|Home | About | Journals | Submit | Contact Us | Français|
Predation risk has negative indirect effects on prey fitness, partly mediated through changes in behaviour. Evidence that individuals gather social information from other members of the population suggests that events in a community may impact the behaviour of distant individuals. However, spatially wide-ranging impacts on individual behaviour caused by a predator encounter elsewhere in a community have not been documented before. We investigated the effect of a predator encounter (hawk model presented at a focal nest) on the parental behaviour of pied flycatchers (Ficedula hypoleuca), both at the focal nest and at nearby nests different distances from the predator encounter. We show that nest visitation of both focal pairs and nearby pairs were affected, up to 3 h and 1 h, respectively. Parents also appeared to compensate initial disrupted feeding by later increasing nest visitation rates. This is the first evidence showing that the behaviour of nearby pairs was affected away from an immediate source of risk. Our results indicate that the impacts of short-term predator encounters may immediately extend spatially to the broader community, affecting the behaviour of distant individuals. Information about predators is probably quickly spread by cues such as intra- and heterospecific alarm calls, in communities of different taxa.
Predation risk affects prey via direct mortality and by indirect negative effects on different fitness components. At the individual level, increased perceived predation risk negatively affects reproductive success through sub-optimal territory choice and parental care [1,2]. Exposure to predators may also elevate physiological stress [3–5] (but see ), modify offspring growth rates [7,8] and influence the composition of maternal effects transferred to the developing embryo [9–11].
Predation risk impacts spatial habitat selection decisions; settling individuals avoid high-risk areas [12–15]. For example, the nests of avian predators, a point source that may anchor predation risk in a landscape, spatially alters bird community structure [16,17] and affects prey behaviour [18,19]. These effects stem from predator nests being a fixed source of threat in the environment for the entire breeding season. But brief individual encounters with predators that elevate perceived risk are probably frequent events in nature and cues elicited by individuals alert the community to these predator encounters [20–23]. Therefore, it seems probable that the influence of even short-term predator risk such as a predator encounter may extend spatially in communities and will not be limited to a small area. Nevertheless, little is known about the spatial extent of impacts on individual behaviour resulting from a predator event taking place elsewhere in the community.
Individuals of many taxa gather information on predator events. Birds cue on predator signs or use social cues of other birds, especially from acoustic alarms [24–26]. Even nestling birds cue on the alarm calls of others . Individuals often mob a detected predator by repeating loud and easily localizable calls signalling to the predator that it might be more productive to hunt elsewhere . Acoustic cues also attract other individuals from the vicinity, which is beneficial because the larger the mobbing group the more likely it is that the predator will move on [24,27]. Mobbing is both a selfish and reciprocal behaviour restricted by territory boundaries [21,28–30]. Experiments with breeding pied flycatchers (Ficedula hypoleuca) have shown that nearby pairs often mob together . Moreover, individuals can complement their knowledge of a current threat by inspecting mobs as a source of information about local predators  and prospect the nest contents of nearby nests after predator events . The reliability of the information and the type of predation, such as threat to adults or nestlings, is also important with regard to parental behaviour [31,33].
We examined the spatial extent to which a brief predator event can impact parental behaviour in a breeding population. Anti-predator behaviours are adaptive in terms of improving survival of adults, but may disrupt normal nestling provisioning behaviour, potentially affecting the quality of young through reduced feeding rates [7,34,35]. We measured the effects of a predator encounter (presenting a predator model) on parental behaviour both at focal nests experiencing the predator event, and at nearby nests at different distances from the focal nest. We expected that (i) both focal and nearby breeding pairs will decrease their nest visiting behaviour after the exposure to a predator when compared with control stimulus, and that (ii) nest visitations of nearby pairs will be negatively affected by the intensity of mobbing at the focal nest. We also predicted (iii) that the expected effects of a short-term predation risk will decrease with increasing distance from the predator presentation.
The study was carried out in 2012 on the island of Ruissalo (southwest Finland 60°25′ N 22°10′ E). Approximately 430 nest-boxes were available to cavity nesting passerines in forest dominated by oak (Quercus robur) and Scots pine (Pinus sylvestris). Our study focused on breeding pied flycatchers settled in a subset of 180 boxes. The pied flycatcher is a small (12–13 g) migratory cavity-nesting passerine that breeds in most of Europe and western Siberia .
All nests were monitored every 4–5 days from the settlement period in early May until chick fledging in July as part of ongoing long-term studies . Female pied flycatchers were generally captured during mid-incubation, whereas males were captured either while prospecting nest-boxes soon after their arrival to the study site or during the early nestling phase. All captured individuals were ringed or their existing ring numbers noted. PIT tags were implanted in each bird. We used TROVAN ID 100A passive integrated transponders (2.12 × 11.5 mm, 0.1 g) and an IID-102 implanter to insert the tags. The tags were implanted subcutaneously in the mantle area . Tag insertion and presence is tolerated well by small passerines with no significant negative impacts documented [38,39].
We attached an automated transponder reading system to nest-boxes (Trovan LID665 readers) included in this study. An antenna of the reader system was mounted around the nest-box entrance hole. The system recorded the exact times a transponder-carrying bird entered the nest-box or sat at the entrance hole. A read delay of 30 s for the same code was programmed for all readers; this was necessary to avoid multiple reads for birds sitting for longer periods at the entrance hole. The readers recorded unique transponder codes of each transponder tagged bird with the exact time of visitation. In nearby nest-boxes, parental visit rates were also obtained from the transponder readers.
We selected focal and nearby nests so that the distance between the two nest-boxes varied from 26 to 129 m (median = 85.8 m). The nests were selected so that they were distributed evenly across the study area, and with no overlap between nest pairs (see below). Our study included 25 pairs of focal and nearby nests (i.e. 50 flycatcher nests in total). There were between 0 and 2 other flycatcher nests between the focal and the nearby nest.
At each focal nest we performed two presentations. One was a treatment presentation of a Eurasian sparrowhawk (Accipiter nisus) model to increase the perceived predation risk. The sparrowhawk is a common predator of adults and recently fledged young of passerines in northern Europe [40–42]. Pied flycatchers account for 4.2% of captures by the sparrowhawk , with about 7–10% of adults being predated by sparrowhawks during the breeding season . Given that mobbing is predominantly elicited by predators of adult birds and much less by nest predators , we expected that a short-term encounter with a sparrowhawk may have a stronger effect on the behaviour of distant individuals compared with the response elicited by a nest predator. The other presentation, a common blackbird (Turdus merula) model, served as a control. Models were taxidermist-prepared mounted skins, and we used two different models of each species during the experiments. The predator and control presentations were done in random order on consecutive days (one presentation per day) at approximately the same time of day (mean pairwise difference = 16 min, maximum difference = 77 min) when nestlings were 6–13 days old (hatching day = day 0 post-hatch) between 06.00 and 16.00 h in good weather to minimize the potential effect of time of day on parental activities . Of the 25 nests used for the experiment, 14 focal nests received the predator and 11 focal nests received the control presentation first. Time of day or its interaction with treatment did not influence parental provisioning in focal nests during the first hour after the model presentation in either day of experiment (first day; time of day: F1,20 = 0.12, p = 0.7; treatment: F1,20 = 9.44, p = 0.006; interaction: F1,20 = 0.0, p = 0.99; second day; time of day: F1,20 = 0.30, p = 0.6; treatment: F1,20 = 6.29, p = 0.021; interaction: F1,20 = 2.82, p = 0.11; main effects estimated without interaction) and was omitted from further analysis. Presentations involved quickly approaching the nest and placing the bird model attached to a 1.5 m tall wooden pole around 1–2 m from the nest-box. The observer inserted a transponder through the nest-box entrance to record the start of presentation and immediately retreated and hid in vegetation at least 20 m away from the nest-box. This procedure of placing the bird model took about 15–30 s for each nest. Only one observer (Kadri Moks) performed all procedures related to predator or control presentations, to reduce random noise related to human impact on birds.
We presented the blackbird models for 10 min and sparrowhawk models for 5–15 min (median = 10 min), until both parents started alarm calling. When removing the bird model, the observer's transponder was again inserted through the nest-box entrance to record the end of presentation. Visiting rates of the parents at the focal and nearby nests were documented by readers up to 4 h after the removal of the model.
We recorded the mobbing period and the number of species taking part in mobbing at each predator presentation using a voice recorder (Olympus, DS-50) positioned below the focal nest-box. Mobbing period was calculated as the total length of the mobbing response where flycatchers made a sound in every second. The length of the mobbing period indicated the time period that parents do not enter their nest. The end of the mobbing should mark the point from where parents may start to return to their normal behaviour. We assumed that the number of species joining the mobbing is a proxy for the intensity of the mobbing . We analysed all alarm calls on the sound recordings in order to evaluate how many different species took part in the mobbing in addition to pied flycatchers. The recordings were analysed as sonograms using Avisoft-SASLab Pro Bioacoustics Laboratory Software (Avisoft Bioacoustics, Berlin, Germany). All sounds were displayed as spectrograms, whereby each type of sound produced by different species is identifiable by the specific pattern it produces. Sound spectrograms were analysed frame by frame. We counted the number of species and the number of individuals joining in the mobbing per each frame until the end of alarm calls. Combining information from spectrograms as well as that from the observer, we were able to separate focal pairs from nearby pairs. First, we assumed that alarm calling is initiated by focal parents as other visiting pied flycatchers never mob at a dummy before the nest owners do . Second, the observer presented the predator until both parents started alarm calling. Hence, we assumed that if two pied flycatchers elicited alarm calls and two individuals were also identified from spectrograms, these birds should be focal parents. When more than two conspecifics were simultaneously giving mobbing calls based on spectrograms, these additional birds should be visitors. Multiple predator presentations in the vicinity of a nest may alter the response of provisioning individuals. We controlled for this possible influence by including the trait ‘experience’ in the analyses as a covariate. ‘Experience’ indicated how many times the predator was presented within 160 m before the day of the actual sparrowhawk presentation for a focal-nearby pair. This occurred in 12 focal and 11 nearby nest-boxes. We assumed threats occurring further than 160 m to have little impact on the birds' behaviour . The total dataset contained 25 focal-nearby pairs and focal nest sound recordings from 20 of those pairs.
At five focal nests and two nearby nests, males had not yet received PIT tags and their visits could not be documented. In order to compare these nests with nests for which we have data for both parents, we used the female data for the former nests and mean values of both parents for the latter nests. A previous study  as well as the current study showed that male and female pied flycatchers visit the nest at similar rates (in control phase: GLM, F1,18 = 0.14, p = 0.7, n = 19 nests) and their visitation rates under increased perceived predation threat are similar (in experimental phase: GLM, F1,18 = 0.9, p = 0.4, n = 19 nests).
In focal nests, we calculated mean parental visitation rates for the first, second, third and fourth hour after the removal of the sparrowhawk and blackbird models. For the nearby nests, we extracted visitation rates for the same time periods, but we also analysed the period when the models were presented at the focal nest (nest visits during presentation). In focal nests, parents did not visit their nest during the sparrowhawk presentation.
Parental latency to mob, calculated from the start of the predator model presentation, was log-transformed prior to analysis, to meet the model assumptions of normal distribution of residuals. Residuals of all other dependent variables were normally distributed.
Statistical analyses were conducted with Statistica v. 7.1 (StatSoft, Tulsa, OK, USA). We used repeated-measures GLMs to compare the nest visitation rates of parents at the focal nest after presentations. When analysing parental visitation rates at the focal nests, we used mean parental nest visitation rate of focal birds as a dependent variable (visits per hour), treatment (blackbird versus sparrowhawk) and time (first, second, third, fourth hour after removal of model) as repeated factors, and nestling age on the first experimental day and a previous experience with the sparrowhawk presentation in the 160 m vicinity as covariates in the initial model. A possible interaction between treatment and time was tested as we predicted the difference in nest visitation rate between blackbird and sparrowhawk presentation to be the most pronounced during the first hour after the treatment and this difference to decrease during the following hours. Sequential Bonferroni method was used to adjust p-values in post hoc pairwise comparisons of treatment groups.
Moreover, we noted that mobbing period was positively correlated with the duration of predator presentation (r = 0.67, p = 0.001, n = 20 nests) as well as the latency to mob (r = 0.71, p < 0.001, n = 20), while the duration of predator presentation and latency to mob were highly correlated to each other (r = 0.61, p = 0.003, n = 21). Note that the number of other species taking part in the mobbing was inversely correlated with the latency to mob (r = −0.59, p = 0.005, n = 20). In order to consider the confounding effect of latency, we adjusted the mobbing period and the number of other species for the latency to mob.
Parental visitation rates at the nearby nests were investigated using two repeated-measures GLMs. We measured parental behaviour of nearby pairs at their nest (i) during the predator presentation at the focal nests and (ii) after the removal of the predator. In both models, we used mean parental nest visitation rate of nearby pairs at their nest as a dependent variable, treatment (blackbird versus sparrowhawk) as a repeated factor, the distance to the focal nest, nestling age on the first experimental day and previous experience with the sparrowhawk presentation as covariates, and the interaction between treatment and covariates as predictors in the initial models. In the second model (after removal of the predator), we also used time (first, second, third, fourth hour after removal of model) as a repeated factor in addition to treatment and covariates. There is a strong correlation between the number of nests and the distance between focal and nearby nests (r = 0.72, p < 0.0001). Hence, in order to avoid inter-dependence (collinearity) of these predictors if they were included in the same models, we consider only distance in the analyses.
Upon presenting the predator model, some focal pairs started alarm calling immediately while others delayed responding for up to 8 min. Given that the visits by nearby pairs at their nests during the sparrowhawk presentation depended on the latency to mob (r = 0.58, p = 0.010, n = 20), we calculated the predicted values of provisioning rate (visits per minute) during the predator presentation on the latency to mob to be comparable with control group. Moreover, we studied the effect of the duration of mobbing at focal nest and the number of species participating in the mobbing on visitation rates of nearby pairs during the predator presentation and during the first hour after removal of the predator. Note that the mobbing duration and the number of other species taking part in the mobbing were highly and inversely correlated to each other (r = −0.58, p < 0.01). Therefore, their effects on visitation rates of nearby parents at their nests were tested in separate models to avoid collinearity. The models were simplified by removing the least significant parameters (p > 0.05) until only significant ones were left. In GLM models, partial eta-square (ŋ2) was used as a measure of effect size ranging between 0 and 1 (Statistica).
Blackbird models did not elicit mobbing; alarm calls during these treatments were briefly caused by the human observer but ended soon after the observer left the nest area. Sparrowhawk models instead caused mobbing in 23 of 25 focal nests. The mobbing period lasted up to 15 min (mean = 4.75, s.d. = 3.95) and usually ended within a few minutes after the removal of the sparrowhawk model (mean = 2.8, s.d. = 5.1).
Parental nest visitation rate changed with time as shown by a significant interaction between treatment and time (table 1a, figure 1a). Post hoc tests revealed that following predator presentations, parents of focal nests visited their nests at a 39% lower rate during the first hour (p = 0.0001) and at a 20% higher rate during the third hour (p = 0.004) when compared with controls. Previous experience with the sparrowhawk model did not explain the parental nest visitation rate (table 1a).
Focal and nearby parents visited their nests at similar rates during the 4 h period after treatment (F1,21 = 0.03, p = 0.87), while a three-way interaction between nest category, treatment and time indicated time-specific differences (F3,63 = 5.42, p = 0.002; figure 1). Post hoc tests revealed that focal and nearby pairs differed in nest visitation rates in the first (p = 0.012) and in the third hour (p < 0.001) in the predator group, and in the second hour (p = 0.024) in the control group.
Nearby parents changed nest visitation rate with time after removal of the predator, as shown by a significant interaction between treatment and time (table 1b, figure 1b). They visited their nests at a lower rate in the first hour after the predator presentation compared with the control (post hoc test: p = 0.034). The distance from the focal nest did not differently affect the way the nearby pairs reacted to sparrowhawk versus blackbird presentations (table 1b). Previous experience with the sparrowhawk model was not important (table 1b).
Nearby parents visited their own nest significantly less during the sparrowhawk presentation than during the blackbird presentation at the focal nest (table 1c; mean ± s.d. = 0.38 ± 0.19 for the predator and 0.62 ± 0.31 for the control treatment, given per minute). The effect of distance from the focal nest on the nest visitation rate of nearby pairs did not change depending on the type of presentation (the interaction between treatment and distance was not significant; table 1c). Previous experience with the sparrowhawk model was also not important (table 1c).
We found that six focal pairs out of 23 were not assisted by other pied flycatchers during the mobbing events. Non-mobbing nearby pairs (n = 6 pairs) changed their nest visitation rate similarly to those potentially joining in the mobs with focal parents (n = 16 pairs) during the presentation (treatment × type of nearby pair: F = 0.13, p = 0.7) as well as after removal of the predator model (treatment × type of nearby pair: F = 0.58, p = 0.45; treatment × time × type of nearby pair: F = 0.27, p = 0.85; mean ± s.d. for the mobbing versus non-mobbing nearby pairs: 32.51 ± 9.18 versus 26.86 ± 8.82 visits h−1 for the predator and 37.25 ± 9.82 versus 34.50 ± 7.20 visits h−1 for the control treatment, given for the first hour after the treatment). Moreover, in the sparrowhawk treatment group, we revealed a negative association between the number of other species mobbing at the focal nest and nest visitation rates of nearby pairs at their own nests during the predator presentation (GLM; F1,18 = 6.70, p = 0.021, rpartial = −0.57, n = 20 nests, corrected for the latency to mob; figure 2). Nest visitation rates of nearby pairs were not related to the mobbing duration (GLM, F1,18 = 0.45, p = 0.51; n = 20 nests, corrected for the latency to mob).
Our results show that brief predator events have local spatial effects on parental responses in bird communities. Pulses of predation risk therefore impact not only the behaviour of those individuals in the proximity of the event, but also the behaviour of individuals breeding further away. Our study further suggests that predator-induced acoustic cues may spread information about predation risk through communities that in turn affect parental visitation rates of breeding pairs even when a predator encounter takes place away from their nest.
Parental visitation rates at focal nests were altered by increasing the perceived predation risk. The direction of change in visitation rates (increase or decrease) differed with time periods after the predator presentation. Focal parents visited their nest less during the first hour, there was no difference during the second hour, and during the third hour parents increased their nest visitation rates. We suggest that parents might have visited their nest more frequently during the third hour to compensate for initial reduced nestling provisioning. Compensatory feeding has rarely been detailed in studies previously  and seems the most likely explanation because parental nest visits during nestling phase almost always entail provisioning [19,35]. However, irregular feeding may entail hidden costs because the capacity of nestlings to compensate for a temporary lack of food is limited . Our finding of decreased initial provisioning suggests that an increase in the perceived predation risk disrupts normal parental activities and, if occurring frequently, may have cumulative and potentially harmful consequences on offspring condition.
Our study indicates that a short predator encounter also affects the behaviour of the local population outside the immediate spatial vicinity. A predator event in a community has implications not only at the site of predator encounter but also at nearby nests of the same species, and potentially of other species. Our results revealed that nest visitations of nearby pairs were affected over 100 m from the source of risk. Nearby pairs decreased their nest visitation rates during the first hour after a predator visit at their own nests, returning to normal thereafter. Within our scale of examination, we found no effect of distance from the predator event (measured between 26 and 129 m) on the change in nest visitation rates. Given that it is not known how far the effect can reach, this suggests the spatial impact may extend beyond this distance in the community. Such spatial effect of brief predator encounters has not been documented before and requires further study.
The initial anti-predator response of nearby parents must be evoked through social cues of focal parents mobbing a predator. However, an interesting question is whether the nearby pairs gain additional information on the predator by actively mobbing or whether they rely on eavesdropping. It has been shown that the propensity of nearby pied flycatchers to join in the mobbing decreases considerably with increasing distance from the focal nest . The breeding pairs living in close proximity (20–24 m) always assist in mobbing because it might contribute to the protection of one's own nest . For more distant nearby pairs (36–46 m apart), only 61.5% of them assist nest owners, while those breeding 150 m from the focal nest do not mob the predator at all . Our experiment was not specifically designed to test these alternative possibilities. However, we found that although certain nearby pairs did not join in mobs, they still changed their parental activities similarly to those pairs that probably participated in the mobbing. Hence, we propose that reduction in the nest visits of nearby pairs could be explained by both mechanisms and it is a challenge for future studies to tease apart these possibilities. We still emphasize that despite how exactly this effect may have originated, the novel key finding is that events at the one particular (focal) nest are affecting the behaviour of the community around this nest.
We also found that the diversity of the mobbing flock impacted the response of nearby pairs. The more species that participated in mobbing at the focal nest, the more nearby pairs decreased their visitation at their own nests during the predator presentation. Multi-species mobs may recruit individuals from the neighbourhood [49,50]. The reliability of information increases if more individuals and species provide the same cues [24,31,51]. By contrast, single information sources may be false alarms causing unnecessary loss of foraging or parental care time . It is possible that nearby pairs were more likely to temporarily join multi-species mobs or they recognized the situation as more dangerous when secretly listening to more powerful social cues, leading to a greater decline in their nest visitation behaviour .
Overall, predator encounters may be frequent in nature and can have cumulative effects . Pulses of predation risk may influence the provisioning behaviour of focal as well as nearby pairs. Such effects may also include selective food distribution between differently sized offspring  and reduce brooding activity , potentially facilitating brood reduction mechanisms. Therefore, negative effects on offspring quality and survival are likely when predator encounters occur repeatedly.
Our results suggest quick and spatially moderate impacts on individual behaviour following brief predator encounters in a breeding bird assemblage; this can be a common phenomenon in all animal communities relying on acoustic signalling.
We thank two anonymous reviewers for their helpful comments on the previous version of the manuscript.
The study was conducted with the authorization of the Finnish national board on animal experiments (Animal Experiment Committee of Southern Finland, ID: VARELY/338/07.01/2012).
The article's supporting data can be accessed via Dryad repository.
K.M., V.T., R.L.T., T.L. designed the study; K.M., S.C., P.E.J., W.S., W.V., T.L. collected field data; K.M., V.T. performed data analysis; K.M., V.T., R.L.T. drafted the manuscript; S.C., P.E.J., W.S., W.V., T.L. helped finalizing the manuscript.
We have no competing interests.
The study received financial support from the Estonian Science Foundation (grant no. 8376), the Estonian Ministry of Education and Research (institutional research funding IUT no. 34-8), the European Development Fund (Centre of Excellence FIBIR) and Academy of Finland.