Uneven sampling and the analysis of vocal performance constraints

Area of Science:

• Bioacoustics
• Evolutionary Biology
• Animal Communication

Background:

• Trilled vocalizations illustrate how performance constraints influence mating displays.
• Vocal tract mechanics create a trade-off between repetition rate and frequency bandwidth in trills.
• Sexual selection may favor high-performance trills that maximize both traits.

Purpose of the Study:

• To identify a sampling limitation that confounds the detection of performance trade-offs in trilled vocalizations.
• To reassess existing data sets and quantify the impact of sampling limitations.
• To explore alternative methods, such as quantile regression, for accurately characterizing trade-offs.

Main Methods:

• Reanalysis of 70 data sets from 26 published studies on trilled vocalizations.
• Comparison of traditional upper-bound regression with methods controlling for sampling limitations.
• Application of quantile regression to address sampling limitations.

Main Results:

• Sampling limitations were found to affect 63 out of 70 data sets.
• Traditional methods detected trade-offs in 33 data sets, but only 15 when sampling limitations were controlled.
• Over half of reported performance trade-offs may be confounded by sampling issues.

Conclusions:

• Sampling limitations significantly confound the detection and description of performance trade-offs in trilled vocalizations.
• Traditional regression methods are susceptible to these limitations.
• Quantile regression offers a promising alternative for accurate characterization of evolutionary trade-offs.