Introduction

Signals are transmitted in communication networks in which multiple signalers and receivers are present, but individuals other than those interacting with the signaler can also be informed by these signals (McGregor 2005). In birds, auditory and visual signals are most often used, at the intraspecific level, to compete for a territory or mate, for mate choice and in parent–offspring communication (Kroodsma and Miller 1982; Catchpole and Slater 2008; Bradbury and Vehrencamp 2011). At the interspecific level, they are mostly used in predator–prey interactions (Whitear and Stehlik 2009).

Brood parasites use similar cues to locate nests as avian nest predators (Grieef 1994). They may find nests by habitat searching (Robinson and Robinson 2001) or observation from a nearby perch (Alvarez 1993; Hauber and Russo 2000; Honza et al. 2002; Begum et al. 2011). The success of finding a nest by a parasite may be affected by nest-site characteristics including vegetation type, nest position and height above the ground, distance to habitat edge and to the nearest tree (Honza et al. 1998), as well as by vegetation cover (Moskát and Honza 2000; Fiorini et al. 2012), nest size (Briskie et al. 1990; Soler et al. 1995; McLaren and Sealy 2003; but see Grim et al. 2011; Jelínek et al. 2015) and nest visibility (Jelínek et al. 2014). Moreover, host vocal and visual conspicuousness due to territorial, nest-building or egg-laying behavior may also attract nest parasites (Uyehara and Narins 1995; Clotfelter 1998; Dabelsteen et al. 1998; Banks and Martin 2001). In this respect, some parasitic females even use host nest-defense responses directed toward them as a nest-location cue (Seppä 1969; Smith 1981; Gill et al. 1997).

Apart from using nest-location cues, brood parasites might also search for hosts that are best suited for their offspring (Parejo and Avilés 2007). Thus, they may choose hosts that will rear young parasite(s) successfully (Rothstein 1990). In some hosts, nest size reflects parental ability, and it is positively correlated with the probability of parasitism (Briskie et al. 1990; Soler et al. 1995; McLaren and Sealy 2003). Parasites also prefer high-quality hosts, as suggested from host females’ body condition and egg color (Polačiková et al. 2009). Song output might be an reliable indicator of male condition (Greig-Smith 1982; Poesel et al. 2001; Searcy and Nowicki 2000) or food resources in a territory (Radesäter and Jakobsson 1989; Manica et al. 2014). If so, the parasites may also utilize these signals when searching for suitable nests.

Though there has been considerable effort to identify cues that brood parasites employ to find host nests (Vogl et al. 2002; Antonov et al. 2007; Parejo and Avilés 2007; Patten et al. 2011; and citations above), only a few studies have focused on host singing and movement (e.g., Clotfelter 1998; Banks and Martin 2001; Garamszegi and Avilés 2005). Studies focused on parasitic Cowbirds either supported the host activity hypothesis, i.e., that parasites are attracted by host nest-building or other behavior (e.g., Briskie et al. 1990; Grieef 1994; Banks and Martin 2001; Svagelj et al. 2009) or contradicted it (Fiorini et al. 2009; Steckler and Conway 2012). Studies on parasitic Cuckoo species also had equivocal findings. Soler and Pérez-Contreras (2012), who studied the Great Spotted Cuckoo (Clamator glandarius)—Eurasian Magpie (Pica pica) system, found that the parasite’s observation of host activity near the nest determined its egg-laying decision. In the Common Cuckoo (Cuculus canorus), hereafter “Cuckoo,” this topic has been investigated by Avilés et al. (2009) only. The authors reported that neither nest size nor song output of the Great Reed Warbler (Acrocephalus arundinaceus) hosts predicted their probability of being parasitized. Thus, Avilés et al. (2009) did not support the hypothesis that Cuckoos eavesdrop on signals likely reflecting host parental abilities.

Here we extended the study of Avilés et al. (2009) by marking host males and verifying their identity before we analyzed the effect of host vocal and visual stimuli on the probability of nest parasitism by Cuckoo females. Knowing the identity of male Great Reed Warblers is important because they sometimes change their territories over the course of the breeding season. In places with high breeding density, their territories may also be quite close to each other or partly overlap and males then may make relatively long flights into the territories of their neighbors.

The Great Reed Warbler is a polygynous species that has evolved a conspicuous singing behavior with variable song structures and repertoire size (Catchpole et al. 1985). Unpaired males sing a long advertising song that usually ceases after pair formation but often reappears when the males attempt to attract another female (Hasselquist and Bench 1991). Shorter territorial song is used by both paired and unpaired males in territorial defense (Cramp 1992). Singing males typically perch high on reed stems and often move between different song posts, while exhibiting their song-related behaviors (Kennerley and Pearson 2010). They are noisy, conspicuous, less secretive and much more easily detected than males of most other Acrocephalus warblers. Paired males also sing near nests, guard them and often follow females when they are collecting nest material (Cramp 1992). Thus all the above-mentioned cues might potentially be used by female Cuckoos to locate host nests. Studies on Cowbirds revealed that brood parasitic females use the activity of hosts to locate nests (Clotfelter 1998; Jaramillo and Burke 1999; Banks and Martin 2001). Based on this knowledge, we predicted that nests of more exposed males (i.e., those spending more time singing, moving larger total distances, and spending more time on the top of reed stems) would be parasitized more than the nests of less exposed males. Additionally, we predicted that males producing a higher number and/or lower peak frequencies of kara syllables would also be more parasitized, because these low-frequency song structures are audible at longer distances (Jilka and Leisler 1974).

Methods

Fieldwork

We conducted the study from the end of April up to and including the beginning of July 2010 around fish ponds between Hodonín (48.85°N, 17.12°E) and Mutěnice (48.90°N, 17.03°E), in the Czech Republic. The host population consisted of ~100 breeding pairs of the Great Reed Warbler and experienced a relatively high parasitism rate [47% (Jelínek et al. 2014)]. Birds were mist-netted soon after territory establishment (males) or during nest building, egg laying or incubation (females). After capture, they were ringed with unique combinations of up to three plastic color rings and one aluminum ring. However, since many individuals had been marked in previous years, we did not disturb these unnecessarily and confirmed their identity by re-sighting. Based on our observations, 30 (29%) of 103 Great Reed Warbler males changed their territories over the course of the breeding season studied. The shortest and longest distances of such changes were 40 m and 5.4 km, respectively. Moreover, in some places with high breeding density, territories of some neighboring males were quite close to each other and partly overlapped, while other males made relatively long flights (even hundreds of meters) to change their song posts. Therefore, marking individual males as well as verifying their identity before observation of their behavior and before song recording were important prerequisites for collecting unbiased data.

We systematically searched for nests in male territories. The nests were found at the stage of building or at the beginning of egg laying, and were checked daily until clutch completion. At each visit, we numbered a newly laid egg with a waterproof pen according to its laying order and checked the nest content for the presence of a parasitic egg. Only clutches with a Cuckoo egg were recorded as parasitized.

Observations of color-ringed males started after male settlement and continued through the nesting stages potentially attractive to Cuckoos (i.e., nest building, egg laying) until the ultimate egg was laid (modal clutch size = 5 eggs). Observations were made by M. C., both in the morning (0500–1000 hours) and evening (1700–2100 hours), so that they covered the peaks of the highest Great Reed Warbler singing activity (Capek and Kloubec 2002; Kloubec and Capek 2011). Males were observed for 15 min after they had been detected, and after that their identity was verified using Vortex Viper 10 × 42 binoculars from a distance of 10–20 m. This distance was chosen to minimize observer effect, and to ensure correct identification of the focal male and reliable estimation of his positions. During the observation periods, we recorded the time the focal male spent singing (using a stopwatch), distances of his song-related movements exceeding 0.5 m as well as the time he spent singing at different positions on a reed stem. The position was estimated from the observer’s viewpoint as a value on a scale (1–4) from the bottom to the top quarter of the stem. Sometimes, however, the male could not be properly located, either visually or acoustically. In such cases, we used the mean position value on the scale (i.e., 2.5) to express the unknown position. A song was defined as a continuous sequence of at least four syllables separated by a clear pause. Both types of song (short and long) were taken into account, similarly as in the studies of Catchpole et al. (1986) and Hasselquist and Bench (1991). Since the Great Reed Warblers might be expected to adjust their typical behavioral routines in response to strong wind or heavy rain, we avoided making observations during inhospitable weather, which could also influence males’ detectability.

A total of 159 observations of 42 males belonging to 59 nests (25 of the nests were parasitized) were made. We made 96 observations in the morning and 63 in the evening, with a mean number of four observations per male (range 1–13) and three observations per nest of a particular male (range 1–6). Due to polygyny, some of the males had more than one nest, therefore the numbers of males and nests do not match.

We recorded the songs of 32 color-ringed males (of which 27 were also observed by M. C.) throughout the same nesting stages and periods of the day as above (though not necessarily on the same days). All recordings were made by Z. Š. from a distance 10–20 m, using a Marantz PMD 670 recorder and a Sennheiser ME-66 directional microphone. Again, we covered both types of song (sensu Catchpole 1983). All males included in this study had their nests in the same territories where the observations and/or recordings were made.

For each focal male we recorded the following breeding or nest characteristics (apart from parasitism): clutch initiation day (Julian date), nest status (monogamous—the nest of a monogamous male, primary—the nest belonging to the first female of a polygynous male, secondary—the nest belonging to the second female of the polygynous male), nest distance (in meters) to the nearest potential Cuckoo perch [shrub, tree or electric wire more than 5 m high (Jelínek et al. 2014)] and Cuckoo nest view (0 = no view, i.e., a Cuckoo could not see the nest from a perch nor locate it by host behavior; 1 = indirect view, i.e., a Cuckoo could not see the nest but could locate it by host behavior; 2 = direct view, i.e., the Cuckoo could directly see the nest from a perch; Øien et al. 1996; Jelínek et al. 2014). The last two variables were estimated by eye from the position of a nest.

Variables and statistical analyses

From the observation protocols, we calculated total time a male spent singing, total distance of his movements and conspicuousness score as \(\frac{{\mathop \sum \nolimits t_{i} p_{i} + t_{\text{u}} p_{\text{u}} }}{t}\), where t i is the time spent at a particular position and p i is the position value on the scale 1–4 (see above), t u is the time spent at an unknown position, p u is the mean value on the scale (2.5) and t is the total observation time (15 min). We averaged time spent singing, total distance of movements and conspicuousness score per nest of particular males.

Although we acquired more song recordings per male, we used only the most representative one (i.e., of the highest quality and containing at least 20 songs) for further detailed analyses. The recordings were analyzed in Avisoft SASLab Pro version 4.40 (Specht 2007) by J. C. S. As some temporal and structural parameters of the Great Reed Warbler song, like song length, repertoire size and syllable switching, have no effect on Cuckoo parasitism (Avilés et al. 2009), we focused on those structures and their spectral characteristics that make the song generally most conspicuous (Jilka and Leisler 1974). In each song, we thus counted the number of kara syllables and measured their peak frequencies (hertz), using the linear amplitude spectrum function (Fig. 1). Then we averaged these song-specific values per male.

Fig. 1
figure 1

Example spectrogram of one Great Reed Warbler song with a typical composition; “kara” syllables are indicated by dotted lines, and their peak frequencies by arrows (for the sound recording, see the Electronic supplementary material (ESM)]

Full size image

As we observed and recorded the focal males at different breeding stages and these stages differ in the length of male song (see above), in further analyses we corrected time spent singing and number of kara syllables for the particular stage and used residual values of these song variables.

To investigate whether the parasitism rate could be explained by male singing behavior, we fitted a generalized linear mixed model (GLMM) with a binomial error distribution and logit link function, where Cuckoo parasitism of individual nests was entered as a binary response variable (parasitized = 1, non-parasitized = 0). As fixed effects, we used male characteristics, i.e., mean residual time spent singing, mean total distance of movements and mean conspicuousness score. As covariates, we included nest distance to the nearest potential Cuckoo perch, Cuckoo nest view, nest status, clutch initiation day (centered in each nest status category) and second power of the clutch initiation day (to account for non-linear relationship). Male identifiers were included as a random effect, because more than one nest often belonged to particular males (especially when they were polygynous).

To explore whether the parasitism rate could be explained by conspicuous characteristics of male song, we fitted a generalized linear model (GLM) with a binomial error distribution and logit link function, where Cuckoo parasitism was entered as a binary response variable, mean residual number of kara syllables, their mean peak frequency as fixed effects, and all but one above-mentioned nest characteristics as covariates. We did not include nest status because nest numbers in particular categories were highly imbalanced, which could cause problems when fitting models.

All statistical analyses were conducted in R version 2.15.2 (R Core Team 2012). The GLMM was fitted in the package lme4 version 0.999999.0 (Bates et al. 2008), using Laplace approximation for parameter estimations (Bolker et al. 2009). In both the GLMM and the GLM, all input variables showed pairwise Spearman rank correlations lower than |r S| = 0.34 and variance inflation factors less than 1.65. Thus, the models should not have been biased by multicollinearity. Residuals from the models did not exhibit any unnatural patterns and all Cook’s distances were less than 0.60.

Results

Host males with parasitized nests (n = 25) spent in total 10.6 ± 3.6 min (mean ± SD) singing while males with non-parasitized nests (n = 34) spent 10.9 ± 3.7 min of the 15 min of observation singing. Males with parasitized nests moved a total distance of 52 ± 38 m and reached a conspicuousness score of 2.6 ± 0.5, compared to males with non-parasitized nests, which moved 39 ± 39 m and were almost equally conspicuous (2.6 ± 0.4).

Neither the above-mentioned components of male singing behavior nor the nest characteristics (Cuckoo nest view, nest distance to the nearest Cuckoo perch and nest status, timing of the clutch) significantly explained the variation in the probability of Cuckoo parasitism (Table 1).

Table 1 Estimates, SE and results from the ANOVA table (type III) of fixed effects from the full generalized linear mixed model explaining the variation in the probability of Cuckoo parasitism in relation to Great Reed Warbler singing behavior (time spent singing, total distance of movements and conspicuousness score), nest characteristics (Cuckoo nest view, distance to the nearest Cuckoo perch, nest status) and timing of breeding (clutch initiation day, clutch initiation day2)
Full size table

Males with parasitized nests (n = 14) produced 6.5 ± 2.2 kara syllables per song with a mean peak frequency of 2481 ± 147 Hz while males with non-parasitized nests (n = 18) produced 6.6 ± 2.1 kara syllables per song with a mean peak frequency of 2512 ± 97 Hz. The mean peak frequency of kara syllables was a marginally non-significant predictor of Cuckoo parasitism, with nests where males produced lower-pitched kara syllables showing a slight tendency to be more parasitized than those belonging to males producing higher-pitched kara syllables. The only important predictor, however, was the Cuckoo nest view: visible nests were statistically significantly more parasitized than hidden nests (Table 2).

Table 2 Estimates, SE and results from the ANOVA table (type III) of fixed effects from the full generalized linear model explaining the variation in the probability of Cuckoo parasitism in relation to conspicuous characteristics of Great Reed Warbler song (number of kara syllables and their mean peak frequency), nest characteristics (Cuckoo nest view, distance to the nearest Cuckoo perch) and timing of breeding (clutch initiation day, clutch initiation day2)
Full size table

Discussion

We predicted that the Great Reed Warbler singing behavior and conspicuous song structures (i.e., kara syllables) may be used by Cuckoo females as nest-location cues. In general, the song of male passerines is critical for their reproductive success because it is used in territorial defense and mate attraction (McDonald 1989; Catchpole and Slater 2008). Thus, for males, the benefits of singing should outweigh the potential cost of reduced reproductive success if their nests are parasitized (Banks and Martin 2001).

Contrary to our expectations, neither the singing behavior of more exposed host males nor their potentially conspicuous song characteristics were significant predictors of parasitism. Only nest visibility from the Cuckoo’s viewpoint, which we treated as a covariate, proved to be significant, at least in one of our analyses. More visible nests were more often parasitized than less visible nests. Our findings thus indicate that the parasitic females use nest visibility rather than male singing behavior as the best cue to locate Great Reed Warbler nests. This may be explained by the fact that the males of this host species do not always sing near their nests (Catchpole et al. 1985). Alternatively, the observation periods in our study might not have been long enough to provide a representative sample of male singing behavior across all the nesting stages potentially attractive to Cuckoos.

Although we failed to support the host activity hypothesis (Smith 1981; Clotfelter 1998; Banks and Martin 2001), Cuckoos may still exploit some aspects of the Great Reed Warbler song not measured by us. For example, Great Reed Warblers might be more conspicuous to Cuckoos if they sing louder or if they produce certain types of calls (e.g., alarm or contact calls). On the other hand, Cuckoos are known to exploit successfully various quiet or secretive species, whereas the Great Reed Warbler song and other vocal communications are generally relatively loud. It is also possible that other host behaviors (e.g., nest building or nest defense) may be relevant to Cuckoos when searching for nests. Once a nest is found, the parasitic females can rely on other signals of host quality, like egg coloration (Polačiková et al. 2009) or nest size (Soler et al. 1995), or simply choose nests on the basis of how well their own eggs match the appearance of host clutches (Antonov et al. 2012; Honza et al. 2014).

Old World parasitic Cuckoo species belong to an ancient avian group that has had enough time to specialize. Many of them lay mimetic eggs to deceive their particular host species (Davies 2000) and use various cues to locate suitable nests (Soler et al. 1995; Øien et al. 1996; Antonov et al. 2007; Begum et al. 2011). In the Cuckoo–Great Reed Warbler system, nest visibility from Cuckoo vantage points and the distance to the nearest parasitic female’s perch substantially affect the likelihood of being parasitized (Moskát and Honza 2000; Jelínek et al. 2014; but see the non-significant effect of the latter in this study). These factors were shown as important also in other Cuckoo hosts (Alvarez 1993; Antonov et al. 2007; Welbergen and Davies 2009); however, host activity as a cue for nest searching has been reported only anecdotally (Löhrl 1950; Enkelaar 1959; Seppä 1969) or was not found at all (Avilés et al. 2009). So far, the host activity hypothesis has been well supported only for the Great Spotted Cuckoo (Soler and Pérez-Contreras 2012).

New World parasitic Cowbirds, on the contrary, are a relatively young lineage, consisting mainly of generalists with no special egg mimicry (Davies 2000). Some investigations on these brood parasites indicated that host activity, including song rates and aggression, may influence host nest detectability (Smith 1981; Arcese and Smith 1988; Banks and Martin 2001; Svagelj et al. 2009; but see Grieef 1994; Clotfelter 1998). Although in Clotfelter’s study there was no evidence that male activity influenced the probability of parasitism, females whose nests were parasitized gave significantly more vocalizations during the egg-laying period than non-parasitized females. This suggests that, in some hosts, female (rather than male) vocalization might serve as a proximate cue for Cowbirds (Clotfelter 1998). Some recent studies, however, show that the more the hosts vocalize or are aggressive near their nests, the lower the probability of parasitism (Fiorini et al. 2009; Steckler and Conway 2012), which contradicts the host activity hypothesis. Host aggressiveness may also be an effective anti-parasite strategy in our study system as Great Reed Warblers are highly aggressive toward nest intruders including the Cuckoo (Bártol et al. 2002; Røskaft et al. 2002; Požgayová et al. 2013; Trnka and Grim 2013), which may even be killed by them (Molnár 1944; Janisch 1948; Mérő and Žuljević 2014).

Despite the considerable effort to identify cues and search modes that brood parasites use to find host nests (e.g., Antonov et al. 2007; Parejo and Avilés 2007; Patten et al. 2011), within the Old World host–parasite systems, there is only one study that explored female Cuckoo preference for certain nests in relation to host singing behavior (Avilés et al. 2009). Avilés et al. (2009) hypothesized that song output differs with host age or quality and thus may be used by Cuckoos as an indicator of host rearing abilities, because young or low-quality hosts are expected to give offspring less care than older or high-quality hosts. Avilés et al. (2009), however, could not support their hypothesis as none of the measured Great Reed Warbler song characteristics (i.e., proportion of time spent singing, song length, repertoire size and syllable switching) possibly reflecting host parental abilities predicted the probability of Cuckoo parasitism. Unfortunately, we could not test for age-related effects of male singing behavior and conspicuous song traits on the probability of parasitism because only a small proportion of males in our dataset were young [i.e., determined as first-year breeders based on Procházka et al. (2012)]. However, as age-specific differences in certain Great Reed Warbler song characteristics have recently been found (Forstmeier et al. 2006; Węgrzyn et al. 2010; Leniowski and Węgrzyn 2013), further studies of this in males of different age cohorts might be desirable.

Conclusion

In the Great Reed Warbler, nest visibility is probably a more important determinant of parasitism than male singing behavior or conspicuous song structures. Well-concealed nests have a better chance of escaping parasitism than nests which are easy to detect (Moskát and Honza 2000). However, the host activity hypothesis might still be valid if other host behaviors (e.g., nest building or aggressiveness) can reveal the nest and thus attract Cuckoos.