Experimentally broadcast ocean surf and river noise alters birdsong

Abstract

Anthropogenic noise and its effects on acoustic communication have received considerable attention in recent decades. Yet, the natural acoustic environment's influence on communication and its role in shaping acoustic signals remains unclear. We used large-scale playbacks of ocean surf in coastal areas and whitewater river noise in riparian areas to investigate how natural sounds influences song structure in six songbird species. We recorded individuals defending territories in a variety of acoustic conditions across 19 study sites in California and 18 sites in Idaho. Acoustic characteristics across the sites included naturally quiet 'control' sites, 'positive control' sites that were adjacent to the ocean or a whitewater river and thus were naturally noisy, 'phantom' playback sites that were exposed to continuous broadcast of low-frequency ocean surf or whitewater noise, and 'shifted' playback sites with continuous broadcast of ocean surf or whitewater noise shifted up in frequency. We predicted that spectral and temporal song structure would generally correlate with background sound amplitude and that signal features would differ across site types based on the spectral profile of the acoustic environment. We found that the ways in which song structure varied with background acoustics were quite variable from species to species. For instance, in Idaho both the frequency bandwidth and duration of lazuli bunting (Passerina amoena) and song sparrow (Melospiza melodia) songs decreased with elevated background noise, but these song features were unrelated to background noise in the warbling vireo (Vireo gilvus), which tended to increase both the minimum and maximum frequency of songs with background noise amplitude. In California, the bandwidth of the trill of white-crowned sparrow (Zonotrichia leucophrys) song decreased with background noise amplitude, matching results of previous studies involving both natural and anthropogenic noise. In contrast, wrentit (Chamaea fasciata) song bandwidth was positively related to the amplitude of background noise. Although responses were quite heterogeneous, song features of all six species varied with amplitude and/or frequency of background noise. Collectively, these results provide strong evidence that natural soundscapes have long influenced vocal behavior. More broadly, the evolved behavioral responses to the long-standing challenges presented by natural sources of noise likely explain the many responses observed for species communicating in difficult signal conditions presented by human-made noise.

Keywords: Acoustic masking; Ambient noise; Birdsong; Natural sound; Sensory ecology; Vocal behavior.

Figure 1. Power spectra of the four acoustic conditions are overlayed with song power spectra for white-crowned sparrow and wrentit in California (A), plus lazuli bunting and yellow warbler (C) and song sparrow and warbling vireo (E) in Idaho.

Spectra of acoustic conditions are approximated to reflect average sound level from by treatment type and songs are amplified to a relative peak amplitude of 85 dB (re 1 dimensionless sample units), reflective of typical song at 1 m from the source (amplification performed in Raven Pro v1.5, Bioacoustics Research Program, 2017). Species-typical song spectrograms for California (B) and Idaho species (D) from individuals recorded under control conditions. Acoustic variables used in the analysis are marked on spectrograms; dB min/dB max = minimum/maximum frequency threshold for trill/song subset, PFC max = maximum peak frequency contour for song subset, BW = frequency bandwidth, and 5%, center, and 95% = 5%, center, and 95% frequencies, respectively (see text for detailed explanation of variables).

Figure 2. Influence of background sound level on song features for California white-crowned sparrows (A–B) and wrentits (C–D).

Points denote individual songs. Only song features with a strong effect of sound level in the top model (ΔAIC_c = 0.00) are shown. Background sound levels reflect received levels at the vocalizing bird and represent acoustic conditions on phantom and shifted sites with and without playback broadcast, plus control and positive control sites.

Figure 3. Influence of treatment type on California white-crowned sparrow (A–B) and wrentit (C–D) song features.

Double asterisks indicate significant contrasts corresponding to 95% CIs and single asterisks correspond to 85% CIs. Only song features with a strong effect of treatment in the top model (ΔAIC_c = 0.00) are shown.

Figure 4. Influence of background sound level on song features for Idaho song sparrows (A–B) and lazuli buntings (C–D).

Points denote individual songs. Only song features with a strong effect of sound level in the top model (ΔAIC_c = 0.00) are shown. Background sound levels reflect received levels at the vocalizing bird and represent acoustic conditions on phantom and shifted sites with and without playback broadcast, plus control and positive control sites.

Figure 5. Influence of treatment type on song features for song sparrows (A–C), warbling vireos (D), lazuli buntings (E), and yellow warblers (F) in Idaho.

Double asterisks indicate significant contrasts corresponding to 95% CIs and single asterisks correspond to 85% CIs. Only song features with a strong effect of treatment in the top model (ΔAIC_c = 0.00) are shown.