Birds - Escape decisions prior to pursuit - Escape and refuge use: theory and findings for major taxonomic groups - ESCAPING FROM PREDATORS: An Integrative View of Escape Decisions(2015)

ESCAPING FROM PREDATORS An Integrative View of Escape Decisions (2015)

Part II Escape and refuge use: theory and findings for major taxonomic groups

IIb Escape decisions prior to pursuit

4 Birds

Anders Pape Møller

4.1 Introduction

The study of flight initiation distance (FID) has a long history. Darwin (1868) described in detail how domestication resulted in loss of fear in birds and mammals, and that a number of different kinds of flight behavior seen in wild animals were lost when animals lived in human proximity. Hediger (1934) synthesized the literature on escape behavior including the distance at which animals became alert and that at which they fled an approaching predator, including humans, in a diverse array of organisms. The review by Hediger explicitly noted that different components of escape such as FID, flight direction, and alert distance (AD) reflected different aspects of flight behavior that could all, or in different combinations, change over the course of evolution. Hediger also pointed out how individuals, populations, and species differed in escape behavior, and that island populations showed less fear than mainland populations of the same or related species. Rand (1964) showed that FID was inversely related to body temperature in a lizard, implying that FID varied with risk of predation. Cooke (1980) noticed that urban birds had much shorter flight distances than rural populations of the same species, and that this difference depended on body size, the difference being larger in small species with high metabolism. This change in behavior was functional in the sense that it allowed birds to coexist with humans even at high human population densities that caused frequent disturbance. Burger (1981), Burger and Gochfeld (1981), and several others noticed that human disturbance affected FID at seabird colonies, and that such disturbance could result in habitat use alteration and reduced reproductive performance. These early studies placed research on escape behavior and FID firmly in the context of conservation biology, where it has remained ever since (Blumstein & Fernández-Juricic 2010).

In this chapter I use the definitions of FID and related terms as described in Chapter 1. Starting distance (SD) and AD are aspects of escape behavior that arise for methodological or biological reasons, and some of these components show significant covariation with FID (e.g., Blumstein 2003). Despite recent theoretical suggestions (Blumstein 2010; Cooper & Blumstein 2013) and empirical evidence (reviewed by Samia et al. 2013) that suggest biological mechanisms underlie covariation between AD (or SD) and FID, at least part of this covariation is artifactual (Dumont et al. 2012). In contrast to AD, SD can be controlled experimentally in the field by keeping a fixed distance to animals before recording FID (e.g., Møller 2012) or by recording SD as part of the approach protocol (Blumstein et al. 2005). The remainder of this chapter focuses on FID and the factors that account for variation in FID within and among individuals, populations, and species.

Here I present an exhaustive review of research findings dealing with FID in birds. I judge the magnitude of effects of the covariates by relying on effect sizes estimated in terms of Pearson product-moment correlation coefficients because they have the intuitively simple property that the value squares reflects the amount of variance explained (Cohen 1988). I evaluate the strength of relationships between variables (preferably partial effects after adjustment for confounding variables) judging than an r = 0.10, accounting for 1% of the variance is small, r = 0.30, accounting for 9% of the variance is intermediate, and r = 0.50, accounting for 25% of the variance is a large effect (Cohen 1988). These three levels can be compared with mean effect sizes in biology in general. Møller and Jennions (2002) showed in a meta-analysis of all meta-analyses in biology that, on average, main effects accounted for 5 to 7% of the variance, thus constituting an intermediate effect. Test statistics were transformed into effect sizes using equations listed in Rosenthal (1994). I also report sample size to allow readers to estimate confidence intervals of the effect sizes. The effect sizes for interspecific comparisons were generally based on phylogenetically controlled analyses accounting for similarity among taxa due to common phylogenetic descent, while intraspecific studies simply represented relationships across individuals. A small number of studies could not be included because test statistics were not reported. I list all effect sizes and the associated sample sizes in the electronic supplementary material, while summary statistics including weighted means and 95% confidence intervals for different categories of effects are reported in Table 4.1. These estimates for categories of effects were derived from a non-phylogenetic, random effects meta-analysis comparing effect sizes for different categories of studies using the MetaWin 2.1 statistical software package (Rosenberg et al. 2000). The main text of this chapter only presents some representative examples of effect sizes, while a complete list of effect sizes can be found in the electronic supplementary material. Finally, readers should note that Pearson product-moment correlations as a measure of effect size may not always be the most efficient way to estimate relationships, and other methods for judging relationships between pairs of variables have been suggested (Koricheva et al. 2013).

Table 4.1 Summary statistics (mean effect size and 95% confidence intervals) for different categories of phenotypic traits for flight initiation distance (FID) based on a non-phylogenetic random effects meta-analyses weighted by sample size (Rosenberg et al. 2000). Effects differing significantly from zero are shown in bold font.

Category

Mean effect size

Lower 95% confidence interval

Upper 95% confidence interval

No. studies

Urbanization

0.62

0.50

0.74

20

Body mass

0.39

0.27

0.51

20

Predation

0.37

0.28

0.47

32

Range

0.35

0.19

0.52

11

Dispersal

0.31

−0.35

0.97

3

Sexual display

0.29

−0.05

0.63

5

Sociality

0.26

0.17

0.34

35

Parasitism

0.22

0.08

0.37

14

Life history

0.20

0.10

0.30

27

Disturbance

0.19

0.04

0.35

13

Habitat

0.17

0.05

0.28

20

Brain

0.09

−0.33

0.50

4

Personality

−0.16

−0.97

0.65

3

To make a comprehensive review of studies of FID in birds, I searched the Web of Science using the key words FID, flight distance, and flight initiation distance. In the review, I start out by investigating (1) the frequency distributions of FIDs and their component parts. This part is based on a data set that I have collected, which comprises 9007 observations belonging to 181 species of European birds. Flight initiation distance was measured by having a fixed SD around 30 m to avoid covariation between FID and SD. Data collected from the same sites without the use of a fixed SD were very similar in terms of FID. Then I analyze (2) the phenotypic variation in FID and its environmental and genetic components, followed by (3) a summary of the different factors affecting FID, with a particular emphasis on life history because FID is squarely placed as the result of trade-offs between components of life history (Blumstein 2006; Møller & Garamszegi 2012). I then (4) analyze underlying assumptions linked to FID, such as its relationship with risk of predation and death, phenotypic adjustment related to predator avoidance, morphological adaptations, sensory organs related with (see also Chapter 12), and their cognitive aspects. Next, I (5) investigate the population consequences of disturbance on FID, and I finish (6) by providing some ideas for future prospects of research.

4.2 Frequency distributions of FID

Using the raw data of 132 to 181 species of European birds (depending on which statistical moment was analyzed; N = 9007 observations) I evaluated the frequency distributions of FID. I found that FID is characterized by a mean, and the moments that reflect variance, skewness, and kurtosis. The characteristics of the frequency distribution of FID are shown in Table 4.2 and Figure 4.1. There is considerable variation among species with a coefficient of variation of 108.

Table 4.2 Characteristics of mean, variance, skewness, and kurtosis of FID for 132 to 181 (depending on variable) species of birds from Europe. N = 9007 observations. Skewness and kurtosis were tested against the null hypothesis of no significant difference from zero (i.e., the parameter-value expected assuming a normal distribution).

Mean

SD

t

N

P

Mean

13.42

80.96

181

Variance

100.09

1827.75

159

Skewness

1.67

8.30

19.07

144

< 0.0001

Kurtosis

5.26

49.43

10.03

132

< 0.0001

(A. P. Møller unpublished data)

Figure 4.1

Frequency distributions of (A) mean (m), (B) standard deviation (m), (C) skewness, and (D) kurtosis of species-specific values of flight initiation distance in birds recorded by A. P. Møller. Sample size was 132 to 181 species depending on the moment. The box plots show means (diamonds), medians, quartiles, 95% confidence intervals, and extreme observations.

The frequency distributions of FID were significantly skewed toward the right and deviated significantly from the expected value of zero for normal distributions. In addition, there was significant leptokurtosis as reflected by the mean kurtosis value across species being positive. Wright (1968) showed, for the special case of a dichotomous individual difference, that the degree of leptokurtosis reflected the ratio of the variances of the two distributions. Hence species on average having leptokurtotic distributions imply that the overall population is composed of different subpopulations that differ inherently in degree of escape behavior. Thus the degree of leptokurtosis reflects heterogeneity in escape behavior among individuals, implying that a more heterogeneous population consists of more different kinds of individuals in terms of escape behavior.

4.3 FID and components of flight

Flying animals such as birds, bats and aquatic organisms are unique by moving in three dimensions, while all other organisms mainly move in two dimensions. This has important consequences for escape behavior (FID) because it is easier to escape in three than in two dimensions, since predators will have to pursue prey that can escape in one additional dimension. We should also expect that the horizontal and vertical components of FID should show design features that exploited these differences in escape propensity (Møller 2010b). Every single FID can be decomposed into its horizontal and vertical components, with the vertical component being zero in organisms moving in two dimensions (Blumstein et al. 2004; Fernández-Juricic et al. 2004). In a comparative study of 69 species of birds, Møller (2010b) showed that four times more variance in FID among species was due to the horizontal than the vertical dimension. The slope of the relationship between horizontal distance and FID (horizontal slope) increased with body mass, whereas the slope for the relationship between vertical distance and FID (vertical slope) decreased with body mass. In other words, large species rely more on the horizontal component of escape than small species, and this may be caused by the higher energy cost of flight in large species. The horizontal slope was negatively related to the vertical slope, albeit this negative relationship had a slope that was less than expected from a perfect trade-off. The horizontal slope decreased with increasing density of the habitat from grassland over shrub to trees, while that was not the case for the vertical slope. If the vertical slope provides a means for avoiding predation, we should expect adult survival rate to decrease with increasing vertical slope, which was indeed the case, while there was no relationship with the horizontal slope. In addition, species with a high rate of senescence had a high vertical slope, suggesting that an increase in vertical escape allowed for individuals of old age to accumulate due to the reduction in risk of predation.

4.4 Sources of variation in FID

4.4.1 Within- and among-individual variation

The variation within and among populations reflects differences among species (e.g., Blumstein et al. 2003). In a sample of 9289 FID estimates from Europe, 55% of the variance in FID occurred among species (F = 59.83, d.f. = 186, 9103, P < 0.0001). A model of mean FID weighted by sample size showed that 53% of the variance occurred among species (F = 195.53, d.f. = 1, 176, P < 0.0001). Thus more than half of the variance occurs among species with an almost similar fraction occurring within species.

4.4.2 Heritability, selection, and response to selection

Phenotypic variance in FID can be partitioned into environmental and genetic components, as well as gene × environment interactions (Falconer & Mackay 1996). Repeatability sets an upper limit to heritability (Falconer & Mackay 1996). Differences in FID among sites can be due to learning, sorting of phenotypes (differential migration of specific phenotypes to particular areas), or local adaptation. Repeatabilities of FID (calculated using intra-class correlation coefficients; Falconer & Mackay 1996) were reported for five species, with all values except one being 0.76 to 0.92 (Table 4.3). Four out of five studies had significant repeatability. The exception was de Jong et al. (2013) who showed an absence of significant repeatability for breeding curlews (Numenius arquata, de Jong et al. 2013). Estimates of repeatability within individuals among years ranged from 0.37 to 0.62 in two species. I (unpublished data) estimated a repeatability of mean FID of 0.60 among individuals within 181species from Europe (SE = 0.04), and a repeatability for the standard deviation in FID among individuals within 159 species of 0.69 (SE = 0.12). This implies that particular species are consistent in their mean and variance in FID.

Table 4.3 Repeatability within and among years and habituation for FID in different species of birds.

Species

Repeatability within years

Repeatability among years

Habituation

Reference

Somateria mollissima

0.76-0.80

0.37-0.69

None

Seltmann et al. 2012

Athene cunicularia

0.84-0.92

None

Carrete & Tella 2010

Athene cunicularia

0.85-0.91

Carrete & Tella 2013

Numenius arquata

0.04

de Jong et al. 2013

Hirundo rustica

0.92

0.62

None

A. P. Møller unpublished data

Passer domesticus

0.79

None

Møller & Garamszegi 2012

Ficedula albicollis

None

Møller & Garamszegi 2012

Melospiza melodia

None

Scales et al. 2011

Dolichonyx oryzivorus

None

Keyel et al. 2012

Cardinalis cardinalis

None

Smith-Castro & Rodewald 2010

There is only weak evidence of habituation, with repeated estimates of FID not showing a decline with repeat tests in eight different species ranging from ducks and owls to passerines (Table 4.3). While some studies may suffer from short intervals between encounters resulting in individuals recognizing the experimenter as not being dangerous, other studies have carefully controlled for such effects. The literature on habituation effects is clearly in need of a general review (Blumstein, 2014).

Broad sense heritability is the proportion of phenotypic variance due to additive genetic effects (Falconer & Mackay 1996). Only a single study has estimated heritability for FID. Parent-offspring resemblance in the barn swallow provided a heritability of FID of 0.38 (0.07) for male parent-offspring and 0.46 (0.07) for female parent-offspring based on 191 individuals, and animal models provided similar estimates (Møller 2014a). Hence there is considerable additive genetic variance in FID.

Selection can be estimated as the change in standardized phenotype before and after selection (Falconer & Mackay 1996). Møller et al. (2013) provided estimates of FID in resident and migratory birds before and after a severely cold winter along a latitudinal gradient in Europe. They provided evidence of a reduction in FID before and after the cold winter, but only among resident species, and more so for rural than urban populations of the same species. This suggests that there has been directional selection on FID. Similarly, rapid change in FID between urban and rural populations of juncos (Junco hyemalis) suggests adaptation to an urban environment (Atwell et al. 2012).

If FID is heritable, and if there is directional selection, we expect the response to selection will be the product of additive genetic variation and selection. Hence, FID should change across selection episodes when the trait is heritable. I am unaware of any animal model or other studies providing evidence of such response to selection. In the scenario where we have prior estimates of FID and animals are subjected to an extremely cold winter, we should expect a reduction in FID after the cold winter, with the phenotype changing over time toward the value recorded before the cold winter, if that original value was close to the optimum.

4.5 Biological causes of variation

4.5.1 Body size

It is an almost trivial finding that body size is the best predictor of FID in interspecific comparative analyses, because it is also the best predictor of life history, anatomy, physiology, behavior, and conservation status (Bennett & Owens 2002). Body size is also the best intraspecific predictor of these characters in species with indeterminate growth, such as fish, amphibians, and reptiles, but not in species with determinate growth such as birds and mammals (Bennett & Owens 2002). Thus a larger flying animal takes longer before take-off (Pennycuick 1989), and I would thus expect longer FID in larger species for this reason alone. Cooke (1980) reported that smaller bird species were more approachable in suburban than in rural areas. Blumstein (2006) found an effect size of 0.21 for 150 species of birds. Møller (2008b) reported an effect size equal to 0.64 for 100 species of European birds and Glover et al. (2011) an effect size of 0.77 for 28 species, while Weston et al. (2012) reported an effect size of 0.69 for 138 species of Australian birds. The allometry coefficient in Weston et al.’s study was 0.29 and in that of Møller (2008b) 0.27, respectively. These coefficients were smaller than isometry (which equals a slope of 1; in the study by Møller 2008b t = 24.33, d.f. = 97, P < 0.0001), implying relatively shorter FIDs for large-sized species. While the previously reported allometry coefficients are not phylogenetically controlled such an analysis showed an even less steep slope of only 0.20 for FID in relation to body mass in European birds (Møller 2008b). Hence FID shows negative allometry (i.e., the allometry coefficient of 0.20 is positive but significantly smaller than one). Thus large species have relatively shorter FID for their body size.

4.5.2 Life history

The decision to flee represents a life history decision because individuals trade foraging efficiency, and hence energy gain, against disturbance and risk of predation. Therefore individuals should optimize their escape behavior relative to their residual reproductive value, which equals the average survival and reproductive output of an individual of a given age (Roff 1992). Thus individuals with low reproductive output or survival prospects (e.g., because they are ill or suffer from parasitism) should take greater risks than healthy conspecifics. Therefore life history components are also expected to correlate with FID.

Old age at first reproduction implies that individuals should take small risks in order not to die before the start of reproduction. Blumstein (2006) showed a small effect size, as did Møller and Garamszegi (2012) for mean and variance in FID.

Clutch size and fecundity were negatively related to mean FID (Blumstein 2006; Møller & Garamszegi 2012). A study investigating the difference in FID and the difference in clutch size between temperate and tropical populations showed a strong effect (Møller & Liang 2013), implying that paired designs that automatically control for confounding variables provide particularly strong effects. Hatching success in curlews (Numenius arquata) peaked at intermediate FID (de Jong et al. 2013), as would be expected if long FID prevented efficient incubation. A similar line of argument can be used for a study by Seltmann et al. (2012), which found longer incubation periods in eiders (Somateria mollissima) with long FID. Interspecific studies of developmental periods only showed weakly related to FID with small effect sizes (Blumstein 2006).

Species with short and variable FID suffer less from disturbance and hence reach a state that allows reproduction more readily than individuals belonging to species with long and invariable FID. Accordingly duration of the breeding season decreased with mean FID and increased with the variance in FID (Møller & Garamszegi 2012).

Both juvenile and adult survival rate increased with mean FID and decreased with variance in FID (Møller & Garamszegi 2012), and this effect was mainly due to the vertical component of FID (Møller 2010b). Therefore, I should expect that longevity and rate of senescence should be related to the vertical component in FID. Indeed, rate of senescence decreased with increasing mean and variance in FID, especially for the vertical component of escape (Blumstein 2006; Møller 2010b; Møller & Garamszegi 2012). This means that individuals belonging to species that mainly escape in the vertical dimension age more slowly than individuals belonging to species that mainly escape in the horizontal dimension.

4.5.3 Urbanization

Urbanization often results in a reduction in fearfulness of animals, and this also applies to FID (Blumstein 2014; Møller 2014a). Cooke (1980) described strong differences in FID between suburban and rural populations of birds, and similar reports exist in the older urbanization literature (Tomialojc 1970; Klausnitzer 1985). Müller et al. (2013) recently discovered in analyses of candidate genes in 12 paired populations of blackbirds (Turdus merula) that a gene involved in harm avoidance differed strongly between urban and rural populations. This implies that there has been divergence between populations due to selection.

Flight initiation distance has been known to relate to urbanization for more than 30 years (Cooke 1980). The magnitude of the difference in FID between urban and rural populations is directly linked to time since urbanization (Møller 2008a). Two studies suggest that both mean and variance in FID relate to urbanization, apparently because a higher variance implies a greater diversity of phenotypes and hence more different behavioral phenotypes (or personalities) (Møller 2010a; Carrete & Tella 2011a,b; Møller & Garamszegi 2012). Urban populations also had shorter FID than rural populations of the same species in a study of patterns of FID across Europe (Díaz et al. 2013). The difference in FID between rural and urban populations decreased with increasing latitude, paralleling trends in raptor abundance. The latitudinal trend also reflects the fact that the history of urbanization of birds is much older in southern than in northern Europe.

Flight initiation distance may play a crucial role in urbanization of birds because relatively short FID will allow birds to coexist in the proximity of humans as shown in a paired comparison of urbanized and closely related non-urbanized species of birds (Møller 2009a). If only a portion of the phenotypes in the ancestral rural population with specific behavior first colonized urban areas, we should expect an initial reduction in variance in FID, because some phenotypes did not enter cities. This should be followed by an increase in variance in FID as more different behavioral phenotypes develop in novel urban habitats to which urban birds only have adapted recently. These scenarios were supported by independent studies comparing rural and urban birds in Europe and South America (Møller 2010a; Carrete & Tella 2011a,b).

Valcarcel and Fernández-Juricic (2009) tested the safe habitat hypothesis and found a strong relationship between FID and habitat. There is also indirect evidence of predators affecting FID of prey in urban areas because many birds seek refuge near human habitation where raptors are rare. In fact, there was a negative relationship between reduction in flight distance between rural and urban habitats and difference in FID between predators and prey (effect size = 0.37) (Møller 2012). The difference in FID between prey species and that of their predator increased with the preference of prey species by sparrowhawks (Accipiter nisus) relative to their abundance (effect size = 0.32) (Møller 2012). Similarly, Guay et al. (2013) recently reported that FID was longer when individual black swans (Cygnus atratus) were farther from water, which acted as a refuge. These findings fit well with the predominant role of predation in predicting urbanization of birds (Møller 2014b).

4.5.4 Song

Males generally compete more strongly for access to mates than females because females are a limiting resource. Thus we should expect that displaying males have reduced FID compared to non-displaying conspecifics. This hypothesis has been investigated by analyzing FID of singing males and males involved in other activities. Males mainly sing in order to attract mates or repel competitors. A second aspect of display that facilitates success is the height at which individuals display because high display sites facilitate transmission of song, but simultaneously expose individuals to predators.

Singing males took greater risks than males that were not singing by reducing their FID with an effect size that was large (effect size = 0.47, N = 40) (Møller et al. 2008). The difference in FID between singing and non-singing males was also related to greater song post-exposure with a large effect size (effect size = 0.50, N = 34). Such exposed display locations may increase the risk of predation. In contrast, sexual dichromatism was only weakly correlated with mean FID in singing males, with a small effect size (effect size = 0.05, N = 32) (Møller et al. 2008).

4.5.5 Hormones and FID

Testosterone is the primary sex hormone in male vertebrates affecting sexual behavior. Therefore high testosterone levels imply an increase in adoption of risky behavior. We might expect that FID would decrease in males relative to females during the breeding season, particularly so in species with high testosterone levels. In a comparative analysis of European birds, P. Tryjanowski and A. P. Møller (unpublished data) found that mean FID decreased with increasing testosterone level (effect size = 0.45). The sex difference in FID was positively correlated with testosterone level (effect size = 0.54, N = 16). Finally, males are expected to have greater variance in FID than females because males should take greater mean risks, but also greater variance in risk than females due to more intense competition for access to mates (Daly & Wilson 1983). Indeed, the sex difference in variance in FID was negatively related to testosterone (effect size = 0.53, N = 18) in a model that included the sex difference in body mass.

These endocrinological findings may also suggest that other hormones, such as corticosterone, could be involved in regulating FID. Indeed, Seltmann et al. (2012) showed a positive correlation between FID and corticosterone levels in incubating eiders. And studies of blackbirds have found lower testosterone and corticosterone levels in urban compared to rural individuals (Partecke et al. 2005, 2006), which is consistent with the hypothesis that hormones modulate risk taking and FID.

4.5.6 Hunting and disturbance

There is an extensive literature on FID and hunting that shows that hunted birds take flight earlier when approached by humans (Madsen 1995, 1998a, b; Madsen & Fox 1995; Weston et al. 2012). For example, Laursen et al. (2005) showed a large effect (0.49) of hunting on FID across 19 species of waterbirds in Denmark. However, this literature is not homogeneous. For example, Møller (2008a) found no significant difference in FID between hunted and other species in an analysis of 55 species of birds from Denmark during the hunting season in a model that controlled for body mass (hunting effect size = 0.004). Likewise, there was little difference in FID between inhabited and uninhabited areas in tropical China in an area with relatively little hunting (Møller & Liang 2013).

4.5.7 Diet

There are only a couple of studies of FID and diet. Carnivorous and omnivorous species of birds were more likely to be flighty than species with other diets (Blumstein 2006). Blumstein suggests that this may reflect a carry-over effect; species that eat live prey may be more generally attuned to movement and may detect threats at a greater distance. Møller and Erritzøe (2010, 2014) did not find evidence for species eating mobile prey having different FIDs than species consuming immobile food.

4.5.8 Sociality

Stankowich and Blumstein (2005) conducted a meta-analysis of risk assessment in animals and found an intermediate-sized effect in most studies. Cooperative breeders were more flighty than species with other breeding systems (Blumstein 2006). Laursen et al. (2005) reported intermediate to large effect sizes for intraspecific relationships between FID and flock size in different species of waterbirds with FID being longer for larger flocks. These findings are inconsistent with dilution effects, because if each individual enjoyed a smaller risk we should then expect a shorter FID in larger flocks. In contrast, the results are consistent with effects of many eyes scanning for the presence of a predator, although differences in phenotypic composition of differently sized flocks may be an alternative explanation. Finally, effect sizes for the interspecific relationship between FID and coloniality were all small (Møller 2008a).

4.5.9 Predation

Flight initiation distance was negatively related to susceptibility to predation by sparrowhawks (Møller et al. 2008). The difference in FID between urban and rural habitats was correlated with susceptibility to sparrowhawk predation, supporting the previously mentioned safe-habitat hypothesis (Møller 2008a). An intraspecific study that used playback of predator calls showed a strong positive effect of predator calls on increasing FID (Zanette et al.2011). Likewise, another study showed that the abundance of mammalian predators has an intermediate positive effect on FID in plover species (St. Clair et al. 2012). Hearing the call of a predator caused crimson rosellas (Platycercus elegans) to reduce their FID (Adams et al. 2006). Finally, a study showed a relationship between perch exposure and FID (Boyer et al. 2006).

4.5.10 Parasitism

Flight initiation distance decreased both with diversity and prevalence of blood parasites in birds, showing a small to intermediate effect. The difference in FID between singing and non-singing males increased strongly with prevalence of Plasmodium that causes malaria (Møller et al. 2008). Independently, the difference in FID between singing and non-singing males increased strongly with the concentration of natural antibodies (Møller et al. 2008). Martín et al. (2006) showed for chinstrap penguins (Pygoscelis antarctica) that FID decreased with strength of T-cell response, which reflects superior body condition.

4.5.11 Habitat

Most studies have shown intermediate to large effects of the relationship between FID and habitat openness and habitat diversity, although Blumstein (2006) found a small effect. In particular, variance in FID was related to being a habitat generalist. This finding is expected because a greater diversity of behavioral phenotypes can exploit more different habitats (Møller & Garamszegi 2012). Moreover, the horizontal component of FID was more strongly related to habitat openness than the vertical component (Møller & Garamszegi 2012).

4.5.12 Range and population size

Flight initiation distance decreased with increasing latitude (Díaz et al. 2013; Møller & Liang 2013). That was particularly the case in paired comparisons investigating the difference in FID between temperate and tropical populations showing large effect sizes for both mean and variance in FID (Møller & Liang 2013). These differences were linked to latitudinal differences in abundance or diversity of predators.

Mean FID decreased and variance in FID increased with increasing range size (Møller & Garamszegi 2012). Thus species with short and variable FIDs have large breeding ranges because these characteristics allow high rates of reproduction and colonization of many different habitats.

Mean FID decreased and variance in FID increased with population size (Møller & Garamszegi 2012). Population density showed quantitatively similar patterns (Lin et al. 2012; Møller & Garamszegi 2012).

4.5.13 Dispersal

Mean FID increased with natal dispersal distance. Variance in FID increased with dispersal distance in one study (Lin et al. 2012), but decreased with variance in FID in another (Møller & Garamszegi 2012).

4.5.14 Overall assessment of effect sizes for different categories of studies

Effect sizes for bivariate associations between FID and other variables were generally adjusted for allometry effects by the inclusion of body mass as an additional predictor. Effect sizes were also generally adjusted for similarity in phenotype among species due to common phylogenetic descent, as reported in the original publications. However, some studies did not remove effects of alert distance and starting distance. This was because AD was not studied due to difficulties of estimating this parameter without error. Starting distance was controlled statistically in many studies, or it was controlled experimentally by keeping SD fixed at approximately 30 m. I analyzed all effect sizes reported in the electronic supplementary material using random effects meta-analysis that relies on Q-statistics that have F-distributions with numerator and denominator degrees of freedom as in analyses of variance (Rosenberg et al. 2000). The 186 effect size estimates varied considerably from −0.81 to +0.90 (see electronic supplementary material) with a mean effect size of 0.30 (95% confidence intervals = 0.26, 0.34) in a random effects meta-analysis weighted by sample. Thus the mean effect size was of an intermediate magnitude, as is the common pattern in biology (Møller & Jennions 2002). Variation in individual effect sizes was slightly larger than expected due to sampling error (Qtotal = 242.10, df = 206, P = 0.0.043). The robustness of mean effects can be estimated from the number of null results required to eliminate the statistical significance of the mean effect size, the so-called fail-safe number (Rosenberg et al. 2000). Rosenthal’s fail-safe number of 21,793 unpublished studies means that the significant mean effect size could not readily be eliminated by unpublished studies. There was significant heterogeneity among categories of effect sizes (Table 4.1; Q = 58.79, df = 12, 194, P = 0.001). Nine out of 13 categories of effects were statistically significant by deviating from zero (Table 4.1). These were urbanization, body mass, predation, range, sociality, parasitism, disturbance, life history, and habitat. Effect sizes were large for urbanization and body mass, explaining more than 25% of the variance. Range size, sociality, parasitism, disturbance, life history, and habitat had intermediate effects, while dispersal, sexual display, and brain size accounted for non-significant, but intermediate to small effects. There was no significant heterogeneity when comparing intraspecific and interspecific studies (Q = 0.72, df = 1, 205, P = 0.73). Likewise, there was no significant heterogeneity when comparing effects sizes based on mean FID and variance in FID (Q = 0.40, df = 1, 205, P = 0.56).

In a meta-regression weighted by sample size I found a significant model that accounted for 23% of the variance as estimated from the pseudo R2 (F = 5.64, df = 13, 193, P < 0.0001). This model fitted the data (F = 1.09, df = 7, 186, P = 0.37). Significant predictors of effect size were category of effect (Table 4.2; F = 5.98, df = 12, 193, P < 0.0001) and marginally whether the study was intraspecific or interspecific (F = 2.68, df = 1, 193, P = 0.10; mean (SE) for intraspecific studies = 0.19 (0.05), interspecific studies = 0.26 (0.04)). Thus effect sizes differed among categories of characters and interspecific studies had larger effects than intraspecific studies.

4.6 Assumptions

4.6.1 FID reflects predation risk

An inherent assumption underlying the estimation of risk-taking from FID is that shorter FID translates into a greater risk of predation. While this assumption may intuitively appear likely, there are few tests of this assumption. The assumption can be tested at both the level of individual risk and average risk across populations or species. Møller (2014a) recorded FID for 2067 adult barn swallows from 1983 to 2012. Of these birds, 18 individuals were captured by predators (domestic cats (Felis catus) or sparrowhawks). There was a significantly shorter FID among de-predated individuals than survivors. There is also evidence of mean FID of different species of birds reflecting risk of predation. Møller et al. (2008) analyzed susceptibility of 63 species of birds in Denmark to predation by sparrowhawks in relation to mean FID, showing a negative relationship accounting for 13% of the variance after adjusting for body mass and body mass squared (Figure 4.2; effect size = 0.36). Predators prefer prey of intermediate body size because such prey is easier to handle while still providing significant resources, hence explaining the inclusion of polynomial effects of body mass. When mean flight distance for different species increased from 6 to 60 m, susceptibility to sparrowhawk predation decreased by a factor ten.

Figure 4.2

Population trend of 115 different species of European breeding birds in relation to mean flight initiation distance. Size of symbols reflects sample size. The partial effect is F = 4.80, d.f. = 1, 113, P = 0.031).

(Adapted from Møller 2008b)

There are two experimental studies of the effects of perceived predation caused by real predators on FID of birds. Adams et al. (2006) showed in playback experiments on crimson rosellas that individuals responded to predator calls. Zanette et al. (2011) provided direct experimental evidence for FID being linked to risk of predation by either playing back predator calls or control calls produced by species that are not predators of song sparrows (Melospiza melodia). This resulted in almost a tripling of FID in the experimental group compared to controls, yielding a large effect size (0.55). Unfortunately this experiment did not allow for discrimination between phenotypic plasticity and phenotypic sorting as the underlying mechanism because individuals were not tested with both treatments. Furthermore, the experiment did not allow discrimination between effects of nest predation and predation on adults as a cause of the change in FID. However, the experiment provides unequivocal evidence for FID being causally linked to predation risk.

An indirect test of the assumption that FID is related to risk of predation is based on basal metabolic rate (BMR). Basal metabolic rate is the minimal metabolic rate in the zone of thermoneutrality required for maintenance. Although many different factors may affect BMR, the fitness costs and benefits in terms of the ability to respond to predators should affect BMR. In other words, BMR is a cost of being wary toward predators. Møller (2009a) analyzed BMR and FID in 76 bird species and found that this positive relationship accounted for 27% of the variance (after inclusion of body mass as a confounding variable; effect size = 0.52). Inclusion of additional potentially confounding variables did not change this conclusion, nor did similarity in phenotype among species due to common phylogenetic descent. Therefore BMR is positively related to risk-taking behavior, and predation is an important factor in the evolution of BMR. It also suggests that the large energy requirements seen in species with high BMR may lead individuals to take greater risks when foraging.

4.6.2 Morphological adaptations to FID

Locomotor efficiency depends on oxygen uptake and transport, large muscles, and appendages that facilitate flight. Møller et al. (2013) predicted that hematocrit, which is a measure of packed red blood cell volume and hence reflects efficiency of oxygen transport, would be the highest in species with short FID. Consistent with this prediction, species with short FID had high hematocrit (effect size = 0.25).

Flight initiation distance could vary with wing and hind limb morphology as they can influence parameters associated with taking-off maneuvers (e.g., acceleration and lift). For instance, in a study on 83 bird species controlling for phylogenetic effects, FID was lower in species with rounded and convex wingtips perhaps due to the greater lift and thrust abilities compared to species with pointed and concave wingtips (Fernández-Juricic et al. 2006). However, the length of the femur and tarsus was not significantly correlated with FID (Fernández-Juricic et al. 2006). In addition, aspect ratio, which reflects maneuverability during flight, was large in species with long FID (effect size = 0.39; Møller et al. 2013). Finally, FID increased with wing area, which reflects low costs of lifting a bird with a given body size (effect size = 0.44). These effects were independent of potentially confounding variables (e.g., body size) and similarity due to common phylogenetic descent. These results suggest that physiological and morphological adaptations to FID have evolved as a means of reducing the costs of flight. Alternatively, species that have evolved adaptations to efficient flight have subsequently evolved specific FID. Either way, this suggests that birds with low costs of flight do not await closer approach by a potential predator to compensate for the costs of flight.

4.6.3 Sense organs

Prey rely on their sense organs for monitoring predators and adjusting antipredator behavior to level of risk. Thus well functioning eyes and ears are crucial for escape behavior. An initial study found no significant relationship between eye size and FID (Blumstein et al. 2004), but a more comprehensive analysis by Møller and Erritzøe (2010) analyzed eye size in 97 species of birds in relation to FID, predicting that species with large eyes (i.e., higher ability to resolve visual details) will have relatively longer FID as mediated by longer ADs. This effect size was intermediate (0.34). Visual acuity arises from an effect of eye size and retinal ganglion cells. However, the ability to focus on a predator will also depend on the size of lenses that is likely to mainly be adapted to efficient foraging. Lens size can be broken down into effects of lens size and shape. Indeed, A. P. Møller and J. Erritzøe (unpublished data) weighed and measured the depth and width of lenses in 84 species of birds, showing that FID increased with the interaction between lens depth and lens diameter independent of eye size and body size. This implies that species with long FID had relatively thick and wide lenses in their eyes independent of eye size.

Just as visual information may be crucial for monitoring the behavior of predators, auditory information may independently of visual information play a significant role in predator-prey interactions. The tympanic membrane is crucial for hearing and a relatively larger tympanic membrane for a given body size implies better hearing ability (A. P. Møller & J. Erritzøe unpublished data). Across 37 species of birds, FID was strongly positively related to the size of the tympanic membrane (effect size = 0.74). Likewise, the relationship between the size of the tympanic membrane and the footplate (the flat portion of the stapes, which is set into the oval window of the medial wall of the inner ear) implies better hearing ability, and again there was a significant positive relationship between FID and tympanic membrane/footplate (effect size = 0.35, N = 37). These aspects of sensory ecology for eyes and ears are treated in greater detail in Chapter 12.

4.6.4 Brains and cognition

Brains and cognition should play a role in predator-prey interactions because prey glean information on the whereabouts and predator behavior with sense organs (Chapter 12). Information acquired by the sense organs is processed by the brain, and this information subsequently may change an individual’s behavior. A large number of studies have investigated the ecological and evolutionary causes of brain size evolution.

This approach has recently been criticized for being superficial and non-scientific (Healy & Rowe 2007). Like all scientific enquiry this avenue of research is based on carefully recorded observations on brain size or size of component parts of the brain, and numerous attempts to verify the reliability of such data have shown a high degree of consistency, even in comparisons of head volume with brain size (Møller et al. 2011). Surprisingly, Healy and Rowe (2007) have not themselves adhered to such approaches in their own research by testing for reliability of their brain size data or adopting rigorous phylogenetic analyses, and they have based their analyses on small and heterogeneous sample sizes.

Guay et al. (2013) in an analysis of 27 bird species from Australia found no significant association between brain size and FID. Although a relatively larger brain (corrected for body mass) may allow for longer FID because of faster reaction to a potential predator, a larger brain may also allow assessment of the intentions of the predator and the likelihood of attack before fleeing. This scenario implies that visual information is gleaned by the eyes, processed by the brain and eventually used for making decisions about FID, with these decisions depending on the relative size of the brain (Møller & Erritzøe 2014). Indeed, in a sample of 107 bird species FID increased with relative eye size (effect size = 0.33), but decreased with relative brain size (effect size = -0.23). In other words, species with relatively larger brains for a given body size stayed put for longer before taking off. Møller and Erritzøe (2014) hypothesized that this relationship came about by species with relatively large brains for a given body size using this “extra” time for monitoring the intentions of the predator before taking off. Furthermore, FID increased independently with size of the cerebellum, which plays an important role in motor control (effect size = 0.37, N = 54). These findings are consistent with cognitive monitoring as an antipredator behavior that does not result in the fastest possible escape, but rather the least expensive escape flight that allows for monitoring of predator behavior. Indeed, attentional monitoring costs have been hypothesized to be one cost of monitoring predators (Blumstein 2010; Cooper & Blumstein 2013; Møller & Erritzøe 2014). Information assessment in the interaction between predators and prey may depend on the behavior of the approaching predator. Birds significantly increased their FID or AD when approaching humans looked directly at them rather than elsewhere, as reported by Eason et al. (2006) for American robins (Turdus migratorius), Bateman and Fleming (2011) for hadeda ibises (Bostrychia hagedash), Clucas et al. (2013) for American crows (Corvus brachyrhynchos) and Lee et al. (2013) for magpies (Pica pica). Hadeda ibises in addition reduced the FID when approached quickly rather than slowly by a human (Bateman & Fleming, 2011), although a study of galahs (Cacatua roseicapilla) only showed a small effect (effect size = 0.017, N = 50; Cárdenas et al. 2005). Thus many bird species can distinguish between potential predators intently looking at a prey individual and those only looking intermittently.

Cars are useful tools for assessing cognitive abilities of birds in relation to risk. Recently, birds were shown to adjust their FID to road speed limits by increasing FID as speed limit increases, while there was no similar effect of the speed of the vehicle as such (Legagneux & Ducatez 2013). This response may reduce the risk of collision and decrease mortality. In a second study, Mukherjee et al. (2013) showed that American crows (Corvus brachyrhynchos) adjusted their risk-taking behavior to the driving direction of cars in a particular lane. Thus crows in the opposite lane to that used by an approaching vehicle stayed put. Additionally, whereas a fifth of crows walked from the driving lane to the safe opposite lane, none walked in the opposite direction. These observations imply that birds are able to assess driving speed and direction, and use this information to reduce risk of collision.

A different way of assessing the effects of cognitive abilities on FID is to determine FID when birds are either approached directly by a human or approached tangentially at a fixed distance. Geist et al. (2005), Fernández-Juricic et al. (2005), Heil et al. (2007) and Bateman and Fleming (2011) showed that birds could distinguish these two kinds of approaches. Møller and Tryjanowski (2014) collected similar data for a large number of bird species in paired populations of rural and urban species. Birds were better able to assess the direction of approach by humans in rural than in urban habitats, apparently because rural birds are more often confronted with human disturbance even at long distances between humans and birds. Thus birds in urban habitats already have very short FID, making it less beneficial to distinguish between direct and tangential approaches.

While prey usually flee in the opposite direction of an approaching predator, there is considerable variation in flight direction among individuals, with some even directly approaching the predator (Domenici et al. 2011a, b). Møller (unpublished manuscript) estimated the angle of flight when recording FID, showing considerable more escapes around directions of 90° and 270° than expected by chance. Such escapes are basically escapes from the direction at which a potential predator (in this case a human) is approaching, resulting in energy savings by the bird. The proportion of individuals escaping at angles of 90° and 270° in 84 species of birds increased with brain size (effect size = 0.25), decreased with body mass (effect size = 0.38) and increased with urbanization (effect size = 0.46). A different way of adjusting flight to disturbance is to run away rather than fly. Rodriguez-Prieto et al. (2008a) showed for blackbirds (Turdus merula) exposed to frequent disturbance by humans fled on foot, rather than engaging in potentially expensive flight.

4.7 Population consequences

The individual consequences of frequent disturbance and the resulting flights to evade risk of predation are not well known. However, the high metabolic cost of short flights can be considerable. Tatner and Bryant (1986) measured costs of short flights typical for human disturbance of the European robin (Erithacus rubecula) using doubly labeled water. This species is common in forests, but also in urban habitats including gardens throughout Europe, and hence humans disturb robins at a high rate. The flight costs were extremely high at 23 times basal metabolic rate, implying that they could have consequences for reproduction and during cold spells even survival. Because robins typically have short FIDs (mean = 5.9 m, SD = 2.9 m, N = 247, A. P. Møller unpublished data), the flight costs estimated by Tatner and Bryant (1986) can be considered biologically relevant. Robins living in urban areas may be disturbed by humans, dogs, and cats numerous times per day with the consequence of significant energetic costs. Therefore it is reasonable to ask whether frequent short FIDs have population consequences. Indeed, some hypotheses predict changes in habitat/patch use based on the frequency of disturbance (e.g., number of disturbance events per day; Fernández-Juricic 2002; Frid & Dill 2002).

Costly defensive strategies can reduce population density of animals through indirect effects (Bolnick & Preisser 2005; Preisser et al. 2005). The population consequences of FIDs can be investigated by relying on extensive monitoring efforts by amateur ornithologists across Europe (Møller 2008b), but also in other continents. We may expect species with long FIDs for their body size to show declining population trends because humans disturb such species more often. Among 56 species of European birds, FID accounted for 33% of the variance in population trend with effect sizes ranging from 0.36 to 0.58 in different analyses. Therefore species with long FIDs for their body size had declining populations while species with short FIDs had increasing populations. That was also the case when controlling statistically for potentially confounding effects that are known to account for population trends such as migration distance, latitude, farmland, and brain size. Thaxter et al. (2010) analyzed population trends in the UK in relation to predictors, but found no significant effect of FID. The reason for this difference in conclusions between the two studies remains unknown. However, stronger effects of brain size on population trends in birds in North-Western Germany, compared to an area from Eastern Germany to the Czech Republic, have shown that effects can be context dependent (Reif et al. 2011). It is possible that bird populations have evolved shorter FIDs in countries with higher human population density, such as in the UK, and that such shorter FIDs reduce the impact of human disturbance. The apparent lack of a significant effect of FID on population trends in the UK also implies that the effect is stronger in other countries since the European population trend is based on trends in different countries weighted by the size of the country. The Thaxter et al. (2010) study had some shortcomings: they did not use FID data obtained in the UK, nor did they weight their analyses by sample size, which is required because estimates of FID differ in precision depending on sample size, nor did they report the effect size from their analyses.

A more detailed way of investigating the relationship between population trend and FID is to analyze trend and FID data for different countries. Díaz et al. (2013) analyzed 329 populations with information on both variables in the same country and found a strong negative correlation between population trend and FID. That was even the case when confounding variables such as migration distance, body mass, and brain mass were included in the models.

These findings raise the possibility that FID can be a useful tool in conservation including assessment of levels of disturbance and susceptibility to disturbance (Madsen 1995, 1998a, b; Tarlow & Blumstein 2007; Weston et al. 2012; see also Chapter 17).

4.8 Future prospects

There are numerous open research questions dealing with the causes and consequences of variation in avian FID. Many predator communities are dynamic with rapid spatial and temporal changes. That is, for example, the case with cyclical populations of predators, population declines caused by DDT and other pesticides, and urban environments being colonized by raptors. Such changes in risk of predation should have predictable consequences for risk-taking behavior and hence FID. Islands are interesting laboratories because communities of animals typically are impoverished, and predators are often rare or even completely absent (Blumstein & Daniel 2005). It would be interesting to compare FID of island and mainland bird populations of the same species. Likewise, it would be interesting to investigate FID on islands without mammalian, but with avian, predators. Such situations could potentially be used to make crosses between populations of prey species with and without predators, thereby testing the prediction that crosses should have intermediate FIDs between those of the two parental populations. Furthermore, there are no studies testing whether FID varies with respect to the location of predator nests or roosts. Is it the case the prey that live close to the nest of a raptor adjust their FID to the presence of predators by behaving more cautiously or more frequently use refuges to reduce the risk of predation? Finally, there is a need for studies of the fitness consequences of FID for banded individuals with a focus on risk of predation and reproductive success.

References

Adams, J. L., Camelio, K. W., Orique, M. J. & Blumstein, D. T. (2006). Does information of predators influence general wariness? Behavioral Ecology and Sociobiology, 60, 742-747.

Atwell, J. W., Cardoso, G. C., Whittaker, D. J. et al. (2012). Boldness behavior and stress physiology on a novel urban environment suggest rapid correlated evolutionary adaptation. Behavioral Ecology, 23, 960-969.

Bateman, P. W. & Flemming, P. A. (2011). Who are you looking at? Hadeda ibises use direction of gaze, head orientation and approach speed in their risk assessment of a potential predator. Journal of Zoology, 285, 316-323.

Bennett, P. M. & Owens, I. P. F. (2002). Evolutionary Ecology of Birds. Oxford: Oxford University Press.

Blumstein, D. T. (2003). Flight-initiation distance in birds is dependent on intruder starting distance. Journal of Wildlife Management, 67, 852-857.

Blumstein, D. T. (2006). Developing an evolutionary ecology of fear: How life history and natural history traits affect disturbance tolerance in birds. Animal Behaviour, 71, 389-399.

Blumstein, D. T. (2014). Attention, habituation, and anti-predator behaviour: Implications for urban birds. In D. Gil & H. Brumm (eds). Avian Urban Ecology. Oxford: Oxford University Press, pp. 41-53.

Blumstein, D. T. & Daniel, J. C. (2005). The loss of anti-predator behaviour following isolation on islands. Proceedings of the Royal Society of London - Series B, 272, 1663-1668.

Blumstein, D. T. & Fernández-Juricic, E. (2010). A Primer of Conservation Behavior. Sunderland, MA: Sinauer.

Blumstein, D. T., Anthony, L. L., Harcourt, R. & Ross, G. (2003). Testing a key assumption of buffer zones: Is flight initiation distance a species-specific trait? Biological Conservation, 110, 97-100.

Blumstein, D. T., Fernández-Juricic, E., LeDee, O. et al. (2004). Avian risk assessment: Effects of perching height and detectability. Ethology, 110, 273-285.

Bolnick, D. I. & Preisser, E. L. (2005). Resource competition modifies the strength of trait-mediated predator-prey interactions: A meta-analysis. Ecology, 86, 2771-2779.

Boyer, J. S., Hass, L. L., Lurie, M. H. & Blumstein, D. T. (2006). Effect of visibility on time allocation and escape decisions in crimson rosellas. Australian Journal of Zoology, 54, 363-367.

Burger, J. (1981). The effect of human activity on birds at a coastal bay. Biological Conservation, 21, 231-241.

Burger, J. & Gochfeld, M. (1981). Discrimination of the threat of direct versus tangential approach to the nest by incubating herring and great black-backed gulls. Journal of Comparative Physiology and Psychology, 95, 676-684.

Cárdenas, Y. L., Shen, B., Zung, L. & Blumstein, D. T. (2005). Evaluating temporal and spatial margins of safety in galahs. Animal Behaviour, 70, 1395-1399.

Carrete, M. & Tella, J. L. (2010). Individual consistency in flight initiation distances in burrowing owls: A new hypothesis on disturbance-induced habitat selection. Biology Letters, 6, 167-170.

Carrete, M. & Tella, J. L. (2011a). Inter-individual variability in fear of humans and relative brain size of the species are related to contemporary urban invasions in birds. Public Library of Science One, 6, e18859.

Carrete, M. & Tella, J. L. (2013). High individual consistency in fear of humans throughout the adult lifespan of rural and urban burrowing owls. Scientific Reports, 3, 3524.

Clucas, B., Marzluff, J. M., Mackovjak, D. & Palmquist, I. (2013). Do American crows pay attention to human gaze and facial expressions? Ethology, 119, 1-7.

Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence Erlbaum.

Cooke, A. S. (1980). Observations on how close certain passerine species will tolerate an approaching human in rural and suburban areas. Biological Conservation, 18, 85-88.

Cooper, W. E. & Blumstein, D. T. (2013). Novel effects of monitoring predators on costs of fleeing and not fleeing explain flushing early in economic escape theory. Behavioral Ecology, 25, 44-52.

Daly, M. & Wilson, M. (1983). Sex, Evolution, and Behavior, 2nd edn. Boston, MA: Willard Grant.

Darwin, C. (1868). The Variation of Animals and Plants under Domestication. London: John Murray.

de Jong, A., Magnhagen, C. & Thulin, C.-G. (2013). Variable flight initiation distance in incubating European curlew. Behavioral Ecology and Sociobiology, 67, 1089-1096.

Díaz, M., Møller, A. P., Flensted-Jensen, E. et al. (2013). The geography of fear: A latitudinal gradient in anti-predator escape distances of birds across Europe. Public Library of Science One, 8, e64634.

Domenici, P., Blagburn, J. M. & Bacon, J. P. (2011a). Animal escapology I: Theoretical issues and emerging trends in escape trajectories. Journal of Experimental Biology, 214, 2463-2473.

Domenici, P., Blagburn, J. M. & Bacon, J. P. (2011b). Animal escapology II: Escape trajectory case studies. Journal of Experimental Biology, 214, 2474-2494.

Dumont, F., Pasquaretta, C., Reale, D., Bogliani, G. & von Hardenberg, A.(2012). Flight initiation distance: Biological effect or mathematical artefact? Ethology, 118, 1051-1062.

Eason, P. K., Sherman, P. T., Rankin, O. & Coleman, B. (2006). Factors affecting flight initiation distance in American robins. Journal of Wildlife Management, 70, 1796-1800.

Falconer, D. S. & Mackay, T. F. C. (1996). Introduction to Quantitative Genetics, 4th edn. New York, NY: Longman.

Fernández-Juricic, E. (2002). Can human disturbance promote nestedness? A case study with birds in an urban fragmented landscape. Oecologia, 131, 269-278.

Fernández-Juricic, E., Vaca, R. & Schroeder. N. (2004). Spatial and temporal responses of forest birds to human approaches in a protected area and implications for two management strategies. Biological Conservation, 117, 407-416.

Fernández-Juricic, E., Venier, P., Renison, D. & Blumstein, F. T. (2005). Sensitivity of wildlife to spatial patterns of recreationist Biological Conservation, 125, 225-235.

Fernández-Juricic, E., Blumstein, D. T., Abrica, G. et al. (2006). Relationships of anti-predator escape and post-escape responses with body mass and morphology: A comparative avian study. Evolutionary Ecology Research, 8, 731-752.

Frid, A. & Dill, L. M. (2002). Human-caused disturbance stimuli as a form of predation risk. Conservation Ecology, 6, www.consecol.org/Journal/vol6/iss1/art11/print.pdf.

Geist, C., Liao, J., Libby, S. & Blumstein, D. T. (2005). Does intruder group size and orientation affect flight initiation distance in birds? Animal Biodiversity and Conservation, 28, 69-73.

Glover, H. K., Weston, M. A., Maguire, G. S., Miller, K. K. & Christie, B. A. (2011). Towards ecologically meaningful and socially acceptable buffers: Response distances of shorebirds in Victoria, Australia, to human disturbance. Landscape and Urban Planning, 103, 326-334.

Guay, P.-J., Weston, M. A., Symonds, M. R. E. & Glover, H. K. (2013). Brains and bravery: Little evidence of a relationship between brain size and flightiness in shorebirds. Austral Ecology, 38, 516-522.

Healy, S. D. & Rowe, C. (2007). A critique of comparative studies of brain size. Proceedings of the Royal Society B, 274, 453-464.

Hediger, H. (1934). Zur Biologie und Psychologie der Flucht bei Tieren. Biologisches Zentralblatt, 54, 21-40.

Heil, L., Fernández-Juricic, E., Renison, D. et al. (2007). Avian responses to tourism in the biogeographically isolated high Córdoba Mountains, Argentina. Biodiversity and Conservation, 16, 1009-1026.

Keyel, A. C., Peck, D. T. & Reed, J. M. (2012). No evidence far individual assortment by temperament relative to patch area or patch openness in the bobolink. Condor, 114, 212-218.

Koricheva, J., Gurevich, J. & Mengersen, K. (2013). Handbook of Meta-analysis in Ecology and Evolution. Princeton, NJ: Princeton University Press.

Laursen, K., Kahlert, J. & Frikke, J. (2005). Factors affecting escape distances of staging waterbirds. Wildlife Biology, 11, 13-19.

Lee, S., Hwang, S., Joe, Y. et al. (2013). Direct look from a predator shortens the risk-assessment time by prey. Public Library of Science One, 8, e64977.

Legagneux, P. & Ducatez, S. (2013). European birds adjust their flight initiation distance to road speed limits. Biology Letters, 9, 20130417.

Lin, T., Coppack, T., Lin, Q. et al. (2012). Does avian flight initiation distance indicate tolerance towards urban disturbance?Ecological Indicators, 15, 30-35.

Madsen, J. (1995). Impacts of disturbance on migratory waterfowl. Ibis, 137, S67-S74.

Madsen, J. (1998a). Experimental refuges for migratory waterfowl in Danish wetlands. I. Baseline assessment of disturbance effects of recreational activities. Journal of Applied Ecology, 35, 386-397.

Madsen, J. (1998b). Experimental refuges for migratory waterfowl in Danish wetlands. II. Tests of hunting disturbance effects. Journal of Applied Ecology, 35, 398-417.

Madsen, J. & Fox, A. D. (1995). Impacts of hunting disturbance on waterbirds: A review. Wildlife Biology, 1, 193-207.

Martín, J., de Neve, L., Polo, V., Fargallo, J. A. & Soler, M. (2006). Health-dependent vulnerability to predation affects escape responses of unguarded chinstrap penguin chicks. Behavioral Ecology and Sociobiology, 60, 778-784.

Møller, A. P. (2008a). Flight distance of urban birds, predation and selection for urban life. Behavioral Ecology and Sociobiology, 63, 63-75.

Møller, A. P. (2008b). Flight distance and population trends in European breeding birds. Behavioral Ecology, 19, 1095-1102.

Møller, A. P. (2009a). Basal metabolic rate and risk taking behavior in birds. Journal of Evolutionary Biology, 22, 2420-2429.

Møller, A. P. (2009b). Successful city dwellers: A comparative study of the ecological characteristics of urban birds in the Western Palearctic. Oecologia, 159, 849-858.

Møller, A. P. (2010a). Interspecific variation in fear responses predicts urbanization in birds. Behavioral Ecology, 21, 365-371.

Møller, A. P. (2010b). Up, up, and away: Relative importance of horizontal and vertical escape from predators for survival and senescence. Journal of Evolutionary Biology, 23, 1689-1698.

Møller, A. P. (2012). Urban areas as refuges from predators and flight distance of prey. Behavioral Ecology, 23, 1030-1035.

Møller, A. P. (2014). Life history, predation and flight initiation distance in a migratory bird. Journal of Evolutionary Biology, 27, 1105-1113.

Møller, A. P. (2014a). Behavioural and ecological predictors of urbanization. In D. Gil and H. Brumm (eds.) Avian Urban Ecology. Oxford: Oxford University Press, pp. 54-68

Møller, A. P. (2014b). Urban birds use cheap escape: Variance in escape direction from predators by prey.

Møller, A. P. & Erritzøe, J. (2010). Flight distance and eye size in birds. Ethology, 116, 458-465.

Møller, A. P. & Erritzøe, J. (2014). Predator-prey interactions, flight initiation distance and brain size. Journal of Evolutionary Biology, 27, 34-42.

Møller, A. P. & Garamszegi, L. Z. (2012). Between individual variation in risk taking behavior and its life history consequences. Behavioral Ecology, 23, 843-853.

Møller, A.P. & Jennions, M.D. (2002). How much variance can be explained by ecologists and evolutionary biologists? Oecologia, 132, 492-500.

Møller, A. P. & Liang, W. (2013). Tropical birds take small risks. Behavioral Ecology, 24, 267-272.

Møller, A. P. & Tryjanowski, P. (2014). Direction of approach by predators and flight initiation distance of urban and rural populations of birds. Behavioral Ecology, 25, 960-966.

Møller, A. P., Nielsen, J. T. & Garamszegi, L. Z. (2008). Risk taking by singing males. Behavioral Ecology, 19, 41-53.

Møller, A. P., Bonisoli-Alquati, A., Rudolfsen, G. & Mousseau, T. A. (2011). Chernobyl birds have smaller brains. Public Library of Science One, 6, e16862.

Møller, A. P., Vágási, C. I. & Pap, P. L. (2013). Risk-taking and the evolution of mechanisms for rapid escape from predators. Journal of Evolutionary Biology, 26, 1143-1150.

Mukherjee, S., Ray-Mukherjee, J. & Sarabia, R. (2013). Behaviour of American crows (Corvus brachyrhynchos) when encountering an oncoming vehicle. Canadian Field-Naturalist, 127, 229-233.

Müller, J. C., Partecke, J., Hatchwell, B. J., Gaston, K. J. & Evans, K. L. (2013). Candidate gene polymorphisms for behavioural adaptations during urbanization in blackbirds. Molecular Ecology, 22, 3629-3637.

Partecke, J., Schwabl, I. & Gwinner, E. (2006). Stress and the city: urbanisation and its effects on the stress physiology in European blackbirds. Ecology, 87, 1945-1952.

Partecke, J., Van’t Hof, T. J. & Gwinner, E. (2005). Underlying physiological control of reproduction and forest-dwelling blackbirds Turdus merula. Journal of Avian Biology, 36, 295-305.

Pennycuick, C. J. (1989). Bird Flight Performance. Oxford: Oxford University Press.

Preisser, E. L., Bolnick, D. I. & Benard, M. F. (2005). Scared to death? The effects of intimidation and consumption in predator-prey interactions. Ecology, 86, 501-509.

Rand, A. S. (1964). Inverse relationship between temperature and shyness in the lizard Anolis lineatopus. Ecology, 45, 863-864.

Reif, J., Böhning-Gaese, K., Flade, M., Schwarz, J. & Schwager, M. (2011). Population trends of birds across the iron curtain: Brain matters. Biological Conservation, 144, 2524-2533.

Rodriguez-Prieto, I., Fernández-Juricic, E. & Martín, J. (2008a). To run or to fly: Low cost versus low risk strategies in blackbirds. Behaviour, 125, 1125-1138.

Roff, D. (1992). Life History Evolution. New York, NY: Chapman & Hall.

Rosenthal, R. (1994). Parametric measures of effect size. In Cooper, H. & Hedges, L. V. (eds). The Handbook of Research Synthesis. New York, NY: Russel Sage Foundation, pp. 231-244.

Rosenberg, M. S., Adams, D. C. & Gurevitch, J. (2000). MetaWin: Statistical Software for Meta-analysis. Version 2.1. Sunderland, MA: Sinauer Associates.

St. Clair, J. J. H., García-Peña, G. E., Woods, R. W. & Székely, T. (2012). Presence of mammalian predators decreases tolerance to human disturbance in a breeding shorebird. Behavioral Ecology, 21, 1285-1292.

Samia, D. S. M., Nomura, F. & Blumstein, D. T. (2013). Do animals generally flush early and avoid the rush? A meta-analysis. Biology Letters, 9, 20130016.

Scales, J., Hyman, J. & Hughes, M. (2011). Behavioral syndromes break down in urban song sparrow populations. Ethology, 117, 1-9.

Seltmann, M. W., Öst, M., Jaatinen, K. et al. (2012). Stress responsiveness, age and body condition interactively affect flight initiation distance in breeding female eiders. Animal Behaviour, 84, 889-896.

Smith-Castro, J. R. & Rodewald, A. D. (2010). Behavioral responses of nesting birds to human disturbance along recreational trails. Journal of Field Ornithology, 81, 130-138.

Stankowich, T. & Blumstein, D. T. (2005). Fear in animals: A meta-analysis and review of risk assessment. Proceedings of the Royal Society of London B, 272, 2627-2634.

Tarlow, E. M. & Blumstein, D. T. (2007). Evaluating methods to quantify anthropogenic stressors on wild animals. Applied Animal Behaviour Science, 102, 429-451.

Tatner, P. & Bryant, D. M. (1986). Flight cost of a small passerine measured using doubly labeled water: Implications for energetic studies. Auk, 103, 169-180.

Thaxter, C. B., Joys, A. C., Gregory, R. D., Baillie, S. R. & Noble, D. G. (2010). Hypotheses to explain patterns of population change among breeding bird species in England. Biological Conservation, 143, 2006-2019.

Tomialojc, L. (1970). Quantitative studies on the synanthropic avifauna of Legnica town and its environs. Acta Ornithologica, 12, 293-392.

Valcarcel, A. & Fernández-Juricic, E. (2009). Antipredator strategies of house finches: Are urban habitats safe spots from predators even when humans are around? Behavioral Ecology and Sociobiology, 63, 673-685.

Weston, M. A., McLeod, E. M., Blumstein, D. T. & Guay, P.-J. (2012). A review of flight-initiation distances and their application to managing disturbance to Australian birds. Emu, 112, 269-286.

Wright, S. (1968). Evolution and the Genetics of Populations. Vol. 1. Genetic and Biometric Foundations. Chicago, IL: University of Chicago Press.

Zanette, L. T., White, A. F., Allen, M. C. & Clinchy, M. (2011). Perceived predation risk reduces the number of offspring songbirds produce per year. Science, 334, 1398-1401.