We present the results of a deep, wide-area, optical and near-IR survey of massive high-redshift galaxies. The Prime Focus Camera (Suprime-Cam) on the Subaru telescope was used to obtain BRIz ′ imaging over 2 × 940 arcmin 2 fields, while JK s imaging was provided by the SOFI camera at the New Technology Telescope (NTT) for a subset of the area, partly from the ESO Imaging Survey (EIS). In this paper, we report on the properties of K-band-selected galaxies, identified from a total area of ∼ 920 arcmin 2 to K Vega = 19, of which 320 arcmin 2 are complete to K Vega = 20. The BzK selection technique was used to assemble complete samples of about 500 candidate massive star-forming galaxies (sBzKs) and about 160 candidate massive passively evolving galaxies (pBzKs) at 1.4 < ∼ z < ∼ 2.5; and the (R − K) Vega > 5 color criterion was used to assemble a sample of about 850 extremely red objects (EROs). We accurately measure surface densities of 1.20 ± 0.05 arcmin −2 and 0.38 ± 0.03 arcmin −2 for the sBzKs and the pBzKs, respectively. Both sBzKs and pBzKs are strongly clustered, at a level at least comparable to that of EROs, with pBzKs appearing more clustered than sBzKs. We estimate the reddening, star formation rates (SFRs) and stellar masses (M * ) for the ensemble of sBzKs, confirming that to K Vega ∼ 20 typical (median) values are M * ∼ 10 11 M ⊙ , SFR ∼ 190 M ⊙ yr −1 , and E(B − V ) ∼ 0.44. A correlation is detected such that the most massive galaxies at z ∼ 2 are also the most actively star-forming, an effect that can be seen as a manifestation of downsizing at early epochs. The space density of massive pBzKs at z ∼ 1.4 − 2 that we derive is 20%±7% that of similarly massive early-type galaxies at z ∼ 0. Adding this space density to that of our massive star forming class, sBzKs, in the same redshift range produces a closer comparison with the local early-type galaxy population, naturally implying that we are detecting star formation in a sizable fraction of massive galaxies at z > 1.4, which has been quenched by the present day. Follow-up optical and near infrared spectroscopy is in progress at the ESO Very Large Telescope (VLT) and at the Subaru telescope, in order to elucidate more thoroughly the formation and evolution of massive galaxies. 3 The relevant conversions between Vega and AB magnitudes for this paper are B AB = B Vega − 0.08, R AB = R Vega + 0.22, z AB = z Vega + 0.53, and K AB = K Vega + 1.87.

With the HDS (High Dispersion Spectrograph) on the Subaru telescope, we obtained high resolution optical region spectra of three red giant stars (cos 4, cos 82, and cos 347) in the Ursa Minor dwarf spheriodal galaxy. Chemical abundances in these stars have been analysed for 26 elements including α-, iron-peak, and neutron capture elements. All three stars show low abundances of α-elements (Mg, Si, and Ca) and two stars (cos 82 and cos 347) show high abundance of Mn compared to Galactic halo stars of similar metallicity. One star (cos 4) has been confirmed to be very metal deficient ([Fe/H]= −2.7) and found to show anomalously low abundances of Mn, Cu, and Ba. In another star cos 82 ([Fe/H]= −1.5), we have found large excess of heavy neutron-capture elements with the general abundance pattern similar to the scaled solar system r-process abundance curve. These observational results are rather puzzling: low abundances of α -elements and high abundance of Mn seem to sugggest a significant contribution of SNe Ia at low metallicity, while there is no hint of s-process (i.e., AGB stars) contribution even at [Fe/H]= −1.5, suggesting a peculiar nucleosynthetic history of the UMi dSph galaxy.

Abstract. Star formation and chemical enrichment histories of the dwarf spheroidal galaxies (dSphs) Draco, Sextans, and Ursa Minor are investigated by means of chemical evolution models and a simulation code for colour-magnitude diagrams (CMDs). The CMD simulation code is designed to fully consider effects of the chemical evolution on stellar evolution and photometric properties. For this aim, star formation and chemical enrichment histories are calculated consistently in the code. Comparisons between the chemical evolution models and the observed abundance patterns reveal that the star formation rates were very low (1-5% of that of the solar neighbourhood disc) and that the initial star formation continued for a long duration ( >3.9-6.5 Gyr) in these dSphs. This star formation history can reproduce morphologies of the observed CMDs, such as narrow red giant branches and red horizontal branches and succeeds in solving the second parameter problem of the dSph Draco. Hence, both of the abundance patterns and the morphologies of the CMDs can be explained by the star formation histories characterised by the low star formation rate and the long duration of the star formation period. Because of the low star formation rates, plenty of gas remains at the final epoch of star formation. We suggest that gas stripping by the Galaxy results in termination of star formation in the dSphs.

We have studied Leo A -the isolated and extremely gas rich dwarf irregular galaxy of very low stellar mass and metallicity. Ages of the stellar populations in Leo A are ranging from ∼10 Myr to ∼10 Gyr. Here we report the discovery of an old stellar halo and a sharp stellar edge. Also we find the distribution of stars extending beyond the gaseous envelope of the galaxy. Therefore, Leo A by its structure as well as stellar and gaseous content is found to resemble massive disk galaxies. This implies that galaxies of very low stellar mass are also able to develop complex structures, challenging contemporary understanding of galaxy evolution.

We have carried out spectroscopic observations in four cluster fields using Subaru's FOCAS multislit spectrograph and obtained spectra for 103 bright disc field and cluster galaxies at 0.06 z 1.20. 77 of these show emission lines, and 33 provide reasonably secure determinations of the galaxies' rotation velocity. The rotation velocities, luminosities, colours and emission-line properties of these galaxies are used to study the possible effects of the cluster environment on the star formation history of the galaxies. Comparing the Tully-Fisher relations of cluster and field galaxies at similar redshifts we find no measurable difference in rest-frame B-band luminosity at a given rotation velocity (the formal difference is 0.18 ± 0.33 mag). The colours of the cluster emission line galaxies are only marginally redder in rest-frame B − V (by 0.06 ± 0.04 mag) than the field galaxies in our sample. Taken at face value, these results seem to indicate that bright star-forming cluster spirals are similar to their field counterparts in their star formation properties. However, we find that the fraction of disc galaxies with absorption-line spectra (i.e. with no current star formation) is larger in clusters than in the field by a factor of ∼3-5. This suggests that the cluster environment has the overall effect of switching off star formation in (at least) some spiral galaxies. To interpret these observational results, we carry out simulations of the possible effects of the cluster environment on the star formation history of disc galaxies and thus their photometric and spectroscopic properties. This allows us to create mock samples of unperturbed 'field' galaxies [with approximately constant star formation rates (SFRs)] and perturbed 'cluster' galaxies with different star formation histories, including star formation truncation, with or without an associated starburst. We show that, if we select only bright galaxies with current star formation (i.e. with emission lines strong enough for rotation-curve measurements), the average colours and luminosities of the 'cluster' galaxies may not be very different from those of galaxies in the 'field' sample, even though their star formation histories may be significantly different. However, the fraction of emission and absorption-line galaxies would change significantly. We also use these simulations to estimate the size of field and cluster galaxy samples that would allow us to differentiate the different star formation scenarios considered. Finally, we find that the rest-frame absolute B-band magnitude of the field galaxies in our sample shows an evolution of −1.30 ± 1.04 mag per unit redshift at fixed rotation velocity. This indicates that the average SFR of bright disc galaxies evolves more slowly than the universal SFR as determined from ultraviolet, Hα, far-infrared and radio studies. This suggests the evolution of the universal SFR density is not dominated by bright star-forming disc galaxies, in agreement with previous studies.