EDP Sciences
Free Access
Issue
A&A
Volume 584, December 2015
Article Number A125
Number of page(s) 11
Section Extragalactic astronomy
DOI https://doi.org/10.1051/0004-6361/201527136
Published online 04 December 2015

© ESO, 2015

1. Introduction

The character of the galaxy assembly in the constellation of Antlia is not as clear as, for example, in the cases of the Virgo or Fornax galaxy clusters. Some authors used the term “galaxy group” (Ferguson & Sandage 1990), others “galaxy cluster” (Hopp & Materne 1985; Smith Castelli et al. 2008), and Hopp & Materne (1985) identified five “clusters” in the Antlia region (α = 10h–10h 50m, δ = −42°−30°). The most striking central cluster is called “Antlia II”. Its morphological appearance is that of two groups concentrated around the giant ellipticals NGC 3258 and NGC 3268 in a projected distance of 220 kpc.

Both dominant galaxies are extended X-ray sources (Nakazawa et al. 2000; Pedersen et al. 1997) that exhibit rich globular cluster systems. The globular cluster system of NGC 3258 contains about 6000 members. The system of NGC 3268 is somewhat poorer, but is still typical of a giant elliptical galaxy, with almost 5000 globular clusters (Bassino et al. 2008). Hopp & Materne (1985) and, more recently, Hess et al. (2015) estimated velocity dispersions of ~500 km s-1 for the bright population of Antlia, higher than those in clusters like Fornax (Drinkwater et al. 2001). The globular cluster luminosity function indicated that NGC 3258 could be a few Mpc nearer than NGC 3268 (Dirsch et al. 2003b; Bassino et al. 2008), which agrees with the distances obtained with surface brightness fluctuations by Blakeslee et al. (2001). However, Cantiello et al. (2005) quoted the same distance moduli for both galaxies.

Despite being nearby, a thorough radial velocity survey of Antlia’s galaxy population has not been done yet. In this paper, we present new radial velocities of galaxies located in Antlia. We supplement our galaxy sample with literature data, ending up with 105 galaxy velocities. These data will help us to better understanding the structure of Antlia.

In the following, we keep the term “Antlia cluster” for simplicity. To maintain consistency with earlier papers (e.g. Smith Castelli et al. 2012; Caso et al. 2013, 2014; Calderón et al. 2015), we adopt a distance of 35 Mpc, which means a scale of 169.7 pc/arcsec.

2. Observations and reductions

We performed multi-object spectroscopy with VLT-VIMOS of the galaxy population in six fields located in the inner part of the Antlia cluster. The observations were carried out under the programmes 60.A-9050(A) and 079.B-0480(B) (PI Tom Richtler), observed during the first semesters of 2007 and 2008. Figure 1 shows the four quadrants for each one of the six VIMOS fields, using different colours.

The grating was HR blue, and the slit width was 1′′. For each science field, the integration time was 1 h, split into three individual exposures. This configuration implied a wavelength coverage spanning 3700 Å–6600 Å (depending on the slit positions) and a spectral resolution of ~2.5 Å. The data were reduced with esorex in the usual manner for VIMOS data. First, a master bias was obtained for each field with the recipe VMBIAS from five individual bias exposures. The normalised master flat field was created with the recipe VMSPFLAT from a set of dome flat-field exposures. The recipe VMSPCALDISP was used to determine the wavelength calibrations and spectral distortions. Typically, more than 20 lines were identified for each slit. Afterwards, the bias and flat-field corrections were applied to each science exposure, together with the wavelength calibration. This was done with the recipe VMMOSOBSSTARE. Individual exposures were then combined with the iraf task IMCOMBINE to achieve a higher S/N. The spectra were extracted with the task APALL, also within iraf. We measured the heliocentric radial velocities using the iraf task FXCOR within the NOAO.RV package. We used synthetic templates, which were selected from the single stellar population (SSP) model spectra at the MILES library1 (Sánchez-Blázquez et al. 2006). We selected SSP models with the metallicities [M/H] = 0.71 and [M/H] = 0.4, a unimodal initial mass function with slope 1.30, and an age of 10 Gyr. The wavelength coverage of these templates is 3700 Å–6500 Å, and their spectral resolution is 3 Å FWHM.

thumbnail Fig. 1

Positions of the six VIMOS fields (each one composed of four quadrants). The black circle is centred on the midpoint between the projected positions for the two gEs, and its radius is 20′. North is up, east to the left.

Open with DEXTER

We also obtained GEMINI-GMOS multi-object spectra from programme GS-2013A-Q-37 (PI J. P. Calderón). The grating B600_G5303 blazed at 5000 Å was used with a slit width of 1 arcsec. The wavelength coverage spans 3300 Å–7200 Å, depending on the position of the slits. The data were reduced using the GEMINI.GMOS package within IRAF. We refer to Caso et al. (2014) for more information about the reduction.

We could determine heliocentric radial velocities (VR,h) for 67 galaxies located in the inner 20′ of the Antlia cluster (i.e., the inner 200 kpc for our adopted distance). In previous studies (Smith Castelli et al. 2008, 2012; Caso et al. 2013), those galaxies with VR,h between 1200 km s-1 and 4200 km s-1 have been assigned to Antlia (which already raised doubts owing to the large velocity interval). In our enhanced sample, we find the lowest velocity to be VR,h = 1150 km s-1, and there are no velocities between 4300 km s-1 and ~7600 km s-1. Galaxies with higher VR,h than the latter limit were rejected from our sample and are listed in Table A.1. It can be noticed that several galaxies from the Ferguson & Sandage (1990) catalogue are indeed in the background. In these cases, Ferguson & Sandage (1990) classified them as “likely members” or “probable background”. From the 67 galaxies with VR,h measurements, 45 are thus members of our sample.

Additional VR,h measurements were collected from the literature (Smith Castelli et al. 2008, 2012) and the NED2. The final sample of Antlia members consists of 105 galaxies, which are listed in Table A.2. A fraction of the present galaxies had been measured earlier. In these cases, we find good agreement between our measurements and the literature values (Fig. 2). The mean VR,h difference and dispersion are 20 km s-1 and 50 km s-1, respectively. In comparison, the uncertainties of our measurements are typically in the range 1040 km s-1.

thumbnail Fig. 2

Comparison between heliocentric radial velocity measurements from this paper and the literature for Antlia members.

Open with DEXTER

Ferguson & Sandage (1990) provided the photometrical catalogue of galaxies in Antlia with the largest spatial coverage, while Calderón et al. (2015) have carried out a deeper survey of the early-type galaxies in a region that contains our VIMOS fields. To supplement the Ferguson & Sandage (1990) catalogue, we derived B magnitudes for the galaxies measured by Calderón et al. (2015) in the Washington (C,T1) photometric system (Canterna 1976). As a result, we applied equation 4 from Smith Castelli et al. (2008) to transform (CT1)0 into (BR)0 colours. Then we obtained B magnitudes, considering that R and T1 filters only differ in a small offset (Dirsch et al. 2003b). Figure 3 shows the completeness for those galaxies located within 30′ of the cluster centre (see Sect. 3.2), which roughly matches the region observed with our VIMOS fields and previous spectroscopic studies (Smith Castelli et al. 2008, 2012). The bin width is 1 mag. When we exclude galaxies with Ferguson & Sandage (1990) membership status “3” (which are the less probable members), the 80% completeness is reached at BT = 17 mag and the 60% at BT = 19 mag (small red circles).

thumbnail Fig. 3

Completeness analysis for the spectroscopic sample considering the entire photometric catalogue (large light-blue circles), and excluding galaxies with Ferguson & Sandage (1990) membership status “3” (small red circles). The bin width is 1 mag.

Open with DEXTER

3. Results

3.1. VR,h velocity distribution

The histogram of the VR,h distribution for the galaxies in our Antlia sample is shown in Fig. 4 with a bin width of 150 km s-1. The green solid curve represents the smooth velocity distribution, obtained with a Gaussian kernel. The distribution resembles a Gaussian, which would be expected if the spatial velocity for the Antlia members is described by a Maxwellian distribution.

thumbnail Fig. 4

Histogram of the VR,h distribution for the galaxies in our Antlia sample. The bin width is 150 km s-1. Overplotted are the smooth velocity distribution, obtained with a Gaussian kernel (green solid curve), and the normal profile fitted by least squares (brown dashed curve).

Open with DEXTER

A Shapiro-Wilk normality test returns a p value of 0.9 (Shapiro & Wilk 1965; Royston 1995, hereafter S-W test), meaning that we cannot reject the hypothesis that our sample is drawn from a normal distribution. Considering this, we fitted a normal profile to the VR,h distribution by least-squares, assuming Poisson uncertainties for the bin values. The best fit corresponds to a Gaussian distribution with a mean velocity of 2708 ± 42 km s-1 and a dispersion of 617 ± 46 km s-1 (dashed brown curve in Fig. 4), which agree with the values from optical data by Hess et al. (2015). This dispersion fits massive clusters like Virgo (Conselice et al. 2001). Nakazawa et al. (2000) obtained a gas temperature of 2 keV for the surroundings of NGC 3268, which does not exclude a dispersion of 617 km s-1, considering the σ-T-relation of Xue & Wu (2000) and its scatter. However, whether this high velocity dispersion obtained for our Antlia sample represents a single virialized system, is an interesting question, which we try to answer here.

Assuming that the individual VR,h uncertainties are the sigma values of a normal distribution, we simulated the VR,h for each one of the 105 galaxies by a Monte-Carlo method. The procedure was repeated 100 times, in all cases obtaining a non-rejection of the normality hypothesis. This demonstrates that the intrinsic uncertainties of the measurements do not play a central role in the result.

thumbnail Fig. 5

Left panel: projected density distribution for galaxies, split into three percentile ranges. Middle panel: projected positions for bright galaxies. The colour palette ranges from red to blue, spanning the VR,h range defined in Sect. 2. Right panel: similar plot for dwarf galaxies in the sample. The black asterisks indicate the positions of NGC 3258 (south-west) and NGC 3268 (north–east). In both panels, the origin of the black circle is the bona fide centre of the cluster, and its radius is 30′.

Open with DEXTER

Several studies in galaxy clusters found different velocity dispersions for the dwarf and bright galaxies populations (e.g. Drinkwater et al. 2001; Conselice et al. 2001; Edwards et al. 2002). To test whether this is also true for the Antlia cluster, we subdivided our sample in giants and dwarfs using the morphological classification from Smith Castelli et al. (2008, 2012) and, when this is not available, from Ferguson & Sandage (1990). Normal distributions were fitted by least squares to the dwarf and bright galaxies, separately. The resulting mean values for the bright and dwarf populations were 2768 ± 104 km s-1 and 2698 ± 108 km s-1, respectively. The corresponding dispersions are 698 ± 122 km s-1 and 682 ± 120 km s-1. From a K-S test, we cannot reject the hypothesis that both samples are drawn from the same distribution with 90% confidence.

3.2. Morphological types and spatial distribution

Assuming the groups of galaxies dominated by the gEs as the main substructures of the cluster and that both haloes present similar masses (Pedersen et al. 1997; Nakazawa et al. 2000), we consider the midpoint between them as its approximate centre (α = 10h29m27s, δ = −35°27′58′′).

The 105 galaxies in our sample occupy a region of ~7 deg2. Therefore, we obtained the mean surface density, ~15 deg-2, and the mean distance between galaxies, ~9 arcmin. For each galaxy, we calculated the number of neighbours nearer than this value, and we used it to split the galaxies in four percentiles. The left-hand panel of Fig. 5 shows the regions occupied by the 25, 50, and 75 percentiles of galaxies with a larger number of neighbours. The white asteriks correspond to the position of NGC 3258 (south-west) and NGC 3268 (north-east). The origin of the black circle is the bona fide centre of the cluster, and its radius is 30′ (i.e. ~300 kpc at Antlia distance). The 25 percentile seems to clearly represent the overdensities of galaxies around both gEs, which are also identified in the 50th percentile. The lower percentile shows a more extended spatial distribution.

In the middle panel of Fig. 5, the projected spatial distribution for bright galaxies (i.e., all those galaxies not classified as dwarfs) is plotted. Different symbols indicate the Hubble type (i.e., spirals, lenticulars, or ellipticals). The colour palette ranges from red to blue, spanning the VR,h range defined in Sect. 2. The origin and radius of the black circle are identical to those in the left-hand panel. The two groups around each of the gEs can be clearly identified. There are also several galaxies located in the extrapolation of the straight line joining the two gEs, mainly to the north-east. These galaxies, as well as those located in the central part of the cluster, mainly present intermediate VR,h. There are a few galaxies with VR,h< 2000 [km s-1], whose projected positions agree with the general scheme of galaxies with intermediate velocities. However, the galaxies with the higher VR,h seem to present a different projected spatial distribution. The majority of them were classified as spirals, despite the relatively isolated elliptical FS90-152 (see Table A.2). On the other hand, the few other ellipticals have a projected distance to the centre of the cluster (dp) lower than 25′ and present 1775 <VR,h [km s-1 ] < 3000.

The right-hand panel of Fig. 5 shows the projected spatial distribution for dwarf galaxies in the sample and the positions of NGC 3258 and NGC 3268. The origin and radius of the black circle are identical to the previous panels. Their concentration within the region restricted by dp = 30′ is due to a bias in the spectroscopic survey. Therefore, we cannot arrive at any conclusion regarding their distribution when dp> 30′. The dwarfs near the centre present a wide range of VR,h. They seem to be more concentrated towards the two gEs and do not seem to be aligned with the projected position of the two gEs.

thumbnail Fig. 6

Heliocentric radial velocities (VR,h) as a function of projected distances to the galaxy centre (dp). Different symbols represent dwarf (red circles), ellipticals (blue squares), lenticulars (blue-violet triangles), and spirals (green diamonds). The dashed line indicates dp = 30′, and the dotted one represents the VR,h for both gEs, ~2800 km s-1. The grey regions indicate the caustic curves obtained with the Praton & Schneider (1994) infall model for two sets of parameters.

Open with DEXTER

Figure 6 shows the VR,h as a function of dp for the galaxies in the sample, discriminated by morphological types. The grey regions represent caustic curves obtained with the infall model of Praton & Schneider (1994), assuming Ω0 = 0.3 and two sets of parameters. For the inner region, we proposed typical parameters for the Fornax cluster rvir = 0.7 Mpc and σv = 374 km s-1 (Drinkwater et al. 2001). Considering the high mass derived for the Antlia cluster by Hess et al. (2015), we plotted the second region assuming the Virgo cluster virial radius, rvir = 1.8 Mpc (Kim et al. 2014) and σv = 617 km s-1, which were previously derived in Sect. 3.1. While the first set of parameters implies that a large number of galaxies lie outside of the caustic curves, for the second one all galaxies present VR,h-values lower than the escape velocity at their corresponding projected distances from the assumed centre of the cluster.

Dwarf galaxies seem to be spread all over the VR,h range for dp< 20′. Just a few dwarf galaxies outside this limit were measured, but all of them present intermediate values of VR,h. The picture is quite different when we look at the bright galaxy population. As expected, spiral galaxies are mainly at larger distances than 20′ from the bona fide cluster centre. Moreover, their velocity distribution also distinguishes them from the rest of the galaxies in the sample. From the thirteen spirals with dp> 30′, seven of them present VR,h around 3200 km s-1 and relatively low dispersion. In fact, the mean VR,h and its dispersion for these spirals are 3190 ± 40 km s-1 and 110 km s-1.

If we consider all the spirals with dp> 30′, their mean VR,h is 2980 ± 120 km s-1. For comparison, the eleven lenticulars in the same radial regime have a mean velocity of 2550 ± 125 km s-1. A Student’s test between the two samples rejects the hypothesis that both samples belong to the same population with 90% of confidence.

To study the central part of the cluster, which is expected to be dominated by the gEs haloes, we restricted our sample to include the 50th percentile of galaxies with larger number of neighbours. The upper panel of Fig. 7 shows the VR,h distribution for these galaxies, adopting a bin width of 180 km s-1. It seems to represent three groups of galaxies with VR,h around ~2000 km s-1, ~2800 km s-1, and ~3700 km s-1. To the last group belong the three bright lenticulars NGC 3267 (FS90-168), NGC 3269 (FS90-184), and NGC 3271 (FS90-224), near to NGC 3268 in projected distance. Considering the VR,h for both gEs, the galaxies in the group with intermediate velocities have the higher probability of belonging to the Antlia cluster. We applied a S-W test to this sample and obtained a p value of 0.32. Then, we ran GMM (Muratov & Gnedin 2010) for a trimodal case. GMM calculated three peaks with means of 2020 ± 160 km s-1, 2750 ± 100 km s-1, and 3650 ± 50 km s-1. For the unimodal hypothesis, we obtained χ2 = 13.9 with eight degrees of freedom (pχ2 = 0.1) and a distribution kurtosis k = −0.9. These results point to a multi-modal distribution. The parameter DD, which measures the relevance of the peak detections, was 2.8 ± 0.7 (DD> 2 is required for a meaningful detection; Ashman et al. 1994; Muratov & Gnedin 2010).

In the middle panel we present the VR,h distribution for galaxies up to the 75th percentile with the same bin width as in the previous plot. The three different groups also seem to be present, but smoothed. In this case, the p value obtained from the S-W test is higher, ~0.80. For the trimodal case, the mean from GMM were 2060 ± 200 km s-1, 2780 ± 100 km s-1 and 3600 ± 130 km s-1. In this case, χ2 = 7 (pχ2 = 0.5), k = −0.5, and DD = 2.5 ± 0.8. The parameters are less conclusive, but point to a multi-modal distribution in VR,h.

thumbnail Fig. 7

Upper panel: VR,h distribution for galaxies belonging to the 50 percentile with larger number of neighbours. Middle panel: VR,h distribution for galaxies belonging to the 75 percentile with larger number of neighbours. Lower panel: VR,h distribution for Antlia galaxies, separated by morphology.

Open with DEXTER

The lower panel of Fig. 7 shows the VR,h distribution for Antlia galaxies, separated between dwarf and bright galaxies, and the latter group between early-types (ellipticals and lenticulars) and late-types (spirals). Spirals fill a wide range of VR,h, but seem to avoid velocities similar to those of the gEs (~2800 km s-1).

3.3. Mass estimations

Pedersen et al. (1997) used X-ray observations (ASCA) to measure the mass enclosed within a radius of 240 kpc from NGC 3258 and obtained 0.9−2.4 × 1013M. NGC 3268 was studied in X-rays (ASCA) by Nakazawa et al. (2000), who determined ~2 × 1013M internal to ~260 kpc. Here we compare our galaxy velocities with the results from X-ray observations. We excluded spirals because of the differences that we found between their VR,h distribution and those of early-type bright galaxies in Sect. 3.2. Two subsets of our sample were made, selecting from the galaxies contained in the 50th percentile those with shorter projected distance to NGC 3258 or NGC 3268. These results in samples of 17 (up to 10 arcmin) and 24 members (up to 18 arcmin), respectively.

We applied the “tracer mass estimator” (MTr) from Evans et al. (2003), (1)where Ri and VLOS,i are the projected distances from the corresponding gE and velocities relative to it, respectively. Here, G is the constant of gravitation, and N the number of tracers. In the case of isotropy, the constant C is calculated through (2)We assume that the three-dimensional profile of the tracer population is represented well by a power law between an inner radius rin and outer radius rout with exponents γ and α, respectively (e.g. see Mamon & Łokas 2005a,b). Under the assumption of constant circular velocity, α = 0. In the case of the tracer population, it is difficult to measure the power law exponent precisely, but it should not differ too much from γ = 3 (Dekel et al. 2005; Agnello et al. 2014; Courteau et al. 2014). Taking this into account, we obtain the mass that corresponds to γ ranging from 2.75 to 3.25. In the case of a shallower profile, the resulting masses are somewhat lower, but not by an order of magnitude.

For the sample surrounding NGC 3258 we obtained MTr = (8–11) × 1013M. This has to be compared with the X-ray mass of NGC 3258. Pedersen et al. (1997) obtained kTe ≈ 1.7 keV, assuming an isothermal gas. The parameters of their beta model are rc = 8.2 arcmin, corresponding to 83.5 kpc, and β = 0.6.

With the assumption of spherical symmetry, we apply the expression (Grego et al. 2001) (3)Adopting μ = 0.6, the derived mass within the sphere with radius 10′ is ~7 × 1012M. In the case of the sample surrounding NGC 3268, the result from the Evans et al. (2003) estimator is MTr = (5–8) × 1013M.

We again applied Eq. (3) with the parameters derived by Nakazawa et al. (2000): kTe = 2 keV, rc = 5 arcmin, and β = 0.38. Then the mass enclosed within 18′ around NGC 3268 is ~1.4 × 1013M. In both cases, the mass derived from X-ray observations is significantly lower than the mass estimated from the VR,h measurements. It is unlikely that the mass enclosed within ~100 kpc around both gEs reaches values of a few times 1013M. We therefore conclude that the galaxies with the most deviant velocities in the two samples are probably not gravitationally bound to either of the two groups.

From the VR,h histograms in Fig. 7, a large number of the galaxies is symmetrically grouped around 2750–2800 [ km s-1 ], spanning approximately the velocity range 2200 <VR,h [ km s-1 ] < 3400. This group corresponds to ~50% of the galaxies that belong to the 50th percentile, and ~60% of the galaxies for the 75th percentile. This velocity constraint reduces the size of the samples around NGC 3258 and NGC 3268 to 8 and 13 members, respectively. Applying the mass estimators to these limited samples, we obtain, for the galaxies, around NGC 3258 and NGC 3268 MTr = (1.4−2) × 1013M up to 9′ and MTr = (1.7−2.4) × 1013M up to 15′, respectively.

These estimations are in reasonable agreement with the masses derived from X-ray observations. The mass that we derived from the analysis of an UCD sample around NGC 3268 is 2.7 × 1012M within 47 kpc. Assuming a constant circular velocity, the extrapolated mass within 101 kpc (corresponding to 10′) is 6 × 1012M, and therefore also is in reasonable agreement with our mass estimation from both companion galaxies and X-rays. To avoid the need to explain peculiar radial velocities of the order of 1500 km s-1, we conclude that the large velocity dispersion of our complete sample is caused by recessional velocities of galaxies in the near foreground or background, which are mixed in.

3.4. Tests for cluster substructure

In this section, we try a different approach. The nearest-neighbour tests are commonly used for detecting subgroups in the environment of clusters of galaxies (e.g. Boschin et al. 2010, 2006; Burgett et al. 2004; Owers et al. 2009; Hou et al. 2012). Our aim is to search for substructure in VR,h besides the obvious existence of the two groups dominated by NGC 3258 and NGC 3268.

The Δ test (Dressler & Shectman 1988) was intended for probing deviations from the local mean velocities and dispersions compared with the global cluster values. To achieve this, the local mean velocity and dispersion are calculated for each galaxy, restricting the sample to the galaxy itself and its ten nearest neighbours. Then, their deviation from the global cluster values are computed, and their sum is defined as the Δ parameter. Colless & Dunn (1996) proposed the κ test, similar in intent to the Δ test. The possibility that the scale in which the substructure is more obvious differs from what is imposed by the ten nearest neighbours condition is considered, leaving the number of neighbours n as a free parameter. Arguing that distributions cannot be characterised by the first two moments in general, they proposed a statistic ruled by the probability of the K-S two-sample distribution (Kolmogorov 1933; Smirnov 1948). For both tests, the significance of their statistics are estimated by Monte Carlo simulations, in which the velocities of the cluster galaxies are shuffled randomly. The significance level p was obtained after repeating the previous procedure 1000 times.

Pinkney et al. (1996) found that the existence of radial gradients in the velocity dispersion of galaxies could produce artificial substructure when applying 3D statistics. We therefore ran the tests on galaxies whose projected distances to the equidistant point to both gEs was less than 20′, because velocity dispersions seem to remain constant up to that limit (see Fig. 6). This restriction reduced our sample to 57 members. The area matches the field of view of our VIMOS survey, which guarantees a homogeneous sample. For the κ test, we selected n = 10,15. The results were Δ = 66.9 and pΔ = 0.12, κ10 = 14.4 and p10 = 0.16, κ15 = 15.6 and p15 = 0.06.

To investigate the significance of these numbers, we performed 1000 Monte Carlo simulations. We randomly generated the VR,h for the galaxies in the sample, assuming a normal distribution with mean and dispersion of 2800 km s-1 and 275 km s-1, respectively. Then, we applied the tests to these samples. For the Δ test, fewer than 15% of the cases produced a Δ-parameter higher than the value obtained from the observations, and this percentage is reduced to 6% for the κ15 test. Therefore, the tests point to underlying substructure, most probably due to the existence of non-member galaxies in our sample.

4. Discussion

4.1. Extreme radial velocities

To use radial velocities as a discriminant for cluster membership is often difficult if the depth along the line of sight is considerable. An extreme example is the interesting pair of Virgo galaxies IC 3492 with a radial velocity of 575 km s-1 and IC 3486 with 1903 km s-1, having a separation of only 1.4 arc min. The deviation from a strict Hubble law may be a mix of peculiar velocities resulting from large scale structure and local gravitational fields. Is it possible that the extreme radial velocities in Fig. 6 are infall velocities? To answer this, it is useful to consider the highest possible velocities. We can represent the total mass distribution around NGC 3268 quite well by an NFW-mass distribution with a scale length of 22 kpc and a characteristic density of 0.05 M/pc3. A mass probe on an exactly radial orbit, initially at rest and falling in from a distance of 1.5 Mpc, crosses the centre after 6.8 Gyr with a velocity of 1550 km s-1. However, after an additional 0.06 Gyr, its velocity has already declined to 1000 km s-1. It is obvious that such a configuration is highly artificial and not suitable to explaining the broad velocity interval. The more plausible explanation is therefore the mix of recessional and Doppler velocities.

4.2. Internal structure of the two dominant groups

What we called the Antlia cluster seems to be mainly formed by two subgroups, each one dominated by a gE. Despite that, both galaxies present similar VR,h, it is uncertain whether they are located at the same distance (e.g. Blakeslee et al. 2001; Cantiello et al. 2005; Bassino et al. 2008). A recent study in HI (Hess et al. 2015) indicates that the subgroup dominated by NGC 3258 could be falling in to NGC 3268, which they estimated is the actual cluster centre. It is clear that both gEs dominate their own subgroups, and their peculiar velocities relative to the group systemic velocity should not differ significantly. For this reason, if the difference in distance between both galaxies was 2–4 Mpc, we would expect a VR,h difference of ~140–280 km s-1. As a result, the measured VR,h might agree with the scenario suggested by Hess et al. (2015), where both subgroups are in a merging process.

It is interesting to have a closer look at the group around NGC 3268. Strikingly, all three S0’s around the central galaxy (FS90-224, FS90-168, and FS90-184) have positive velocity offsets in the range of 900–1000 km s-1. If these velocities were due to the potential of NGC 3268 (and its associated dark matter), very radial orbits would be needed for surpassing the circular velocity that is about 300 km s-1 and probably constant (Caso et al. 2014). The high velocity offsets are, moreover, related to fairly large projected spatial offsets (which in the case of FS90-224 is about 65 kpc), whereas the highest velocities are found at the bottom of the potential well. Furthermore, the galaxies do not show any morphological indication of being tidally disturbed. Moreover, there is no indication that the globular cluster system of NGC 3268 is tidally disturbed (Dirsch et al. 2003a), nor does the X-ray structure show any abnormality (Nakazawa et al. 2000). We therefore prefer the interpretation that the three S0s have to be placed in the background of NGC 3268. As Fig.7 suggests, more dwarf galaxies can be associated with this group, but some cases are ambiguous. There is the dwarf FS90-195 that hosts about ten point sources, which may be globular clusters. The radial velocity is 3495 km s-1, but it is located close to NGC 3268 (projected distance 15 kpc), so it may be bound and demonstrates how globular clusters can be donated to the system of NGC 3268. In that sense it could resemble SH2 in NGC 1316 (Richtler et al. 2012).

We note that NGC 3269 (FS90-184), together with NGC 3267 (FS90-168) and NGC 3271 (FS90-224), has been identified as the “Lyon Group of Galaxies (LGG)” 202 of Garcia (1993, 1995) and Barnes & Webster (2001), who already suspected its location behind the Antlia cluster.

Galaxies of the low velocity group are mostly dwarf-like galaxies and might be bona fide in the foreground. All look quite discy, which may also indicate a less dense environment. The situation is somewhat different for the group around NGC 3258. There are also dwarfs (ANTLJ102914-353923.6 and FS90-137) with positive velocity offsets of more than 1000 km s-1 to NGC 3258, but the brighter galaxies do not show such extreme radial velocities. There is also a group around FS90-226 that is less striking than the two dominant groups, which is also identified as a subgroup by Hess et al. (2015).

Therefore the likely explanation for Fig. 6 is that we are looking along a filament of galaxies and that at least a fraction of the galaxies in the three velocity groups apparent in Fig. 6 are separated by their recession velocities.

4.3. NGC 3268/3258 as fossil groups?

If the crowding of galaxies around NGC 3268 is mainly a projection effect, this galaxy fulfils the criteria for being a fossil group. After applying the definition of Jones et al. (2003), it has the required X-ray luminosity of more than 1042 erg s-1, and the next galaxy in a luminosity ranking would be FS90-177, which is fainter than 2 mag in the R-band.

The brighter galaxies around NGC 3258 (FS90-105 and FS90-125) may be companions. FS90-125 is perhaps a magnitude fainter than NGC 3258, so only the X-ray criterium applies, and it cannot be called a fossil group. However, its globular cluster system is even richer than that of NGC 3268, so the idea that a galaxy group collapsed at very early times is plausible.

4.4. Emission line galaxies

Only three galaxies from our VIMOS sample present emission lines. One of them, FS90-131, was classified as a spiral. The other galaxies, FS90-220 and FS90-222, were classified as lenticulars. Evidence of strong star formation would not be expected if the environment of the Antlia gEs was dense: FS90-131 is located at 5′ from NGC 3258, FS90-220 is ~6′ from NGC 3268, and FS90-222 is ~3′ from NGC 3273. Moreover, the VR,h in the three cases is lower than 1200 km s-1, which disqualifies them from dynamically belonging to these groups. Their velocities are better explained by recession velocities.

5. Summary and conclusions

We have presented new radial velocities, measured with VLT-VIMOS and Gemini GMOS-S, for galaxies in the region normally addressed as the Antlia cluster. The fields are located in the surroundings of NGC 3258 and NGC 3268, the dominant galaxies of the cluster. Together with the literature data, our list of Antlia galaxies with measured radial velocities now embraces 105 galaxies. Large numbers of these objects are projected onto either of the two subgroups around NGC 3258 or NGC 3268. Because the gravitational potentials of these galaxies are constrained by X-ray studies and stellar dynamical tracers, we could compare the observed radial velocities with those expected. It turned out that the total range of velocities seems to be too high to be generated by the gravitational potential of NGC 3258 or NGC 3268. There are three groups of galaxies (not always clearly separated) characterised by radial velocities of about 1800 km s-1, 2700 km s-1, and 3700 km s-1. We interpreted these velocities as recession velocities, which places the bright S0s around NGC 3268, in particular, into the background. Therefore, NGC 3268 might qualify as a fossil group, resembling, for example, NGC 4636 in the Virgo cluster in its properties. The intermediate velocity group (i.e. VR,h around 2700 km s-1) is the most populated one. The distances between NGC 3258 and NGC 3268 are not well constrained in the literature, and we cannot discard that the groups dominated by them are in the early stages of a merging process.

What we originally called the Antlia cluster is therefore characterised by several groups that, at least in some cases, are not gravitationally bound. We may be looking along a filament of the cosmic web.


2

This research has made use of the NASA/IPAC Extragalactic Database (NED), which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

Acknowledgments

We thank Ylva Schuberth for discussions about the radial velocities of Virgo galaxies, Mike Fellhauer for permission to use his orbit program, Francisco Azpillicueta for discussions of statistical issues, and Lilia Bassino for discussions about the Antlia cluster. We thank the refereee for suggestions that improved this article. This work was based on observations made with ESO telescopes at the La Silla Paranal Observatory under programmes ID 60.A-905 and ID 079.B-0480, and on observations obtained at the Gemini Observatory, which is operated by the Association of Universities for Research in Astronomy, Inc., under a cooperative agreement with the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), the National Research Council (Canada), CONICYT (Chile), the Australian Research Council (Australia), Ministério da Ciência, Tecnologia e Inovação (Brazil) and Ministerio de Ciencia, Tecnología e Innovación Productiva (Argentina), and on observations acquired through the Gemini Science Archive. This research has made use of the NASA/IPAC Extragalactic Database (NED), which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. This work was funded with grants from Consejo Nacional de Investigaciones Científicas y Técnicas de la República Argentina, Agencia Nacional de Promoción Científica y Tecnológica (BID AR PICT-2013-0317), and Universidad Nacional de La Plata (Argentina). T.R. is grateful for financial support from FONDECYT project Nr. 1100620, and from the BASAL Centro de Astrofísica y Tecnologías Afines (CATA) PFB-06/2007.

References

Online material

Appendix A: Additional tables

Table A.1

Heliocentric radial velocities for background galaxies measured in this paper.

Table A.2

Heliocentric radial velocity for Antlia members up to date.

All Tables

Table A.1

Heliocentric radial velocities for background galaxies measured in this paper.

Table A.2

Heliocentric radial velocity for Antlia members up to date.

All Figures

thumbnail Fig. 1

Positions of the six VIMOS fields (each one composed of four quadrants). The black circle is centred on the midpoint between the projected positions for the two gEs, and its radius is 20′. North is up, east to the left.

Open with DEXTER
In the text
thumbnail Fig. 2

Comparison between heliocentric radial velocity measurements from this paper and the literature for Antlia members.

Open with DEXTER
In the text
thumbnail Fig. 3

Completeness analysis for the spectroscopic sample considering the entire photometric catalogue (large light-blue circles), and excluding galaxies with Ferguson & Sandage (1990) membership status “3” (small red circles). The bin width is 1 mag.

Open with DEXTER
In the text
thumbnail Fig. 4

Histogram of the VR,h distribution for the galaxies in our Antlia sample. The bin width is 150 km s-1. Overplotted are the smooth velocity distribution, obtained with a Gaussian kernel (green solid curve), and the normal profile fitted by least squares (brown dashed curve).

Open with DEXTER
In the text
thumbnail Fig. 5

Left panel: projected density distribution for galaxies, split into three percentile ranges. Middle panel: projected positions for bright galaxies. The colour palette ranges from red to blue, spanning the VR,h range defined in Sect. 2. Right panel: similar plot for dwarf galaxies in the sample. The black asterisks indicate the positions of NGC 3258 (south-west) and NGC 3268 (north–east). In both panels, the origin of the black circle is the bona fide centre of the cluster, and its radius is 30′.

Open with DEXTER
In the text
thumbnail Fig. 6

Heliocentric radial velocities (VR,h) as a function of projected distances to the galaxy centre (dp). Different symbols represent dwarf (red circles), ellipticals (blue squares), lenticulars (blue-violet triangles), and spirals (green diamonds). The dashed line indicates dp = 30′, and the dotted one represents the VR,h for both gEs, ~2800 km s-1. The grey regions indicate the caustic curves obtained with the Praton & Schneider (1994) infall model for two sets of parameters.

Open with DEXTER
In the text
thumbnail Fig. 7

Upper panel: VR,h distribution for galaxies belonging to the 50 percentile with larger number of neighbours. Middle panel: VR,h distribution for galaxies belonging to the 75 percentile with larger number of neighbours. Lower panel: VR,h distribution for Antlia galaxies, separated by morphology.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.