On between-individual and residual (co)variances in the study of animal personality: are you willing to take the “individual gambit”?

Abstract

A recent discussion in this journal (Dingemanse et al. Behav Ecol Sociobiol 66: 1543–1548, 2012; Garamszegi and Herczeg Behav Ecol Socibiol 66:1651–1658, 2012) deals with a core issue in animal personality research: Can animal personality research quantify correlated behaviors on the between-individual level, or is this too demanding in terms of design and analysis of the data, and should behavioral ecologists therefore take the “individual gambit” and work on the phenotypic level only. Taking this gambit implies accepting that the between-individual correlation in behavioral traits (which is the correlation of interest) may be masked by a residual correlation of different magnitude or sign. Understanding (co)variances on different levels is the main thrust of quantitative genetics, and animal personality research can make good use of the plethora of ideas and analytical approaches developed in this field. I, here, outline reasons why the “individual gambit” may or may not work out and its relationship to the quantitative genetic “phenotypic gambit”. I especially emphasize the meaning of residuals and phenotypic plasticity which has not been fully appreciated in the debate thus far. I conclude that instead of a priori assuming that between-individual correlations are captured sufficiently well by the phenotypic correlation, animal personality researchers should set up more ambitious data collection and analysis designs to critically test this conjectured equality.