Data collection
We collected acoustic and phylogenetic data for the species belonging to the subfamily Cuculinae, in which 56 parasitic and 32 nonparasitic species are currently recognized (Payne and Sorenson 2005). Among these we obtained acoustic data for 80 species from online repositories (see below). Phylogenetic tress were generated using BirdTree (birdtree.org, http://www.birdtree.org, Jetz et al. 2012) for 76 species. As a result, both acoustic and phylogenetic data were available for 67 species (42 parasitic and 25 nonparasitic species), with which we conducted further analyses.
Acoustic samples of the typical male calls were collected from mainly Xeno-Canto (www.xeno-canto.org) and AVoCet (Avian Vocalizations Center, www.avocet.zoology.msu.edu). We referred to Payne and Sorenson (2005) to determine a typical male call type for each of the species. On average, we collected 13.57 ± 6.19 different high quality calls across species (see Additional file 1: Table S1). In this study, we followed the scientific names used by Payne and Sorenson (2005) because these are consistent with those used in Xeno-Canto.
Acoustic data measurement
Acoustic parameters were measured using the software Raven Pro 1.4 (Cornell Lab of Ornithology 2010). We measured five parameters for each sample: high frequency (HF), low frequency (LF), delta frequency (DF), delta time (DT), and number of notes (NN) (Fig. 1). HF and LF represent the highest and lowest pitch of selected ranges of syllables, respectively, and DF is the difference between them. DT is the time duration of the syllable, and NN, the number of notes involved in a single syllable. All definitions of parameters follow the manual of Raven Pro 1.4. (Charif et al. 2010).
Statistical analysis
Analyses were conducted in two ways, with and without phylogenetic consideration. For the original data without phylogenetic consideration, the distribution patterns of vocal parameters between the parasitic and nonparasitic groups were compared using the Kolmogorov–Smirnov tests. The Wilcoxon rank sum tests were applied to examine the median difference between the two groups. On the other hand, phylogenetic independent contrast (PIC) analyses were conducted for each parameters using the function “brunch” in the package “Caper” (Orme et al. 2013) in R 3.2.1 (R Development Core Team 2011), which calculates the contrast values of the continuous variable on the nodes where two different categorical variables diverge. For phylogenetic information used in the analysis, we first obtained a total of 1000 possible phylogenetic trees of the 67 Cuculinae species from BirdTree (Rubolini et al. 2015), and then we randomly selected a tree from these to meet the condition of the phylogenetic tree format for the analysis (Additional file 1: Fig. S1). The choice of trees may not change the results qualitatively as shown in other studies (e.g., Medina and Langmore 2015). Statistical significance of the difference in the mean of the contrast values from zero was tested using a one-sample t test. Models were checked for the robustness of the contrast using ‘caic.diagnostics’ function in ‘Caper’. Acoustic data were averaged according to species, and log-transformed before the analysis. All statistical analyses were performed in R 3.2.1 (R Development Core Team 2011).