It is convenient to divide models for the formation of GC systems into three broad categories: pre-, proto- and post-galactic. While these divisions are useful for heuristic purposes, they should not be taken too literally since the classification is, in some cases, rather subjective. It is also worth noting that some models that have historically been considered quite distinct (e.g., Ashman & Zepf 1992 and Côté et al. 1998) actually rely on similar mechanisms - mergers and accretions - to explain the formation of GC systems. These models differ mainly the assumed nature of the merger/accretion process: most notably in the assumed mass spectrum of the progenitors, the epoch of galaxy assembly, and the amount of dissipation and star/GC formation (if any) induced by the merger/accretion process.
In this picture, the formation of GCs predates that of their host galaxy. First suggested by Dicke & Peebles (1968), who noted that the Jeans mass at recombination was similar to the mean mass of Galactic GCs, this model was later revised by Peebles (1984) to explain GC formation in a cold dark matter (CDM) dominated universe. We now recognize the GC mass distribution to have a roughly double power-law form with a break at M ~ 105 M, weakening the original Jeans mass argument.
Recently, Cen (2001) has proposed a variant of this model in which the MP GCs surrounding galaxies form during reionization, when the external radiation field produces inward shocks in low-mass, low-spin dark matter halos. Such shocks, he argues, will compress the baryons and lead to the formation of GCs. While this model gives roughly the correct mass spectrum for GCs more massive than about 105 M, the mechanism itself is quite unlike that observed in star- and cluster-forming molecular clouds in the local universe (and leaves the formation of the MR GCs unexplained). Moreover, it seems inevitable that this model should predict many intergalactic GCs in the local universe. Although the properties of some GCs associated with a few galaxies like M87 are consistent with an intergalactic origin (see Côté et al. 2001), there is still no case in which even a single GC can be said with certainty to be truly intergalactic in nature.