Astron. Astrophys. 353, 479-486 (2000)

4. Discussion and conclusions

We have presented the near-infrared LF of a nearby cluster of galaxies, Coma, down to faint magnitudes ( mag, i.e. mag corresponding roughly to mag). This has been made possible due by the relatively deepness of the present images and by the small background variance associated with the large surveyed area. The computed LF is the deepest ever measured in the near-infrared, on any type of environment.

The shape of the Coma LF in the region studied seems not to depend on wavelength, at least in b and H bands, and with less confidence, in R. The similarity of the LF implies that in the central region of Coma there is basically no new population of galaxies which disappears or becomes too faint to be observed in the optical bands (because of the presence of dust, for instance), down to the magnitudes of dwarfs. Furthermore, if the H band LF traces the galaxy mass function, also the blue LF traces the mass in this case. This is in apparent contradiction with the results by Gavazzi et al. (1996), who found that for spiral galaxies the M/L is approximatively constant in the near-infrared but not in the optical filters. Since our finding is based on just one sample in one particular environment, although selected with well understood selection criteria (volume-complete), it has to be verified on other samples of nearby galaxies, possibly spiral-rich clusters or groups, before any dangerous generalization.

The bright part of the Coma H band LF, i.e. the brightest three magnitudes, agrees with the expectations based on optical LFs and usual colors for galaxies, and with what is observed in shallower near-infrared surveys of clusters of galaxies and also on the field. This confirms that the shape of the tip of the mass function seems environment-independent and therefore environmental effects have a minor impact on the luminosity of bright galaxies (), and possibly on their masses. Coma and the field population differ by a factor of 100 in galaxy density. The extension of this sentence to faint () galaxies still awaits a determination of the field LF in the dwarf regime.

The Coma near-infrared LF presents a real dip at a luminosity corresponding to that observed in the optical LF. This is the first detection of such a feature in the near-infrared. The existence of a dip in the Coma LF in the H band implies the presence of a dip also in the galaxy mass function. To our knowledge, there is presently no simulation of cluster formation which is able to produce such a feature in the galaxy mass or luminosity function. This feature, being distinctive, will set a strong constraint for the future simulations.

Kauffmann & Charlot (1998) have shown that the apparent passive evolution and the slope of the color-magnitude relation can be accommodated within a hierarchical model, even if the galaxies themselves grow by mergers until late times. One of the important remaining issues is the comparison between the predicted and the observed LF, in particular, the distribution of galaxies as a function of their morphological type, at least for early-type galaxies. Probably the main limitation till now has been the lack of suitable observational data to compare with model expectations. Our near-infrared catalog, published in Paper I, joint to the morphological types for the Coma galaxies, available from Andreon et al. (1996, 1997), fills this observational gap.

The overall slope of the Coma LF is intermediate (). The slope is measured down to the dwarfs regime: we reach using our own data alone and even fainter magnitudes (, roughly equivalent to ) when including Mobasher & Trentham (1998) data and under their assumptions. When comparing Coma and the field LFs in the near-infrared, we have to take into account that both LFs have been derived with completely different data and methods, because of the different selection criteria for the two samples: the field LF is computed on a flux-limited sample, whereas the cluster LF is computed on a volume-limited sample. In particular, the field LF suffers from a 10% redshift incompleteness (or is based on an optical selection, as for the Szokoly et al. (1998) LF), and a poor sampling of faint luminosities, because of the small volume explored at that luminosities. Nevertheless, and even if the environments sampled are quite different, the bright tail of the Coma and the field LFs are in close agreement. In our opinion, this excludes the possibility of large systematic errors in the derivation of field LFs, and therefore indirectly confirms the disagreement between the observed near-infrared LF and that expected on theoretical grounds in the present simulations of a hierarchical Universe (Kauffmann et al. 1999). This also suggests that more detailed models are needed to reproduce the observed properties of galaxies, such as the LF. In particular, as mentioned before, the existence of a dip in the present LF and the large range on which this LF is computed (7 mag) provides a strong constraint to future simulations. It is worth to noting that theoretical predictions on the behaviour of high order statistics, such as the color distribution or the galaxy evolution, use a particular realization of the LF as a "weight". Thus, increasing the accuracy on the determination of the LF will certainly contribute to the improvement of the theoretical knowledge on galaxy formation and evolution.

© European Southern Observatory (ESO) 2000

Online publication: December 17, 1999
helpdesk.link@springer.de