Associations between Manakin Display Complexity and Both Body and Brain Size Challenge Assumptions of Allometric Correction: A Response to Gutierrez-Ibanez et al. (2016).

plexity. This conclusion is conceptually similar to our own, and it little alters the broader impacts of our published findings; still, the general relevance of the issues warrants a response. In their reanalysis, Gutierrez-Ibanez et al. [2016] perform a ‘residual analysis’, which regresses body weight on brain weight with phylogenetic generalized least squares (PGLS) to obtain size-corrected relative brain size , a method that is widely criticized [Newman, 1993; Smith, 1999; Darlington and Smulders, 2001; GarciaBerthou, 2001; McCoy et al., 2006; Hoverman and Relyea, 2008; Revell, 2009; Berner, 2011]. An implicit assumption of relative brain size analysis is that selection is neutral regarding body size and, accordingly, that the portion of brain tissue needed to control bodily functions can be statistically removed [Deaner et al., 2007] – an improbable assumption for many sexually selected traits in sexually dimorphic species [García-Peña et al., 2013] and a particularly dubious assumption when the trait in question is a physically demanding display. Avoiding these pitfalls, we corrected for body size using general linear modeling (GLM) to obtain estimated marginal means Essentially, all models are wrong, but some are useful. Box and Draper, 1987