Annu. Rev. Astron. Astrophys. 1997. 35: 389-443 Copyright © 1997 by Annual Reviews. All rights reserved

4.4. A Distinct Population of Rapidly Fading Objects?

The difficulties encountered with the traditional PLE models (Section 4.3) can be overcome by invoking a separate population of blue objects that undergo strong luminosity evolution (Phillipps & Driver 1995). Unlike the traditional PLE models, this hypothesis invokes a delayed star formation in a subset of the galaxy population followed by a remarkably rapid decline in activity thereafter. The present-day remnant of this population would presumably occupy the faint end of the local LF. Although physical models have been proposed to support this picture (Cowie et al 1991, Babul & Rees 1992, Babul & Ferguson 1996), the original hypothesis was proposed only to fit the data. Nevertheless, there is a strong theoretical motivation because ab initio models tend to overpopulate the faint end of the local LF (Cole et al 1994a). Evidence in support of this empirical picture includes the very different evolutionary behavior of field LFs characterized by color and spectral features (Lilly et al 1995, Heyl et al 1997) and the rapid increase in the number of irregulars in the MDS data (Glazebrook et al 1995b, Driver et al 1995b). A high fraction of the HST irregulars are also blue [O II]-strong sources (LeFevre et al 1996a). However, the key to testing this model further would be to verify that a distinct population exists and to demonstrate that it fades according to the rate determined from the LF studies.

Some progress might be made by studying the supposedly rapidly evolving component in more detail, as well as by finding local examples. A typical, rapidly evolving blue galaxy at z > 0.3 has M_B -19.5 + 5log h, B-I 1-1.5, a rest-frame equivalent width of [O II] of 20-40 Å, and an irregular appearance with knots and occasionally a compact core. Although a few giant irregulars are found locally (Gallagher et al 1989), the bulk of the blue sources with comparable [O II] strengths at low redshift are gas-rich metal-poor dwarfs believed to have suffered sporadic bursts of star formation (Telles et al 1997, Gallego et al 1995, Ellis et al 1996a). Leitherer (1996) provides a comprehensive review of the physical processes responsible for sudden star formation in galaxies. There are two broad categories: (a) H II region-like spectra superimposed on that of an older population (Telles et al 1997) and (b) nuclear starbursts. Both may be induced by interactions.

Surprisingly little is known about the astrophysical properties of the faint blue galaxies, which at first sight seems remarkable given spectra are available for several thousand such galaxies! With the exception of Broadhurst et al' (1988) original data taken at a spectral resolution of 4 Å (which they interpreted as supporting signs of short-term star formation in the distant population), both the signal/noise and resolution of more recent, fainter data are optimized for little more than measuring redshifts; a resolution of 20-40 Å is more typical. Heyl et al (1997) Hammer et al (1997) discuss the spectral properties of co-added spectral datasets in the Autofib/LDSS and CFRS surveys, respectively, but disagree on the degree to which short-term star formation may be occurring in their samples. Higher quality spectra are required to make progress.

A tentative connection between the faint blue population and active galaxies is reported by many groups. On the basis of emission line ratios, Tresse et al (1996) claimed between 8 and 17% of the z < 0.3 CFRS sample could be Seyfert 2 or LINERs, depending on uncertain corrections for underlying stellar absorption. This is about four times higher than the local rate and would suggest a dramatic evolution in the proportion of active sources over quite recent epochs. Likewise, Treyer & Lahav (1996) claimed B < 23 galaxies with z < 0.3 could contribute 22% of the 0.5- to 2-keV X-ray background. At fainter limits, Windhorst et al (1995) found that the micro-Jy radio sources overlap significantly with the faint blue population. Windhorst et al argue that the nonthermal activity is more likely a by-product of interactions and mergers rather than the direct output of classical active galactic nuclei.

How might fading in a given population be demonstrated convincingly? We return to the question of selecting some physical parameter that does not change in the process. Koo et al (1995), Guzman et al (1996) have studied a particular subclass of compact narrow emission line galaxies (CNELG) selected on the basis of very small effective radii (1 kpc) and have suggested that these may be the ancestors of local dwarf spheroidals. Although only a minority of the faint blue population lies in this category, this suggestion is important because, if correct, it would demonstrate for a subset of the faint population that significant fading has occurred. Spectroscopic velocity dispersions from the Keck HIRES are typically 30-50 km s^-1, which indicates low mass/light ratios and bursting star-formation histories. Substantial fading (4-7 mag in M_B in 3 Gyr) would be needed to reduce such systems to the present luminosities of dwarf spheroidals.

The CNELG study illustrates an important way forward because, if local dwarf galaxies are faded versions of their high z cousins, the surface brightnesses within some fixed metric scale must also have declined by the same amount as indicated in the LF studies. The crucial test for the fading hypothesis is thus to examine the surface brightness distributions of various subsets of the field population, taking due care to allow for possible selection biases. The availability of HST images for subsets of the large redshift surveys (Ellis 1995, Schade et al 1995, LeFevre et al 1996a) makes this an important direction for future study.