Can ’ t get there from here : inferring kinship from pairwise genetic relatedness

K inship among animals has important implications for their behaviour, as it determines their opportunities for nepotism and their risk of inbreeding (e.g. Hamilton 1963; Wade 1979; Trontti et al. 2005). The most accurate way to assess kinship is through complete knowledge of pedigrees (we use the term kinship to refer to categorical levels of coancestry, as would be derived from pedigree construction). However, precisely identifying kinship among all individuals in a population is rarely, if ever, possible from observations alone, and researchers often either use genetic data alone (see reviews in Blouin 2003; Garant & Kruuk 2005) or use genetic data to complement their behavioural data (e.g. Taylor et al. 1997; Kappeler et al. 2002; Kruuk et al. 2002; Carpenter et al. 2005; Cutrera et al. 2005; Randall et al. 2007). Several methods are available by which genetic data can be used to estimate relatedness, as a continuous estimate of allele sharing due to coancestry. However, each of these methods (e.g. Queller & Goodnight 1989; Lynch & Ritland 1999; Wang 2002) has inherent limitations (see discussions in van de Casteele et al. 2001; Thomas et al. 2002; Wang 2002; Blouin 2003; Thomas 2005; Fernández & Toro 2006). In particular, the accuracy and precision of these genetic estimators of relatedness are known to be influenced by characteristics of themarkers and the sample