7. Properties of galaxies in groups
In this section we examine the luminosities and the spectral features of galaxies in different environments: the "field", groups, and clusters. The dependence of these properties on the environment offers insights into the processes of galaxy formation and evolution and on the dynamical status of groups.
7.1. The luminosity function of members
Here we investigate the possible difference between the luminosity functions of member and non-member galaxies. We compute the luminosity function with the STY method (Sandage, Tamman & Yahil 1979). We assume a Schechter (1976) form for the luminosity function and follow the procedure described in detail in Zucca et al. (1997).
We find that galaxies in groups have a brighter with respect to non-member galaxies; the slope does not change significantly in the two cases. In particular the parameters we obtain are and for the 1250 members, and and for the 1835 non-members.
In Fig. 8 we draw (dotted lines) the confidence ellipses of the and parameters obtained in the two cases of member and non-member galaxies. The two luminosity functions differ at the level. In Fig. 8 we also plot the confidence ellipses for the parameters of the total sample (solid lines) derived in the same volume of ESP where we identify groups.
The fact that galaxies in groups are brighter than non-member galaxies is a clear demonstration of the existence of luminosity segregation in the ESP survey, a much deeper survey than those where the luminosity segregation has been previously investigated (Park et al. 1994, Willmer et al. 1998). Our finding is consistent with the results of Lin et al. (1996), who find evidence of a luminosity bias in their analysis of the LCRS power spectrum.
In further support of the existence of a luminosity segregation, we also find that becomes brighter for members of groups of increasing richness. As before, the parameter remains almost constant. Only in the case of the richest groups, , we find a marginally significant steepening of the slope .
7.2. Emission/absorption lines statistics
One interesting question is whether the environment of a galaxy has a statistical influence on the presence of detectable emission lines in the galaxy spectrum. Because emission lines galaxies are mostly spirals (Kennicutt 1992), the answer to this question is relevant to the investigation of the morphology-density relation in systems of intermediate density.
The fraction of ESP galaxies with detectable emission lines within the redshift range km s-1 is 44% (1360/3085). Of these -galaxies , (34 2)% (467/1360) are members of groups. The fraction of galaxies without detectable emission lines, -galaxies , that are members of groups is significantly higher: 783/1725 or (45 2)%. We note that our detection limit for emission lines correspond to an equivalent width of about 5
We consider three types of environments: a) the "field", i.e. all galaxies that have not been assigned to groups, b) poor goups, i.e. groups with 3 4, and c) rich groups with 5 . We find that the fraction of -galaxies decreases as the environment becomes denser. In the "field" the fraction of -galaxies is = 49%, it decreases to = 46% for poor groups and to = 33% for richer groups. In Fig. 9 we plot as a function of . We also indicate the values of of the two richest Abell clusters in our catalog, A2840 ( = 21%) and A2860 ( = 19%).
The significance of the correlation between environment and can be investigated with a 2-way contingency table (Table 2). For simplicity, we do not consider triplets and quadruplets.
Table 2. Frequency of Emission Line Galaxies in Different Environments
The contingency coefficient is C=0.15 and the significance of the correlation between environment and frequency of emission line galaxies exceeds the 99.9% level.
Our result indicates that the morphology-density relation extends over the whole range of densities from groups to clusters. Previous indications of the existence of the morphology-density relation for groups are based either on very local samples (Postman & Geller 1984) or on samples that are not suitable for statistical analysis (Allington-Smith et al. 1993). Very recently, Hashimoto et al. (1998) also confirm the existence of a morphology-density relation over a wide range of environment densities within LCRS.
Examining our result in more detail, we note that the fraction of -galaxies , , in triplets and quadruplets is very similar to the value of for isolated galaxies. Triplets and quadruplets are likely to correspond, on average, to the low-density tail of groups. Moreover, Ramella et al. (1997) and Frederic (1995a) estimate that the FOFA could produces a large fraction of unbound triplets and quadruplets. These "pseudo-groups" dilute the properties of real bound triplets and quadruplets with "field" galaxies, artificially increasing the value of . This effect, in our survey, is partially counter-balanced by the triplets and quadruplets that are actually part of richer systems cut by the OPTOPUS mask. Considering that rich systems are significantly rarer than triplets and quadruplets, we estimate that the value of we measure for triplets and quadruplets should be considered an upper limit.
Our catalog also includes ESP counterparts of 17 clusters listed in at least one of the ACO, ACOS (Abell et al. 1989) or EDCC (Lumsden et al. 1992) catalogs (Sect. 8, Table 3). For these clusters = 0.25 (63 -galaxies out of 256 galaxies). The number of members of these systems is not a direct measure of their richness because of the apparent magnitude limit of the catalog and because of the OPTOPUS mask. However, because they include all the richest systems in our catalog and because they are counterparts of 2-D clusters, it is resonable to assume that they are intrinsically rich. We remember here that Biviano et al. (1997) find = 0.21 for their sample of ENACS clusters.The fact that for ESP counterparts of clusters we find a lower value of than for the other rich groups ( = 0.36 without clusters), further supports the existence of a morphology-density relation over the whole range of densities from clusters to the "field".
Many systems of our catalog are not completely surveyed, therefore the relationship between and the density of the environment we find can only be considered qualitative. However, while incompleteness certainly increases the variance around the mean result, we do not expect severe systematic biases. In order to verify that incompleteness does not affect our results, we consider the subsample of 67 groups that contain no ESP objects without measured redshift. We obtain for this subsample the same relationship between group richness and fraction of emission line galaxies we find for the whole catalog.
7.3. Seyfert galaxies
Within ESP we identify 12 Seyfert 1 galaxies and 9 Seyfert 2 galaxies. We identify type 1 Seyferts visually on the basis of the presence of broad (FWHM of a few km s-1) components in the permitted lines. Our list is complete with the possible exception of objects with weak broad lines which are hidden in the noise.
The identification of type 2 Seyferts is not straightforward, because it is based on line ratios and usually requires measurements of emission lines which fall outside our spectral range: only the F([O III]5007)/F(H) ratio is available from our spectra, and it is therefore impossible to draw a complete diagnostic diagram (Baldwin et al. 1981, Veilleux & Osterbrock 1987). We classify tentatively as type 2 Seyferts all emission line galaxies with : this threshold cuts out almost all non-active emission line galaxies, but also many narrow-line AGN with a medium to low degree of ionization. Thus the list of possible Seyfert 2 galaxies is almost free of contamination, but should by no means be considered complete.
The origin of the Seyfert phenomenon could be related to the interaction with close companions (Balick & Heckman 1982, Petrosian 1982, Dahari 1984, MacKenty 1989), or to a dense environment (Kollatschny & Fricke 1989, De Robertis et al. 1998). Observational evidence is, however, far from conclusive. For example, Seyfert 1 and Seyfert 2 galaxies have been found to have an excess of (possibly) physical companions compared to other spiral galaxies by Rafanelli et al. (1995). Laurikainen & Salo (1995) agree with Rafanelli et al. (1995) about Seyfert 2 galaxies, but reach the opposite conclusion about Seyfert 1 galaxies.
In our case, 7 (33%) out of 21 Seyferts are group members. For comparison, 460 (34%) emission line galaxies (not including Seyfert galaxies) are group members and 879 are either isolated or binaries. Clearly, within the limits of our relatively poor statistics, we find that Seyfert galaxies do not prefer a different environment than that of the other emission line galaxies.
In order to test the dependence of the Seyfert phenomenon on the interaction of galaxies with close companions rather than with the general environment, we compute for all Seyferts and emission line galaxies the projected linear distance to their nearest neighbor, the nn-distance. We limit the search of companions to galaxies that are closer than 3000 km s-1 along the line of sight.
We find that the distribution of nn-distances of the sample of Seyfert galaxies is not significantly different from that of all emission line galaxies.
We also consider the frequency of companions at projected linear distances Mpc. We have 7 Seyfert galaxies with such a close companion (33%) and 315 (23%) emission lines galaxies. One of the 7 Seyferts is a member of a binary systems, the remaining six Seyferts are members of groups. Even if, taken at face value, the higher frequency of close companions observed among Seyfert galaxies supports a causal connection between gravitational interaction and the Seyfert phenomenon, these frequencies are not significantly different.
We note that members of close angular pairs ( arcsec) in the original target list for ESP, are more frequently missing from the redshift survey than other objects (Vettolani et al. 1998). This bias, due to OPTOPUS mechanical constraints, could hide a possible excess of physical companions of Seyfert galaxies.
In order to estimate how strongly our result could be affected by this observational bias, we identify the nearest neighbors of Seyfert and emission line galaxies from a list including both galaxies with redshift and objects that have not been observed. When we compute projected linear distances to objects that have not been observed, we assume that they are at the same redshift of their candidate companion galaxy. As before, we do not find significant differences between the new nn-distributions of Seyferts and -galaxies .
This result demonstrates that the higher average incompleteness of close angular pairs does not affect our main conclusions: a) Seyfert galaxies within ESP are found as frequently within groups as other emission line galaxies, b) Seyfert galaxies show a small but not significant excess of close physical companions relative to the other emission line galaxies. We point out again that the sample of Seyferts is rather small and the statistical uncertainties correspondingly large.
© European Southern Observatory (ESO) 1999
Online publication: December 22, 1998