Astron. Astrophys. 364, 26-42 (2000)

7. Discussion

7.1. Field and cluster objects

Since our sample includes galaxies both from the field and from a cluster environment, we can investigate the eventual differences between these two subsamples. Following the conclusions of the authors, we will consider as cluster members the objects previously studied by Liu et al. (2000) and S97, and assume the rest of the sample to be representative of the ERO field population; the two subsamples include 14 and 27 objects respectively.

The most remarkable difference between them is that most of the galaxies classified as irregular (5 out of 6) belong to the field population. On the other hand, the only irregular cluster object (n. 5 in Table 1) is neither diffuse nor characterized by two interacting components, but is rather a compact galaxy with an irregular core; also, its spectrum does not exhibit features typical of ongoing star formation (S97; again, one example of an apparently "old" object that does not resemble local ellipticals). We conclude that, if some starburst galaxies are present among the EROs, they are not likely to be found in clusters. Considering only the field population, the fraction of irregular galaxies is only slightly increased (19%) with respect to our previous estimate.

For what concerns the fraction of non-exponential profiles, it is roughly the same for the two subsamples, again close to 80% for the high-signal objects.

In Table 4 we report a summary of the sample statistics. In the upper half of the table we consider the whole sample, divided into irregular galaxies, exponentials, and de Vaucouleurs. In the lower half we consider only the high-signal subsample, for which we distinguish between and .

Table 4. ERO morphology: a summary

7.2. Red exponential galaxies

As we have seen, some of our galaxies appear to be compact exponentials (see also Stiavelli & Treu 2000). of course, these objects cannot be classified as typical bright ellipticals if we use the local objects as a reference; on the other hand, the regularity of their surface brightness distributions tends to exclude the hypothesis of heavily reddened objects. The existence of this subclass, therefore, implies that the ERO population is apparently more composite than previously thought, a conclusion also emerging from the work by Liu et al. (2000), Stiavelli & Treu (2000) and Corbin et al. (2000). The possibility that such objects are undergoing an intermediate post-merging phase that eventually ends up in an elliptical galaxy is in contrast with the simplest monolithic collapse model, in which all ellipticals are formed at high redshift. Its implications in the framework of the different scenarios for galaxy formation certainly deserve further, more detailed investigation. At the same time, we cannot exclude that some ongoing star formation, suitably distributed throughout the galaxy, might transform an elliptical-like bringhtness distribution into one of the kind observed. Again, this hypothesis could be tested by a better characterization of the stellar content of these exponential objects.

7.3. Morphology and colors

In Fig. 9 we plot the median colors, computed for the different morphological classes introduced: irregular objects (corresponding to the point), exponential distributions, and the ones with . To include the galaxies with only available, we adopted a color to convert the color into . The adopted is an average value estimated for elliptical galaxies using the synthetic spectral-energy distributions of Bruzual & Charlot (1997) for a set of passively evolving models at (see the caption of Fig. 1). Since the color depends strongly on where the 4000 Å falls, the adopted has an associated uncertainty of mag.

Fig. 9. Median colors for the different morphological classes; the point corresponds to the irregular objects of the sample. The horizontal dotted line is the average color for the galaxies. Irregular galaxies are characterized, on average, by the reddest colors.

The dotted line represents the average for galaxies. Although each morphological bin is characterized by a significant scatter, there seems to be a tendence for the irregular objects to be characterized by the most extreme colors, with an average exceeding the mean color of likely ellipticals () by about 1 magnitude. Inverting the argument, we find that among the 6 reddest EROs () only one exhibits the typical morphology of an elliptical galaxy.

The adopted is most likely appropriate for high redshift ellipticals, but might be somewhat overestimated for possible starburst galaxies; thus, if we adopt a smaller for the irregular objects, the difference between them and the rest of the sample is furtherly increased. This result does not change also if we consider only the field subsample, and it is broadly consistent with the recent findings that the EROs detected in the submm show the reddest colors (Cimatti et al. 1998; Dey et al. 1999; Smail et al. 1999; Gear et al. 2000). A possible connection - to be confirmed by further, more detailed investigations - is therefore suggested between morphology, submm emission, and optical/infrared colors (in particular ): high z ellipticals and starburst might in fact exhibit a different behaviour with respect to each of the three parameters.

7.4. The distribution of n

We can consider the subsample of galaxies whose best shape index n is as a set of likely high-z ellipticals. For these objects we can compare the distribution observed for n with the results found for local samples of elliptical galaxies. Caon et al. (1993) determined a non-integer index n for a sample of local early-type galaxies in the Virgo cluster, so we have considered the galaxies from their sample with , and rebinned them in the range 2-8. More recently, the same kind of distribution was derived by Khosroshahi et al. (2000) for a sample of elliptical galaxies in the Coma cluster.

In the top panel of Fig. 10 we plot our distribution for the high-signal subsample with no correction applied to the derived n values (thick solid line). The dotted line is the same distribution, corrected for the systematic effect described in Sect. 5 (Eq. 1). In the bottom panel we plot the distributions for the Virgo cluster and the Coma cluster. Quite surprising, the two "local" histograms appear rather different, with the Virgo distribution extended up to large n values, and clearly peaked at , and the Coma distribution characterized by a peak and an upper cutoff at . Due to this diversity a comparison with our data is rather difficult; we just note that our "high-z" distribution looks somewhat intermediate between the two, being quite flat between and , and confined to this range. Without attempting any deeper comparison, we limit ourselves to consider this result as consistent with our claim that we are actually looking at a population of elliptical galaxies.

Fig. 10. Top panel: The distribution of the shape index n derived from our sample, both uncorrected (thick solid line) and corrected (dotted line) for the systematic effect described in Sect. 5. Bottom panel: the distributions derived by Caon et al. (1993) (dotted line), and by Khosroshahi et al. (2000) (dashed line).

7.5. The Kormendy relation

Scaling relations represent a powerful tool to investigate the evolution of galaxies at high redshift; in the case of elliptical galaxies, the Kormendy Relation between effective surface brightness and radius is relatively easy to build for a sample of ellipticals, when a detailed analysis of the brightness distributions and an accurate measure of the redshifts are available. Previous studies of this kind (for example Fasano et al. 1998; Ziegler et al. 1999 - Z99 hereafter; Roche et al. 1998) observed, as expected, an increase of the rest-frame surface brightness with redshift, but the type of evolution implied (passive or partially active) is not yet well constrained by the models.

For 6 of our compact galaxies a spectroscopic measure of the redshift has been published (in particular, 4 galaxies from S97 and 2 from Liu et al. 2000). All these galaxies are likely to reside in a cluster environment, and all of them are approximately at . As a consequence, they make up a particularly homogeneus set, well suited to pinpoint a particular time in the luminosity evolution of cluster ellipticals. Four more elliptical candidates (the ones in the HDFS) have photometric redshifts measured, but the two estimates available (Chen et al. 1998; Bénitez et al. 1999) are quite discrepant and they have been excluded from the following analysis.

We have used the 6 spectroscopic redshifts to derive the rest-frame parameters of the relative galaxies, following the prescriptions outlined in Z99; in particular, the observed F814W and H160W surface brightnesses have been corrected for the cosmological dimming, and transformed to the rest-frame B band. The corrections (Pozzetti, private communication) are evaluated for different cosmologies and spectral templates, using the models described in Pozzetti et al. (1998). The observed luminosities have been corrected for the Galactic extinction using the results by Schlegel et al. (1998). In Fig. 11 we plot the rest-frame Kormendy Relation for the 6 selected galaxies, derived adopting , , and a single stellar population model with redshift of formation . As a local reference, the solid line is the relation reported by Z99 for the whole sample studied by Jorgensen et al. (1995):

Fig. 11. The B-band, rest-frame Kormendy Relation for a subsample of "cluster" EROs at . The rest-frame parameters are evaluated assuming km s-1 Mpc-1 and . The solid line represents the local relation; the dotted line is the best fit to the data, with the slope fixed at the local value; the dashed line is a fit with both parameters free. The typical uncertainty on the data points is shown in the upper right corner of the plot.

An upwards shift of the data points with respect to the local relation is evident: a line with the same slope fitted to the data (the dotted one in Fig. 11) yields a difference of 1.5 0.4 mag. A value as low as 1.1 can be obtained by choosing a different local template (see Z99 for the details), or increasing the adopted value of up to 0.5. The best fit with both parameters free (dashed line) is

consistent with the constant-slope hypothesis. The possible slight steepening of this relation should be considered with caution, both because of the very few data points, and because of possible selection effects (for example, a set of galaxies selected in a narrow range of redshift and luminosity necessarily tends to exhibit a slope of 5).

A comparison with published results shows that the measured shift in surface brightness is consistent with the predictions of evolutionary models for elliptical galaxies (for example, with the Pure Luminosity Evolution models considered by Roche et al. 1998), as well as with the trends of luminosity and surface brightness vs. z observed at lower redshifts. Z99, for example, find a difference of 0.80.9 B-mag for two clusters at , whereas Schade et al. (1999) estimate a luminosity evolution of about 1 mag for a sample of field galaxies at redshift between 0.75 and 1.

© European Southern Observatory (ESO) 2000

Online publication: December 15, 2000
helpdesk.link@springer.de