Introduction

One of the major thrusts in modern cosmology is to trace the formation and evolution of galaxies from high redshift to the present. This effort involves mapping the evolution of star formation and metal abundances, as well as galaxy morphology, in order to gain both a direct picture of the galaxy evolution process, and to constrain models and simulations.

The star formation history is most easily studied in the optical/UV mainly because of the sensitivity of current telescopes (e.g. Keck and HST) in these bands. However, even in our own Galaxy, which has a modest star formation rate (SFR), stars mainly form in dusty regions. As a result, it is not clear whether optical/UV observations faithfully measure the SFR since dust extinguishes radiation in these bands, and re-emits it into the far-IR (FIR) and sub-mm bands.

The first indication that star formation in dust-enshrouded regions may be dominant on a cosmic scale came from observations of the sub-mm extragalactic background light (SEBL) with the COBE satellite (Puget et al. 1996). However, since the SEBL represents an integrated history of star formation, these observations do not provide information on the redshift evolution of obscured star formation. This crucial information only became accessible with the advent of SCUBA on JCMT, which allowed a large fraction of the SEBL to be resolved into individual sources. Targeted and blank-field SCUBA surveys have uncovered a population of very luminous, high redshift galaxies, which host a significant fraction of the cosmic star formation (e.g. Smail et al. 1997; Barger et al. 1998; Blain et al. 2002).

Still, despite the pioneering work of SCUBA (and other sub-mm instruments), an exact determination of the evolution of obscured star formation has been confounded by the unknown redshift distribution of the sub-mm population. This is due to the lack of precise localizations of sub-mm sources (typical beam-sizes are ~10 arcsec), and the lack of sufficiently bright optical counterparts in the majority of cases. As a result, there is still an on-going debate regarding the fraction and redshift evolution of obscured star formation, and whether it is really missing from optical/UV surveys (Madau et al. 1996; Blain et al. 1999; Adelberger & Steidel 2000). SIRTF and ALMA will contribute significantly (and perhaps decisively) to these questions.

The Host Galaxies of Gamma-Ray Bursts

An alternative, and perhaps more timely, way of breaking the deadlock is to study a different population of high redshift galaxies, which is potentially devoid of some of the biases inherent in the current samples. The host galaxies of gamma-ray bursts (GRBs) provide such a sample for the following reasons:

1. Redshifts and Luminosities: Thanks to the bright optical afterglows of GRBs, the redshifts of the host galaxies can be determined regardless of their brightness, via absorption spectroscopy; as a result, even hosts with ~29 mag have known redshifts.

2. Immunity to Dust: The immense dust-penetrating power of the gamma-ray emission from GRBs results in a sample that is independent of the dust properties of the individual galaxies. Thus, the sample is potentially representative of the general population of star-forming galaxies.

3. Very High Redshift: GRBs are so bright that they are detectable to redshifts >20 (should they exist; Lamb & Reichart 2000). At present, the redshift record-holder is GRB000131 at z=4.5 (Andersen et al. 2000).

The first point in particular offers a significant advantage over current sub-mm surveys since accurate redshifts and positions allow us to trace the evolution of star formation in detail and perform multi-wavelength studies of these galaxies. In addition, the immunity of GRBs to dust results in a sample of galaxies which possibly bridges the extremely dusty SCUBA-selected galaxies, and the relatively dust-free optical/UV-selected galaxies.

At present, the main limitation of the GRB sample is that it is small, numbering about 30 sources (the sample is currently growing at a rate of one per month, but is expected to increase to one per few days after the 2003 launch of SWIFT). However, the severity of this problem diminishes when one realizes that the number of securely-identified sub-mm galaxies is several times smaller (e.g. Frayer et al. 1998).

We note in passing that the bright optical afterglows of GRBs are also powerful tools for studying the evolution of metal abundance in the disks of high-redshift galaxies, as well as the intergalactic medium. In particular, unlike quasars, which shine continuously and ionize their surroundings, GRB afterglows illuminate the line-of-sight without modifying the material on galactic scales, thus providing unique information on the chemical composition at all redshifts.

A SCUBA Survey of GRB Host Galaxies: Results to Date

Initial work with SCUBA concentrated on target-of-opportunity observations of the afterglows themselves, often in less than ideal observing conditions. These observations, summarized by Smith et al. (1999 & 2001), provided only loose constraints (typical 1 sigma=1-3 mJy) on the combined emission from the afterglow and host. Recently, this work has been superseded by the detections of the host galaxies of GRB980703 (VLA) and GRB010222 (SCUBA) by us and colleagues. In the case of GRB980703 we detected a non-fading radio source at the position of the burst (Berger et al. 2001a), while SCUBA observations of GRB010222 revealed a constant sub-mm source (Frail et al. 2002).

Motivated by these detections, and the unique potential of GRB host studies, we have begun an intensive program of host galaxy observations at JCMT, with a synergistic program at the Very Large Array. Over the past several months we observed thirteen host galaxies in the standard photometry mode and we are happy to report another detection, the host galaxy of GRB000418 (see also Berger et al. 2001b). Three additional hosts have fluxes ~ 2 mJy. Including the previous detections, this means that approximately 20-40% of GRB hosts have bolometric luminosities and star formation rates, which classify them as Ultra Luminous Infra-Red Galaxies (ULIRGs). This fraction matches the prediction of Ramirez-Ruiz et al. (2002), which relies on the assumption that GRBs trace massive stars.

In Figure 1 we plot the spectral energy distributions of the three securely-detected GRB hosts, in comparison to the local ULIRG Arp220, and the z=1.44 extremely red object HR10. The GRB hosts are more luminous than Arp220, and have a comparable luminosity to HR10. Optical imaging and spectroscopy of these host galaxies do not reveal such vigorous star formation activity, implying significant dust obscuration.

However, despite the resemblance to Arp220 and HR10 in the radio/sub-mm regime, the optical properties of the GRB hosts reveal that they represent a somewhat different population than the SCUBA-selected galaxies. In Figure 2 we plot a histogram of R-K$ color for SCUBA-selected galaxies (Chapman et al. in prep.), radio-selected galaxies (Haarsma et al. 2000), and several GRB hosts. All of the sub-mm selected galaxies have R-K > 3, and 2/3 have R-K > 4. This is expected since the high dust content in these galaxies tends to redden the optical colors. On the other hand, close to 80% of the GRB hosts have R-K < 3. Thus, the host galaxies that have been detected with SCUBA so far represent a population that would have been missed in the traditional SCUBA surveys, and at the same time would not be classified as ULIRGs based on optical data.

At present, it is difficult to interpret these results, but it is possible that GRB hosts represent younger starbursts, while the typical SCUBA-selected galaxies host more evolved population of stars. Regardless of the exact explanation, it appears that current sub-mm surveys miss a certain fraction of the cosmic star formation, which can be possibly recovered by observations of GRB hosts.

Future Work

The field of sub-mm and radio observations of GRB host galaxies, as well as other high-redshift samples, is still in its infancy. Over the past several years great strides have been made in our understanding of the cosmic star formation history, especially thanks to SCUBA, and the results of our initial SCUBA observations of GRB hosts indicate that this new sample brings a fresh and unique perspective into the field. In the near future, our continued program will rely on the growing sample of GRB host galaxies, and current sub-mm and radio facilities. On a longer timescale, we anticipate a revolution in the capabilities of sub-mm and radio instrumentation, and a substantial growth in the number of GRB host galaxies with accurate localizations and redshifts. The advent of instruments and observatories such as SCUBA-2, the Expanded VLA, the Square Kilometer Array, SIRTF, and ALMA, will allow us to study the properties of these hosts with ever-increasing sensitivity and resolution.

One of the issues that may be addressed by these improved observations is whether the subset of optically-dark GRBs lack an optical afterglow because they are located within the dense environs of giant molecular clouds. If so, the fraction of GRBs with no optical afterglow will provide an independent estimate of the fraction of obscured star formation at cosmic scales.

Thus, as more GRB host galaxies are detected and studied in detail in the sub-mm and radio, we will be able to address a large number of issues pertaining not only to the nature of GRBs and their hosts, but also to the characteristics of galaxies at high redshifts.