Modelling of partial migration in birds has progressed from simple graphical representations to sophisticated analyses that use evolutionary invasion analysis to determine how the success of the two strategies (stay year round on the breeding grounds, or migrate) can become frequency dependent. Here I build two models to relax two assumptions commonly made in models and often violated in nature: that individuals do not vary in any trait other than their migratory propensity, and that the prior residence effect (which grants priority access of good habitats to non-migrants) operates at maximum strength. The same framework can incorporate and merge aspects of various hypotheses proposed to explain partial migration (dominance, body size, arrival timing, and limited foraging opportunities), and shows that either small (subdominant) or large (dominant) individuals may emerge as the more likely migrants; the latter case occurs when it is easy for socially dominant migrants to win back prime breeding locations upon their arrival. The dynamics of territory acquisition is shown to be an important and understudied topic, as variations in the relative importance of prior residency versus resource holding power can shift a population from complete migration to complete year-round residency. These models also highlight exceptions to a tacit assumption in discussions of evolution of migration under climate change, which is that populations can decline if genetic adaptation or phenotypic plasticity do not occur fast enough. Competition can also yield the opposite pattern where adaptation itself leads to a population decline.