Primordial mass segregation in simulations of star formation?

We take the end result of smoothed particle hydrodynamics (SPH) simulations of star formation which include feedback from photoionization and stellar winds and evolve them for a further 10 Myr using N-body simulations. We compare the evolution of each simulation to a control run without feedback, and to a run with photoionization feedback only. In common with previous work, we find that the presence of feedback prevents the runaway growth of massive stars, and the resulting star-forming regions are less dense, and preserve their initial substructure for longer. The addition of stellar winds to the feedback produces only marginal differences compared to the simulations with just photoionization feedback. We search for mass segregation at different stages in the simulations; before feedback is switched on in the SPH runs, at the end of the SPH runs (before N-body integration) and during the N-body evolution. Whether a simulation is primordially mass segregated (i.e. before dynamical evolution) depends extensively on how mass segregation is defined, and different methods for measuring mass segregation give apparently contradictory results. Primordial mass segregation is also less common in the simulations when star formation occurs under the influence of feedback. Further dynamical mass segregation can also take place during the subsequent (gas-free) dynamical evolution. Taken together, our results suggest that extreme caution should be exercised when interpreting the spatial distribution of massive stars relative to low-mass stars in simulations.