We now consider the biases which affect the various galaxy search techniques discussed in Sections 3 and 4. Each technique provides a shaded view of the deep extragalactic universe; we enumerate these differences in the same order that the various techniques were presented.
Radio galaxies are classical AGNs. This implies, according to the prevailing models of active galaxies, the presence of a supermassive black hole. Nearby radio galaxies are associated with large early-type galaxies, both cDs in the centers of clusters and giant elliptical galaxies. Conventional wisdom states that HzRGs are the precursors to these early-type systems; our bias may simply be that HzRGs sample the most massive systems. At the least, though, HzRGs necessitate the presence of a supermassive black hole in the early universe. The time scale of supermassive black hole formation is not well known (e.g., Loeb 1993). Are black holes the seeds of early galaxy formation, or are they the fruit of it? The radio source itself most likely affects the assembly of the galactic subunits, triggering star formation along the radio jets in at least some cases (e.g., Dey et al. 1997). If HzRGs are indeed found to be the precursors of the most massive early-type systems, radio galaxies at high redshift may prove to be most important for studies of the formation of galaxy clusters and large-scale structure.
Sub-mm and infrared-selected (LIRG) sources require much of their bolometric luminosity to be dust-reprocessed radiation emitted at long wavelength. In an extreme scenario, one could imagine a dust-enshrouded starburst in which no radiation shortward of a few microns escaped a restricted spatial region. Such a source at high redshift would escape detection in contemporary optical/near-infrared surveys; it would perhaps only be detected in the 850 µm sub-mm region. Have any such sources been identified already? Astrometric uncertainties and the large beam size of the SCUBA detector leave the question open. It is conceivable that one or more of the Hughes et al. (1998) sub-mm sources in the HDF lack optical/near-infrared identifications, even to the extremely deep limits of the HDF. Though studies of field sub-mm sources are in their nascent phases, it is likely these systems represent those galaxies in the distant universe undergoing the most vigorous bursts of star formation. We are far from the time when we can reliably relate these distant sources to the local census of galaxies, but conceiving of the sub-mm population as the more distant cousins of ultraluminous infrared galaxies (ULIRGs) seems plausible, if not likely (e.g., Lilly et al. 1999). At the least, sub-mm galaxies are important as they produce a significant fraction ( 15%) of the total bolometric output of the universe averaged over all wavelengths and epochs (Lilly et al. 1999) and largely (entirely?) account for the sub-mm background. Sub-mm galaxies may be the precursors of metal-rich spheroids.
X-ray selection of distant sources likely necessitates the presence of an AGN, implying that studies of the most distant X-ray sources may be most important for understanding the formation of supermassive black holes. Extended cluster X-ray sources derive from the thermal bremsstrahlung in hot, ionized intercluster media. Galaxy clusters sample the deepest potential wells in the universe on Mpc scales. Studies of the most distant X-ray clusters will provide valuable information on the formation of large-scale structure. In particular, several authors have noted the sharp dependence of the evolution of cluster abundances on basic cosmological parameters (e.g., Eke et al. 1996).
The first optical identification of a GRB occurred less than 2 years ago, and our understanding of these intriguing sources, though much improved, is still rather sparse. Are GRBs more prevalent in young stellar systems with many massive remnants? Are they correlated with supernovae explosions? It is premature to comment too deeply on what biases GRB-selected host galaxies may yield on our understanding of the distant universe. At the least, they are not all heavily reddened, which has implications for the prevalence of dust in the universe.
Lyman-break galaxies and photometrically selected young galaxies at high redshift are biased against dusty systems and older populations which would redden the observed spectral energy distributions and likely lead to redshift ambiguities. The main bias is that these techniques only search for galaxies containing a very young, OB star rich, stellar population which dominates the rest-frame continuum in the wavelength range 1216-1700 Å. A small contribution by an older population of stars will not be very noticeable at the observed, deep-UV wavelengths available for study at high redshift. Near-infrared studies with spectrographs on the new generation of 9 ± 1 m telescopes, and later from NGST, will provide critical information on the masses, kinematics, ages, and dust abundance of this population. Are we only skimming the envelope of the most dust-poor young systems? The relation of the Lyman-break galaxies to the local galaxy population remains uncertain.
Ly searches for distant galaxies have a clear, but uncertain, observational bias; strong Ly emission (W > 20 Å) tends to be biased against dusty galaxies, or at least against galaxies whose neutral gas has unfavorable geometry/kinematics. Spectroscopy of the Lyman-break population reveals that approximately half of the galaxies show Ly in absorption. In fact, deep narrowband imaging compared to broadband images can show these systems as negative holes (Steidel 2000)! However, most astronomers involved in Ly surveys are restricting their search to emission-line sources. The surface densities measured (see Table 5) are comparable to those measured for Lyman-break galaxies. The implication is that a large population of strong Ly emitters exist which have fainter continua than the limits of the photometric surveys. If the morphology of the Ly emission primarily depends upon the kinematics of the neutral gas as suggested by UV studies of local metal-poor galaxies (Section 5), then caution should be advised in deriving star formation rates and densities from the Ly-emitting galaxies.