Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 341, 371-384 (1999)

Previous Section Next Section Title Page Table of Contents

5. Spatial and kinematical differences
between early- and late-type galaxies

Biviano et al. (1997) studied the differences between ELG and non-ELG as far as their spatial distribution and kinematical properties are concerned. Combining the data for 75 clusters with at least 20 member galaxies, they found that the line-of-sight velocity dispersion (with respect to the cluster mean velocity), [FORMULA], is [FORMULA] larger for the ELG than it is for the non-ELG. They also found that the spatial distribution of the ELG is significantly less peaked towards the cluster centre than that of the non-ELG.

For a full appreciation of this result it is important to remember that the subsample of ELG consists almost exclusively of late-type galaxies, whereas the subset of non-ELG contains galaxies of all types. In other words: if the late-type galaxies without emission-lines would share the distribution and kinematics of their ELG counterparts, the differences between early- and late-type galaxies could well be more pronounced than between non-ELG and ELG.

On the other hand, it is also quite possible that the less centrally-concentrated distribution and larger [FORMULA] apply only to the late-type galaxies with emission lines. If so, that would provide additional support for the conclusion in Paper III that the ELG are likely to be on fairly radial, first-approach orbits, as suggested by their larger velocity dispersion, their projected spatial distribution, and their rather steep velocity dispersion profile [FORMULA]. The presence of the line-emitting gas would be fully consistent with this picture.

5.1. Kinematics

We have repeated part of the analysis of Paper III, making use of the classification in early- and late-type galaxies on the basis of the spectrum, discussed in this Paper. We start with the same set of 75 clusters as in Paper III. However, our galaxy sample includes only those galaxies for which we could estimate the galaxy morphology from the PCA and ANN. This limits the sample to 2594 galaxies in 66 clusters, of which 399 galaxies are ELG, while [FORMULA] are classified as early-type, and [FORMULA] as late-type. For each galaxy the normalized line-of-sight component of the velocity w.r.t. the cluster centre, [FORMULA], is determined, where [FORMULA] is the average cluster velocity and [FORMULA] is the line-of-sight velocity dispersion of the cluster to which the galaxy belongs. Following Paper III, we construct one large composite cluster by combining the data of all 66 clusters.

Using this sample of 2594 galaxies in 66 clusters, we find that the normalized line-of-sight velocity dispersion [FORMULA] of the ELG is 23% larger than that of the non-ELG, which is fully consistent with the result of Paper III. The values of [FORMULA] for ELG and non-ELG are given in column 3 of Table 5. The value of [FORMULA] for the dominant class of non-ELG is larger than unity because, in constructing the composite cluster, one adds velocity distributions for which the average velocities are known only with a limited accuracy. This leads to the superposition of (approximately Gaussian) velocity distributions with small apparent offsets, which slightly increases the dispersion above the expected value of 1.00. As discussed extensively in Paper III, this effect certainly does not explain the value of [FORMULA] of 1.28 for the ELG, because there is no evidence that the ELG have significant velocity offsets w.r.t. the non-ELG.


Table 5. Line-of-sight velocity dispersion (with respect to the cluster centre) and parameter values of the best-fitting [FORMULA]-model [FORMULA] to the surface density profiles of galaxies. Column 1 gives the subsample of galaxies. Column 2 gives the number of galaxies in this subsample. All values are averages ([FORMULA] r.m.s. values around these averages) over 10 realizations of the ANN. The best-fitting model parameters are not listed for the early-type ELG, as these are very uncertain.

In Table 5 we also give the values of [FORMULA] for several other subsets of the total sample. It appears that the [FORMULA] of the late-type galaxies is [FORMULA] larger than that of the early-type galaxies. This difference is significantly smaller than it is for ELG versus non-ELG, which makes it unlikely that the non-ELG spirals have the same kinematics as the ELG (mostly late spirals). This is indeed confirmed by the value of [FORMULA] for the non-ELG late-type galaxies (mostly early spirals) of [FORMULA]. Although this is somewhat higher than the value of 1.04 for all non-ELG, it is also very much smaller than the value of 1.28 found for all ELG, and for the subset of late-type ELG.

The intermediate value of [FORMULA] for the non-ELG late-type galaxies may mean one of three things. First, and most simply, it may be a statistical fluke, i.e. a 2[FORMULA] excursion of a value that is not fundamentally different from the [FORMULA] that we find for the early-type galaxies. Secondly, the separation of the late-type galaxies into ELG and non-ELG may not be perfect. This could be a result of our observational limit for the detection of emission lines, which need not correspond exactly to a kinematical distinction. In other words: the non-ELG late-type galaxy category may contain a fraction (which must be significant) of intrinsic ELG, for which the emission lines were not detectable in the ENACS. In that case, the true [FORMULA] of the non-ELG late-type galaxies is smaller and closer to the value of [FORMULA] found for the early-type galaxies. Thirdly, the non-ELG late-type galaxies may be a dynamically `pure' class, with kinematics intermediate between that of the early-type galaxies and that of the late-type ELG.

One might have a slight worry that the results in Table 5 are somewhat influenced by the fact the separation between e.g. early- and late-type galaxies on the basis of the spectrum is not perfect. In other words: the value of [FORMULA] for the early-type class may have been somewhat overestimated because the early-type class contains a non-negligible contribution of late-type galaxies. Similarly, the value of [FORMULA] for the late-type class may be somewhat underestimated. However, these effects are small.

Using the success rates in Table 3 for the two-class system, we estimate that at most 1 out of 4 galaxies in the early-type class is a misclassified late-type galaxy. Because essentially all galaxies in the early-type class (i.e. including the misclassified late-type galaxies) are non-ELG, the value of [FORMULA] of the early-type class is not overestimated very much. Using the value of [FORMULA] of 1.09 for the late-type non-ELG galaxies (which is a slight underestimate, see below), we estimate the bias in [FORMULA] of the non-ELG early-type galaxies to be at most a few percent. With this result, we can estimate that the value of [FORMULA] of the late-type non-ELG is more likely to be about 1.13 rather than 1.09, but this is still considerably smaller than the value of 1.28 of the late-type ELG.

Therefore, the data in Table 5 support a picture in which there is a clear correlation between the presence of emission lines and a high velocity dispersion. Rather unexpectedly perhaps, the ratio between the [FORMULA] of ELG and that of non-ELG does not appear to depend on whether the ELG or non-ELG are early- or late-type galaxies. The ELG among the early- and late-type galaxies have a value of [FORMULA] that is about 18% larger than that of the non-ELG of the corresponding galaxy type. In view of the large uncertainty in the estimate of [FORMULA] for the early-type ELG, this may be totally fortuitous, however, and we certainly should not overinterpret this result.

In summary, the basic factor driving the difference in kinematics seems to be the presence or absence of emission lines, whereas the distinction between early- and late-type galaxies is less important, while the class of late-type non-ELG presents an intrigueing cross-breed which may hold important clues to the physical meaning of the results.

5.2. Projected distributions

In view of the results in Table 5 and in Paper III, it is interesting to see how the kinematics and the projected spatial distribution are related. We therefore determined the surface density profiles of all subsamples, which we show in Fig. 6. The profiles are averages over the 10 realizations of the ANN (for the samples based on the distinction between early- and late-type). The profiles are shifted vertically such that at [FORMULA] Mpc the fitted profiles have the same values. The lines show the best-fitting [FORMULA]-model,


where [FORMULA] is the surface density, rc is the core-radius and [FORMULA] is the logarithmic slope at large radii. The best-fitting values of [FORMULA] and [FORMULA] are given in columns 4 and 5 of Table 5. Note that, as a result of the details of the OPTOPUS observations, the spatial completeness of the galaxy samples may not be uniform, so that the estimate of [FORMULA] may be biased. The errors in the estimates, determined from the comparison of the 10 realizations of the ANN, are small, of the order of 10%. Only for the early-type ELG the errors are substantially larger because the number of galaxies in this subsample is small.

[FIGURE] Fig. 6. Surface density profiles for the various subsamples of galaxies. The lines are the best-fitting [FORMULA]-models. The parameters of these are listed in Table 5.

The non-ELG are significantly more centrally peaked than the ELG, as was already concluded by Biviano et al. Although we find the same value of [FORMULA] for both subsamples, the ELG population has a much larger core-radius [FORMULA] than that of the non-ELG. The difference between early- and late-types is similar to that between non-ELG and ELG, the former being more centrally concentrated than the latter. The subsample of late-type ELG has a value of [FORMULA] that seems different from that of all other subsamples, but the difference probably is not significant, as [FORMULA] is quite large.

Apparently, the late-type ELG are distributed much more towards the cluster outskirts than all other galaxies, including the late-type non-ELG. For the early-type ELG, the values of [FORMULA] (-0.58) and [FORMULA] (0.02) are not very reliable because of the small number of galaxies involved. However, from a comparison between all ELG and the late-type ELG, one may conclude that both [FORMULA] and [FORMULA] are probably quite small for the early-type ELG. So the distribution of early-type ELG probably also deviates from that of the other galaxies, in the sense that they are more centrally concentrated. As we have seen in Sect. 4.2, the early-type ELG are often AGN, and this result therefore is not too surprising.

However, the early-type ELG may also contain a contribution from central dominant galaxies with emission lines from cooling flows (e.g. Heckman et al. 1989, and Crawford et al. 1995), which might give an important contribution to the high surface density of early-type ELG in the innermost bin in Fig. 6. Yet, it is not clear that the line ratios of the lines we observe are consistent with this explanation, and from our present data it is not easy to estimate this contribution.

5.3. What does it mean?

Combining the results of the spatial and kinematical properties of the different galaxy populations, we conclude that the late-type non-ELG have properties that resemble more those of the early-type galaxies, i.e. most of the other non-ELG. Yet, their projected distribution is slightly wider than that of the early-type galaxies, with a core radius that is a factor two to three larger, and kinematically they are somewhat `hotter' than the other non-ELG. The (late-type) ELG, which consist mostly of spirals, behave very differently. Their line-of-sight velocity dispersion [FORMULA] is much larger than that of the late-type non-ELG and they are located more towards the outer regions of clusters.

In Paper III the kinematical characteristics of ELG and non-ELG were interpreted as an indication for the ELG to be mostly on fairly (but not necessarily purely) radial orbits, in contrast to the non-ELG. Combining this with the larger velocity dispersion of the ELG and their relative scarcity in the very central regions of the clusters, we were led to the hypothesis that the ELG are mostly on radial, first-approach infall orbits towards the central regions of their clusters. This would be consistent with the presence of the line-emitting gas, as it is likely that that would have been removed from the galaxy on traversing the dense cluster core.

When the ELG are on orbits which are sufficiently radial without, however, traversing the very central regions, it is possible that they have already made several crossings without losing their gas, and will continue to do so until they `get caught'. In other words: their high velocity dispersion need not necessarily imply `first approach' orbits, because in the absence of an encounter they could maintain their velocity, which was due to their `late' infall. We may assume that an ELG which gets too close to the cluster centre (either on its first approach or after several crossings) will probably be `converted' almost instantly into a non-ELG late-type galaxy, as the gas gets stripped.

How the gas gets stripped from the ELG is not totally clear. In principle, ram pressure against intracluster gas could do the trick. However, that probably would not change the kinematics and distribution of the left-overs as drastically as observed. Alternatively, the harassment of galaxies through fairly high-speed and relatively distant encounters, as described by Moore et al. (1996), could be responsible for driving out the gas. Such encounters could be sufficiently frequent (about once per Gyr) to ensure that gas-rich ELG are virtually absent from the central regions. Actually, it is possible that an ELG has to experience a few of those encounters to get rid of its gas.

However, it is not immediately clear that such encounters will `instantly' reduce their velocity dispersion of 1.28 to 1.09, the value observed for their non-ELG counterpart. One factor which may contribute to this large apparent reduction is projection. If the ELG are indeed on fairly radial orbits, and their gas-robbed encounter products are on less radial orbits, this geometric effect might be responsible for most of the apparent reduction of [FORMULA].

Thus, it is possible that a slight change of the orbit characteristics (in particular the anisotropy parameter [FORMULA]), which results from the encounter which strips the ELG from its gas, is sufficient to considerably reduce [FORMULA] and produce a more centrally concentrated distribution. Note, however, that the kinematics and distribution of these stripped ELG may be different from those of the early-type galaxies which suggests that the latter are a more advanced product of encounters in the central cluster region.

The AGN among the ELG are a special class. They are predominantly ellipticals which, probably because of their central location, show AGN characteristics. Our data unfortunately do not allow us to determine convincingly how their velocity dispersion compares to those of the other types of galaxies in the cluster (see Table 5), but they seem to be at least as centrally concentrated as the non-AGN early-type galaxies.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: December 4, 1998