3.2 The Morphology of Galaxies in Compact Groups

The situation for compact groups of galaxies is not as clear. The space densities are as high or higher in compact groups than even the densest regions in clusters so we might expect the compact groups to consist of essentially all elliptical galaxies, based on the morphology-density relation. This is not the case, however (see Mamon 1986, Figure 2, and Hickson, Kindl, and Huchra 1988; hereafter HKH). The fraction of ellipticals is only about 20% in the Hickson (1982) compact groups (HKH; Rood and Williams 1989), only slightly above the fraction found in the outer regions of a cluster ( 12%; Dressler 1980), or the loose groups surrounding the compact groups ( 7%; Rood and Williams 1989). Perhaps the lack of a large percentage of ellipticals is caused by the small number of galaxies in the compact groups, generally only four or five. Although the space density is similar to the densest regions in clusters, the small number of galaxies cannot produce the deep potential well found in clusters. This suggests that the global conditions may be more fundamental than the local conditions for the production of elliptical galaxies.

Although the fraction of elliptical galaxies in compact groups is much lower than at comparable densities in clusters, the morphology-density relation still appears to hold, but offset to higher densities. HKH find a 97.5% chance that the two parameters are correlated. Much of the reason the correlation is not better is due to the four highest density groups which are all very nearby and tend to be dominated by spirals. If we trim the sample to include only groups in the range 0.015 z 0.05, in an attempt to minimize selection effects, the probability that they are correlated increases to 99.99% (using the unbinned data rather than the binned data in HKH).

The strongest correlation HKH find is between spiral fraction and velocity dispersion of the group, with the higher dispersion groups containing more elliptical galaxies. This is contrary to what we might expect if mergers are the dominant mechanism since low-velocity interactions should be most effective. However, if the higher velocity dispersion is indicative of a deeper potential well, then this correlation may again suggest that global conditions are more important than local conditions.

One concern is that higher dispersion clusters are also the most distant, introducing the possibility of a systematic bias. At higher redshifts the brighter, more massive galaxies would be easier to detect, implying a higher velocity dispersion. If there is also a systematic tendency to misclassify the more distant galaxies toward earlier type galaxies because of the poorer spatial resolution, the trend between morphology and velocity dispersion could be an artifact. HKH address a similar point and conclude that the relationship between morphology and velocity dispersion is the fundamental correlation, since the statistical significance is much higher than for the morphology-luminosity relationship.

We can make an independent check of potential biases caused by the wide range in distance by trimming the sample to only include galaxies in the range 0.02 z 0.04. In this range, the average velocity dispersion for 11 elliptical-rich groups (i.e., more than 50% early type galaxies) is 229 ± 43 km s^-1. The average velocity dispersion for 19 spiral-rich groups (i.e., more than 50% spirals) is 145 ± 40 km s^-1. The corresponding numbers when the whole sample is used are 240 ± 19 km s^-1 for the elliptical-rich groups and 105 ± 20 km s^-1 for the spiral-rich groups. So it appears that although a systematic bias related to the distance may affect the results, the morphology-velocity dispersion relation is not dominated by this effect. Higher dispersion groups do tend to be dominated by ellipticals.

Recently, Barnes (1989) constructed self-consistent N-body simulations of galaxies within compact groups which indicated that they should merge into a single elliptical galaxy within a few billion years. At intermediate stages the merged galaxies would also appear to be ellipticals, so we might expect the first-ranked galaxy in these compact groups to tend to be ellipticals. HKH find that the morphology of the first-ranked galaxy does not differ significantly from that of the general population found in the groups. A Kolmogorov-Smirnov test, with the galaxies arranged in the normal E-S0-Sa-Sb-Sc-Sd-Im sequence, indicates that the distribution of first-ranked galaxies could arise from the total distribution 15% of the time. This might indicate that rapid merging is not occurring, as also suggested by the fact that the morphology-velocity dispersion relation goes in a sense which is opposite to what would be predicted by simple merger models.

Before we write off the merger hypothesis in compact groups completely we should note that the fraction of first-ranked galaxies which are ellipticals actually does rise, from 22% of the total population in compact groups to 35% of the first-ranked galaxies in the groups. What keeps the significance of the K-S test low is that in the next bin, the S0 galaxies fall from 26% of the total population to 11% of the first-ranked galaxies. If the types of galaxies were placed in order of how likely a merger is to produce a first-ranked galaxy of that type, S0 galaxies might be placed at the end of the distribution. We might expect merging galaxies to form ellipticals, and in some cases perhaps spirals if interactions spark large global spiral structures that brighten the galaxy. However, we would not expect brighter S0 galaxies to be produced by mergers, since the strong interactions would probably disrupt the fragile S0 disk, and a new purely stellar disk would take longer than the collapse time of the group to form. This might explain the dramatic decrease in the number of S0 galaxies. With the S0 galaxies arranged at the end of the distribution (and the few Im and cI galaxies removed) the chance that the first-ranked galaxies come from the same distribution as the total sample drops to 0.02%. The tentative finding of an anomalous population of bluish ellipticals in the Hickson groups by Zepf and Whitmore (1990a) also suggest the presence of merging activity at some level.

Another potential tool for determining whether initial conditions or late evolution is responsible for determining morphological types is the possible existence of morphological concordance between galaxies in groups. HKH find that in 20 of the 58 quartets, all four of the galaxies are of the same type (early or late). They calculate the probability of this occurring by chance is 10^-5. Yamagata, Nouguchi, and Iye (1989) have examined the CfA redshift survey and also conclude that nearest neighbor galaxy pairs tend to have the same morphological type. Similar studies of morphological concordance in pairs have been made by several other groups (e.g., Noerdlinger 1979) with similar results.

This morphology concordance is often taken as evidence that initial conditions are responsible for determining the morphology of the galaxies. Galaxies formed near each other would probably have similar initial conditions, and would therefore evolve in similar manners. However, White (1990) has recently shown that this effect may simply be the result of the morphology-density relation (or more generally, a morphology-"anything" relation). Most studies have calculated the probability by assuming the different types of galaxies are evenly mixed. But if certain groups contain higher fractions of one type of galaxies than another due to an effect such as the morphology-density relation, we would automatically expect more cases of morphology concordance.

The morphology-density relation might be caused by either initial conditions or late evolution, so the observed morphology concordance cannot tell us anything about the origins of galaxies until this effect is sorted out.