Conspecific cues and breeding territory selection in Whinchat on abandoned fields

1.   Introduction

Breeding territory selection has important fitness consequences (Kelly and Schmidt, 2017). During territory establishment, birds often integrate information from multiple sources, including social cues and personal information obtained through direct assessment of the environment (Szymkowiak, 2013). Social information can be derived from observing other individuals' interactions with the environment, including their actions, decisions, and performance (Danchin et al., 2004). For birds, vocalisations are the most obvious social cue for quickly assessing the presence and performance of conspecifics (Ahlering et al., 2010; Szymkowiak, 2013; Buxton et al., 2020). Acoustic cues are used by migratory species in landfall decisions (Mukhin et al., 2008; Lehnardt and Sapir, 2024), breeding habitat selection and territory selection (Buxton et al., 2020; Valente et al., 2021). At all these stages, heterospecific acoustic cues may be used in addition to conspecific acoustic cues (Mukhin et al., 2008; Szymkowiak et al., 2017; Kelly et al., 2018). Different types of acoustic signals, such as territorial male songs (Ahlering et al., 2010) and fledgling calls (Parejo et al., 2007), can effectively attract individuals to breeding habitats or territories. Broadcast playback of these signals has been used to study local recruitment and density in many bird species (Ahlering et al., 2010; Valente et al., 2021).

Many studies have shown that migrant passerines perceive social information about territory quality during the pre-breeding period, and the most common conspecific cue is the song of territorial males (Ahlering et al., 2010). Playback of male songs during the pre-breeding period resulted in both occupation of unoccupied habitats and increased abundance in habitats already occupied (Valente et al., 2021). However, the use of pre-breeding cues carries the risk of ecological traps, as pre-breeding cues may not provide information on potential reproductive success (Doligez et al., 2003; Szymkowiak et al., 2016). Performance-based cues may provide more accurate estimates of territory quality than conspecific presence alone (Kelly and Schmidt, 2017). Fledgling begging calls are a good indicator of reproductive success and could be used by birds as performance-based post-breeding cues (Parejo et al., 2007; Kelly and Schmidt, 2017; Brandl et al., 2019). Although the use of pre- and post-breeding sources separately has been well studied, few studies have examined both sources of social information simultaneously for a single species (Andrews et al., 2015).

Despite the general consensus that social information plays an important role in the establishment of territories, it is clear that different bird species differ in the extent to which they use these cues, and the external and internal causes of these differences remain incompletely studied (Valente et al., 2021). One hypothesis is that the characteristics of the breeding habitat have a strong influence on the importance of social information. Early successional and ephemeral habitats, which can be used by birds almost immediately after they're formed, seem to be of particular interest. Although some authors have suggested that social cues are more important for birds inhabiting early successional habitats (Ahlering et al., 2010), the majority of studies show that vegetation structure cues may be more reliable than social cues for these species due to the highly ephemeral nature of optimal habitat (Nocera et al., 2006; Ward et al., 2010; Albrecht-Mallinger and Bulluck, 2016). For example, Andrews et al. (2015) showed that Grasshopper Sparrows (Ammodramus savannarum) do not rely on social cues when arriving at the breeding sites, and that adding social cues did not increase density between years. The authors suggested that the sparrows quickly discover new habitats, probably via habitat cues, as conspecific social cues would not be present in newly created sites.

All studies of conspecific attraction during breeding territory selection in migratory passerines in early successional habitats have been conducted in the Western Hemisphere; information on the use of conspecific social information by Palaearctic birds is only available for a few forest species (Alatalo et al., 1982; Szymkowiak et al., 2016; Grendelmeier et al., 2017). At the end of the 20th century, large areas of abandoned fields, pastures and meadows appeared in Eastern Europe (Lesiv et al., 2018). In these areas, specific bird communities have developed that are distinct from those of natural and semi-natural grasslands (Oparin, 2008; Herzon et al., 2014; Kamp et al., 2018). An important feature of these habitats is a high level of nest predation (Frankiewicz, 2008; Shitikov et al., 2015). The high species diversity of nest predators determines the unpredictability of the reproductive success of ground-nesting birds in abandoned areas, which can decrease to critically low levels in some years (Samsonov et al., 2022). Under these conditions, indicators of habitat quality available to birds during the pre-breeding period may be ineffective, and abandoned fields may become ecological traps for birds (Lameris et al., 2016). It is therefore of great interest how and when birds select breeding territories in abandoned fields, as the ephemeral nature of the habitat implies the use of pre-breeding cues (personal or social), while the high nest predation rate may determine the importance of post-breeding performance-based cues.

Here, we used a field experiment to test two hypotheses describing the role of conspecific attraction in breeding territory selection in Whinchat (Saxicola rubetra) in abandoned fields in northwestern European Russia. The pre-breeding conspecific attraction hypothesis predicts that Whinchats use the territorial songs of conspecifics as habitat quality cues during the pre-breeding season. The post-breeding conspecific attraction hypothesis predicts that Whinchats use fledglings' calls as habitat quality cues during the post-breeding season. We evaluated whether Whinchat abundance changed in response to the simulated presence of conspecifics. We conducted playback experiments with pre-breeding (male songs) and post-breeding (fledglings’ calls) social information. We estimated the difference in Whinchat abundance in two consecutive years and tested whether this value differed between treatments (post- or pre-breeding) and control plots.

2.   Materials and methods

2.1   Study species and site

Whinchat is a small, territorial, ground-nesting songbird that inhabits a variety of open habitats including farmland, grassland and peat bogs. Whinchat is a long-distance migrant with wintering grounds in sub-Saharan Africa (Winkler et al., 2020). In northwestern Russia, Whinchat spring arrival and settlement starts in early May. Males usually start singing after arrival and stop after the first egg-laying (Vaytina and Shitikov, 2019). Whinchat is socially monogamous and single-brooded.

The study was conducted in 2023 and 2024 in “Russky Sever” National Park, northwestern Russia. Open habitats are patchily distributed in the region and consists mostly of small patches, usually no larger than 100 ha, separated by large areas of forest. The majority of agricultural fields, pastures and hayfields were completely abandoned since the beginning of the 21st century (Shitikov et al., 2015). Our study covered two patches of abandoned agricultural lands (Topornya, 59.767° N, 38.367° E, S = 700 ha and Sukhoverkhovo, 59.852° N, 38.504° E, S = 46 ha) located 12 km apart. Both sites were covered by mainly grassland vegetation dominated by Cock's-foot (Dactilus glomerata), Timothy (Phleum pratense) and Cow Parsley (Anthríscus sylvéstris), with patches of Large-leaved Lupine (Lupínus polyphyllus) and Raspberry (Rubus idaeus). All of the abandoned fields are encroached by Willow (Salix sp.), young Scots Pine (Pinus sylvestris) and Norway Spruce (Pícea abies). Sukhoverkhovo is a single field whereas Topornya is a complex mosaic of abandoned fields with varying degrees of overgrowth, separated by small patches of forest, villages, dirt roads and drainage ditches.

2.2   Whinchat abundance estimation

We counted territorial Whinchats from 25th May until June 25, 2023 in five different fields (21–87 ha) in the Topornya study site and in the whole Sukhoverkhovo study site (46 ha). The total area covered by the counts was 258 ha. Censuses involved repeated weekly field visits during which the exact locations of territorial individuals or pairs were georeferenced using a handheld GPS device (Garmin GPSMap 64). Following Szymkowiak et al. (2016), we classified a territory as occupied if at least one of the three following conditions is sufficient: (a) a singing male was recorded at least three times in the same location; (b) we observed a pair at least twice on subsequent visits; (c) a nest was found (n = 10 in 2023 and n = 23 in 2024). The results of the 2023 pre-treatment surveys were used to assign experimental plots (see below). In 2024, surveys were repeated in all five fields in Topornya study site and in the whole of Sukhoverkhovo study site at the same time as the previous year. The observer (TMV) was blind to the distribution of treatment types between plots.

2.3   Playback preparation

In June 2023, we placed a Rode NTG 2 shotgun microphone ∼0.5 m from nests containing 10–11-day old nestlings to record nestling begging vocalisations. For further processing, the first 15 min of audio were removed to exclude adult alarm calls. Recordings of nestlings from two nests were mixed with recordings of two Whinchat fledglings in 15-min fragments separated by an acoustic pause of the same duration. Recordings of Whinchat fledglings were obtained from https://xeno-canto.org/. Four different 15-min fragments separated by pauses were combined into a single recording, which was used as a post-breeding cue in the experiment. Recordings of vocalisations of four territorial unpaired Whinchat males, obtained in the local population in 2012–2013 (see Vaytina and Shitikov, 2019 for details), were combined into 15-min fragments separated by an acoustic pause of the same duration. Four different 15-min fragments separated by pauses were combined into a single recording, which was used as a pre-breeding cue in the experiment. All recordings were prepared as 16-bit wav files (sampling rate: 44.1 kHz) in Audacity 3.3.2 (https://audacityteam.org).

2.4   Playback experiment

Based on the results of the pre-treatment (2023) surveys, we selected 27 experimental plots (24 at the Topornya study site, 3 at the Sukhoverkhovo study site). Each plot was 3.14 ha in size, corresponding to a circle with a radius of 100 m, which would allow up to 5 Whinchat pairs to settle within the experimental plot (Shitikov and Fedchuk, 2008). The nearest plot centres were >200 m apart (mean: 323.0, max: 627.1, min: 213.1). There are two to four Whinchat breeding pairs per plot (mean 2.7 pairs), habitat patches with low (0–1 pairs per plot) and high (5 pairs per plot) breeding numbers were excluded from the experiment. Thus, our experiment was conducted on plots with moderate Whinchat densities, as the response to conspecific attraction may be strongest in areas of moderate density prior to treatment (Fletcher, 2007). We assigned each plot to one of three treatments: post-breeding treatment (T1), pre-breeding treatment (T2) and silent control. As vegetation structure varied slightly across the Topornya study site, we distributed treatments so that each field had an equal number of plots of each treatment type. Within each field, the distribution of treatment types was randomised. Consistent with many other studies using playback to investigate conspecific attraction in songbirds (Fletcher, 2007; Andrews et al., 2015; Albrecht-Mallinger and Bulluck, 2016; Szymkowiak et al., 2016; Kelly and Schmidt, 2017), we did not broadcast any playbacks on control plots.

Playback stations consisted of a digital audio player with loaded playback recordings, 12 V battery and Bluetooth speaker (Tronsmart trip 10, Tronsmart T2 or JBL Flip 5). The audio player and battery were placed in a plastic box and covered by dry grass. The speaker was mounted on the wooden pole 30–40 cm above the ground. Based on volume levels used in other studies (Fletcher, 2007) playbacks were calibrated to c. 90 dB using the XY1359 Sound Level Meter (Dongguan XinJiaYi electronics Co., Ltd.) as measured 1 m from the speaker. In each T1 plot, four playback points were selected at the corners of a square whose center coincided with the center of the plot. The distance from each playback point to the plot center was 30 m. The playback station was placed at one of the playback points for three days, then removed for recharging (24 h). The recording was played continuously in repeat mode. After recharging, the system was installed at the next playback point. This avoided habituation of birds and simulated movement of fledglings around the plot. If a system switched off prematurely due to power problems, it was removed for recharging and reinstalled at the next playback point within the plot after 24 h. All audio systems were removed from the plot during heavy rainfall. Playback began simultaneously at all nine T1 plots on June 24, 2023 and was completed on July 29, 2023. This resulted in between 19 and 24 days of playback (mean = 22 days) per plot. In each T2 plot, we randomly selected two playback points 60 m apart, with the centre of the plot in the middle. At each of the two points, the playback stations were placed for three days, after which both playback stations were removed for recharging (24 h). The recording was played continuously in repeat mode. After recharging, the systems were installed at the previous playback points. This simulated the permanent presence of two territorial males at the plot, i.e., the lower level of moderate Whinchat density. Playback started simultaneously at all nine T2 plots on May 4, 2024 and was completed on May 30, 2024, which resulted in 18 days of playback at each plot.

2.5   Statistical analysis

We used a linear mixed effect models (LMM) in lme4 package (Bates et al., 2015) to test the hypotheses for using social cues in habitat selection by Whinchat. The change in estimated Whinchat abundance at experimental plots between the 2023 and 2024 breeding seasons (abundance₂₀₂₄ – abudance₂₀₂₃) was used as the response variable. The treatment type (T1, T2 or control) was used as a fixed effect. We considered models with specified treatment contrasts, i.e. where control plots formed a reference group and parameter estimates for T1 and T2 treatments were compared with this baseline. Because birds can use the density of conspecifics as a post-breeding cue (Szymkowiak et al., 2016) we use a pre-treatment density per each plot as a numeric independent variable. We use the field identifier (FT1–FT5 for Topornya and FS for Sukhoverkhovo) as a random factor. The dependent variable was normalized with mean 0 and SD of 1. We started from the general model included two independent variables (treatment type and pre-treatment numbers), its interaction and the random effect. Then we used the function “dredge” from MuMIn package (Barton, 2022) to rank all possible reduced models. We computed the marginal pseudo-R²-value for mixed-effect models (Nakagawa and Schielzeth, 2013) using partR2 package (Stoffel et al., 2021) to estimate the percentage variation in the model explained by the fixed effects.

We used an information-theoretic approach to model selection and multi-model inference (Burnham and Anderson, 2002). For each model in a particular candidate set, the value of Akaike information criterion adjusted for small sample sizes (AICc) was calculated. Models within a given set were compared based on ΔAICc values. We used model averaging over a confidence set of most supported (ΔAICc ≤4) models containing the variable to obtain beta estimates and their confidential intervals (Burnham and Anderson, 2002). The fixed effect was considered strong if the 95% CI of the estimate did not span zero. To assess the strength of the effect and to compare our results with those of other studies, we calculated Hedges' d and its variance using the formula provided by Valente et al. (2021). The value of Hedge's d indicates the number of pooled standard deviations separating the average treatment and control sites. Values below 0.5 are considered small to medium effect sizes, while values of 0.8 or greater are considered large and obvious (Cohen, 1988). Statistical analysis was performed in R v. 4.2.1 (R Core Team, 2022).

3.   Results

Changes in Whinchat numbers between years differed between treatments (Fig. 1). Whinchat abundance increased on T1 plots (mean difference between years was 0.67 ± 0.29 pairs per plot, 95% CI: 0.001–1.33) and did not change significantly on T2 (mean difference between years was −0.22 ± 0.32 pairs per plot, 95% CI: −0.97–0.52) and control (mean difference between years was −0.22 ± 0.22, 95% CI: −0.73–0.29) plots. Of the 9 T1 plots, the number of Whinchats increased in the post-treatment year in 4 plots and remained unchanged in 5 plots. The confidence intervals for all three means overlap (Fig. 1). Model selection did not clearly identify a top model, but the treatment model and the intercept model had ΔAICc <2 (Table 1). The model averaging revealed a positive response of Whinchat abundance to the post-breeding treatment (β = 0.97 ± 0.43, 95% CI 0.07, 1.88), whereas there was no effect of the pre-breeding treatment on Whinchat abundance (β = 0.01 ± 0.44, 95% CI −0.90, 0.91). The model considering the additive effect of abundance in the pre-treatment year had an ΔAICc within four points, also close to the best model, but the 95% CI for the estimate of the density effect included zero (β = −0.33 ± 0.35, 95% CI −1.05, 0.40). The Hedges’ d for the T1 effect was d = 1.13 and its variance was v = 0.26, indicating a large effect of the post-breeding treatment on Whinchat abundance.

Figure  1.  The year-to-year difference in Whinchat numbers for treatment (post-breeding and pre-breeding) and control plots. Mean changes in abundance and 95% confidence intervals are shown.

DownLoad: Full-Size Img PowerPoint

Table  1.  Comparison of models describing the effect of treatment on the numbers of Whinchat.

Model	K	ΔAICc	wi
Treatment	5	0.00	0.42
Intercept	3	0.51	0.33
Density	4	2.58	0.12
Density + Treatment	6	2.86	0.10
Density × Treatment	8	5.30	0.03
Models were ranked based on the difference from the top model in Akaike’s Information Criterion corrected for small sample size (ΔAICc). K – number of parameters; wi – model weight. The pseudo-R2 value (Nakagawa and Schielzeth, 2013) for general model was 0.28.

 | Show Table

DownLoad: CSV

4.   Discussion

We did not find any evidence that Whinchats use the pre-breeding social information during territory establishment. This result contradicts the popular view that passerines are influenced by pre-breeding conspecific cues when selecting a breeding territory (Ward and Schlossberg, 2004; Ahlering et al., 2010; Valente et al., 2021). One the other hand, our results in line with studies indicating that birds of early successional and ephemeral habitats do not use the social information during pre-breeding period (Nocera et al., 2006; Ward et al., 2010; Albrecht-Mallinger and Bulluck, 2016). Whinchat benefits greatly from the abandonment of farmland, with densities rising sharply in newly formed abandoned fields (Oparin, 2008; Shitikov et al., 2015; Kamp et al., 2018). So, this species needs to have an evolutionarily evolved mechanism to quickly discover newly created territories. At the same time, it has strict requirements for all occupied habitats (both ephemeral and permanent), the most important of which is the presence of singing and feeding perches above the general level of grass cover (Fischer et al., 2013; Border et al., 2017). Therefore, we assume that Whinchats discover new territories via habitat cues, and pre-breeding conspecific social cues do not influence this choice. A similar mechanism of habitat selection has been described for passerines in early successional grasslands in North America (Andrews et al., 2015).

It is necessary to consider that the design of our experiment does not allow us to test unequivocally whether pre-breeding conspecific cues influence the decisions of Whinchat during the settlement process. The effect of treatment on settlement increased with size of the area the researchers examined for settlement (Valente et al., 2021). Moreover, negative density dependence could affect individuals at territory scales within a habitat patch (territory despotism; Fretwell and Lucas, 1970; Rodenhouse et al., 1997). The limited amount of open habitat in our study area precluded the use of large plots. Our experimental plots contained 1–5 Whinchat breeding territories, so our results suggest a lack of conspecific attraction only at the territory scale. This does not completely exclude the possibility that conspecific pre-breeding cues have a significant effect on Whinchat at larger spatial scales (i.e., during habitat selection).

Our results provide limited evidence that Whinchats use post-breeding social information from conspecific fledgling calls when establishing breeding territories in the subsequent year. This finding is in line with other studies showing that fledgling calls can be an important cue for establishing breeding territories in songbirds (Parejo et al., 2007; Kelly and Schmidt, 2017; Brandl et al., 2019). Whinchat abundance was higher in the part of the plots treated with fledgling calls than in the silent controls. There was a slight increase in pair numbers, and not in all the post-breeding treatment plots. The LMM results also indicate a high uncertainty around the effect revealed, primarily due to the small sample size. This uncertainty may also have a biological explanation. There is a considerable time lag between the perception of a post-breeding cue and the establishment of a breeding territory the following year. During this time, Whinchats undertake two transcontinental migrations and spend the winter in sub-Saharan Africa. The apparent survival rate of adult Whinchats after successful breeding in our study area is no more than 40% (Shitikov et al., 2015; Fay et al., 2021), and we have no reason to believe that this may be much higher for birds that have chosen new territories in the post-breeding period. If more than half of these birds do not return from the wintering grounds, the post-breeding treatment could not increase the number of Whinchats in all plots. In recent metanalysis Valente et al. (2021) showed that in 43 studies of conspecific attraction effect breeding bird abundance was on average 0.76 (95% CI: −0.059 to 1.73) standard deviations higher in treatment than in control sites. In our study, using the formula from Valente et al. (2021), we obtained a value for the post-breeding treatment of d = 1.13, indicating that the effectiveness of conspecific attraction is even higher for Whinchat than for many other bird species.

Whinchat is a species whose population dynamics are largely determined by breeding success (Fay et al., 2021). While haying and grazing are the main causes of Whinchat nest mortality on intensively used agricultural land (Grüebler et al., 2015; Tome et al., 2020), nest predation is the main cause of nest mortality on abandoned farmlands (Frankiewicz, 2008; Shitikov et al., 2015). High predation pressure, combined with unstable weather conditions, reduced Whinchat annual reproductive success by up to 10% in some years in our study area (Grudinskaya et al., 2022; Samsonov et al., 2022). A significant number of adult Whinchats change their breeding territory after unsuccessful breeding and may do so during the season (Grüebler et al., 2015; our unpublished data) or the following year (Shitikov et al., 2015). When birds leave territories after unsuccessful breeding, they need to have a strategy for finding new territories in order to maintain high levels of reproductive success. Using social information that indicates reproductive success (i.e., post-breeding social information) seems to be the best way to make this choice. Whinchat males almost completely stop singing during the incubation stage (Vaytina and Shitikov, 2019), during the nestling and fledgling stages adult Whinchats only produce alarm calls, which may indicate the presence of predators, i.e., potentially low territory quality. Thus, Whinchat could not use male songs in post-breeding period as shown for other passerine species (Betts et al., 2008; Andrews et al., 2015) and fledglings’ calls may be the only source of post-breeding social information for this species.

The increased number of Whinchats in the post-breeding treatment plots could be explained by using other sources of social information besides conspecific acoustic cues, such as the density of conspecifics in the previous year (Szymkowiak et al., 2016). We did not find any effects of pre-treatment year abundance on Whinchat response to simulated conspecific cues. This suggests that information on conspecific density collected during the previous breeding season did not influence breeding territory selection and Whinchats used performance-based indicators only. Breeding territory selection in migratory passerines may also be determined by heterospecific density or heterospecific acoustic cues (Szymkowiak et al., 2017; Kelly et al., 2018). Heterospecific cues, especially for habitat specialists inhabiting fragmented wetlands, have also been shown to be very important for birds to cease migration (Mukhin et al., 2008). Similarly, it is conceivable that heterospecific cues could significantly increase the efficiency of breeding habitat selection for birds of early successional habitats, such as Whinchats. In our study system, there is only one other species (Booted Warbler Iduna caligata) with similar habitat requirements and densities to Whinchat. It is possible that Whinchat could use information on the distribution of Booted Warblers acquiring from acoustic cues. However, Booted Warblers arrive and start breeding at least 14 days later than Whinchats (Grudinskaya et al., 2022). In addition, the breeding success of Whinchat in our study area decreased at high densities of heterospecifics (Shitikov et al., 2018). This makes the use of heterospecific cues by Whinchats in territory selection unlikely in our study system.

Pärt et al. (2011) showed that using multiple cues (social and personal) reduced the negative effect of stochasticity on the reliability of social cues at small spatial scales (e.g., territories) in Northern Wheatear (Oenanthe oenanthe) in farmland. We propose that a similar strategy is used by Whinchat. Firstly, experienced adults after successful breeding show a high degree of site fidelity (Shitikov et al., 2015; Fay et al., 2021). Some of them may also retain their previous breeding territories (Shitikov and Fedchuk, 2008). Habitat cues are also very important for Whinchat (Fischer et al., 2013; Border et al., 2017) and are certainly used when searching for territories in both newly established and long-term habitats. Finally, the results of the present study allow us to propose that Whinchats use performance-based post-breeding cues to select the following year's territory.

CRediT authorship contribution statement

Dmitry Shitikov: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Tatiana Vaytina: Writing – original draft, Methodology, Investigation, Data curation. Polina Lebedyanskaya: Writing – review & editing, Investigation, Data curation.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The fieldwork in Russky Sever National Park was made possible through the support of the administration of the Park, particularly A.L. Kuznetsov and L.V. Kuznetsova. The students and graduates of the Department of Zoology and Ecology, Moscow Pedagogical State University, were involved in the fieldwork.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.avrs.2024.100212.