Acoustic analyses
To analyze the effect of time on acoustic measurements, we analyzed 10,707 solo songs. We used a principal component analysis (PCA) using the “FactoMineR” R package (Lê et al., 2008) to condense the five song measurements into three variables with eigenvalues > 1, that explained 76% of total variance. The first principal component (PC1) correlated with duration and number of elements, the second principal component (PC2) correlated with minimum and maximum frequency, and the third principal component (PC3) correlated with the frequency of maximum amplitude (Table 1). We performed linear mixed models (GLMMs) using the “lme4” package (Bates et al., 2015) to determine the effect of populations and time periods on the variation of acoustic traits. Specifically, we included in the model two independent variables: Populations (four levels: MTV, HDA, UCR, and JBL), time periods (two levels: period 1 and period 2), and the interactions between both variables. Each of the three principal components (PC1, PC2, and PC3) of the song was included as a response variable in the model, and the territory inside each population as random factor. We additionally carry out post hoc tests when the model showed us significant differences on pairwise comparisons between the main effects and the two-factor interactions.