Flexibility in European starlings’ use of social information: experiments with decoys in different populations

Animal Behaviour 80 (2010) 965e973

Contents lists available at ScienceDirect

Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav

Flexibility in European starlings’ use of social information: experiments with decoys in different populations Alexandra Rodriguez a,1, Martine Hausberger b, 2, Philippe Clergeau c, * a

INRA ScribeeUMR Ecologie et Santé des Ecosystèmes CNRSeUniversity of Rennes 1, UMR 6552 Ethologie animale et humaine c Museum National d’Histoire Naturelle, UMR 5173 Conservation des Espèces, Restauration et Suivi des Populations b

a r t i c l e i n f o Article history: Received 29 April 2009 Initial acceptance 15 June 2009 Final acceptance 5 August 2010 Available online 9 October 2010 MS. number: 09-00277R Keywords: biological invasion colonization decoy European starling local enhancement social cue social information Sturnus vulgaris

Individuals confronted with novel environments, for instance recently modified environments such as towns or recently occupied habitats such as colonization fronts, have to cope with a lack of information about the location of resources. We hypothesized that, under these conditions, individuals of a social species such as European starlings, Sturnus vulgaris, would be more responsive to social cues indicating food presence than individuals belonging to populations already established in a well-known habitat. To test this hypothesis, we displayed starling decoys with various age, group size and interindividual distance characteristics in feeding postures to evaluate their attractiveness to starlings from towns and colonization fronts and noted whether attracted individuals joined particular decoy groups. Our results supported our hypothesis that sensitivity to social cues varied according to the population’s history, and we suggest that these cues enhance the success of starlings in occupying new habitats. Ó 2010 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

Before occupying and breeding in a new environment, animals must explore it and collect information. Individuals need to have at least partial knowledge of their environment’s food, habitat resources (Beauchamp & Ruxton 2005) and suitable breeding sites (Doligez et al. 2002). Information about habitat characteristics may have various origins: (1) ‘preharvest information’ gained before arriving on a site (Valone 1991); (2) ‘private’ or individually gained information collected by exploration and trial-and-error learning (Chesler 1969; Valone 1989); and (3) ‘social information’ obtained through social facilitation by watching other individuals foraging, testing new food or solving new problems (Clayton 1978; Coussi-Korbel & Fragaszy 1995; Galef & Laland 2005; Jackson & Ruxton 2006). The proportion of each kind of information and its characteristics depend on

* Correspondence: P. Clergeau, Museum National d’Histoire Naturelle, UMR 5173 Conservation des Espèces, Restauration et Suivi des Populations, 55 rue Buffon, 75005 Paris, France. E-mail address: [email protected] (P. Clergeau). 1 A. Rodriguez is at INRA ScribeeUMR Ecologie et Santé des Ecosystèmes, Campus de Beaulieu, 35042 Rennes cedex, France. 2 M. Hausberger is at CNRS-University of Rennes 1, UMR 6552 Ethologie animale et humaine, Campus de Beaulieu, 35042 Rennes cedex, France.

accessibility, individuals’ experience and social context (Valone & Giraldeau 1993; Templeton & Giraldeau 1996; Valone & Templeton 2002). Social information is usually also divided into social cues and public information. Social cues are simple indicators of the presence of conspecifics in a potentially interesting site, whereas public information involves more precise information such as the feeding or breeding success (indicated by the number of prey captured by conspecifics in a patch or the number of surviving young in a breeding area; Danchin et al. 2004; Parejo et al. 2008). Public information thus gives more precise information about patch quality. In the study conducted here we focused on social cues. Individuals that leave their population of origin and disperse into new habitats can be confronted with a lack of preharvest and private information. Nevertheless, it is crucial to get information about resource distribution efficiently to avoid starvation and ensure survival. In ‘new’ areas such as those individuals have colonized after an increase in resources or where they have been translocated, individuals have to cope with unpredictability and the potential dangers of novelty. Taking account of conspecific activities and experience can thus be a strategy to compensate for these difficulties. On the one hand, Ryer & Olla (1992) demonstrated that juvenile walleye pollocks, Theragra chalcogramma, exploit spatially variable, ephemeral food patches more successfully when foraging

0003-3472/$38.00 Ó 2010 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.anbehav.2010.08.010

966

A. Rodriguez et al. / Animal Behaviour 80 (2010) 965e973

in groups by local enhancement. On the other hand, red crossbills, Loxia curvirostra, avoid exploring patches where other individuals have already foraged unsuccessfully and thus spend less time foraging on poor patches and longer on rich patches (Smith et al. 1999). Joining successful foraging conspecifics on suitable sites and avoiding unsuccessful ones can thus be a good mechanism to cope with unpredictability in ‘new’ habitats. It can be especially useful to be able to recognize experienced individuals that could give reliable information about resource distribution. For example, young canaries, Serinus canaria, eat more seeds after seeing their fathers eating seeds than when they have no tutor, and they tend to eat the same kinds of seeds (Cadieu et al. 1995). Adret-Hausberger & Cumming (1987) and Wauters et al. (2002) demonstrated that young chickens, Gallus gallus domesticus, are attracted to older individuals at feeding places and that they choose the same type of food as their mothers or older peers. These observations suggest that information from adults or experienced peers can be important for naïve individuals. In the same way, Slaa et al. (2003), who studied social stingless bees, found that in the species Trigona amalthea, the decision to join other foragers was affected by the experience of individuals with a food source: newly recruited foragers showed local enhancement, whereas experienced foragers showed social avoidance conducive to local inhibition. We hypothesized that social information is thus more important for populations in newly colonized areas than for well-established populations and that particular visible characteristics of animal groups may influence their attraction. Not only the age and activity of conspecifics but also the number and form of the group could be informative (Clergeau 1982; Burger 1988). We chose the European starling, Sturnus vulgaris, as our biological model since it has colonized a variety of habitats successfully, becoming an invasive species in many regions (Feare 1984). We investigated the attraction to ‘foraging’ starling decoys of starlings from: (1) long-established populations in rural areas; (2) populations that have recently colonized new environments such as towns; and (3) populations that have invaded new areas at the edge of the species’ European range in the colonization front. As the positions and the numbers of individuals on a feeding site may influence the landing decisions of birds flying over the area, we also tested the influence of social aspects such as groups’ sizes and interindividual distances by modifying the presentation of the decoys. The birds’ age and experience (naïve or informed, adult or young) may also influence the decisions to join different types of groups. We suggest that naïve individuals should more frequently rely on the information provided by experienced ones (as adults) and thus should tend more frequently to join these kinds of individuals. For this reason, we also tested the effect of the age of the birds on their reaction towards three groups of decoys representing different ages. METHODS Studied Populations We compared the reactions of starlings from four areas where the populations had different colonization histories and that reflected different situations in which individuals have different access to information. We chose to test starlings on 10e15 sites in each area rather than to focus on only one group per population, thus accounting for the distribution and density of colonies. Two populations and their respective geographical areas were selected in western France. (1) Rural populations in Brittany (Rural Brittany) where starlings have been established for several centuries (Richard 1826). This

population includes experienced adults which know the habitat well and are capable of transmitting information to younger individuals, reducing the proportion of naïve individuals. Sixteen sites in agricultural areas were selected in the country near Rennes and Saint-Brieuc (towns approximately 100 km apart; 175 birds were tested during the breeding period and 439 during the postbreeding period). (2) Urban populations in Brittany (Urban Brittany). Starlings have been breeding in these urban areas for 30e100 years (Clergeau 1981). This population uses habitats frequently disturbed by anthropogenic activities and with scarce resources particularly difficult to get during the breeding period (Mennechez & Clergeau 2006). Twelve sites within Rennes and Saint-Brieuc were selected (153 birds were tested during the breeding period and 123 during the postbreeding period). We also selected two Italian populations, located in the propagation front in the south where individuals exploited some cities and cliffs along the Adriatic Sea less than 30 years ago. Recent touristic activities and new irrigation practices have changed the landscape structure and availability of new food resources (G. La Gioia, personal communication). These populations can also receive individuals from the north, increasing the proportion of naïve individuals in them. (3) Rural populations in the Apuglia region (Rural Italy) described as one of the most southern and recent colonization fronts in Europe (Castiglia & Tabarrini 1982). Eleven sites between San Cataldo and Otranto (70 km along the Adriatic coast) were selected (152 birds were tested during the breeding period and 91 during the postbreeding period). (4) Urban populations of Apulia (Urban Italy), also on the colonization front (Pasquali 1984). Nine sites in Bari and Mola di Bari were selected (130 birds were tested during the breeding period and 66 during the postbreeding period). In addition, we experimented in an area of buildings and gardens (Suburban Brittany) on the periphery of Rennes during the postbreeding period to evaluate the attraction gradient from urban to rural areas (224 birds tested). All experimental sites were open short-grass fields measuring between 60
45 m and 120
90 m. We chose lawn parks, football grounds and grazed grass fields so that sites were as similar as possible. Experimental Procedure Decoys have proven to be good attractors facilitating aggregation (Williams & Schwab 1973; Jeffries & Brunton 2001) and useful tools to study patch selection and lek composition (Pius & Leberg 2002; Master et al. 2005; Jiguet & Bretagnolle 2006). Most studies using decoys to test their visual attractiveness to birds have shown that stuffed decoys are better attractors than twoor three-dimensional plastic decoys (Fancher 1984; Kotliar & Burger 1984; Harvey et al. 1995; Crozier & Gawlik 2002). Decoys with their heads pointing down (as if they were eating) test attraction to foraging sites more effectively than upright decoys that represent vigilant individuals (Drent & Swierstra 1977; Inglis & Isaacson 1977; Clergeau 1981). For these reasons, we used stuffed starlings with their heads pointing down and beaks touching the ground. Dead birds were obtained from hunters, as this species is a game bird in France (construction details in Clergeau 1982). To increase local enhancement (Guyomarc’h 1995), decoys in all experiments were placed in circular groups (Fig. 1), as this is the most attractive distribution of decoys (Clergeau 1981, 1982). To ensure that the attraction effect of the decoys was independent of the nutritive quality of the patch, we conducted a preliminary experiment testing the attraction of decoys to starlings flying over a football ground covered with artificial plastic turf. During this

A. Rodriguez et al. / Animal Behaviour 80 (2010) 965e973

967

(a)

Experiment 1 : Interindividual distance effect

A

B 12 m

(Interindividual distance = 0.6 m in A, 1.2 m in B)

(b)

Experiment 2 : Group size effect

A

B 12 m

(Interindividual distance = 0.6 m in A and B) (c)

Experiment 3 : Status-age effect

C: adults

(Interindividual distance = 0.6 m in A, B and C)

12 m

12 m A: young

B: immatures

12 m

Figure 1. Ground positions of decoys for (a) the interindividual distance experiment, (b) the group size experiment and (c) the statuseage experiment (black triangle ¼ decoy).

preliminary test, nine birds flew over the ground, and four of them landed near the decoys, thus confirming an attraction to the decoys. To evaluate the influence of season on starling reactions and to investigate relations between their behaviour and physiological state, we observed study populations during two periods: the first breeding period (6 MarcheApril 2007) and the postbreeding period (25 Junee24 July 2007). To test the effect of distances between decoys, of group size and of the decoys’ morphological aspect (which gives information about their age) on visual attractiveness to flying birds, we conducted three experiments (Fig. 1). (1) Experiment 1 (interindividual distance effect): starlings could choose between a group of six widely spaced adult decoys (interindividual distance: 120 cm) and a group of six adult decoys placed closer to each other (interindividual distance: 60 cm; see Fig. 1); these distances were chosen according to observed behaviours, in which mean interindividual distances commonly measured about 60 cm on high-quality feeding sites (Clergeau 1981). (2) Experiment 2 (group size effect): starlings could choose between a large group of 13 adult decoys and a small group of six adult decoys, all with the same interindividual distance (60 cm). (3) Experiment 3 (statuseage effect): starlings were given a choice between groups of six young (brown plumage; individuals born in the year), six adult (black plumage, yellow beak) and six immature (black-and-white spotted plumage, grey beak; individuals in their first year) decoys. Since this experiment was conducted during the postbreeding period, only adults and young starlings could be tested because few immature starlings were present. We observed no young starlings in Italy during the postbreeding period since they had probably already dispersed.

For each experiment we used a folding rule to measure the distances between decoys. The observations were performed during periods of intense foraging (Clergeau 1981). The two experiments that took place in the breeding period were conducted only in the morning (0630e1200 hours) and experiment 3, in the postbreeding period, was conducted in the morning and in the evening (0630e1200 and 1500e1900 hours). The same observer (A.R.) performed all 126 experimental trials in this study. During a trial, the observer sat 20 m from the test area, counted the individuals flying over the observation site and noted the behaviour of each bird. Data Collection We considered that a starling had been attracted by the decoys only when it changed its flight direction to fly towards the decoys or when it landed in a decoy group, but not when it reacted to the behaviour of another bird in its own group that had already landed or was flying towards the decoys. We considered that a starling had chosen a particular decoy group if it landed within 60 cm of the edge of the group (Fig. 1). If it landed more than 60 cm from the edge of any group, we considered that it had been attracted and landed but had not chosen a group. During the postbreeding period, the observer also noted the plumage colour of the attracted birds. Once the responses of birds were clear (on the ground for at least 3 s), the observer scared away all starlings remaining on the ground to clear it for potential new groups. To avoid recounting the same birds, observations lasted only 30 min, and the observer looked at the flying direction of birds

968

A. Rodriguez et al. / Animal Behaviour 80 (2010) 965e973

and observed if they came back or not. Observations were stopped if birds already counted remained in the observation area. Various sites were used for the study, but no site was used twice for the same experiment, although it could be used for two different experiments. For example, considering sites A, B, C, D, E, F, sites A, C, D could be used for the group size experiment (using each site only once in this experiment) and then sites C, E, F could be used in the interindividual distance experiment. Thus, for instance, site C was used only once in the experiment testing the group size effect and only once in the experiment testing interindividual distances. Statistical Analysis

RESULTS

We measured the percentage of birds attracted by the decoy population in each experiment and calculated the standard error for each case using the mathematical formula to obtain standard deviation for discrete grouped variables. To obtain the standard error we conducted the following calculation:

SD SE ¼ pffiffiffiffi N

SD ¼

pffiffiffiffi V

where V ¼ variance.

Pn V ¼

set of the statuseage experiment conducted during the postbreeding period (morning and evening) and the number of attracted and nonattracted birds as a function of the time of day. We calculated the exponential value of the estimates to obtain the odds ratios values between pairwise populations and established the 95% confidence intervals. To test the preferences of landing birds, we performed chisquare tests to compare the number of birds attracted to the various decoy groups. We also observed whether flock size influenced the decision to join decoys or follow the flying individuals’ own group. We implemented Bonferroni correction for multiple comparisons.

n ðx  xÞ2 Pni i i ¼ 1 ni

i¼1

ni ¼ number of birds tested at site i, xi ¼ proportion of birds attracted to site i, x ¼ proportion of birds attracted in the whole population tested, N ¼ number of sites used in the experiment. We used generalized linear mixed models (GLMM) implementing a binomial distribution family and the logit link function to compare the levels of visual attractiveness (Faraway 2006). We used the R software version 2.8.1 (R Development Core Team, Vienna, Austria) and package lme4 to test the following four models (M1eM4). (1) M1: to test whether the number of birds attracted during the interindividual distance experiment depended on their population of origin. We used the data set of interindividual distance experiment conducted during the breeding period and the number of attracted and nonattracted birds as a function of the origin of the population. (2) M2: to test whether the number of birds attracted during the group size experiment depended on the population of origin. We used the data set of the group size experiment conducted during the breeding period and the number of attracted and nonattracted birds. (3) M3: to test whether the number of birds attracted during the mornings of the statuseage experiment depended on the population of origin. We used the data set of the statuseage experiment conducted in the mornings of the postbreeding period and the number of attracted and nonattracted birds as a function of the origin of the population. (4) M4: to test whether the number of birds attracted during the evenings of the statuseage experiment depended on the population of origin. We used the data set of the statuseage experiment conducted in the evenings of the postbreeding period and the number of attracted and nonattracted birds as a function of the origin of the population. In the four models we implemented site as a random factor because sites were considered as repetitions within populations. We also conducted a generalized linear model (GLM) to test whether the number of birds attracted during the statuseage experiment depended on the time of day. We used the whole data

Temporal and Spatial Factors Affecting Attractiveness Responses to decoys differed between populations, both during and after the breeding period. During the breeding period, less than 11% of individuals from the longer-established population (Rural Brittany) were attracted to decoys, whereas more than 28% of individuals were attracted in recently settled Urban Brittany and Italian populations (Tables 1 and 2, Fig. 2). During the postbreeding period, attractiveness also appeared greater to more recently settled populations than to the longer-established one. The probability of being attracted was 1.5 times greater in the morning than in the afternoon (1/0.66 from the results shown in Table 2), and this difference is significant (P < 0.02). Attractiveness was also stronger in the postbreeding period than in the breeding period (Table 1, Fig. 2), but limitations of the experimental design meant that we could not determine whether season or the type of experiment was a determinant factor. The levels of attractiveness in the interindividual distance experiment were tested with GLMM analyses (Table 2). Odds ratios results underline that, during the breeding period, individuals from Urban Brittany, Rural Italy and Urban Italy were 208, 85 and 62 times more attracted to decoys than those from Rural Brittany (all P < 0.02; 362 birds, 24 sites). The results of GLMM analysis on attractiveness in the group size experiment are similar, but the differences observed between

Table 1 Percentage of birds attracted for each population in each experiment Number of birds tested

Percentage of birds attracted

SE

8 9 5 6

11 33 45 69

6.4 6.4 8.1 12.5

6 6 6 6

0 28 33 48

0 9.9 12.7 11.9

Statuseage experiment (AM) Rural Brittany 210 Rural Italy 71 Urban Italy 30 Urban Brittany 80 Suburban Brittany 182

10 9 3 8 15

25 52 50 81 80

8.7 1.7 3.7 2.5 1.6

Statuseage experiment (PM) Rural Brittany 229 Rural Italy 20 Urban Italy 36 Urban Brittany 43 Suburban Brittany 42

6 3 3 9 4

19 20 28 84 79

1.6 2.9 1.7 1.5 1.2

Group size experiment Rural Brittany Rural Italy Urban Italy Urban Brittany

46 67 73 62

Interindividual distance experiment Rural Brittany 129 Rural Italy 85 Urban Italy 57 Urban Brittany 91

Number of sites N

A. Rodriguez et al. / Animal Behaviour 80 (2010) 965e973 Table 2 Results of the generalized linear mixed models M1, M2, M3, and M4 using a binomial distribution function link Compared variables

Odds ratios

Z value

GLMM results in M1: interindividual distance Rural Brittany/Urban Brittany 207.9 Rural Brittany/Rural Italy 85.5 Rural Brittany/Urban Italy 62.2

P

Confidence interval 95%

experiment df¼3 3.401 <0.001 [9.6; 4501] 2.82 <0.001 [3.9; 1893] 3.52 <0.02 [2.5; 1547.5]

GLMM results in M2: group size experiment df¼3 Rural Brittany/Urban Brittany 11.72 2.21 Rural Brittany/Rural Italy 4 1.43 Rural Brittany/Urban Italy 7.7 2.12

0.027 0.15 0.034

[1.3; 103.8] [0.6; 27] [1.16; 50.5]

GLMM results in M3: statuseage experiment Rural Brittany/Suburban Brittany 21.2 Rural Brittany/Urban Brittany 22.9 Rural Brittany/Rural Italy 4.7 Rural Brittany/Urban Italy 4.4

AM df¼4 3.71 3.67 2.045 1.912

0.031 0.002 0.041 0.05

[4.2; 106.2] [4.3; 121.5] [1.1; 20.3] [1; 20.5]

GLMM results in M4: statuseage experiment PM df¼4 Rural Brittany/Suburban Brittany 54.1 2.48 Rural Brittany/Urban Brittany 65.5 2.94 Rural Brittany/Rural Italy 0.9 0.04 Rural Brittany/Urban Italy 0.3 0.48

0.013 0.003 0.97 0.62

[2.3; 1257] [4; 1061] [0.02; 48.1] [0.001; 56.16]

populations are less marked than in the interindividual distance experiment. We obtained no significant difference between Rural Brittany and Rural Italy (P > 0.05). In contrast, birds from Urban Brittany and from Urban Italy were 12 and 8 times more attracted than birds from Rural Brittany and this difference was significant (P < 0.04; 248 birds, 28 sites). The probability of being attracted in the morning was significantly higher than in the evening (GLM: Akaike’s information criterion, AIC ¼ 62, df ¼ 9, all P < 0.02). Consequently, we tested separately the comparisons between populations (we tested first the differences in the morning and then the potential differences in the populations’ reactivity in the evening). In the morning, birds from Suburban Rennes, from Urban Rennes and from Rural Italy were significantly more attracted than birds from Rural Brittany (P < 0.05). Birds from Suburban Brittany and from Urban Brittany were 21 and 22 times more attracted than those from Rural Brittany.

969

Birds from Rural Italy were five times more attracted than those from Rural Brittany. Birds from Urban Italy tended to be more attracted than those from Rural Brittany (P ¼ 0.05; 573 birds and 45 sites were used for these comparisons). In the evening, the differences were significant only between Rural Brittany and Suburban Brittany and between Rural Brittany and Urban Brittany (P < 0.02). In Suburban Brittany birds were 54 times more attracted than in Rural Brittany. Birds from Urban Brittany were 65 times more attracted than birds from Rural Brittany. Birds from Italy behaved similarly to those from Rural Brittany in the evening (370 birds and 25 birds were used for these comparisons). Social Factors Affecting Attractiveness Since few individuals were attracted to decoys in Rural Brittany during the breeding period, data were insufficient to determine the preferred kinds of groups for the group size and interindividual distance experiments. In Urban Brittany and in Rural Italy, individuals tended to choose the loose group with larger interindividual distances more frequently (Fig. 3a) but this was significant only in the Rural Italian population. Urban Brittany and Rural Italy individuals tended to choose the big group containing 13 decoys more frequently, but this was significant only for Urban Brittany individuals (Fig. 3a). The other populations did not express a clear preference for one of the groups (Fig. 3b). Globally, a greater attractiveness was observed in the group size experiment (total of 19 decoys) than in the interindividual distance experiment (12 decoys): the number of decoys seen by flying starlings may influence their decision to land (Fig. 2). The reaction of adults was similar throughout all populations. Adult starlings from all populations were significantly more attracted by adult and immature groups than by young ones (Fig. 3c). In contrast, reactions of young birds differed across populations. In Rural Brittany, young starlings were significantly more attracted by adult and immature groups (Fig. 3d). In Suburban Brittany, young birds tended to join groups of adults and of young decoys more frequently. In Urban Brittany, young starlings tended to choose the group of young decoys more frequently. Except for Rural Brittany, these differences in young birds’ choices were not significant.

***

% Birds attracted

100

*

***

**

* 80

* *

*** *

60

* **

40 20 0

RB RI UI UB (46) (67) (73) (62)

RB RI UI UB (129) (85) (57) (91)

RB RI UI UB SB (210) (71) (30) (80) (182)

AM

AM

AM

Gs experiment

Id experiment

Breeding period

RB RI UI UB SB (229) (20) (36) (43) (42) PM

StA experiment Postbreeding period

Figure 2. Visual attractiveness of decoys to starlings from populations with different histories in the three experiments conducted during the breeding and postbreeding periods. Id: interindividual distance experiment; Gs: group size experiment; StA: statuseage experiment; the numbers of birds tested by population are shown in parentheses; *P < 0.05; **P < 0.01; ***P < 0.001.

970

A. Rodriguez et al. / Animal Behaviour 80 (2010) 965e973

**

(a)

(b) 100

80

80 No choice Tighter group Loose group

60

Not enough data

0

40

(c)

**

***

Ru It ral a (1 ly 8)

**

U r It ban a l (2 y 3)

***

Br Rur itt al (3 an ) y

Ru It ral a (1 ly 1)

0 U r It ban (1 aly 3)

100 (d) * 80

40

al aly + (3 Ur 5) ba

It

ur (R

)

0

Not enough data Br Rur it a (4 tan l 0) y Su Br bu it rb (2 tan an 7) y

0 n)

20 Br Rur it a (2 tan l 2) y Su Br bu it rb (6 tan an 0) y U Br rb itt an (5 an 5) y

20

I al taly +U rb an

40

Young Immature Adults

60

ur

Young Immature Adults

60

(R

80

U Br rb itt an (1 an 5) y

100

No choice Group of 6 Group of 13

60

20

Br Rur itt al an y U Br rb it a (4 tan n 2) y

% Birds making choice

40 20

*

U Br rb itt an (4 an 0) y

100

Figure 3. Interindividual distance choices and group size choices in relation to individual origin. (a) Individuals’ choices in the interindividual distance experiment (Rural Brittany: not enough data; Urban Brittany: c2 ¼ 5.28; Urban Italy: c2 ¼ 1.08; Rural Italy: c2 ¼ 12.18). (b) Individuals’ choices in the group size experiment (Rural Brittany: not enough data; Urban Brittany: c2 ¼ 8.6; Urban Italy: c2 ¼ 1.39; Rural Italy: c2 ¼ 1). (c) Adults’ landing choices (Rural Brittany: c2 ¼ 12.09; Suburban Brittany: c2 ¼ 39.9; Urban Brittany: c2 ¼ 25.78; Italy: c2 ¼ 9.66). (d) Young birds’ landing choices (Rural Brittany: c2 ¼ 6.65; Suburban Brittany: c2 ¼ 2.89; Urban Brittany: c2 ¼ 5.2; Italy: not enough data). The numbers of birds tested by population are shown in parentheses. *P < 0.05; **P < 0.01; ***P < 0.001.

Concerning flock sizes, we noticed (Fig. 4, Table 3) that they varied depending on the origin of birds and the period of the year. In all the populations, we tested mostly individuals flying alone. Nevertheless, we could observe and test groups of two and more individuals. We observed that the population from Rural Brittany presented the broader variety of flock sizes whereas almost all flocks from recent populations were composed of fewer than eight individuals in both periods. During the breeding period, we observed flocks of 2e18 birds in all populations and, during the postbreeding period, we observed flocks of up to 25 individuals in Rural Brittany. When we compared the levels of attractiveness in the breeding period between populations, taking account of the flock sizes, we observed that individuals from Rural Brittany reacted significantly less to decoys than the individuals from the other populations (P < 0.001; Table 3) and that was observed both for individuals flying alone and for individuals flying in flocks of two to seven individuals. That means that individuals flying in flocks in Rural Brittany almost always prefer following their social group than joining the decoys, whereas in recent populations we observed the opposite. We also observed in the breeding period that single individuals behave similarly to groups of two to seven individuals in all the populations (P > 0.05; Table 3). In the Rural Brittany population, we observed that groups of two to seven individuals behaved similarly to individuals of groups of 8e18 individuals. Only in the postbreeding period did we find that, in the Rural Brittany population, the size of the flocks could influence attractiveness, as flocks of two to seven individuals reacted significantly less than lone individuals (P < 0.001) and as flocks of 8e25 individuals reacted significantly less than individuals in flocks of two to seven individuals. During this period, we found that lone

Urban and Suburban Brittany birds reacted significantly more than Rural Brittany ones (P < 0.001) whereas single individuals from Italian populations behaved similarly to single individuals observed in Rural Brittany (P > 0.05). DISCUSSION Differences in Attractiveness Levels The use of decoys revealed that starlings from populations confronted with new environments (in cities or colonization fronts) were more sensitive to social information given by decoys than those from established populations with a longer familiarity with their environment. In an interspecific study of Ardeidae, Green & Leberg (2005) showed that more territorial and less social species were less attracted by decoys than gregarious species. We could thus think that the higher levels of attractiveness observed in recently settled populations reflect a higher degree of gregariousness or of sociability. Nevertheless, we observed more big flocks in Rural Brittany than in the recent populations. It thus seems that all populations are very gregarious but the element that differs is the choice between following individuals from other groups or individuals from one’s own group. In fact, individuals from the Rural Brittany population more frequently followed their own group and recently settled populations more frequently joined the ‘feeding’ decoys. Starlings in colonization fronts in unpredictable or suboptimal habitat would benefit by following individuals that have found food (Boyd & Richerson 1988; Galef & Laland 2005). In contrast, in wellknown environments, starlings would probably do better by going

Number of birds

A. Rodriguez et al. / Animal Behaviour 80 (2010) 965e973

40 35 30 25 20 15 10 5 0 100 90 80 70 60 50 40 30 20 10 0

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 (c)

70 60 50 40 30 20 10 0 120 100

971

(b)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 (d)

80 60 40 20 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Flock size

Figure 4. Reaction of birds observed by size flock category during the experiments conducted in the breeding period. (a) Rural Brittany; (b) Urban Brittany; (c) Rural Italy; (d) Urban Italy. Black bars: number of birds observed by flock size category; grey bars: number of birds attracted by decoys.

directly to sites that they know provide a permanent food source (Tinbergen 1981) or following experienced individuals of their own group than to risk following unknown individuals. Moreover, Mennechez & Clergeau (2006) showed that the closer starlings were to the centre of Rennes, the more difficult it was for adults to find prey items to feed their young and the higher was the juvenile mortality. The Apulia region of southern Italy is dry, and foraging grounds are scarce and probably poor in prey items. In contrast, rural areas in Brittany have many pastures and corn silos used for feeding cows. Starlings feed mainly on these sites (Feare 1984). We suggest that difficulty in finding food increased reactions to social cues in Italy and Urban Brittany. This can explain why starlings in colonization fronts were more attracted by groups of decoys and why their attraction increased in more urbanized zones. Starlings in colonization fronts could also be confronted with novel food items; thus, copying innovative feeding behaviour could be a good strategy (Galef 1993; Galef & Laland 2005). Social and Physiological Factors Influencing Attractiveness Our results reveal that adults are mainly attracted by adults (black feathers) in Brittany or by adults and immature birds (black-and-white feathers) in southern Italy, but not by young Table 3 Comparison of visual attractiveness (attracted versus nonattracted, df ¼ 1) taking account of flock sizes and population origin Population comparisons

Period

c2

P

Rural Rural Rural Rural Rural Rural Rural Rural Rural Rural Rural Rural Rural Rural

Breeding period Breeding period Postbreeding Postbreeding Breeding Breeding Breeding Breeding Breeding Breeding Postbreeding Postbreeding Postbreeding Postbreeding

1.99 0.2 12.1 29.1 16.9 83.6 12.2 19.8 19.7 11.9 44.4 64.0 1.8 0.6

>0.05 >0.05 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 >0.05 >0.05

Brittany Brittany Brittany Brittany Brittany Brittany Brittany Brittany Brittany Brittany Brittany Brittany Brittany Brittany

(1)/Rural Brittany (2e7) (2e7)/Rural Brittany (8e25) (1)/Rural Brittany (2e7) (2e7)/Rural Brittany (8e25) (1)/Urban Brittany (1) (2e7)/Urban Brittany (2e7) (1)/Rural Italy (1) (2e7)/Rural Italy (2e7) (1)/Urban Italy (1) (2e7)/Urban Italy (2e7) (1)/Urban Brittany (1) (1)/Suburban Brittany (1) (1)/Rural Italy (1) (1)/Urban Italy (1)

Flock size is given in parentheses.

individuals (brown feathers). Adults tend to follow other adults and subadults, probably because they identify them as experienced individuals able to give better information about the quality of feeding sites. Senar & Escobar (2002), who studied siskins, Carduelis spinus, showed that only individuals that were able to find enough food resources developed yellow bands on their wings. This characteristic is an honest signal indicating the capacity of its bearer to find food and therefore indicates good foragers. Siskins without these bands were more attracted to decoys with bands than to other decoys. Starlings are possibly more sensitive to social information from adults or experienced individuals, which can act as social references motivating tasting of new types of food, eating in new places or performing novel tasks, as observed in rodents, primates and other birds (Galef 1993; Visalbergi & Fragazsi 1994; Wauters et al. 2002; Dindo et al. 2008). Our results did not show clear preferences in young starlings, only choices as a function of origin. The young birds in the established (Rural Brittany) population chose groups of adults, but young birds from Urban Brittany tended to choose groups of young decoys. This may support the hypothesis that individuals join similar individuals, an antipredator strategy decreasing the risk of being detected as conspicuous, since we observed magpies, Pica pica, attacking the brown decoys (also reported by Fernandez-Juricic et al. 2004; Morgan et al. 2005). We found a clear choice of starlings in Urban Brittany populations for large group sizes and a similar tendency in the Rural Italy colonization front. These reactions have been observed previously in starlings and in other species such as least terns, Sternula antillarum, and Laysan albatrosses, Phoebastria immutabilis (Clergeau 1982; Burger 1988; Podolsky 1990). These authors suggested that individuals benefit from being in large groups as their probability of being predated decreases because of dilution and confusion effects. Moreover, ideal-free distribution theory and empirical observations suggest that individuals tend to join larger groups of foraging individuals because they act as indicators of the presence of abundant or high-quality resources (Inman & Krebs 1987; Wrona & Dixon 1991; Gotceitas & Colgan 1991). Concerning interindividual distances, individuals from Urban Brittany and from Rural Italy tended to prefer the loose group with larger interindividual distances. This choice can probably indicate a gathering behaviour by using social information, but at the same

972

A. Rodriguez et al. / Animal Behaviour 80 (2010) 965e973

time an avoidance of very close competition (Fernandez-Juricic & Kacelnik 2004). Finally, starlings tended to be more attracted during the postbreeding period than during the breeding period. Likewise, Clergeau (1982) found that starlings were more attracted in winter than in spring. These different levels of attractiveness might be explained by variation in sociability levels and physiological needs during the starlings’ seasonal cycle or by the seasonal changes in vegetation that make resources unpredictable (Bautista et al. 1998). Another argument that can indicate the importance of the physiological status of starlings is the fact that they were more reactive in the morning than in the afternoon, probably because they had spent the night fasting in roosts. Our study corroborates the influence of social information on animals in novel contexts such as urban habitats or colonization fronts. This mechanism appears to be an important factor enabling successful establishment of new populations and can explain the success of various gregarious birds in colonizing towns. This flexibility in the use of social cues could be a key to explain the worldwide success of starling introduction in different regions of the world, as other flexible behaviours have been suggested to enhance invasion success (Sol et al. 2002). However, complementary studies need to be conducted to confirm whether these observations can be generalized to other old and recent starling populations and to other species whose populations are expanding. Numerous questions remain, such as the variability and adjustment of the use of social information during the different phases of the colonizationeinvasion process (dispersion, arrival and establishment). The first step seems the most interesting to analyse, both in the laboratory and in the field, as it involves minimization of neophobia (Martin & Fitzgerald 2005) and the use of interspecific information (Clergeau 1990; Hazlet et al. 2003; Parejo et al. 2004). Acknowledgments We thank Patricia Le Quilliec (INRA Scribe), Giuseppe La Gioia (Museum of Natural History of Salento) and Giovanni Ferrara (University of Bari) for their help in the field, Gregory Guernec (INRA Scribe) and Laurent Mizrahi for their help with statistical analysis, and Ann Cloarec and Michael Corson, who improved the English of the manuscript. We thank the five referees for their comments and help in improving the manuscript. The study benefited from a grant from INRA Scribe Ecology of Biological Invasions and A.R. was supported by a graduate scholarship from the French Ministry of Research. References Adret-Hausberger, M. & Cumming, R. B. 1987. Social attraction to older birds by domestic chicks. Bird Behavior, 7, 44e46. Bautista, L. M., Tinbergen, J., Wiersma, P. & Kacelnik, A. 1998. Optimal foraging and beyond: how starlings cope with changes in food availability. American Naturalist, 152, 543e561. Beauchamp, G. & Ruxton, G. 2005. Harvesting resources in groups or alone: the case of renewing patches. Behavioral Ecology, 16, 989e993. Boyd, R. & Richerson, P. J. 1988. An evolutionary model of social learning: the effects of spatial and temporal variation. In: Social Learning: Psychological and Biological Approaches (Ed. by T. Zentall & B. G. Galef), pp. 29e48. Hillsdale, New Jersey: L. Erlbaum. Burger, J. 1988. Social attraction in nesting least terns: effects of numbers, spacing, and pair bonds. Condor, 90, 575e582. Cadieu, J. C., Cadieu, N. & Lauga, J. 1995. Local enhancement and seed choice in the juvenile canary, Serinus canarius. Animal Behaviour, 50, 793e800. Castiglia, G. & Tabarrini, G. 1982. Stazioni di nidificazione dello Storno. Uccelli d’Italia, 7, 93e104. Chesler, P. 1969. Maternal influence in learning by observation in kittens. Science, 16, 901e903. Clayton, D. A. 1978. Socially facilitated behaviour. Quarterly Review of Biology, 53, 373e391.

Clergeau, P. 1981. Comportements liés à l’alimentation de l’etourneau Sturnus vulgaris en Bretagne: rôle joué par certaines variables environnementales et sociales. Ph.D. thesis, University of Rennes 1. Clergeau, P. 1982. Attractivité et utilisation du milieu chez des étourneaux en alimentation. Oecologia Applicata, 3, 303e320. Clergeau, P. 1990. Mixed flocks feeding with starlings: an experimental field study in Western Europe. Bird Behavior, 8, 95e100. Coussi-Korbel, S. & Fragaszy, D. M. 1995. On the relation between social dynamics and social learning. Animal Behaviour, 50, 1441e1453. Crozier, G. & Gawlik, D. 2002. The use of decoys as a research tool for attracting wading birds. Journal of Field Ornithology, 74, 53e58. Danchin, E., Giraldeau, L., Valone, T. & Wagner, R. 2004. Public information: from nosy neighbours to cultural evolution. Science, 305, 487e491. Dindo, M., Thierry, B. & Whiten, A. 2008. Social diffusion of novel foraging methods in brown capuchin monkeys (Cebus apella). Proceedings of the Royal Society B, 275, 187e193. Doligez, B., Danchin, E. & Clobert, J. 2002. Public information and breeding habitat selection in a wild bird population. Science, 297, 1168e1170. Drent, R. & Swierstra, P. 1977. Goose flocks and food finding: field experiments with barnacle geese in winter. Wildfowl, 28, 15e20. Fancher, J. 1984. A technique for making least tern decoys. Journal of Field Ornithology, 55, 241e243. Faraway, J. J. (Eds). 2006. Extending the Linear Model with R. Generalized Linear, Mixed Effects and Nonparametric Regression Models. Boca Raton: Chapman & Hall/CRC. Feare, C. 1984. The Starling. Oxford: Oxford University Press. Fernandez-Juricic, E. & Kacelnik, A. 2004. Information transfer and gain in flocks: the effects of quality and quantity of social information at different neighbour distances. Behavioral Ecology and Sociobiology, 55, 502e511. Fernandez-Juricic, E., Jokimäki, J., Chase, M. J., Melado, F., Toledano, A., Mayo, C., Martin, B., Fresneda, I. & Martin, V. 2004. Effects of opportunistic predation on anti-predator behavioural responses in a guild of ground foragers. Oecologia, 140, 183e190. Galef, B. G. 1993. Functions of learning about food: a casual analysis of effects of diet novelty on preference transmission. Animal Behaviour, 46, 247e265. Galef, B. G. & Laland, K. N. 2005. Social learning in animals: empirical studies and theoretical models. Bioscience, 55, 489e499. Gotceitas, V. & Colgan, P. 1991. Assessment of patch profitability and ideal free distribution. The significance of sampling. Behaviour, 119, 65e76. Green, M. C. & Leberg, P. 2005. Flock formation and the role of plumage colouration in Ardeidae. Canadian Journal of Zoology, 83, 683e693. Guyomarc’h, J. C. 1995. Ethologie. 2nd ed. Abrégés: Masson. Harvey, W., Hindman, L. & Rhodes, W. 1995. Vulnerability of Canada geese to taxidermy-mounted decoys. Journal of Wildlife Management, 59, 474e477. Hazlet, B., Burba, A., Gherardi, F. & Acquistapace, P. 2003. Invasive species of crayfish use a broader range of predation-risk cues than native species. Biological Invasions, 5, 223e228. Inglis, I. I. & Isaacson, A. J. 1977. The responses of dark-bellied brent geese to models of geese in various postures. Animal Behaviour, 26, 953e958. Inman, A. J. & Krebs, J. 1987. Predation and group living. Trends in Ecology & Evolution, 2, 31e32. Jackson, A. & Ruxton, G. 2006. Toward an individual-level understanding of vigilance: the role of social information. Behavioral Ecology, 17, 532e538. Jeffries, D. S. & Brunton, D. H. 2001. Attracting endangered species to ‘safe’ habitats: responses of fairy terns to decoys. Animal Conservation, 4, 301e305. Jiguet, F. & Bretagnolle, V. 2006. Manipulating lek size and composition using decoys: an experimental investigation of lek evolution models. American Naturalist, 168, 758e768. Kotliar, N. & Burger, J. 1984. The use of decoys to attract least terns (Sterna antillarum) to abandoned colony sites in New Jersey. Colonial Waterbirds, 7, 134e138. Martin, L. B. & Fitzgerald, L. 2005. A taste for novelty in invading house sparrows, Passer domesticus. Behavioral Ecology, 16, 702e707. Master, T., Leiser, J., Bennett, K., Bretsch, J. & Wolfe, H. 2005. Patch selection by snowy egrets. Waterbirds, 28, 220e224. Mennechez, G. & Clergeau, P. 2006. Effect of urbanisation on habitat generalists: starlings not so flexible? Acta Oecologica, 30, 182e191. Morgan, D., Wass, J. & Innes, J. 2005. Magpie interactions with other birds in New Zealand: results from a literature review and public survey. Notornis, 52, 61e74. Parejo, D., Danchin, E. & Avilés, J. 2004. The heterospecific habitat copying hypothesis: can competitors indicate habitat quality? Behavioral Ecology, 16, 96e105. Parejo, D., Pérez-Contreras, T., Navarro, C., Soler, J. J. & Avilés, J. M. 2008. Spotless starlings rely on public information while visiting conspecific nests: an experiment. Animal Behaviour, 75, 483e488. Pasquali, R. 1984. Le colonie nidificanti di storno, Sturnus vulgaris, nell’Italia CentroMeridionale. Rivista Italiana di Ornitologia Milano, 54, 221e229. Pius, S. & Leberg, P. 2002. Experimental assessment of the influence of gull-billed terns on nest site choice of black skimmers. Condor, 104, 174e177. Podolsky, H. 1990. Effectiveness of social stimuli in attracting Laysan albatross to new potential nesting sites. Auk, 107, 119e125. Richard, M. A. 1826. L’étourneau. In: Oeuvres de Buffon (Ed. by F. Baudouin) Paris: F. Baudouin. Ryer, C. H. & Olla, B. L. 1992. Social mechanisms facilitating exploitation of spatially variable ephemeral food patches in a pelagic marine fish. Animal Behaviour, 44, 69e74.

A. Rodriguez et al. / Animal Behaviour 80 (2010) 965e973 Senar, J. C. & Escobar, D. 2002. Carotenoid derived plumage coloration in the siskin Carduelis spinus is related to foraging ability. Avian Science, 2, 19e24. Slaa, J., Wassenberg, J. & Biesmeijer, J. 2003. The use of field-based social information in eusocial foragers: local enhancement among nestmates and heterospecifics in stingless bees. Ecological Entomology, 28, 369e379. Smith, J., Benkman, C. & Coffey, K. 1999. The use and misuse of public information by foraging red crossbills. Behavioral Ecology, 10, 54e62. Sol, D., Timmermans, S. & Lefebvre, L. 2002. Behavioural flexibility and invasion success in birds. Animal Behaviour, 63, 495e502. Templeton, J. & Giraldeau, L. A. 1996. Vicarious sampling: the use of personal and public information by starlings foraging in a simple patchy environment. Behavioral Ecology and Sociobiology, 38, 105e114. Tinbergen, J. M. 1981. Foraging decisions in starlings. Ardea, 69, 1e67. Valone, T. J. 1989. Group foraging, public information and patch estimation. Oikos, 56, 357e363. Valone, T. J. 1991. Bayesian and prescient assessment: foraging with pre-harvest information. Animal Behaviour, 41, 569e577.

973

Valone, T. J. & Giraldeau, L. A. 1993. Patch estimation by group foragers: what information is used? Animal Behaviour, 45, 721e728. Valone, T. J. & Templeton, J. 2002. Public information for the assessment of quality: a widespread social phenomenon. Philosophical Transactions of the Royal Society, 357, 1549e1557. Visalbergi, E. & Fragazsi, D. 1994. The behaviour of capuchin monkeys, Cebus apella, with novel food: the role of social context. Animal Behaviour, 49, 1089e1095. Wauters, A. M., Richard-Yris, M. A. & Talec, N. 2002. Maternal influences on feeding and general activity in domestic chicks. Ethology, 108, 529e540. Williams, R. & Schwab, R. 1973. Tests of a potential method for decoying starlings to bait stations. Wildlife Damage Management Internet Center for Bird Control Seminars Proceedings, University of Nebraska. http://digitalcommons.unl.edu/ icwdmbirdcontrol/117/. Wrona, R. & Dixon, J. 1991. Group size and predation risk: a field analysis of encounter and dilution effects. American Naturalist, 137, 186e201.