1) Most of the GRB hosts discovered so far belong to the population of faint and blue star-forming galaxies at high redshift. They have low masses as suggested by their faint luminosity in the near-infrared. They are also sub-luminous sources at optical wavelengths. Most of them are characterized by intrinsic R-K colours even bluer than those displayed by the starburst galaxies observed in the nearby Universe.
2) The lack of GRB detection toward luminous starbursts and/or reddened sources such as those observed in the infrared and submillimeter deep surveys seems to indicate a possible bias of the currently-known GRB host sample against this type of objects. This could be explained by the fact that the selection of GRB host galaxies, so far, had to rely on the identification of optical GRB afterglows likely probing unobscured star-forming galaxies. The follow-up of optically-dark and radio-dark GRBs, and the use of their X-ray afterglows to obtain their localization with a sub-arcsecond error box will be likely necessary to probe dust-enshrouded star-forming galaxies in the early Universe with these particular phenomena. On the other hand, the hypothesis that such a bias of selection is purely intrinsic to the GRB host properties can not be rejected, assuming that GRBs preferentially occur within young and blue starbursts.
3) The observed GRB host galaxies seem to be statistically less luminous than the faint blue sources which mostly contributed to the B-band light emitted at high redshift. This could reveal an intrinsic bias of the GRB selection toward star-forming regions with very low luminosities, and might be explained taking account of particular environmental properties (e.g., metallicity) favouring the formation of gamma-ray burst events. In this context, this could also indicate that GRBs can not be used as unbiased probes of star formation. A larger statistics of the GRB host absolute Bmagnitudes and a detailed study of the chemical composition of the gas within GRB host galaxies will be however required to further confirm this result.
Note added in proof: A mistake in Table .1 of
Appendix A has been brought to
our attention regarding the R magnitude quoted for the host galaxy
of GRB 011121. The value in our table does not reflect the total
emission of this source, but its contribution to the flux measured in
a small aperture of 1
in diameter centered on the position of
the optical transient. The correct magnitude of the galaxy should be
revised to R=18.95 mag (dereddened from foreground Galactic
extinction, Greiner et al. 2003, A&A, submitted). We found that this
new value does not modify the general conclusions of our work.
Acknowledgements
We would like to specially thank the teams of the NTT and ESO-3.6 m at La Silla for their kind and efficient assistance during the observations in visitor mode. We have also appreciated the work of the ESO-Paranal and Gemini-North staff regarding the acquisition of the VLT and Gemini data in service mode. We acknowledge F. Mannucci for publicly providing his optical/NIR spectral templates via a user-friendly web interface, as well as A. Fernández-Soto for maintaining his HTML access to the HDF photometric redshift catalog. We are grateful to C. Lidman for his advice in the NIR data reduction techniques, as well as H. Aussel, L. Cowie, D. Elbaz, R. Chary, F. Combes, E. Feigelson and C. de Breuck for useful discussions related to this work. We finally thank our referee, S. A. Eales, for interesting comments and suggestions on this paper. This research project was partially supported by CONICET/Argentina and Fundacion Antorchas. DM is supported by FONDAP Center for Astrophysics 15010003.
Copyright ESO 2003