APPENDIX 2. NONBREEDING SEASONS AND THE EVOLUTION OF MIGRATION Seasonal migration has evolved many times within diverse taxa (e.g., Pascual et al. 2001, Alerstam et al. 2003) and is expressed through a wide range of movement behaviours (Quinn and Myers 2004, Dingle 2006, Bolger et al. 2008). Migration will be favoured where the benefit gained from moving between habitats outweighs the fitness risks and energetic costs of the journey; seasonal and spatial variations in environmental conditions are therefore primary forces in its evolution (Fretwell 1972, Alerstam and Enckell 1979). Some theory suggests that migratory behaviour evolved primarily in response to the advantages of breeding in regions of seasonally high resource availability and safety for juveniles (Corkeron and Connor 1999, Friedland et al. 2005), especially where resources in the nonbreeding area were limited (Fretwell 1972, Cox 1968, 1985). Indeed, there is evidence today for competition during the nonbreeding season among many seasonal migrants (e.g., Pienkowski and Evans 1984, Perez-Tris and Telleria 2002, Ruggerone et al. 2003). Alternatively, the principal selection pressure might have been the survival advantage gained by individuals that migrated to less harsh, safer or more resource-rich environments after breeding (Lack 1968, Ketterson and Nolan 1982, Hebblewhite and Merrill 2007). Both scenarios involve the balancing of benefits and costs, in terms of energy acquisition and mortality risks, between breeding and nonbreeding seasons. There is evidence that migratory behaviour continues to evolve in response to environmental changes or variation in nonbreeding habitats (Berthold et al. 1992). For instance, selection favours individual birds whose spring migration synchronizes breeding with peak resource availability (Kokko 1999, Drent et al. 2003, Bêty et al. 2004); heritability of migratory traits could thus allow migrants to adapt to changing climatic conditions experienced prior to breeding (Berthold and Pulido 1994, Both et al. 2005). Natural selection has also favored local adaptation to nonbreeding areas in migrating fish (e.g., Fraser and Bernatchez 2005), which could ultimately lead to speciation if it favored reproductive isolation between populations (Wood and Foote 1996). Nevertheless, the ability of migrants to adapt to rapid environmental changes may be also shaped by seasonal features such as migration distance, severity of nonbreeding habitat loss, correlations between seasonal environmental changes, or the strength of cross-seasonal migratory connectivity (Dolman and Sutherland 1994, Sutherland 1998, Webster et al. 2002, Lemoine and Bohning-Gaese 2003, Both et al. 2005).