In short-lived animals, innate immunity is an important component of fitness and quality. Although receivers cannot generally assess a signaler's immune function directly, sexually selected displays such as birdsong may reflect past or current condition. We investigated the degree to which song complexity and consistency, thought to reflect condition over different developmental timescales, predict multiple aspects of innate immunity in male song sparrows (Melospiza melodia). We also investigated correlations among immune measures. Non-cellular components of innate immunity (soluble blood proteins including natural antibody and other protective proteins) were negatively related to cellular (phagocytosis-based) components, suggesting trade-offs within innate immune protection. This pattern underscores the risk of inferring "immunocompetence" from a single metric. Song complexity, a permanent trait in this species, was positively related to noncellular relative to cellular immune components and may thus provide information as to the singer's innate immune strategy (investment in non-cellular versus cellular activity). Such a relationship could arise through shared timing of song learning and antibody repertoire development in early life. Singing consistency, thought to track variation in current condition and measured at both whole-song and syllable scales, did not predict any immune measures. Developmental timing of signals thus appears to influence their information content.
Between April 16 and May 18, 2010, we recorded song repertoires from 20 of the 38 study birds. Songs were recorded at 16-bit, 44.1-kHz resolution on Marantz PMD 671 recorders with Telinga Twin Science parabolic microphones (Uppsala, Sweden). We considered a repertoire complete after recording 300 consecutive or 450 nonconsecutive songs (Pfaff et al. 2007). For another 14 birds, repertoires had been recorded in previous years. We included these birds in analyses of song complexity, which remains fixed throughout life (Nordby et al. 2002), but not in analyses of song consistency, which may change over time (Botero et al. 2009). The remaining four birds were not recorded but were included in correlational and principal components analysis of immune function.
Song Complexity. For all 34 birds with repertoires available, we generated spectrograms in SyrinxPC (http:// syrinxpc.com). We sorted each bird’s complete set of recordings into song types and for each identified its component syllables, defined as one or more traces on a spectrogram that always occur together. To estimate song complexity, we noted syllable repertoire size as the number of distinct syllable types in a bird’s repertoire (range p 28–51). Syllable repertoire size is correlated with song repertoire size, the number of song types in a repertoire (Pearson’s r33 p 0.45, P p .008) but is more continuously and hence more normally distributed.
Song Consistency. We used two complementary measures of song consistency. First, whole-song consistency measures spectral similarity across multiple renditions of the same song type (Gilman et al. 2007; Schmidt et al. 2013b), and reflects consistency in song structure (e.g., number of syllable repetitions; fig. 1). Second, syllable consistency measures similarity across multiple renditions of a single syllable (Botero et al. 2009) and likely reflects neuromuscular coordination or fine motor control.