Subspecies status and methods explain strength of response to local versus foreign song by oscine birds in meta-analysis

Abstract

To understand the implications of geographical variation in vocal culture in songbirds, researchers have often compared territorial responses to playback of local songs versus responses to playback of songs from ‘foreign’ conspecifics. This body of work has the potential to help us move towards a general understanding of factors driving divergence in signal recognition. We conducted a systematic review and meta-analysis of 57 playback studies to explain variation in strength of response to local versus foreign songs. Studies with incomplete reporting of results had elevated effects due to selective reporting. Studies that used small numbers of stimuli as exemplars (pseudoreplication) had more variable effects than studies without severe pseudoreplication. Whether or not we controlled for pseudoreplication, we found greater response to playback of local song than to foreign song. In investigating potential biological drivers of the variation in strength of experimental effects, we found that the difference in territorial response to local versus foreign song was stronger if the foreign song was recorded from another subspecies than if the foreign song was recorded from the same subspecies as the focal individuals. Indexes of risk of accidental response to heterospecific song did not coherently explain response to foreign conspecific songs, nor did factors expected to influence individual experience with foreign conspecific songs. Thus, although oscine songbirds clearly react more aggressively to local song than to foreign song and variation in the strength of this effect is influenced by methodological choices and subspecies status, considerable variation in the strength of response to local versus foreign song playback remains to be explained.

Subspecies status and methods explain strength of response to local versus foreign song by oscine birds in meta-analysis

“…We recognize that some species exhibit dialects that appear to be relatively uniform across 100's of km, and it may be biologically plausible to consider such distant songs to be 'local' in some species. However, we excluded all cases in which some or all 'local' songs were recorded >100 km from the playback site (or where we could not exclude this possibility) in the interests of adopting a consistent decision rule and avoiding judgments for which we often lacked sufficient empirical basis (for details of excluded studies, see Parker et al 2018). We also excluded studies in which the 'local' song recording was taken from an adjacent neighbour, as there is a well-demonstrated tendency for reduced aggression towards playback of known-neighbour song, presumably due to the 'dear-enemy' effect (Brooks and Falls 1975;Godard 1993;Hyman 2005;Wei et al 2011;Wilson and Vehrencamp 2001).…”

“…In these cases, we conducted Fisher's Exact tests in R (using the function fisher.test in R) to generate p-values for conversion to Fisher's Z. We provide details of the derivation of each effect size in our archived data (Parker et al 2018).…”

“…We analysed data with R version 3.4.2 (R_Core_Team 2017). All R code is publicly archived (Parker et al 2018). We conducted meta-analysis with the metafor (version 1.9.9) package (Viechtbauer 2010).…”

“…This classification was subjective in some cases, but it allowed us to test for, and then if needed control for, a relationship between effect size and distinctiveness of local from foreign song. We report the basis for each classification in our archived data (Parker et al 2018).…”

“…In such playback experiments, it seems that the most common result is for the territorial individual to respond more strongly to playback of the local song than to playback of the foreign song (e.g., Bradley et al 2013;Lemon 1967;McGregor 1983;Podos 2007), though this has never been quantified. However, these playback experiments (Appendix 1) reveal substantial variability in response to foreign and local song as measured in hundreds of statistical tests from dozens of species and populations studied around the world (Parker et al 2018). Thus, these experiments are an excellent resource for seeking to understand variability in response to geographically divergent signals.…”