2016MNRAS.461.3353S

A debate has arisen concerning the fundamental nature of luminous blue variables (LBVs) and their role in stellar evolution. While Smith & Tombleson proposed that their isolated environments indicate that LBVs must be largely the product of binary evolution, Humphreys et al. have recently expressed the view that the traditional single-star view still holds if one appropriately selects a subsample of LBVs. This paper finds the claim of Humphreys et al. to be quantitatively unjustified. A statistical test of 'candidate' as opposed to 'confirmed' LBVs shows no significant difference (<1σ) between their environments. Even if the sample is further subdivided as proposed, the three most luminous LBVs are spatially dispersed similar to late O-type dwarfs, which have much longer median lifetimes than expected for classical LBVs. The lower luminosity LBVs have a distribution associated with red supergiants (RSGs), but these RSGs are dominated by stars of 10-15 M_☉ initial mass, with much longer lifetimes than expected for those lower luminosity LBVs. If one's view is restricted to the highest luminosity LBVs, then the appropriate comparison is with early O-type stars that are their presumed progenitors; when this is done, it is clear that even the high-luminosity LBVs are more dispersed than expected. Humphreys et al. also suggest that velocities of LBVs support the single-star view, being inconsistent with runaways. A quantitative analysis of the radial velocity distribution of LBVS in M31 and M33 contradicts this; modest runway speeds expected from mass gainers in binary evolution are consistent with the observed velocities, although the data lack the precision to discriminate.