• Fstatistics;
  • genetic population structure;
  • species-level;
  • systematics isolation by distance;
  • enzyme electrophoresis

We used hierarchical and pairwise F-statistics to describe genetic differentiation and infer gene flow (M) on local and regional scales within and among parapatric European butterfly taxa in the Pieris napi (L.) group. Within-population allozyme variability is consistently high, and local effective population sizes are inferred to be in the thousands of individuals. The pairwise analysis yields an average neighbourhood area of radius 3.5 km. Among populations within most regions, differentiation is low and M > 2 effective individuals population-1generation-1. Pairwise comparisons within the brilannica group show a disjunction indicating that it is out of equilibrium, perhaps as a result of secondary contact between highland and lowland groups. Comparison between meridionalis groups on mainland Italy and Corsica yields M > 12; this is surely loo high and lack of equilibrium resulting from initial colonization is suspected. The hierarchical analysis indicates that 23 ≤0020M≤ 88 among the taxa napi, bryoniae and meridionalis that meet in hybrid zones; no effective gene flow barrier exists among them. This high estimate could also result from recent primary contact, but such a genetic barrier should produce the ‘edge effects' seen in population genetic simulations, and no evidence of this was found among geographically close samples of napi and bryoniae populations from Switzerland. Studies of gene flow among geographic regions are greatly limited by the equilibrium assumption, though studies of local differentiation are much less so. Population studies of gene flow on local scales at regional boundaries provide limited means of testing the equilibrium assumption, and both regional and local analyses provide testable predictions about local population structure. When the equilibrium assumption is not upheld, local patterns at regional boundaries can provide historical information about primary vs. secondary contact.