|Annu. Rev. Astron. Astrophys. 1997. 35:
Copyright © 1997 by Annual Reviews. All rights reserved
A key question that remains is the position of compact groups in the clustering hierarchy. Are compact groups distinct entities (Sulentic 1987) or an intermediate stage between loose groups and triplets, pairs and individual galaxies (Barnes 1989, White 1990, Cavaliere et al. 1991, Rampazzo & Sulentic 1992, Diaferio et al. 1994). Some compact groups are purely projection effects, others may be small clusters (Ebeling et al. 1995), but most appear to be real. It seems that they can arise naturally from subcondensations in looser groups, but further studies are needed to better determine both the observed space density of groups as a function of population and the timescales involved in the evolutionary process.
This question is related to that of the formation mechanism of compact groups. Two mechanisms have been discussed in the literature. Diaferio et al. (1994) conclude that compact groups form continually from bound subsystems within loose groups. This gains some support from the observation (see Section 3) that most HCGs are embedded in loose groups, although it is not obvious that these loose groups are sufficiently rich (Sulentic 1997). Governato et al. (1996), proposed a model in which merging activity in compact groups is accompanied by infall of galaxies from the environment. This naturally explains the observed mix of morphological types, and it allows compact groups to persist for longer times.
Where do the Shakhbazian groups fit in this picture? Recent studies (Tikhonov 1986, Amirkhanian & Egikian 1987, Amirkhanian et al. 1988, 1991, Amirkhanian 1989, Kodaira et al. 1988, 1990, 1991, Stoll et al. 1993a, 1993b, 1994a, 1994b, 1996a), 1996b show that these objects are typically compact clusters or groups of early-type galaxies. Although the systems were selected on the basis of red colors and compact appearance of their galaxies, both of these factors result from their large distances because K-corrections and contrast effects become significant. The galaxies are in fact relatively normal, although luminous (Del Olmo et al. 1995). However, the number of blue (gas rich) galaxies in these systems does seem to be very small. Thus it appears the Shakhbazian groups are mostly small clusters, possibly intermediate in physical properties between classical compact groups and clusters.
7.2 Galaxy evolution and merging
If the groups are dynamically bound, galaxy mergers should commence within a few dynamical times (Carnevali et al. 1981, Ishizawa et al. 1983, Barnes 1985, Ishizawa 1986, Mamon 1987, 1990, Zheng et al. 1993). Both N-body and hydrodynamic simulations indicate that the dark matter halos of individual galaxies merge first, creating a massive envelope within which the visible galaxies move (Barnes 1984, Bode et al. 1993). Kinematic studies of loose groups (eg. Puche & Carignan 1991) indicate that the dark matter is concentrated around the individual optical galaxies. In contrast, the X-ray observations indicate that in most compact groups, the gas and dark matter is more extended and is decoupled from the galaxies. This may explain the observation that galaxies in compact groups typically have mass-to-light ratios 30% to 50% lower than more isolated galaxies (Rubin et al. 1991).
Is there any observational evidence that galaxies in compact groups are merging? By 1982 it was evident that first-ranked galaxies in compact groups did not appear to be merger products, because the fraction of first-ranked galaxies that are type E or S0 is the same as for the general population of HCG galaxies (Hickson 1982). If mergers were a dominant effect, the first-ranked galaxies would be expected to be more often elliptical. The same conclusion was reached by Geller & Postman (1983) who found that the luminosities of first-ranked galaxies were consistent with a single luminosity distribution for all group galaxies. Of course this may just mean that in small groups the first-ranked galaxy is not necessarily the most evolved. Rather, one should ask if any galaxies in compact groups show indications of merging. The relative paucity of merging galaxies in compact groups was first noted by Tikhonov (1987), from a visual inspection of optical images. Zepf & Whitmore (1991) realized that elliptical galaxies formed by recent mergers of gas-rich systems should have bluer colors than normal. Examining the HCGs, they found only a small enhancement in the fraction of early-type galaxies having blue colors, a conclusion reinforced by an independent study by Moles et al. (1994). On the other hand, Caon et al. (1994) argued that the large effective radii of compact group elliptical galaxies is indicative of an origin by merging or accretion of companions.
Zepf (1993) estimated that roughly 7% of the galaxies in compact groups are in the process of merging. This conclusion was based on roughly consistent frequencies of (a) optical signatures of merging, (b) warm far-infrared colors, and (c) sinusoidal rotation curves. However, few galaxies show all of these effects simultaneously. The merging fraction may thus be as high as 25% if one allows that any one of these criteria would be considered to be sufficient to indicate a merger (Hickson 1997). Given the small numbers of objects in these studies, it is fair to say that the fraction of merging galaxies is highly uncertain at present. It seems safe to conclude that current observations do not rule out a significant amount of merging in compact groups.
Detailed studies of individual compact groups can be quite revealing. Many galaxies that at first appear normal are revealed to have peculiar morphology or spectra when examined more closely. Many, perhaps most, compact groups clearly contain galaxies that are dynamically interacting. However, the groups likely span a range of evolutionary states. At the extreme end are high-density groups like Seyfert's Sextet, HCG 31, HCG 62, HCG 94 (Pildis 1995) and HCG 95 (Rodrigue et al. 1995) in which we find strong gravitational interactions. At the other end are lower density compact groups, such as HCG 44, which most likely are in a less-advanced stage of evolution. This picture is supported by radio observations: Seyfert's Sextet and HCG 31 are both embedded in extended HI clouds whereas in HCG 44 the HI is associated with individual galaxies (Williams et al. 1991).
It seems clear that the groups as we now see them can persist for only a fraction of a Hubble time. Simulations indicate that merging should destroy the group on a time scale tm that is typically an order of magnitude larger than td, (Cavaliere et al. 1983, Barnes 1984, Navarro et al. 1987, Kodaira et al. 1990), although longer lifetimes are possible depending on the distribution of dark matter (Athannasoula et al. 1997) and initial conditions (Governato et al. 1991). Assuming that the groups are in fact bound dynamical systems, we can draw two conclusions: (a) There must be an ongoing mechanism for forming or replacing compact groups, and (b) there must be a significant population of relics of merged groups.
What are the end-products of compact groups? It is tempting to identify them with field elliptical galaxies, following a suggestion first made by Toomre (1977). Simulations (Weil & Hernquist 1994) indicate that multiple mergers in small groups of galaxies best reproduce the observed kinematical properties of elliptical galaxies. The resulting galaxies are predicted to possess small kinematic misalignments, which can be detected by detailed spectroscopic and photometric studies. Neverthess, it remains to be demonstrated that these merger remnants can reproduce the tight correlation between size, luminosity and velocity dispersion found in present-day elliptical galaxies.
If compact groups have lifetimes on the order of tm, and form continuously, then the number of relics, per observed group, is expected to be on the order of (H0tm)-1. Thus, the number of relics could exceed that of present day groups by as much as an order of magnitude. Mamon (1986) estimated that, if all HCGs are real, then the relics would account for about 25% of luminous field elliptical galaxies. As we have seen, the true space density of compact groups is uncertain by at least a factor of two, and may be underestimated because of selection biases. There is then the potential problem of producing too many relics.
A second problem is the fact that the integrated luminosities of compact groups are typically a factor of three to four times greater than luminosities of isolated elliptical galaxies (Sulentic and Rabaça 1994). It is possible that interaction-induced star formation has boosted the luminosities of some compact-group galaxies, and that some degree of fading of the merger product is expected. However, at this point it is not clear whether or not the relics can be identified with isolated elliptical galaxies.
Despite these problems, a fossil compact group may have actually been found. Ponman et al. (1994) have detected a luminous isolated elliptical galaxy surrounded by diffuse X-ray emission which is consistent with the expected end-product of a compact group. If more objects like this are found, it may be possible to compare their space density with that expected for compact group relics.
7.3 Role in galaxy formation and evolution
Interactions are often implicated in the development of active nuclei in galaxies (eg. Freudling & Almudena Prieto 1996). The HCG catalog includes several examples of compact groups containing both starburst galaxies and AGN. Several recent examples of associations between starburst galaxies or AGN and what appear to be compact groups have been reported: Del Olmo & Moles (1991) have found a broad-line AGN in Shakhbazian 278; Zou et al. (1995) find that the luminous infrared source IRAS 23532 coincides with a compact group that includes a Seyfert 1 as well as a starburst galaxy. If this association extends to QSOs, one would expect to find numerous compact groups at redshifts z ~ 2, where the comoving number density of QSOs peaks (eg. Hartwick & Schade 1984). The tendency for QSOs to have close companions has been known for some time (eg. Stockton 1982, Bahcall et al. 1997). Recently, several examples of compact groups associated with luminous infrared galaxies, AGN and QSOs at z 2 have been found using HST (Pascarelle et al. 1996, Francis et al. 1996, Matthews et al. 1994, Tsuboi & Nakai 1994, Hutchings 1995, Hutchings et al. 1995).
These observations provide support to the idea that tidally-triggered star formation is a predominant factor in the galaxy formation process (Lacey & Silk 1991, Lacey et al. 1993). In this model disk star formation occurs relatively late, after the compact group has formed and tidal interactions are strong. This seems at least qualitatively consistent with the fragmentary nature of high-redshift galaxies observed with the Hubble Space Telescope (Schade et al. 1995), although these fragments appear to be much less luminous and more irregular than most present-day compact group galaxies. The model also offers a possible explanation for the excess numbers of faint blue galaxies found in field galaxy count as dwarf galaxies undergoing star formation at a redshift of z 1.
Compact groups may possibly play a role in the formation of other systems. We have seen that giant galaxies may be formed as the end product of compact-group evolution. At the other end of the scale, dwarf galaxies have physical properties distinct from normal galaxies, which suggests a unique formation mechanism. One possibility is that they form during gravitational interactions from tidal debris (Duc & Mirabel 1994). If this is the case, one would expect to find evidence for this in compact groups of galaxies. From an examination of condensations in tidal tails, Hunsberger et al. (1996) concluded that the fraction of dwarf galaxies produced within tidal debris in compact groups is not negligible. There is also evidence that star clusters form from tidal debris. Longo et al. (1995) have found an excess population of unresolved blue objects around HCG 90 which appear to be recently formed star clusters. These may be similar to the population of new star clusters recently reported in the merger remnant NGC 7252 (Whitmore et al. 1993).
7.4 Gravitational lensing
Because compact groups have a high galaxy surface density, they may form effective gravitational lenses. Gravitational amplification of background field galaxies was proposed by Hammer and Nottale (1986) as a possible explanation for the presence of the high-redshift discordant member of this group. Mendes de Oliveira and Giraud (1994) and Montoya et al. (1996) find that most HCGs are too nearby to produce strong lensing effects. However, because the critical mass density required for strong lensing depends reciprocally on distance, analogous systems 5-10 times more distant should produce a non-negligible fraction of giant arcs.
Studies of small groups may provide clues to the overall structure of the universe. The baryon fractions found in clusters of galaxies appear to be inconsistent with a density parameter = 1, unless the dark matter is more prevalent outside clusters (White 1992, Babul & Katz 1993). Compact groups provide a means to study dark matter in such regions. The baryon fractions found for compact groups do appear to be lower than those for clusters. David et al. (1995) argued that the gas is the most extended component; galaxies being the most compact and the dark matter being intermediate. They concluded that the baryon fraction approaches 30% on large enough scales, which is comparable to the values found for clusters. Given the constraints of standard Big-Bang nucleosynthesis this would imply that the density parameter is at most 0.2. On the other hand, the infall picture of compact-group evolution (Governato et al. 1996) requires a high-density ~ 1 universe. In a low-density universe the infall rate is insufficient. As there is at present no other clear mechanism for avoiding the overproduction of relics by merging compact groups, this may be a strong argument for a high-density universe.
During the last two decades we have seen a resurgence of interest in compact groups. While initially little more than a curiosity, these systems are now viewed as potentially important sites of dynamical evolution, shaping the structure of many galaxies. It now seems clear that while many compact groups are contaminated by projections, a large fraction of at least the high-surface-brightness HCGs are physically dense. They form by gravitational relaxation processes within looser associations of galaxies. The densest are generally in an advanced stage of evolution characterized by strong interactions, starburst and AGN activity, stripping of stellar and dark matter halos, and merging. They contain large amounts of dark matter and primordial X-ray-emitting gas trapped within the gravitational potential well.
Despite this progress, many questions remain unanswered. What are the end
products of compact group evolution, and do they have properties consistent
with any know population of objects? What is the space density of such relics?
Where do compact groups fit in the overall clustering hierarchy? What is
their role in the evolution of galaxies both past and present? Given the
current interest and research activity in this area, it is likely that
many of these questions may soon be addressed.
It is a pleasure to thank the Observatories of Brera and Capodimonte for
hospitality during the initial work on this review. I have benefitted from
discussions with many individuals, but I would like to acknowledge
the contributions of A Iovino, E Kindl, G Longo, G Mamon, C Mendes de Oliveira,
TK Menon, G Palumbo, H Rood, and J Sulentic. I thank G Mamon, A Sandage
and J Sulentic for providing helpful comments on an earlier version of the
manuscript. Financial support was provided by the Natural Sciences and
Engineering Research Council of Canada and NATO.
It is a pleasure to thank the Observatories of Brera and Capodimonte for hospitality during the initial work on this review. I have benefitted from discussions with many individuals, but I would like to acknowledge particularly the contributions of A Iovino, E Kindl, G Longo, G Mamon, C Mendes de Oliveira, TK Menon, G Palumbo, H Rood, and J Sulentic. I thank G Mamon, A Sandage and J Sulentic for providing helpful comments on an earlier version of the manuscript. Financial support was provided by the Natural Sciences and Engineering Research Council of Canada and NATO.