On the Association of Hickson Compact Groups with Loose Groups

The problem of the nature of compact groups (CGs) of galaxies is a very intriguing one. According to numerical simulations (Barnes 1985, 1989; Bode et al. 1993; Mamon 1986), the member galaxies in very dense Hickson compact groups (HCGs) should interact and eventually merge into one giant elliptical galaxy after several crossing times. Their dynamical timescales are, thus, so short that they should be very rare if observable at all. However, since HCGs are in fact observed, Mamon (1986) and Walke & Mamon (1989) suggested that HCGs are not real physical entities but are chance alignments of galaxies in loose groups (LGs). According to another point of view, HCGs are filaments seen end‐on (Hernquist, Katz, & Weinberg 1995; Ostriker, Lubin, & Hernquist 1995). However, some arguments have been presented showing that HCGs are real physical groups (Hickson & Rood 1988; Mendes de Oliveira & Giraud 1994; Mendes de Oliveira 1995; Ponman et al. 1996; Hickson 1997).

A possible solution to the problem comes from the fact that CGs are small subsystems in larger LGs (Rose 1977; Sulentic 1987; Rood & Williams 1989, hereafter RW; Rood & Struble 1994). By studying redshift space, Vennik, Richter, & Longo (1993) and Ramella et al. (1994) found that many HCGs are associated with LGs. Ramella et al. (1994) concluded that we see HCGs because they are being continually formed in collapsing LGs, as predicted by N‐body simulations of rich groups of galaxies (Diaferio, Geller, & Ramella 1994).

However, if strong interaction and merging processes are really going on in HCGs, then facts such as the lack of luminous radio sources (Menon 1995), absence of strong far‐infrared sources (Sulentic & de Mello Rabaça 1993; Hickson, Kindl, & Auman 1989), large spiral fraction, and lack of blue elliptical galaxies (Zepf & Whitmore 1991; Zepf 1993; Pildis, Bregman, & Schombert 1995) are hard to explain.

Another solution to the problems of the existence of HCGs as physical systems and of the absence of facts showing that the processes of interaction and merging are extensively taking place in HCGs has been given by Tovmassian, Martinez, & Tiersch (1999), Tovmassian & Chavushyan (2000), and Tovmassian, Yam, & Tiersch (2000). Tovmassian et al. (1999) showed that radial velocity dispersions (RVDs) of HCGs correlate with the elongation of groups: the higher the elongation of the group, the smaller the RVD. It is known that HCGs have the shape of a triaxial spheroid (Hickson et al. 1984; Malykh & Orlov 1986). The found dependence of the RVDs of HCGs on their elongation shows that HCGs are real, physical formations, members of which rotate around the gravitational center of the group. Tovmassian & Chavushyan (2000) showed that the accordant redshift galaxies in the environments of a dozen studied HCGs and poor groups of galaxies are distributed in narrow strips of a few hundred kiloparsecs in width. Moreover, they found that accordant redshift galaxies in the environment of HCGs and poor groups of galaxies obey the same correlation of RVDs with the elongation of the whole system. The location of the detected members of LGs in narrow strips oriented along the direction determined by the principal members of HCGs shows that the HCG+LG systems must also have the shape of a triaxial spheroid. Tovmassian et al. (2000) showed also that RW and Ramella et al. (1994) detected LGs mostly in the environment of those HCGs whose orientation of elongation is close to ∼45°, i.e., around those HCGs which are more favorable for detection of LGs.

The dependence of the RVD of faint galaxies in the environment of HCGs on the elongation of the group shows that LG members are gravitationally bound with corresponding HCGs. As Tovmassian & Chavushyan (2000) mentioned, there could be three options to account for the dependence of RVDs of galaxies on the elongation of the system: (1) infall of field galaxies from opposite directions toward the center of the group along its longer axis; (2) ejection of galaxies from the central galaxy in opposite directions; and (3) rotation of the LGs members around the common gravitational center of the corresponding HCG+LG system. HCG+LG systems, like HCGs, apparently have the shape of a triaxial spheroid, or a "cigar," as it has been stated by Oleak et al. (1995, 1998) for Shakhbazian compact groups. Tovmassian & Chavushyan (2000) concluded that the third option is the most realistic. Indeed, since movements of member galaxies in such elongated systems are generally parallel to the elongation of the system and are very high near the center of the system, the rate of coalescence and merging will be significantly reduced. This may explain the absence of strong signs of interaction and merging, the absence of strong radio sources, and so on, in HCGs.

Hence, LGs in the environment of HCGs must be very elongated triaxial ellipsoids, the large axis of which coincides approximately with the elongation of the corresponding HCG. Then LGs could be detected more easily if we search for their possible members in a direction along the elongation of the corresponding HCGs.

To test this suggestion I searched the neighborhood of HCGs. The search was made in a different way than that of RW. I found that a majority of HCGs are associated with LGs, the members of which are, indeed, located along the direction in which the principal members of HCGs are distributed.

The counts of galaxies in the neighborhood of HCGs were made by using the Palomar Observatory Sky Survey Automated Plate Scanner (APS) catalog.¹ The groups which are sufficiently elongated, i.e., b/a<0.5 (Tovmassian et al. 1999), were chosen for the search. There were 45 such groups out of 92 accordant redshift HCGs with available data in the APS as of 2000 January 1.

Since members of the associated LG are expected to be distributed along the elongation of the corresponding HCG (Tovmassian & Chavushyan 2000; Tovmassian et al. 2000), I made counts of galaxies in the narrow strip centered on each HCG and with position angle determined by the direction in which the accordant redshift members of the HCG are distributed. The length of the strip is 0.5 Mpc, which exceeds the size of the HCG by several times, and the width is 0.125 Mpc.² In angular scale the length of strips is within ∼10 ^'–45 ^'. One may argue that some uncertainty about the complete membership of an HCG may introduce a certain uncertainty in the computed ellipticity of the group and in the deduced position angle. However, the found correlation of the RVD of HCGs and HCG+LG systems on the elongation (Tovmassian et al. 1999; Tovmassian & Chavushyan 2000) shows that, in fact, there are no large errors in the deduced values b/a and, therefore, also in the deduced position angles of strips.

To see if there is any correlation of the distribution of galaxies in the environment of HCGs on their elongation, these counts were compared with counts made in orthogonal strips of the same length and width. The limiting magnitude of the counted galaxies is 22 mag on E prints. The differences Δ_c between numbers N_c of galaxies in the strip oriented along the elongation of the HCG (first strip) and numbers N_o in the orthogonal strip (second strip) are determined. Apparently, galaxies in the central squares with sizes 0.125 Mpc are not included in the counts. The members of HCGs which are outside the central square also were not included in the counts. Numbers of galaxies in the outer wings of corresponding strips with sizes 0.125 × 0.375 Mpc² are presented in columns (6) and (7) of Table 1.³ The values Δ_c are presented in column (8).⁴ In the columns (1) and (2) of Table 1 the designations of groups and their redshifts (Hickson 1994) are given. The positions of the centers of the searched strips are presented in columns (3) and (4). The centers of the searched areas either coincide with those given in Hickson (1982 or 1994) or are very close to them. The position angles of the first strips are given in column (5).

In the absence of any correlation of the distribution of galaxies in the environment of HCGs on the orientation of groups, the differences Δ_c and Δ_we should be determined by a noise and should have a normal distribution around zero. According to Yoshi (1993) there are in the mean ∼1500 galaxies of 22 mag deg⁻². In ∼25–450 arcmin² strips, where counts for nearby groups are made, there should be on average ∼10–180 galaxies. Hence, the expected Poissonian rms should be ∼3–13.

In Figure 1 the Δ_c values are plotted versus redshifts. Since members of LGs are generally dwarf galaxies, one would expect that LG members would be detected mostly in nearby galaxies and that the number of detected LG members would decrease with the distance. Figure 1 shows that the numbers of galaxies in the central strips oriented along the elongation of the group, indeed, outnumber those in orthogonal strips in the case of nearby groups with z smaller than ∼0.025. The regression line drawn for groups with z<0.03 shows that Δ_c gradually decreases with the redshift, as expected. The decrease in the number of galaxies in the first strips of nearby groups shows that the excess of galaxies here is not due to noise. The regression line drawn for groups with z>0.03 is strictly horizontal. Hence, LGs are, apparently, detected in the environments of nearby HCGs with z<0.025. Although the excess of galaxies in the outer wings of the first strips of nearby HCGs is not large and for most groups is ∼σ, it is notable, however, that the average of negative Δ_c values is −2.6 ± 2.5, while that of positive Δ_c values is 15.5 ± 9.4. If we exclude the group HCG 90 with the largest Δ_c, 39, then the average of positive Δ_c values is 13.4 ± 6.1. A Kolmogorov‐Smirnov test showed that Δ_c values for all 17 nearby groups with z<0.025 have a normal distribution (P<0.01) with an average value 10.2 ± 11.6.

Zoom In Zoom Out Reset image size

For nearby HCGs with z<0.025 the counts were made also in two half‐width narrow strips on both sides of the first strip. The numbers of galaxies on the western and eastern side strips are given in columns (9) and (10) of Table 1. The differences Δ_we of these numbers are given in column (11). In one case, HCG 90, Δ_we is very large, 56, and differs appreciably from all other values. With its exclusion the average of Δ_we values of 16 groups, which also have a normal distribution (P<0.01), is close to zero (−0.1 ± 6.8), as expected. The two‐sample Mann‐Whitney rank‐sum test showed that the average of the Δ_c values and Δ_we values are different at greater than the 99.92% significance level.

I compared also the integral number of galaxies N_s in the two side strips with the number N_o of galaxies in the central orthogonal strip. In this case the galaxies in the overlapping parts of the second strips are excluded to make equal the areas in which counts are made. Differences Δ_s between N_s and N_o are presented in column (12) of Table 1. These differences have a normal distribution centered at 3.0 ± 10.9. The average of the Δ_s values does not differ from that of the Δ_we values with a probability greater than 99%. This means that the small excess of galaxies in the side strips is not statistically significant. Hence, the excess of galaxies is found only in very narrow central strips oriented along the elongation of the corresponding HCGs.

For groups with z>0.025 the distribution of Δ_c values has an average of −0.1 ± 4. Thus, no significant difference of the number of galaxies in the two orthogonal strips is observed in this case. LGs are, apparently, not detected in more distant groups because of the faintness of their possible members and because of the smallness of the searched area in angular scale (about 25–50 arcmin²).

If we consider the distributions of Δ_s and Δ_we values as a control samples, then one may assume that a LG is detected if Δ_c exceeds the dispersion of the normal distributions of Δ_s and Δ_we values, i.e., ∼5. Then LGs are detected in the environments of 12 out of 17 nearby groups (z<0.025) in common with the RW list. The detection rate is 0.71. RW detected seven LGs in the environment of the same 17 HCGs; i.e., the detection rate is 0.41. Hence, although the searched area in the present paper is much smaller, the detection rate is by 1.7 times higher than that in RW. The average number of galaxies in seven LGs detected by RW is 3.9 and that in 12 LGs detected in the present work is 15.5. Both results are based on the statistical counts of galaxies in the environments of HCGs. Hence, our statistical approach allowed us not only to detect more LGs associated with HCGs but also to detect more members of corresponding LGs. The detection rate for 11 out of 13 nearby HCGs is higher, 0.85, in Ramella et al. (1994), who considered the redshift distribution of galaxies around HCGs.

For HCGs with z<0.025 I made counts also in two central orthogonal strips with limiting magnitudes of 23, 20, and 18 mag. The average Δ_c deduced in the case of the limiting magnitude of 23 mag is equal to 9.6 ± 10.2 and in practice does not differ from the one deduced with the limiting magnitude of 22 mag. Hence, inclusion of fainter galaxies in the counts does not, in the mean, increase the excess of galaxies in the first strips. Meanwhile, counts made with brighter limiting magnitudes show that the excess of galaxies in the first strips is decreasing. The average Δ_c is equal to 4.4 ± 8.2 in the case of the limiting magnitude of 20 mag and is equal to 2.3 ± 4.2 in the case of the limiting magnitude of 18 mag. Thus, the excess of galaxies in the first strips is observed at magnitudes between 18 and 22 mag.

The limiting absolute magnitude of the faintest galaxy with m = 22 mag detectable in the nearest group HCG 90 at a distance of ∼35 Mpc is ∼−10.7 mag. The limiting absolute magnitude of the faintest galaxy in the farthest of the sample of nearby groups, HCG 91, is −13 mag. No excess of galaxies is observed in the range of absolute magnitudes from −15 to −17 mag. Hence, absolute magnitudes of the LG galaxies range from −10.7 to −17 mag.

The vicinities of 44 elongated HCGs are searched for associated galaxies. Counts of galaxies are made in two orthogonal strips passing through the center of the group. The width of the strips is 0.125 Mpc, and the length is 0.5 Mpc. For groups with z<0.025 two half‐width strips on both sides of the first strip are also inspected. Counts are made with limiting magnitudes of 23, 22, 20, and 18 mag.

An excess of 5–39 galaxies in the range of absolute magnitudes from −10.7 to −17 mag is detected in the environment of about 70% of nearby HCGs with z<0.025. The decrease in the number of detected LG galaxies with the distance of the HCG allows one to suggest that distant groups should also be associated with LGs members of which were not detected for their faintness. Thus, it is apparent that most HCGs are associated with an LG, as found by Vennik et al. (1993) and Ramella et al. (1994). In this paper it is shown, in addition, that LGs associated with HCGs are elongated systems, oriented along the elongation of the corresponding group. It was shown earlier that accordant redshift members of LGs are distributed along the elongation of the corresponding HCG, and RVDs of members of the compact group and also of the associated LG correlate with the elongation of the compact group (Tovmassian et al. 1999; Tovmassian & Chavushyan 2000). These facts allow one to suggest that LG members most probably rotate together with principal members of HCGs around a common gravitational center (Tovmassian & Chavushyan 2000). The specific distribution of LG members revealed in this paper is evidence that they must be gravitationally bound with corresponding HCGs.

Thus, HCGs are not simply located in dense environments (Rose 1977; Sulentic 1987; Vennik et al. 1993; Ramella et al. 1994; Rood & Struble 1994). Members of LGs in the environment of HCGs are physically associated with them. They apparently rotate in sufficiently elongated orbits around a common gravitational center. The high velocities with which galaxies are passing near the center of the group should apparently decrease the efficiency of the processes of interaction and merging. Hence, such systems are, evidently, more stable than it has been assumed. If HCG+LG systems are, thus, sufficiently stable configurations, then the known puzzles related to HCGs may be solved.

We observe such an elongated "cigar‐like" system as a compact group if, at first, its large axis is close to the line of sight or, otherwise, if its bright members are projected close to each other on the sky. The latter may occasionally happen as the result of a regular rotation of the group members around the gravitational center of the group. Earlier, Mamon (1986) and Walke & Mamon (1989) suggested that HCGs are a result of chance alignments of galaxies within LGs. We see now that compact groups are, indeed, a result of a chance projection of galaxies but not of gravitationally unrelated members of LGs. The probability of such a chance orientation of the group itself or its bright members is apparently very small. It means that the number of such groups in the universe may be much larger than is detected. The vast number of groups, including those in which there are no bright members, thus remains undetected. They, most probably, form the background of galaxies on the sky.