2015MNRAS.447..598S

Based on their relatively isolated environments, we argue that luminous blue variables (LBVs) must be primarily the product of binary evolution, challenging the traditional single-star view wherein LBVs mark a brief transition between massive O-type stars and Wolf-Rayet (WR) stars. If the latter were true, then LBVs should be concentrated in young massive clusters like early O-type stars. This is decidedly not the case. Examining locations of LBVs in our Galaxy and the Magellanic Clouds reveals that, with only a few exceptions, LBVs systematically avoid clusters of O-type stars. In the Large Magellanic Cloud, LBVs are statistically much more isolated than O-type stars, and (perhaps most surprisingly) even more isolated than WR stars. This makes it impossible for LBVs to be single `massive stars in transition' to WR stars. Instead, we propose that massive stars and supernova (SN) subtypes are dominated by bifurcated evolutionary paths in interacting binaries, wherein most WR stars and Type Ibc supernovae (SNe Ibc) correspond to the mass donors, while LBVs (and their lower mass analogues like B[e] supergiants, which are even more isolated) are the mass gainers. In this view, LBVs are evolved massive blue stragglers. Through binary mass transfer, rejuvinated mass gainers get enriched, spun up, and sometimes kicked far from their clustered birth sites by their companion's SN. This scenario agrees better with LBVs exploding as SNe IIn in isolation, and it predicts that many massive runaway stars may be rapid rotators. Mergers or blue Thorne-Zytkow-like objects might also give rise to LBVs, but these scenarios may have a harder time explaining why LBVs avoid clusters.