It is a widely observed fact that starbursts are concentrated in nuclei, and in particular the strongest ones (ULIRGs). But there can be exceptions, such as:
M82 is a good opportunity to study the evolution of starburst location: in the M82B fossil region (see figure 1), stars formed 100 Myr ago, with a comparable amplitude than the present starburst in the center. De Grijs et al. (2001) find there an important (113) number of evolved super star clusters (SSC). Their detailed age study conclude that the starburst begun 2 Gyr ago, with a peak 600 Myr ago, and stopped about 30Myr from now. This episode could coincide to a previous passage/interaction of the companion M81. The evolution of the SSCs is compatible with them being progenitors of globular clusters.
Figure 1. HST-WFPC2 image (V-band) of M82 (in the center), and PC-field images in B, V, I and NICMOS in J & H, from de Grijs et al. (2001). The PC and NICMOS images are centered on the fossil starburst (region B), while regions A and C indicate the present on-going nuclear starburst. In M82B, a large system of evolved super star clusters has been found.
This evolution of the starburst location could correspond to ring concentration and ring evolution. Indeed, in barred galaxies star formation is frequently in nuclear rings (cf Buta et al 2000, NGC 1326; Maoz et al 2000, NGC 1512, NGC 5248), and in particular bright knots in the rings.
Sometimes star formation can occur even farther from the center: in tidal dwarfs (e.g. Duc et al. 2000), shells, garlands, large HII complexes in the outer regions, as in M101 or NGC 628 (Lelièvre & Roy 2000). Or the nucleus does not concentrate the star formation activity, which is more randomly distributed, as in dwarf irregulars. A recently studied example is NGC 4214 (Beck et al. 2000; MacKenty et al. 2000), where interferometric CO observations (Walter et al 2001) reveal that the star formation is not always coinciding with the gaseous concentrations. If one CO complex is indeed the site of a starburst, a comparable one, at the same distance from the center, is completely quiecent.
An interesting question is to estimate the relative importance of starburst, and more quiescent or "steady-state" star formation in the global rate of star formation of the Universe. If a starburst is defined as having a rate larger than 50 M/yr, an estimation from NICMOS images in the Hubble Deep Field conclude that both processus appear similar in importance (Thompson 2000).