue lines). Receivers also significantly increased their vigilance level in response to alarm calls from bolder alarm callers (Table 1, Figure 2b). By randomly playing back one of 2 different calls for most callers (16/20), we ensured that differential reactions were due to the individual characteristics of the callers and not only to the context of recording of their calls. An unconditional use of the social information Receivers used the social information provided by the alarm caller’s boldness independently of their own boldness (nonsignificant interaction between the receiver’s and caller’s boldness; Table 1, Figure 2b). In addition, familiarity between the receiver and the caller did not affect the receiver’s response to the caller’s boldness (nonsignificant interaction between familiarity and caller’s boldness, Table 1, Figure 2b). DISCUSSION Using a carefully designed field playback experiment that integrated multiple levels of natural individual variation, we showed that personality affects the interpretation of social information in eastern chipmunks. Receivers adapted their response to alarm calls based on their own tendency to take risk, but also on the risk-taking behavioral profile of the signaler. This result reinforces the importance of considering individual variation in response to signals. Our results emphasize that behavioral tendencies represent important characteristics of individuals. Our results showed that receivers did not react more intensely towards calls from familiar neighbors or potential relatives (Figure 3). This suggests that familiar signals are not considered to be more reliable or to represent more imminent danger, in contrast to what has been reported for some social mammals (Hare 1998; O’Connell-Rodwell et al. 2007, but see Blumstein and Daniel 2004; Hare and Warkentin 2012). Familiarity may convey less valuable information on the imminence of the threat in more spatially dispersed territorial species such as chipmunks. However, alarm calls could still have evolved to be directed primarily towards relatives, as preferential response given to their calls may not be required in cases where the population/species displays spatial clustering of relatives. Moreover, granting more importance to every neighbor or relative is not necessarily a beneficial strategy if these individuals vary in their intrinsic call reliability depending on their personality. Figure 3 View largeDownload slide Main findings arising from the playback experiment, and their implications for underlying relationships. The effects on receiver’s vigilance response to alarm calls, tested through the playback experiment, are represented by black arrows, with thick arrows for significant results and thinner dashed arrows for nonsignificant ones. Double arrows represent interactions. Underlying findings are indicated in blue: the brace emphasizes how the boldness of the caller reflects its reliability, as calls from bolder individuals appeared to be judged as being more reliable by eliciting more intense reactions in the receivers; the dashed-dotted line indicates that the information on caller boldness is decoded directly from the acoustic features of the calls, as receiver reactions to the caller boldness did not depend on their level of familiarity. (Online version in color.) Figure 3 View largeDownload slide Main findings arising from the playback experiment, and their implications for underlying relationships. The effects on receiver’s vigilance response to alarm calls, tested through the playback experiment, are represented by black arrows, with thick arrows for significant results and thinner dashed arrows for nonsignificant ones. Double arrows represent interactions. Underlying findings are indicated in blue: the brace emphasizes how the boldness of the caller reflects its reliability, as calls from bolder individuals appeared to be judged as being more reliable by eliciting more intense reactions in the receivers; the dashed-dotted line indicates that the information on caller boldness is decoded directly from the acoustic features of the calls, as receiver reactions to the caller boldness did not depend on their level of familiarity. (Online version in color.) In fact, we show that individuals consistently differed in their basal vigilance and that these individual differences in risk-taking behavior (i.e., boldness) play a role in the way receivers perceive and react to alarm calls. Receivers responded to calls according to their own boldness, with shyer individuals being more vigilant that bolder ones in response to alarm calls (Figure 3). This suggests that, in an alarm context, interindividual behavioral variation can constrain individuals’ decisions as much as other individual characteristics such as age, sex, or relationship (Hare 1998; Blumstein and Daniel 2004; Hollén and Radford 2009). Receivers use the available social information on threat detection as they increased their vigilance level towards alarm calls (Clermont et al. 2017, see also Weary and Kramer 1995; Baack and Switzer 2000), but their relative level of reaction is dependent on their own risk perception and reflects their own boldness. A receiver’s response level is thus constrained by its personality, but the response can still be plastic. Personality constraints and behavioral plasticity are not mutually exclusive (Dingemanse et al. 2010), as an individual’s behavioral phenotype is a combination of its unique behavioral tendencies and its responsiveness to environmental variation. We thus found that receivers also respond plastically to the specificity of the information, as they showed reduced responsiveness towards calls from shyer individuals (Figure 3). Behavioral responses are therefore adjusted to the boldness of the callers, which implies that chipmunks could assess and respond to the alarm caller’s reliability (Figure 3). This would suggest that the ability to assess the reliability of the signaler is not restricted to social mammals (Cheney and Seyfarth 1988; Hare 1998; Hare and Atkins 2001; Blumstein et al. 2004). More importantly, we show here for the first time that individuals can perceive information related to a signaler’s personality and respond accordingly. While individuals can evaluate the quality of an alarm signal from fixed characteristics of the signaler (e.g., age or relationship level, Hollén and Radford 2009), or from its instantaneous behavior (e.g., in dwarf mongooses, Helogale parvula, the elevation position of an alarm caller can reflect its reliability to detect predators, Kern et al. 2017), we showed here that they can also respond to a signaler’s intrinsic behavioral characteristics (i.e., risk-taking profile; though the proximate basis of that assessment is presently unknown). In other contexts, individuals respond to the reliability of the social information by observing social foraging cues from conspecifics. In barnacle geese (Branta leucopsis), for instance, individuals who observe other foraging individuals will ignore the social information when it is made incorrect (Kurvers et al. 2010). Also, bumble bees (Bombus impatiens) adjust their decisions to the social information depending on its reliability to predict a reward (Dunlap et al. 2016). Although the cost-benefit balance of evaluating the reliability of the social information seems quite profitable in a foraging context, responding to all alarm calls may appear more optimal in an alarm context. The “better-safe-than-sorry” strategy—or “adaptive gullibility”—may indeed prevail in risky situations, given the potential fatal cost of not responding to a genuine alarm (Wiley 1994; Searcy and Nowicki 2005; Beauchamp 2010). However, individuals would limit feeding opportunities by responding to erroneous alarm signals, and should thus reduce time-consuming antipredator behaviors if the social information eliciting them is untrustworthy (Hare and Atkins 2001; Searcy and Nowicki 2005). We show that, just as the vigilance response to alarm calls is predicted to decrease in a group with a high rate of false alarms (Beauchamp and Ruxton 2007), the same holds true at the individual level: vigilance also decreased in response to shyer alarm callers more prone to cry wolf because of their high level of risk-sensitivity. We further found that the receivers all showed “social competence” (Taborsky and Oliveira 2012), as individuals of different boldness similarly adjusted their response to the social information provided by the alarm caller’s boldness (Figure 3). Although it has been shown in some situations that shyer individuals are more prone to plastically adjust their behavior to the behaviors of conspecifics (Harcourt et al. 2009; Kurvers et al. 2010), in an alarm context plastic responses may be selected for and considered adaptive since receivers may benefit from placing less trust in individuals that are overly risk-sensitive. In addition, we found that receivers did not need to know the callers to evaluate their reliability (as in Cheney and Seyfarth 1988; Hare and Atkins 2001), as they adjusted their vigilance level to the boldness of both familiar and unfamiliar callers. This implies that the assessment of callers’ risk-taking tendencies is not necessarily based on social experience, and that cues about caller’s boldness could be decoded directly from the vocalizations (Figure 3). The signals could indeed convey information on the tendency of an individual to take risks, especially since alarm calls can encode information about the situation’s riskiness (Hollén and Radford 2009) or urgency (Manser 2001). Acoustic characteristics could thus be modulated by the individuals’ levels of excitation, which are intrinsically related to stress physiological constraints. Individual variation in sympathetic system reactivity (see POLS concept, Réale et al. 2010) could particularly affect the rate of emission of repetitive calls such as chucking, in the same way it affects heart rate (more proactive individuals have high sympathetic response reflected in high heart rates, Montiglio et al. 2012). A next research step will be to explore this possibility in chipmunks, as they seem to produce chucking at variable rates and are more responsive to chucking played at higher rates (Weary and Kramer 1995). The presence of trills within chucking bouts also requires further investigation, as these calls are thought to signal highly threatening situations (Burke da Silva et al. 1994) and could be predominantly emitted by overly risk-sensitive individuals. CONCLUSION The social environment is a crucial component of an individual’s environment and the use of social information can greatly impact decision-making, not only in group-living species but also in so-called solitary species, such as chipmunks, which interact with conspecifics on a daily basis. The integration of personality traits to social networks increasingly sheds light on our understanding of complex social behaviors (Aplin et al. 2013) and here we showed that individual variation in behavior can play a significant role in the perception and interpretation of alarm signals. The individual-based approach, which takes into account widespread interindividual behavioral differences, has brought considerable insights in many research areas (Réale et al. 2007; Réale et al. 2010; Wolf et al. 2012) and could help us understand variation in alarm communication strategies. Studies using playback experiments should thus carefully consider behavioral variation for individuals chosen as receivers and signalers. We advocate the integration of individual behavioral variation in studies of alarm communication systems as personality could broadly affect the way animals communicate about danger in both solitary and social species. Considering individual variation emphasizes the ability of animals to perceive and respond accurately to social information in many ecological contexts. FUNDING The study was supported by a research team grant from the Fonds de Recherche du Québec - Nature et Technologies (FRQNT) to D.G., D.R. (and Murray Humphries, Fanie Pelletier), by Discovery grants from the Natural Sciences and Engineering Research Council of Canada (NSERC) (D.G., D.R.) and by a Canada Research Chair (D.R.). C.C. was supported by a scholarship from the FRQNT and through grants from the UQAM, and by a National Geographic Young Explorers grant (#9422-13) and a Student Research grant from the Animal Behavior Society. J.C. was supported by an Undergraduate Summer Research Award from NSERC and a Biodiversity Science Discovery award from the Quebec Center for Biodiversity Science. We thank the Green Mountains Nature reserve of the Nature Conservancy of Canada (CNC) for allowing us to access the study sites, as well as all the people that helped in the field to maintain the long-term project over the last 10 years. We are also grateful to Arnaud Béchet, Anne Charmantier, Christelle Couchoux, Carolyn Hall, Joël Jameson, Rupert Marshall, James Hare, and an anonymous reviewer for useful comments on previous versions of the manuscript. We also thank Torben Dabelsteen, Élise Huchard, and Rupert Marshall for their advice on the experimental design, and Raimund Specht for his advice on Avisoft’s acoustic settings. Data accessibility: Analyses reported in this article can be reproduced using the data provided by Couchoux et al. (2017). REFERENCES Amy M, Sprau P, de Goede P, Naguib M. 2010. Effects of personality on territory defence in communication networks: a playback experiment with radio-tagged great tits. Proc Biol Sci . 277: 3685– 3692. Google Scholar CrossRef Search ADS PubMed Andersson M. 1994. Sexual selection . Princeton (NJ): Princeton University Press. Aplin LM, Farine DR, Morand-Ferron J. 2013. Individual personalities predict social behaviour in wild networks of great tits (Parus major). Ecol Lett . 16: 1365– 1372. Google Scholar CrossRef Search ADS PubMed Baack JK, Switzer PV. 2000. Alarm calls affect foraging behavior in eastern chipmunks (Tamias striatus, Rodentia: Sciuridae). Ethology . 106: 1057– 1066. Google Scholar CrossRef Search ADS Bates D, Mächler M, Bolker BM, Walker SC. 2015. Fitting linear mixed-effects models using lme4. J Stat Softw . 67: 1– 48. Google Scholar CrossRef Search ADS Beauchamp G, Ruxton GD. 2007. False alarms and the evolution of antipredator vigilance. Anim Behav . 74: 1199– 1206. Google Scholar CrossRef Search ADS Beauchamp G. 2010. Determinants of false alarms in staging flocks of semipalmated sandpipers. Behav Ecol . 21: 584– 587. Google Scholar CrossRef Search ADS Blumstein DT, Daniel JC. 2004. Yellow-bellied marmots discriminate between the alarm calls of individuals and are more responsive to calls from juveniles. Anim Behav . 68: 1257– 1265. Google Scholar CrossRef Search ADS Blumstein DT, Patton ML, Saltzman W. 2006. Faecal glucocorticoid metabolites and alarm calling in free-living yellow-bellied marmots. Biol Lett . 2: 29– 32. Google Scholar CrossRef Search ADS PubMed Blumstein DT, Steinmetz J, Armitage KB, Daniel JC. 1997. Alarm calling in yellow-bellied marmots: II. The importance of direct fitness. Anim Behav . 53: 173– 184. Google Scholar CrossRef Search ADS Blumstein DT, Verneyre L, Daniel JC. 2004. Reliability and the adaptive utility of discrimination among alarm callers. Proc Biol Sci . 271: 1851– 1857. Google Scholar CrossRef Search ADS PubMed Blumstein DT. 2007. The evolution, function, and meaning of marmot alarm communication. Adv Stud Behav . 37: 371– 401. Google Scholar CrossRef Search ADS Botero CA, Pen I, Komdeur J, Weissing FJ. 2010. The evolution of individual variation in communication strategies. Evolution . 64: 3123– 3133. Google Scholar CrossRef Search ADS PubMed Burke da Silva K, Kramer DL, Weary DM. 1994. Context-specific alarm calls of the eastern chipmunk, Tamias striatus. Can J Zool . 72: 1087– 1092. Google Scholar CrossRef Search ADS Burke da Silva K, Mahan C, da Silva J. 2002. The trill of the chase: eastern chipmunks call to warn kin. J Mamm . 83: 546– 552. Google Scholar CrossRef Search ADS Cheney DL, Seyfarth RM. 1988. Assessment of meaning and the detection of unreliable signals by vervet monkeys. Anim Behav . 36: 477– 486. Google Scholar CrossRef Search ADS Clary D, Skyner LJ, Ryan CP, Gardiner LE, Anderson WG, Hare JF. 2014. Shyness–boldness, but not exploration, predicts glucocorticoid stress response in Richardson's ground squirrels (Urocitellus richardsonii). Ethology , 120: 1101– 1109. Google Scholar CrossRef Search ADS Clermont J, Couchoux C, Garant D, Réale D. 2017. Assessing anti-predator decisions of foraging eastern chipmunks under varying perceived risks: the effects of physical and social environments on vigilance. Behaviour . 154: 131– 148. Google Scholar CrossRef Search ADS Couchoux C, Clermont J, Garant D, Réale D. 2017. Data from: signaler and receiver boldness influence response to alarm calls in eastern chipmunks. Dryad Digital Repository . http://dx.doi.org/10.5061/dryad.vq021 Dall SR, Giraldeau LA, Olsson O, McNamara JM, Stephens DW. 2005. Information and its use by animals in evolutionary ecology. Trends Ecol Evol . 20: 187– 193. Google Scholar CrossRef Search ADS PubMed Dammhahn M, Landry-Cuerrier M, Réale D, Garant D, Humphries MM. 2017. Individual variation in energy-saving heterothermy affects survival and reproductive success. Funct Ecol . 31: 866– 875. Google Scholar CrossRef Search ADS Danchin E, Giraldeau LA, Valone TJ, Wagner RH. 2004. Public information: from nosy neighbors to cultural evolution. Science . 305: 487– 491. Google Scholar CrossRef Search ADS PubMed Dingemanse NJ, Kazem AJ, Réale D, Wright J. 2010. Behavioural reaction norms: animal personality meets individual plasticity. Trends Ecol Evol . 25: 81– 89. Google Scholar CrossRef Search ADS PubMed Dunlap AS, Nielsen ME, Dornhaus A, Papaj DR. 2016. Foraging bumble bees weigh the reliability of personal and social information. Curr Biol . 26: 1195– 1199. Google Scholar CrossRef Search ADS PubMed Elliot L. 1978. Social behavior and foraging ecology of the eastern chipmunk (Tamias striatus) in the Adirondack Mountains. Smithson Contrib Zool . 265: 1– 107. Google Scholar CrossRef Search ADS Emmering QC, Schmidt KA. 2011. Nesting songbirds assess spatial heterogeneity of predatory chipmunks by eavesdropping on their vocalizations. J Anim Ecol . 80: 1305– 1312. Google Scholar CrossRef Search ADS PubMed Frommolt KH, Goltsman ME, Macdonald DW. 2003. Barking foxes, Alopex lagopus: field experiments in individual recognition in a territorial mammal. Anim Behav . 65: 509– 518. Google Scholar CrossRef Search ADS Fuong H, Maldonado-Chaparro A, Blumstein DT. 2015. Are social attributes associated with alarm calling propensity? Behav Ecol . 26: 587– 592. Google Scholar CrossRef Search ADS Giraldeau LA, Valone TJ, Templeton JJ. 2002. Potential disadvantages of using socially acquired information. Philos Trans R Soc Lond B Biol Sci . 357: 1559– 1566. Google Scholar CrossRef Search ADS PubMed Harcourt JL, Ang TZ, Sweetman G, Johnstone RA, Manica A. 2009. Social feedback and the emergence of leaders and followers. Curr Biol . 19: 248– 252. Google Scholar CrossRef Search ADS PubMed Hare JF, Atkins BA. 2001. The squirrel that cried wolf: reliability detection by juvenile Richardson’s ground squirrels (Spermophilus richardsonii). Behav Ecol Sociobiol . 51: 108– 112. Google Scholar CrossRef Search ADS Hare JF, Warkentin KJ. 2012. The song remains the same: Juvenile Richardson’s ground squirrels do not respond differentially to mother’s or colony member’s alarm calls. Curr Zool . 58: 773– 780. Google Scholar CrossRef Search ADS Hare JF. 1998. Juvenile Richardson’s ground squirrels, Spermophilus richardsonii, discriminate among individual alarm callers. Anim Behav . 55: 451– 460. Google Scholar CrossRef Search ADS PubMed Hollén LI, Radford AN. 2009. The development of alarm call behaviour in mammals and birds. Anim Behav . 78: 791– 800. Google Scholar CrossRef Search ADS Kern JM, Laker PR, Radford AN. 2017. Contextual variation in the alarm call responses of dwarf mongooses, Helogale parvula. Anim Behav . 127: 43– 51. Google Scholar CrossRef Search ADS Kurvers RH, van Oers K, Nolet BA, Jonker RM, van Wieren SE, Prins HH, Ydenberg RC. 2010. Personality predicts the use of social information. Ecol Lett . 13: 829– 837. Google Scholar CrossRef Search ADS PubMed Lima SL, Dill LM. 1990. Behavioral decisions made under the risk of predation: a review and prospectus. Can J Zool . 68: 619– 640. Google Scholar CrossRef Search ADS Manser MB. 2001. The acoustic structure of suricates’ alarm calls varies with predator type and the level of response urgency. Proc Biol Sci . 268: 2315– 2324. Google Scholar CrossRef Search ADS PubMed Maynard-Smith J. 1965. The evolution of alarm calls. Am Nat . 99: 59– 63. Google Scholar CrossRef Search ADS McNamara JM, Stephens PA, Dall SR, Houston AI. 2009. Evolution of trust and trustworthiness: social awareness favours personality differences. Proc Biol Sci . 276: 605– 613. Google Scholar CrossRef Search ADS PubMed Messier GD, Garant D, Bergeron P, Réale D. 2012. Environmental conditions affect spatial genetic structures and dispersal patterns in a solitary rodent. Mol Ecol . 21: 5363– 5373. Google Scholar CrossRef Search ADS PubMed Montiglio PO, Garant D, Pelletier F, Réale D. 2012. Personality differences are related to long-term stress reactivity in a population of wild eastern chipmunks, Tamias striatus. Anim Behav . 84: 1071– 1079. Google Scholar CrossRef Search ADS Nakagawa S, Schielzeth H. 2010. Repeatability for Gaussian and non-Gaussian data: a practical guide for biologists. Biol Rev Camb Philos Soc . 85: 935– 956. Google Scholar PubMed O’Connell-Rodwell CE, Wood JD, Kinzley C, Rodwell TC, Poole JH, Puria S. 2007. Wild African elephants (Loxodonta africana) discriminate between familiar and unfamiliar conspecific seismic alarm calls. J Acoust Soc Am . 122: 823– 830. Google Scholar CrossRef Search ADS PubMed Pinheiro JC, Bates DM. 2000. Mixed-effects models in S and S-Plus . New York (NY): Springer Verlag. Google Scholar CrossRef Search ADS Pollard KA. 2011. Making the most of alarm signals: the adaptive value of individual discrimination in an alarm context. Behav Ecol . 22: 93– 100. Google Scholar CrossRef Search ADS R Development Core Team. 2013. R: a language and environment for statistical computing . Vienna (Austria): R Foundation for Statistical Computing. Réale D, Garant D, Humphries MM, Bergeron P, Careau V, Montiglio PO. 2010. Personality and the emergence of the pace-of-life syndrome concept at the population level. Philos Trans R Soc Lond B Biol Sci . 365: 4051– 4063. Google Scholar CrossRef Search ADS PubMed Réale D, Reader SM, Sol D, McDougall PT, Dingemanse NJ. 2007. Integrating animal temperament within ecology and evolution. Biol Rev Camb Philos Soc . 82: 291– 318. Google Scholar CrossRef Search ADS PubMed Schielzeth H, Forstmeier W. 2009. Conclusions beyond support: overconfident estimates in mixed models. Behav Ecol . 20: 416– 420. Google Scholar CrossRef Search ADS PubMed Searcy WA, Nowicki S. 2005. The evolution of animal communication: reliability and deception in signaling systems . Princeton (NJ): Princeton University Press. Sloan JL, Hare JF. 2008. The more the scarier: adult Richardson’s ground squirrels (Spermophilus richardsonii) assess response urgency via the number of alarm signallers. Ethology . 114: 436– 443. Google Scholar CrossRef Search ADS Sloan Wilson D, Clark AB, Coleman K, Dearstyne T. 1994. Shyness and boldness in humans and other animals. Trends Ecol Evol . 9: 442– 446. Google Scholar CrossRef Search ADS PubMed Taborsky B, Oliveira RF. 2012. Social competence: an evolutionary approach. Trends Ecol Evol . 27: 679– 688. Google Scholar CrossRef Search ADS PubMed Ward P, Zahavi A. 1973. The importance of certain assemblages of birds as information-centres for food-finding. Ibis , 115: 517– 534. Google Scholar CrossRef Search ADS Weary DM, Kramer DL. 1995. Response of eastern chipmunks to conspecifics alarm calls. Anim Behav . 49: 81– 93. Google Scholar CrossRef Search ADS Wiley RH. 1994. Errors, exaggeration, and deception in animal communication. In: Real L, editor. Behavioral mechanisms in evolutionary ecology . Chicago (IL): University of Chicago Press. p. 157– 189. Wolf M, Van Doorn GS, Weissing FJ. 2011. On the coevolution of social responsiveness and behavioural consistency. Proc Biol Sci . 278: 440– 448. Google Scholar CrossRef Search ADS PubMed Wolf M, Weissing FJ. 2012. Animal personalities: consequences for ecology and evolution. Trends Ecol Evol . 27: 452– 461. Google Scholar CrossRef Search ADS PubMed © The Author(s) 2017. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: email@example.com
Behavioral Ecology – Oxford University Press
Published: Jan 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera