Assessing the effect of pseudoreplication when individual identities are unknown: a reply to Gratton & Mundry.

Gratton and Mundry (2019, Animal Behaviour , 000, 000-000) statistically evaluate an approach I had suggested in an earlier paper that relied merely on verbal arguments (Garamszegi, 2016, Animal Behaviour , 120, 223-234). In particular, their simulation study challenges a methodology I proposed to account for pseudoreplication when the identity of individuals is unknown in conservation studies. Gratton and Mundry (2019) conclude that the random assignment of observations to individuals along the lines of my original idea produces a considerable type I error rate. Here, I suggest that although this evaluation is straightforward, it is still incomplete because it neglects two aspects of my original suggestion. To fill this gap, I have performed additional simulations that show that my proposed approach provides biased estimates even when the random assignment of identity tags is combined with multimodel inference and/or is applied to cases when within-individual correlations exist. These additional results support the conclusions of Gratton and Mundry (2019) with the caveat that the untested aspects of Garamszegi (2016) should not be dismissed. • Gratton & Mundry statistically examined an approach I suggested in an earlier paper. • By not exactly challenging the proposed methodology, the evaluation was incomplete. • I performed new simulations to deal with issues that Gratton & Mundry disregarded. • These additional results supported the conclusions of Gratton & Mundry. [ABSTRACT FROM AUTHOR]