5.3. Additional Clues to SBH and Galaxy Formation and Further Challenges
I conclude this review with some general comments about the early stages of galaxy and black hole formation. The M - relation (and, even more so, the M - MDM relation) implies a causal connection between the evolution of black holes and their host galaxies. But what came first: the stars or the black holes? And was the M - relation imprinted during the early stages of galaxy formation? The answer to the latter question is generally assumed to be affirmative, but in fact we have no direct proof of it. The most distant galaxy in the M - plot (NGC 6251, Ferrarese & Ford 1999) is at ~ 100 Mpc. Studies of reverberation-mapped galaxies (Ferrarese et al. 2001, Pogge et al. 2002) have reached two times farther, and it might be technically possible to push the envelope up to z ~ 1. It seems unlikely that we will ever be able to build an M - relation at the redshift corresponding to the optically bright phase of the QSOs (z ~ 2 - 3), let alone at redshifts at which the first protogalactic fragments are believed to have formed, z > 5. Present day dwarf galaxies might very well be relics from such an era (Mateo 1998; Carraro et al. 2001); however, detecting SBHs in these systems requires a spatial resolution well beyond the capabilities of present instrumentation. In fact, the M - relation is defined primarily by bright giant ellipticals which are likely to have an extensive history of merging. In other words, we have no direct information about the "primordial" connection between supermassive black holes and their hosts: what we see is the result of gigayears of evolution.
A scenario in which galaxy formation precedes the formation of supermassive black holes seems to fit more naturally within the current paradigm of hierarchical structure formation (e.g. Miralda-Escude & Rees 1997). For instance, star formation can proceed in halos with virial temperature as low as 104 K, which can form at redshifts z > 10 (e.g. Ostriker & Gnedin 1996). Subsequent stellar evolution in these systems would produce enough energy through stellar winds or supernovae explosions to expel most of the remaining gas from the shallow potential wells (Couchman & Rees 1986; Dekel & Silk 1986), likely inhibiting the formation of supermassive black holes. Deeper potential wells, which are more conducive to SBH formation (e.g. Haehnelt, Natarajan & Rees 1998) would only form at later times. Studies of elemental abundances in high redshift (z > 3) QSOs support this view: most of the metal enrichment and star formation seem to have taken place at least 1 Gyr before the luminous phase of the QSO (Hamann & Ferlan 1999 and references therein; Dietrich et al. 1999).
Fig. 7 shows a comparison between the mass accretion rate onto optically luminous QSOs with M > 7 × 108 (corresponding to the magnitude limit of the SDSS QSO Survey for objects radiating at the Eddington limit), and the star formation rate from Steidel et al. 1999 (see also Abraham et al. 1999; Cowie et al. 1997). Similarities between the two curves, which have been noted many times (e.g. Boyle & Terlevich 1998) are diminished by these recent results, even after the QSO results are corrected for the possible contamination of obscured objects (Barger et al. 2001; Gilli et al. 2001; Salucci et al. 1999). If anything, Fig. 7 supports the conclusion that star formation was well underway by the time the QSOs started shining.
Figure 7. The mass accretion rate onto supermassive black holes with M > 7 × 108 M (thin solid line, units shown on the left axis), compared to the star formation rate from Steidel et al. 1999 (thick solid line, units on the right axis) and the epoch of formation of the DM halos which host such SBHs according to Fig. 5 and 7 (dashed line, from Gottlöber, Klypin & Kravtsov 2001). Eighty percent of the halos which (presumably) host QSO engines are capable of forming stars at redshifts z > 5. The dotted line shows the mass accretion rate onto high redshift QSOs corrected for the contribution of obscured objects, as in Gilli et al. 2001.
The connection between QSO activity and merging rate is also not readily apparent: observations show that the merging rate depends on redshift as (1 + z) with = 2 - 4 (Le Fevre et al. 2000; Burkey et al. 1994; Carlberg et al. 1994; Yee & Ellingson 1995; Abraham 1999). Even in the z < 2.3 range, where both curves decline, the number of mergers declines by at most a factor 30, while the comoving density of QSO declines by three orders of magnitude. Perhaps more telling is the comparison with the merging history of DM halos and the ensuing formation of galaxies. Fig. 7 also shows the distribution of formation redshifts for present day halos with virial velocities > 300 km s-1 taken from the N-body simulation of Gottlöber, Klypin & Kravtsov (2001). According to Figures 5 and 6, these are the halos associated with the black holes sampled by the SDSS, also shown in Fig. 7. Virtually all such halos are able to host a luminous galaxy (a condition reached when the halo progenitor first reaches a virial velocity > 50 km s-1) before a redshift ~ 2.5, i.e. before the optically bright phase of the QSOs.
In the midst of all this, one thing is certain: SHBs can no longer be studied in isolation. Understanding how they form, and how they shape their surroundings, requires a good deal more information from seemingly unrelated fields than could have been anticipated just a few years ago.