Prospectors combine social and environmental information to improve habitat selection and breeding success in the subsequent year


  • Tomas Pärt,

    Corresponding author
    1. Department of Ecology, The Swedish University of Agricultural Sciences, Box 7044, SE 750 07 Uppsala, Sweden
    Search for more papers by this author
  • Debora Arlt,

    1. Department of Ecology, The Swedish University of Agricultural Sciences, Box 7044, SE 750 07 Uppsala, Sweden
    2. School of Agriculture, Policy & Development, Centre for Agri-Environmental Research (CAER), University of Reading, Earley Gate, PO Box 237, Reading RG6 6AR, UK
    Search for more papers by this author
  • Blandine Doligez,

    1. Department of Biometry and Evolutionary Biology, CNRS, UMR 5558, Université Lyon 1, Université de Lyon, Bâtiment Gregor Mendel, 43, Boulevard du 11 Novembre 1918, F – 69622 Villeurbanne Cedex, France
    Search for more papers by this author
  • Matthew Low,

    1. Department of Ecology, The Swedish University of Agricultural Sciences, Box 7044, SE 750 07 Uppsala, Sweden
    Search for more papers by this author
  • Anna Qvarnström

    1. Department of Ecology & Evolution, Animal Ecology, Uppsala University, Norbyv.18d, SE-752 36 Uppsala, Sweden
    Search for more papers by this author

Correspondence author. E-mail:


1. Because habitats have profound effects on individual fitness, there is strong selection for improving the choice of breeding habitat. One possible mechanism is for individuals to use public information when prospecting future breeding sites; however, to our knowledge, no study has shown prospecting behaviour to be directly linked to subsequent choice of breeding site and future reproductive success.

2. We collected long-term data on territory-specific prospecting behaviour and subsequent breeding in the short-lived northern wheatear (Oenanthe oenanthe). Non-breeders established prospecting territories (<2 ha) that overlapped the breeding territories of conspecifics. We tested whether: (i) prospectors used social and environmental cues that predicted territory-specific breeding success in the following year, and (ii) the prospecting territory was tightly linked to the subsequent breeding territory of the prospector, and whether this link would be weakened by intraspecific competition with original territory owners if they also survived.

3. As expected, prospectors were attracted to a combination of site-specific cues that predicted future breeding success, i.e. short ground vegetation, a successfully breeding focal pair and successful close neighbours.

4. Prospecting behaviour was directly linked to the choice of the following year’s breeding territory: 79% of surviving prospectors established a breeding territory at their prospecting site in the following year, with their breeding success being higher than other individuals of the same age. As predicted, fidelity to the prospected site was strongly dependent on whether the original territory owner of the same sex had died or moved.

5. Our findings suggest that the use of multiple cues reduces the negative impact of stochasticity on the reliability of social cues at small spatial scales (e.g. territories) and hence increases the probability of breeding success in the next year. Also, the use of conspecific attraction (i.e. the preference for breeding aggregations) is selectively advantageous because individuals are more likely to find a vacancy in an aggregation as compared to a solitary site. By extension, we hypothesize that species life-history traits may influence the spatial scale of prospecting behaviour and habitat selection strategies.


Breeding habitat selection profoundly affects individual fitness and thus many aspects of the evolution and ecology of organisms: such as population regulation (Pulliam & Danielson 1991), sensitivity to environmental change (Doligez et al. 2003; Fletcher 2006) and speciation (Via 2001). Understanding individual habitat selection strategies and the constraints acting upon them is therefore crucial for fundamental evolutionary, ecological and applied conservation purposes (Clobert et al. 2001; Morris 2003).

Recent studies of breeding habitat selection suggest that individuals use the presence of conspecifics (e.g. conspecific attraction) and their reproductive success (i.e. public information) as a cue of habitat quality for their departure and settlement decisions (see reviews in Danchin, Heg & Doligez 2001; Valone & Templeton 2002; Danchin et al. 2004; Valone 2007). Accumulating evidence suggests that individuals gather this information when prospecting patches with breeding conspecifics one (or more) year(s) before breeding (Reed et al. 1999; Boulinier et al. 2002; Pärt & Doligez 2003; Doligez, Pärt & Danchin 2004a; Ward 2005; Betts et al. 2008). However, almost no study has investigated the direct links between sites prospected and the choice of subsequent breeding site. Still, the direct relationship between prospecting and subsequent breeding site selection may potentially determine the cues used for assessing habitats, the resulting breeding habitat selection patterns and thus the spatial dynamics of populations.

The link between prospecting and subsequent settlement is unlikely to be straightforward. Prospectors may use social cues but fail to settle at high-quality prospected sites because of intraspecific competition with surviving original territory owners (e.g. site dominance; Krebs 1982). The choice of future breeding sites may also depend on environmental signals, such as vegetation structure and food availability, and these may be correlated with social cues, further complicating the role of social cues in determining habitat selection strategies (Ward 2005; Nocera, Forbes & Giraldeau 2006; Betts et al. 2008). Thus, just as individuals are expected to use multiple cues in mate selection (Andersson 1994), we should also expect individuals to integrate multiple information sources when assessing habitat quality in the selection of a suitable breeding site. This should especially be the case for non-breeding prospectors, because they probably pay a low cost of prospecting (e.g. because of no time stress; Reed et al. 1999; Danchin, Heg & Doligez 2001).

We present long-term data on prospecting behaviour in non-breeders (i.e. floaters, unpaired individuals and early failed breeders) of the short-lived migratory and territorial northern wheatear (Oenanthe oenanthe) to test whether: (i) prospectors cue on social (conspecific aggregation size, conspecific reproductive success) and environmental cues (vegetation cue linked to fitness), or to a combination of these cues, (ii) the cues used by prospectors allow them to choose sites with a high probability of breeding success, (iii) prospecting is linked to the choice of next year’s breeding territory and (iv) the subsequent breeding site choice of prospectors is constrained by site dominance and survival of previous owners. Previous studies of territory choice and individual reproductive performance in this population have shown that individuals prefer to settle in territory aggregations (i.e. ‘conspecific attraction’: Stamps 1988) but that aggregation size is not linked to individual fitness (Arlt & Pärt 2007). However, an environmental cue, territory field layer height (i.e. the height of the ground vegetation) during the nestling period is strongly linked to individual fitness with both reproduction and subsequent adult and juvenile survival being higher in short than tall field layer habitats (Pärt 2001a,b; Arlt et al. 2008; Low et al. 2010). Breeding success and territory field layer height during the nestling period predict breeding success and field layer height, respectively, in the subsequent year (Arlt & Pärt 2007). Based on this between-year predictability of site quality in combination with the likelihood that non-breeders pay a low cost of prospecting, we suggest that non-breeding prospectors should assess the suitability of sites based on environmental and social cues when other pairs are breeding. Based on previous studies of prospecting behaviour, we hypothesize that prospectors: (i) prefer sites with short field layers and successfully breeding pairs, (ii) prefer to breed on the prospected sites in the subsequent year, (iii) be prevented from settling in the prospected site in the following year if the same-sex resident bird survives and (iv) prospectors, on average, have a high probability of breeding success in the next year because they should have been able to identify and choose the best sites. We used territory sites as the spatial scale of observation because previous studies on habitat selection in this species suggest the site (i.e. a spatially restricted site defined by the size of the territory of at least one breeding attempt during the years of study) to be an appropriate scale for understanding settlement decisions (Arlt & Pärt 2007, 2008a,b).

Materials and methods

Species and Study Population

The northern wheatear is a small (21 g) insectivorous passerine species wintering in Africa and breeding in Europe, Asia and the northern parts of North America. In our study area, males arrive at the breeding grounds from mid-April to mid-May, a few days before females. Older birds (≥2 years) arrive on average 1 week before yearlings. The nest is usually built in cavities on the ground, often under stones, in stone piles or stone walls. Egg-laying starts mid-May, incubation lasts for 13 days, and nestlings stay in the nest for about 15 days. Wheatears strongly prefer to forage on the ground in grassland habitats with short field layers (Conder 1989; Low et al. 2010). Reproductive success and adult and first-year survival rates are strongly related to territory field layer height during the nestling period, such that pairs breeding in habitats characterised by short field layer are more successful and have higher survival probability than those breeding in tall field layers (Pärt 2001a,b; Arlt & Pärt 2007; Arlt et al. 2008; Low et al. 2010). We defined ‘site’ as any location that had been used as a breeding territory at least once during the years of study, and ‘territory’ as a site occupied by a breeding pair (i.e. breeding territory) or a prospector (i.e. prospecting territory).

The data presented here were collected between 1993 and 2000 in the central part of a large study area situated in farmland south-east of Uppsala, Sweden (59° 50′ N, 17° 50′ E). About 120–180 wheatear pairs breed annually in the study area (for details of standard procedures of data collection, see Pärt 2001a; Arlt & Pärt 2007). All observed males were aged as yearlings or older based on plumage characteristics, while only a proportion of females could be classified into these two age groups (Pärt 2001a). In the very central part (15 km2) of the study area with about 100 sites (i.e. sites with at least one breeding attempt during the 8 years of study) and on average 65 (range = 52–78) breeding pairs per year, most breeding individuals and nestlings were ringed with a unique combination of colour rings and an aluminium ring. Breeding territories were patchily distributed in the landscape, and habitat patches varied in size. About 54% of the habitat patches had only one territory, 22% had two, 16% had three, and 8% had more than three territories. This central part of the study area was visited on a daily basis to collect data on reproductive success and movements of ringed non-breeding birds. Special care was taken to find new individuals in all potential breeding sites, i.e. also sites with no breeding pairs. We are confident that all settled prospectors (as defined below) were detected because all sites were visited on an almost daily basis when the first clutches hatched in the population (i.e. late May/early June). Because we did not expect to observe birds at vacant sites, any prospecting wheatear was more easily detected at those sites than at sites with breeding birds. However, this potential bias was likely to be small and did not affect our main interpretations, because almost no prospectors were observed at vacant sites (see Results). All observations of individuals (i.e. breeders and non-breeders) were positioned on a detailed map. Territory boundaries of breeding pairs and prospectors were estimated by plotting all observations (≥10 independent observations of individual movements separated by at least a day, where each observation was based on 5–15 min mapped movements) between date of egg-laying and fledging of young on a detailed map. Generally, territories were easily mapped because fence poles, stones and other landmarks were often used as boundaries. Fights between close breeding neighbours often confirmed the breeding territory boundaries. For unoccupied sites, we used territory boundaries of the most recent previous breeding attempt. The territories of breeding pairs that had failed were visited several times on the days following the failure, and failed breeders were searched for in the surroundings (i.e. within 1–2 km; the range of post-breeding movements, see Arlt & Pärt 2008a) to check whether they started a replacement breeding attempt. All sites were classified in terms of their ground vegetation structure, i.e. field layer height. Classification of field layer height was based on four visual estimates (between late April/early May and late June) of proportions of short (<5 cm high), medium (5–15 cm high) and high (>15 cm high) field layers (for validation of the method see Pärt 2001b). Sites were classified as having either a permanently short field layer (short on all four occasions within 50 m of the nest site for an area equivalent to the minimum territory size of 0·25 ha; mainly grazed grasslands and farm yards) or a growing field layer (also called ‘tall field layer’; non-grazed or late-grazed grasslands, fallow fields and crop fields). At the time of territory establishment in April, on average, 95% of all sites had a short field layer, but about 50% of these (fields, cultivated grasslands, non-grazed or late grazed pastures) grew tall during the breeding season.

Establishment of Prospecting Territories

In early June, when c. 90% of the pairs had started to breed, some wheatears (hereafter ‘prospectors’) were observed to establish a new non-breeding territory (hereafter a prospecting territory) in habitat patches where these individuals had not been established before. At this prospecting territory, individuals started their complete moult and usually remained until August when migration started. At the establishment of a prospecting territory, prospectors were sometimes involved in aggressive interactions with the breeding pairs at the same site. However, in most cases, breeding pairs ignored the prospectors, probably because females were not fertile at this time in the breeding cycle, and they were too busy feeding their own chicks to defend a foraging area. In late June and early July, some individuals that successfully fledged young also moved to a new site to complete their moult, but these movements are presented elsewhere (Arlt & Pärt 2008a).

An individual was defined to have settled on a prospecting territory when it had been observed at the same site during at least five consecutive days (i.e. the minimum number of days a non-breeding individual was observed at a locally restricted site). Prospecting territories were small (mean ≈1·5 ha, range ≈0·25–2 ha; based on ≥10 days of observations with several observations/day; average size of breeding territories: 1·5 ha), but could overlap several breeding territories and vacant sites. All prospecting territories were located at active breeding sites or at sites that had been used as a breeding territory in other years between 1993 and 2000. The position of a prospecting territory in relation to sites was determined by degree of overlap, where the focal site was that with the greatest overlap.

In total, 60 individuals established permanent prospecting territories on which they remained at least until the start of their moult (i.e. in July after the breeding season). Eleven individuals (five colour-ringed, six unringed), however, were seen to establish a temporary prospecting territory (i.e. during 2–5 days) but abandoned these following breeding failure of the focal pairs (see Results). Some of these temporary birds instantly established a new prospecting site (three colour-ringed and an unknown number of unringed birds) and thus were counted amongst the 60 permanent prospectors. To avoid pseudoreplication, we therefore only analysed patterns of preferences concerning the final choice of prospecting territory. However, the qualitative results did not differ when including the observations of these temporary prospecting territories (results not shown). In a small number of cases (7 of 60), prospecting territories overlapped with each other; thus, the prospecting territory was not a strictly defended area for the exclusive use of the prospector, but the same was also true for breeding pairs at this stage of the breeding cycle (unpublished data). Most prospecting territories were established completely within (35%), or strongly overlapping (58%), the territories of breeding pairs.

Data Selection and Statistics

Because prospectors virtually never settled in territories surrounded by tall-growing vegetation (= 2; strong preference for short field layers; inline image, < 0·0001; Fig. 1), we restricted our analyses to the prospectors that settled in territories with short vegetation (= 58). We used territory as the unit of observation because our focus was on fine-scale spatial choice of territories. Prospector preferences for territory characteristics were analysed by comparing chosen vs. non-chosen short-vegetation territories using generalized linear mixed models (MLwiN version 2.20; Rasbash et al. 2009). We fit models using a binomial distribution and included territory identity as a random effect to control for multiple sampling across years. We created a set of candidate models to examine the relative influence of occupation and breeding success in the focal and neighbouring territories on the probability of a prospector settling at a site, through combinations of the following parameters: (i) whether the focal site (i.e. the site in which the prospector was consistently observed) was occupied by a breeding pair (yes/no), (ii) whether the pair in the focal territory also successfully fledged offspring (unoccupied, occupied but failed breeding, occupied and successful breeding), (iii) the number of neighbouring sites that were occupied (range 0–3) and (iv) the number of neighbouring territories that successfully fledged offspring (range 0–3). Breeding was defined as successful when at least one young fledged from the nest. When calculating the number of neighbours for a prospecting territory, we counted the number of neighbours sharing boundaries with the focal site. Because the two ‘focal’ variables (1 and 2 above) and the two ‘neighbour’ variables (3 and 4 above) were highly correlated, we alternated their inclusion in models and used the relative ranking of models containing these variables to gauge support for whether prospectors only used site occupation or also took into account the breeding success of territories at the prospected site (i.e. when models that accounted for breeding success of the focal and/or neighbouring territories consistently ranked higher than those only containing information about site occupancy). We used AICc and AICc weights (wi) to assess the relative strength of support for models. Because of the overwhelming support for the highest-ranked model from these analyses, we only present parameter estimates from the highest-ranked model (Burnham & Anderson 2002).

Figure 1.

 Raw data on the settlement of prospectors relative to the total number of available sites. The percentage of available sites prospected are represented in four ways based on site characteristics: (a) status of the breeding pair in the focal territory where the prospector settled, (b) the height of the ground vegetation surrounding the prospected site, (c) the number of breeding pairs neighbouring the focal territory and (d) the number of successfully breeding neighbours to the focal territory. Numbers above bars indicate the total number of available sites of that site category across all years of study.

We visualised the relationship between prospected sites in short vegetation and the occupation/breeding success of other wheatears at those sites using a regression tree analysis from the ‘rpart’ package in r (R Core Development Team 2010). Regression trees are a nonparametric recursive partitioning procedure, constructed by continuously dividing data into mutually exclusive groups by comparing binary splits in the explanatory variables and choosing the division that minimises heterogeneity of the dependent variable in the resulting two groups (De’Ath & Fabricius 2000). This process is then repeated on the next grouping level; thus, the output resembles a tree diagram with a single node at the top containing the entire data set, with each branch a decision-rule based on the values of an explanatory variable leading to a subset of the data (Low, Joy & Makan 2006). Recursive partitioning analyses have some advantages over linear modelling approaches in that explanatory variables can be highly correlated, the output shows intuitive relationships between explanatory and response variables, and they are especially useful in demonstrating higher-order interactions (De’Ath & Fabricius 2000; Low, Joy & Makan 2006). For the explanatory variables in our analysis, we included the four variables in the mixed models described above; thus, the splitting algorithm could choose between variables that contained information about occupancy and those that also included information about breeding success. To determine the optimal size of the tree (i.e. to avoid fitting an overly large and complex model), we used cross-validation to estimate a complexity parameter, which calculated the number of divisions in the tree that best balanced model fit against complexity (Therneau & Atkinson 1997).


Establishment of Prospecting Territories

Forty-nine males (45 yearlings and 4 two-year-olds or older) and 11 females (seven yearlings) were observed to establish permanent prospecting territories in early June. In all but one case, individuals settled in their prospecting territories after nestling feeding by the resident or neighbouring pair had begun. Of the 30 colour-ringed prospectors, 14 had attempted to breed elsewhere but had failed because of nest predation, 10 individuals had been observed as floaters (i.e. unsettled individuals without a territory), and six individuals had not been previously observed during that season. Movements between the territory of the failed breeding attempt and the subsequent prospecting territory crossed more than four conspecific breeding territories, and median movement distance was 1750 m (range = 650–3500 m). Assuming an adult sex ratio of 52% males, as suggested by data collected in July when all adults moult (T. Pärt, unpublished data), the sex of individuals establishing a prospecting territory in June was strongly biased towards males (inline image, < 0·0001). There was also a bias towards yearlings (e.g. average proportion of yearlings among breeding males is 38% (Arlt et al. 2008) vs. about 90% among the prospectors), partly because yearlings were over-represented among floaters (T. Pärt, unpublished data) and failed breeders (Pärt 2001a).

Environmental and Social Cues Used

There was overwhelming support for prospectors to prefer sites that were occupied and successful (Table 1; Figs 1 and 2). By comparing models that included only occupancy information to those that considered breeding success, it is clear that prospecting wheatears had a higher preference for sites with successfully breeding pairs, with this preference increasing as the number of successful breeders in the aggregation increased (Table 1; Figs 1 and 2). The model rankings in Table 1 are visually represented by the regression tree model in Fig. 2; the tree that best predicted prospecting included divisions based on the focal site (i.e. occupied sites were more attractive than non-occupied, and territories with successfully breeding pairs were more attractive than those with failed pairs) and the neighbouring territories (i.e. the higher the number of successfully breeding neighbours, the more likely that prospectors would be resident). The tree also highlighted how focal and neighbouring sites interacted in their attractiveness with prospectors: (i) prospectors generally avoided focal site with no successful neighbours regardless of the status of the focal site; (ii) if only one neighbour was successful, the focal site was only attractive to prospectors if it was successful; (iii) if there was two or more successful neighbours, the focal site only needed to be occupied for it to be highly attractive (Fig. 2).

Table 1.   AIC-ranked candidate model set showing the relative influence of occupation and successful breeding in the focal and neighbouring territories on the probability of a prospector taking up residence. The focal territory was modelled in two ways: (1) being occupied (yes or no; focalOCC) or (2) breeding success (unoccupied, occupied failed or occupied successful; focalSUCCESS); neighbouring territories were also modelled in two ways: (1) number of occupied neighbouring territories (neighbourOCC) or (2) number of successfully breeding neighbours (neighbourSUCCESS). Additive effects are shown as +, the number of estimated parameters as K, the AIC corrected for sample size (AICc), the difference in AIC relative to the highest-ranked model (ΔAICc) and AICc weight (wi) showing the relative strength of support for each model
  1. Parameter estimates with (SE) of highest-ranked model:

  2. logit(prospector) = −4·42 (0·59) + 1·29 (0·67) focal occupied failed breeding + 2·03 (0·53) focal occupied successful breeding + 1·12 (0·21) number of successful neighbours.

FocalSUCCESS + neighbourSUCCESS578·0300·91
FocalOCC + neighbourSUCCESS482·584·550·09
FocalSUCCESS + neighbourOCC5109·0731·00
FocalOCC + neighbourOCC4112·7434·70
Intercept only2270·631920
Figure 2.

 Cross-validated regression tree showing relationships between the probability of a wheatear prospecting a site and the number and success of territorial pairs breeding in (i.e. focal) and adjacent to (i.e. neighbour) that site. Probabilities (±SE) are given at each terminal node with sample sizes in parentheses (total = 376).

These correlations could be influenced by some unknown confounding physical habitat cue linked to reproductive success and territory aggregations. There is, however, strong evidence for prospectors to cue on conspecific reproductive success when prospecting. In 11 cases where the prospecting territory was abandoned, this occurred within a few days (mean number of days ±SE = 2·7 ± 1·2) of a focal (= 9) or neighbouring (= 2) nest in the prospecting site being preyed upon; hence, nestling feeding ceased. Three individuals (of five colour-ringed) were later observed establishing a new prospecting territory at a site where pairs were feeding young. Clearly, a prospecting territory containing an established breeding pair (= 66, of which 55 had permanent prospectors and 11 with prospectors abandoning their site) was more likely to be abandoned when the focal pair failed (56% of 16 cases) than when it bred successfully (4% of 50 cases; inline image, < 0·0001). However, in cases where the focal pair failed, the prospector was more likely to remain when neighbouring pairs bred successfully (generalized linear model (number neighbours, neighbour success): number neighbours: inline image, = 0·082; neighbour success effect: inline image, P = 0·015).

Subsequent Breeding Site Selection

Almost all colour-ringed prospectors that returned to the study area in the next season (n = 19) selected a breeding territory at, or close to (i.e. in one of the neighbouring territories), their previous year’s prospecting territory (Fig. 3). Only three (16%) individuals shifted to a more distant breeding territory (280–1600 m away). The probability of breeding at the prospecting territory was strongly dependent on the survival of the original territory owner of the same sex (Fig. 3). When the original owner had died or moved to a new breeding territory, 79% of the returning colour-ringed prospectors bred on their previous year’s prospector territory, while no individual did the same when the original owner had survived and returned to its previous year’s breeding territory (Fisher’s exact test: = 0·0048). Clearly, the selection of a prospecting territory was a selection of a future breeding territory, but the probability of successfully acquiring the preferred prospected site was dependent on the survival of the original owner.

Figure 3.

 Distance from the prospected territory in year t to the breeding territory in year t + 1 for the 19 prospectors that returned in the following year. Distance is represented as the number of wheatear territories away from the prospected territory (0 =  the prospected territory). Black bars are for prospectors where the same-sex owner of the prospected territory had died/moved in the following year, and grey bars are for prospectors where the same-sex owner returned in the following year.

Prospecting and Future Breeding Success

Previously, we showed that both territory field layer height and breeding success significantly predicted reproductive success in the following year (Arlt & Pärt 2007). Further evidence that prospecting territories were located at sites with a high probability of future success was provided by the higher breeding success in the next year at prospected sites (92%) than other sites (77·5%; Logistic regression: inline image, < 0·0001). A similar difference in breeding success was observed among surviving prospectors (94%, = 18 with breeding success data) when compared to other survived birds of the same age class in the population (≥2 years old: 77%; = 233), although it was not statistically significant (inline image, = 0·16; note the small sample of surviving prospectors).


Non-breeding wheatears established prospecting territories, commonly covering one to several breeding territories of conspecifics, at a time when other pairs were feeding nestlings. These prospectors preferred to settle at sites characterised by short field layers and with a successfully breeding focal pair surrounded by other (successfully) breeding pairs (Table 1, Fig. 2); thus, both environmental and social cues were used when prospecting. Most importantly, in the subsequent year, prospectors chose a breeding site at, or in the immediate vicinity of, the prospected site when these sites were available. Successful settling in the previous prospected territories thus depended on the return of the original owner of the same sex, implying a previously overlooked benefit of preferring aggregations of conspecifics, i.e. an increased probability of securing a vacancy (see Discussion below). Our data also show that the choice of prospecting territory was adaptive because returning prospectors tended to have a higher breeding success than other pairs of the same age class. One explanation for this elevated breeding success of prospectors is that they are less constrained in their gathering of information and assessment of site quality and thus are more likely to select the best sites. Although non-breeding prospectors represent only a minority of the population, there is evidence that at least some breeding birds also prospect future territories and possibly use the same cues as shown in this study (Arlt & Pärt 2008a,b). However, because prospecting by breeding birds is generally cryptic brief (T. Pärt & D. Arlt unpublished data) and likely more constrained in time and space, observational data are limited, and thus, studies of non-breeders can give better insights into the cues used by individuals when selecting future breeding sites.

Numerous studies report on prospecting behaviour of non-breeders late in the season when other conspecifics are feeding young (see reviews in Reed et al. 1999; Danchin, Heg & Doligez 2001). Also, habitat patches with successfully breeding conspecifics are visited more frequently by prospectors than those with lower patch success (Boulinier et al. 1996, 2002) even at the small spatial scale of single territories (Pärt & Doligez 2003; Doligez, Pärt & Danchin 2004a; Piper et al. 2006). Studies on emigration and immigration rates at the habitat patch scale suggest that individual habitat selection strategies may be based on conspecific reproductive success as these rates covary with patch reproductive success in the previous year (Danchin, Boulinier & Massot 1998; Doligez et al. 1999, 2004b; Brown, Bomberger Brown & Danchin 2000; Blums et al. 2002; Serrano, Tella & Forero 2001; Ward 2005). The most compelling evidence on the use of social cues in habitat selection, however, comes from experimental studies where either rates of prospecting (Pärt & Doligez 2003; Boulinier, McCoy & Yoccoz 2008) or rates of immigration and emigration (Doligez, Danchin & Clobert 2002; Parejo et al. 2007; Boulinier, McCoy & Yoccoz 2008) are affected by experimental manipulations of reproductive success of conspecifics. However, in many of the above-mentioned studies, the direct link between prospecting behaviour and subsequent breeding habitat (or site) choice, i.e. with individual prospectors returning to their prospected sites, is unknown. At present, we are aware of only a few studies with detailed individual data on prospecting behaviour and its links to subsequent site choice (Boulinier et al. 2002; Pärt & Doligez 2003; Arlt & Pärt 2008a,b; this study). Such detailed knowledge of prospecting behaviour and subsequent choice of breeding site may give new insights into habitat selection strategies.

Conspecific Attraction: Increasing Probability of Establishment

Prospecting wheatears strongly preferred breeding aggregations of three to four conspecific pairs (including focal and the neighbouring territories; Figs 1 and 2) despite aggregation size not predicting breeding success in the next year (Arlt & Pärt 2007). Other studies have shown individuals to be attracted to patches with established conspecifics (Stamps 1988; Brown, Bomberger Brown & Danchin 2000; Nocera, Forbes & Giraldeau 2006; Betts et al. 2008; Fletcher 2008), although the benefit of doing so is usually not clear (Danchin, Boulinier & Massot 1998; Ward 2005; but see Doligez et al. 2004b). Individuals may benefit by using conspecific presence as a location cue reducing search and assessment costs because aggregated breeding sites can be more easily compared (Stamps 2001; King & Cowlishaw 2007).

We suggest that a preference for conspecific aggregations in year t−1 may be advantageous in terms of probability of territory/site establishment in year t. With larger aggregation sizes, there is an increased probability that at least one territory in the aggregation will be vacant the following year. If all breeders in an aggregation survive and display site fidelity, the prospector will have little opportunity to establish a breeding territory within that aggregation because of prior ownership and site dominance (e.g. Krebs 1982). Indeed, we found that no wheatears succeeded in establishing a breeding territory at the prospected territory when the original owner of the same sex returned the next year. However, when the original owner had died or moved to a new site, the vast majority (c. 80%) bred at the prospected territory. As a simple example, assuming a survival probability of 0·5 per year (i.e. average adult local survival rate between years for wheatears; for details on the estimation of survival rates, see Arlt et al. 2008) and complete site fidelity in successful territories (see Arlt & Pärt 2008a,b), the probability of a vacancy increases from 0·5 for a prospected site consisting of only one breeding territory to 0·875 for a prospected aggregation covering three breeding territories. Such a benefit of conspecific attraction has not been investigated before and is not included in previous theoretical studies (e.g. Boulinier & Danchin 1997; Doligez et al. 2003), but has the potential to change predictions concerning the selective advantages of different habitat selection strategies, at least when the spatial scale of habitat patches is small (e.g. one to a few sites). To model this benefit, one would also need to consider: (i) potential variation in adult survival rates relative to aggregation size, (ii) aggregation-specific local recruitment rates and (iii) whether all prospectors ‘play the same game’ of preferring aggregations. In the case of wheatears, the effect of the first two factors on the probability of vacancies is negligible (T. Pärt, unpublished data).

Multiple Cues Reduce the Uncertainty of Social Information

The strong habitat preference by prospecting wheatears for sites characterised by short field layers suggests that factors in addition to social cues are used for habitat selection decisions. This makes sense as vegetation height during nestling feeding but not during the time of territory establishment in early spring is a good predictor of its height at the same time in the subsequent year, i.e. a time period when vegetation structure strongly affects individual fitness of wheatears (Pärt 2001b; Arlt & Pärt 2007; Arlt et al. 2008; Low et al. 2010). Furthermore, choosing to settle at sites with short field layers at the end of the breeding season also ensures an abundance of available food during the energy-demanding period of pre-migration moult (Arlt & Pärt 2008a). Clearly, prospecting wheatears benefit strongly by using vegetation structure as a cue when selecting prospecting and future breeding sites. The view of environmental cues as the primary basis of habitat selection has long dominated (see Cody 1985; Morrison, Marcot & Mannan 1992), but it has also been challenged because such cues may be either unavailable or poor predictors for future breeding success (e.g. food abundance; Orians & Wittenberger 1991; Ward 2005). For example, in our study system, field layer height at the nestling stage is poorly predicted by its height at territory establishment in early spring (Arlt & Pärt 2007). But prospectors overcome this problem by assessing vegetation structure at a time it has predictive power (i.e. when nestlings are fed).

The clear preference for certain social and environmental cues indicates that prospecting northern wheatears used multiple cues when assessing territory quality; however, the relative importance of each cue in settlement decisions is unknown. Social cues are strongly affected by stochasticity of demographic and site-selection processes at small spatial scales (e.g. territories), and as the spatial scale of habitat selection decreases, the degree of uncertainty linked to social cues increases and the value of social information decreases. Consequently, individuals prospecting at the territory level should not be expected to rely only on social information when selecting their future territory. By using a combination of conspecific aggregation, conspecific reproductive success and environmental cues, individuals may reduce this uncertainty and thus increase their likelihood of choosing a high-quality habitat. Other studies have suggested that individuals may use a combination of cues for their settlement (Doligez et al. 2004b; Chalfoun & Martin 2007; Betts et al. 2008) and departure decisions (in particular by combining personal breeding success with patch success; Danchin, Boulinier & Massot 1998; Doligez et al. 1999; Blums et al. 2002; Boulinier, McCoy & Yoccoz 2008).

Spatial Scale, Life-History and Selection Strategies

By interpreting the results in terms of establishment probability and demographic stochasticity, our results suggest that the spatial scale of habitat selection is a key feature in understanding informed selection by prospectors. Although prospectors are known to visit habitat patches containing many breeding conspecifics and/or visit many potential breeding sites (Reed et al. 1999; Danchin, Heg & Doligez 2001), there are few data showing whether prospectors establish a spatially restricted prospecting site or keep on wandering. However, the literature on floating behaviour, which can be considered as a form of prospecting for future breeding sites (Reed et al. 1999), frequently reports the spatial scales of movements and behaviours of non-breeding individuals (reviewed in Zack & Stutchbury 1992). Because the number of vacancies increases as annual adult survival rate decreases, floaters of short-lived species tend to settle in a small-scale floating territory and monitor information from few breeding sites, whereas floaters in long-lived species tend to prospect and monitor more sites (Zack & Stutchbury 1992). Based on this, we suggest a link between species longevity and the spatial scales and cues used by prospectors. Prospecting behaviour and the choice of next year’s breeding site should be best described at small spatial scales (e.g. territory or group of territories) for short-lived species and at larger scales (e.g. habitat patch) for long-lived species. In addition, short-lived species should tend to use a combination of conspecific and environmental cues in their selection of future breeding sites, whereas environmental cues may be less important for long-lived species. Although these predictions are untested, studies on long-lived species suggest that non-breeding individuals prospect many different sites (e.g. Danchin, Boulinier & Massot 1998; Schjørring, Gregersen & Bregnballe 1999; but see Bruinzeel & van de Pol 2004), while limited data on short-lived species suggest that they prospect and settle in ways similar to our observations for the wheatear (see e.g. Lawn 1994; Pärt & Doligez 2003; Ward 2005). To fully test these predictions, there is a need for more data regarding the links between prospecting and subsequent settlement and the use of multiple cues. Furthermore, future modelling of habitat selection strategies should include (i) spatial scale of selection to investigate more formally its effects on the cues used to assess habitat quality and (ii) the probability of acquiring a sampled breeding site, which are likely to strongly influence the evolutionary success of using social information for breeding habitat selection.


We thank all the field assistants for their invaluable help with gathering the basic data and all land owners allowing us to work on their land. The study was funded by The Swedish Research Council (grants to T.P. and A.Q.), FORMAS (grant to T.P. and M.L.), French National Centre for Scientific Research (CNRS, PICS; to B.D.), National Research Agency (ANR-06-JCJC-0082, to B.D.), Oscar and Lili Lamm’s foundation (T.P.) and the Swedish Royal Academy of Sciences (D.A.).