Field work was conducted in a 20 × 30 km area around Apaj, central Hungary (47° 6′ 53.9″ N; 19° 5′ 21.2″ E), ca. 50 km south of Budapest, in May between 2016 and 2019. Here Common Cuckoos parasitize Great Reed Warblers Acrocephalus arundinaceus in the reedbeds of a network of small irrigation and flood relief canals, connected with the river Danube. This host species prefers reedbed edges and other edge-like habitats along the canals for breeding (Moskát and Honza 2002; Mérő et al. 2015). Common Cuckoos typically perch on and move between the trees along the banks of the canals, holding partly overlapping territories (Moskát et al. 2019).
We captured Common Cuckoos with mist-nets using playbacks of male and female cuckoos to attract them. Upon capture, we sexed cuckoos by morphological characters (Svensson et al. 2010; also confirmed by DNA analysis in 2016 and 2017; Moskát et al. 2019), then we measured the following body size metrics: body mass (g; to the nearest 1 g with a Pesola spring scale), wing and tail length (mm; to the nearest 1 mm, with a ruler), and tarsus length (to the nearest 0.1 mm; with a calliper). Following measurements, we tagged each cuckoo with a 1.2 g (1% of the body mass) radio transmitter (Pip3, type 392 by Biotrack Ltd; for more details, see Moskát et al. 2017) and released them at the site of capture.
In the field we followed the tagged cuckoos and identified them individually using a Sika receiver (Biotrack Ltd.) and a flexible 3-element Yagi antenna (Biotrack Ltd.). Calls of male cuckoos were recorded 2–5 days after they had been captured (mean ± SD = 3.21 ± 2.05) and tagged, using a Telinga Universal parabola dish with Rycota Hi Wind Cover, a Marantz PMD-620 MKII sound recorder (48 kHz sampling rate, 24-bit quality, wav format), a Sennhesiser ME62 microphone, with a K6 powering module and a FEL MX mono preamp.
We calculated the natural call rate of cuckoos by measuring the number of continuously uttered calls divided by the length in sec. To obtain a representative sample, we chose sound recordings randomly (n = 18), where the calling sequence was complete, used for assessing our call rate measurements on radio-tagged cuckoos (n = 23). As call rate could be measured in several ways, e.g., by dividing the number of calls per time for a full recording, or a section of continuous calling including short pauses and/or other call types than “cu-coo” (e.g., the 3-note “cu-cu-coo”, c.f. Xia et al. 2019), we chose complete sequences with no pause and containing only the “cu-coo” call type (Fig. 1). We also took care not to sample just after a female conspecific’s bubbling call (Chance 1940; Wyllie 1981) or avoiding any other potential effect might impact the tempo of calling (e.g., the arrival of a new male or female cuckoo).
Wyllie (1981) mentioned that this species’ males uttered 10–20 (up to 270) calls in uninterrupted series, with short inter-series intervals. Møller et al. (2017) measured a range of 1–45 repeats (mean ± SE: 15.6 ± 2.0) in 24 male cuckoos. We did not analyse the maximum number of syllables uttered continuously, which had been suggested to be an indicator of individual quality in male Common Cuckoos (Møller et al. 2016; Tryjanowski et al. 2018). We did not opt for this approach, because cuckoos live at high densities in our study area (Moskát and Honza 2002), and many conspecifically initiated social interactions disrupt continuous callings of individual male cuckoos at our study site (CM pers. obs.). Nonetheless, we observed the longest calling sequences from solitary, newly arriving male cuckoos (over 100 calls per series, where similar series were repeated several times after a pause of few seconds). Consequently, the numbers of syllables uttered continuously are unlikely to function as honest indicators of body size or condition (sensu Maynard Smith and Harper 2003) in our study area with high cuckoo density. Instead, here we used the temporal frequency of the calls uttered during unit of time (i.e., call rate) as a proxy for this metric (e.g., Yorzinski and Vehrencamp 2009; Carlson et al. 2017). We also conducted a playback experiment manipulating this trait to elicit behavioural responses of territorial male cuckoos (see below).
We carried out a field experiment to test the function of one specific acoustic variable (call rate, i.e., the number of calls uttered per unit of time (s)) expected to be associated with body size and/or condition (Podos 1997; Martin et al. 2011; Weiss et al. 2012; Nishida and Takagi 2018), in a territorial context. We manipulated original cuckoo call recordings either by reducing or increasing the length of pauses among syllables, producing “quicker” or “slower” audio files for playbacks (Fig. 1), and we also used behavioural response data to cuckoo call sequences played back at the natural, unmanipulated speed (“normal”) from 2016. As the “cu-coo” call’s main function is territorial defence (Moskát et al. 2017; Tryjanowski et al. 2018), we also expected that the rate of cuckoo calls affected territorial display efficiency. Consequently, the playback file with higher or lower call rates may attract more or fewer conspecific males, respectively.
Here, we utilised the file structure we already used in previous experiments (2-min audio files, containing 3 × 30 s sequence of syllables, and 15 s pauses among the sequences; see more details of the basic call playback file structure in Moskát et al. 2017). The playback files with “normal” speed contained 20.4 ± 0.76 (mean ± SE; range: 9–24) calls in the basic, 30 s, unit of the playback files, the “quicker” files had 28.0 ± 1.3 calls (range: 18–36), and the “slower” files contained 15.2 ± 0.69 calls (range: 9–18). The number of calls per unit of time differed both in the “quicker” and “slower” files when compared to the calling frequency in the “normal” file (Mann–Whitney U test: quicker vs. normal: z17,15 = − 3.888, P < 0.001; slower vs. normal: z16,15 = − 3.959, P < 0.001). Playback files were constructed with the Audacity 2.1.0. program, and we manipulated call rate by reducing or increasing inter-call intervals. The “quicker” vs. “slower” comparison of call numbers was, of course, also highly significant (z17,16 = − 4.843, P < 0.001).
For the call rate experiment we searched for playback sites along the wooded parts of irrigation canals inhabited by host Great Reed Warblers in a slow-moving car. Experimental trials were initiated at sites, where a male cuckoo was heard and seen within 80 m. For playback, we used a JBL Xtreme (40 W) loudspeaker, connected to a Lenovo TAB 2 A7 tablet with a 20 m audio cable (see more technical details of the playback in Moskát et al. 2017). The loudspeaker was placed on a tree ~ 1.5 m height, and two observers handled the equipment and observed wild cuckoos while hiding behind a bush. Observations on cuckoos were dictated onto a Tascam dr-05 ver2 sound recorder. To avoid pseudoreplication (sensu Hurlbert 1984; Kroodsma 1989) we played a sound file only once, and selected the consecutive trial sites for playback at least 1 km distance from each other along the canals to use an individual focal cuckoo only once (e.g., Moskát et al. 2017). Playback experiments with quicker and slower treatments were carried out between May 6 and 11, 2018, in the early hours of the day (between 6 and 11 h). We also used data from playbacks of normal-speed cuckoo calls as a control for the speed manipulation specifically, and a harmless, similarly sized sympatric species, the Eurasian Collared Dove (Streptopelia decaocto) from May 2016 (Moskát et al. 2017) as a positive control, following a similar protocol to the experimental trials described above. All playbacks were conducted under good weather conditions, avoiding rainy, windy, or hot periods of the days. The sex of cuckoos was identified by their sex-specific calls and the partially sexually dichromatic plumage characteristics in this species (e.g., Moskát et al. 2020).
We analysed the following behavioural variables to characterize the cuckoos’ responses during the experimental playback trials in two sets of tests:
(i) A robust comparison of the effects of call rate for the categories “quicker”, “normal”, and “slower”, together with the dove calls used for general control, as these have proven to be the most important responses to playbacks in previous experiments on the Common Cuckoo (e.g., Moskát et al. 2017; Moskát and Hauber 2019).
Movements: A binary variable expressing if the focal cuckoo approached the speaker during the 2-min playback (Y/N).
Closest distance (m): The closest value of distance when the focal bird approached the speaker during a playback trial, either by flying or sitting on a nearby tree.
(ii) A more detailed comparison of call rate modulation (for the categories “quicker” and “slower”) to reveal fine-tuned differences in Common Cuckoos’ responses:
Distance to first detection (m): The distance of a cuckoo from the speaker when it appeared or called in the vicinity of the speaker.
Latency of first detection (s): Time spent from the start of playback until the first visual or vocal appearance of the cuckoo in the vicinity of the speaker.
Closest distance (m): The shortest distance between the cuckoo and the speaker observed during the full playback period. Approaching the speaker closely can be regarded the most important variable indicating positive response to cuckoo playbacks (Moskát et al. 2017).
Latency of closest detection (s): Time spent from the start of playback until the closest appearance of the cuckoo to the speaker during the observational period.
Latency of calling (s): Time spent from the start of playback until the male cuckoo started calling.
Length of continuous calling (s): The longest continuous calling sequence within the observational period.
Number of flights: Number of flights of the focal bird during the observational period. Cuckoos often flew away or towards the speaker, above it, or flew circle-like routes around the speaker, then sat on a tree. Some of them later repeated the same movement(s) once or several times.
Number of birds: The number of male cuckoos observed in the vicinity of the speaker during the playback.
In addition to using body size metrics, to characterise body condition, we used the residual index (Gould 1975), where the body mass is regressed on body size, and the residuals provide an estimate of condition (e.g., Jakob et al. 1996). Using this approach, we offset the size effect per se in the estimation of the physiological condition state of each subject, thus this index reflects to the true body mass of an individual without size constraints; we further we refer as physiological condition index (PCI).
We compared the two sets of call rates with Mann–Whitney U test in SPSS ver. 17.0 (SPSS Inc., Chicago, IL). We also used the glm function in the R 3.6.1 package (R core team 2019) for generalised linear model with quasi-Poisson error term (glm, Bolker et al. 2009), where call rate was the dependent variable, year was the covariate, and body size parameters (mass, PCI, tarsus, wing, and tail lengths) as fixed effects. Year and the linear predictor were adjusted by the number of birds/year. Data on body sizes collected in 2016 (n = 6 from the total 29) were omitted from glm analyses, due to the lack of tarsus data from that year. We also included second order interaction terms in the model, i.e., years × body size parameters.
We also used binary logistic regression to compare cuckoos’ behaviour to the playbacks with call rate manipulation. In the model we used the playback type (“quicker” / “slower”) as dependent variable, and eight variables as independent variables (see list of variables in Table 2). The SPSS Statistics 17.0 program package was used for binary logistic analysis, selecting the method enter. This package was also used for calculating other statistical properties and parametric unpaired t test and non-parametric Kruskall–Wallis test. Principal component analysis (PCA) was used to analyse behavioural response variables of cuckoos to playback. PCA was started from the correlation matrix, and a component was retained if the corresponding eigenvalue was greater than 1.0. For ordination plots the first two components were used with no rotation on component loadings.
Fisher’s exact tests were carried out by the “vassarstats” online calculator (https://vassarstats.net/index.html; accessed on December 22, 2020).