HOT AND COLD GALACTIC GAS IN THE NGC 2563 GALAXY GROUP

The distribution of group members in NGC 2563 allows most of them to be observed with 14 ACIS-I pointings: a 3 × 3 central mosaic to cover the inner 45' × 45' of the group and five outer pointings. A total of 54 of the 64 confirmed group members were observed by Chandra. The majority of the remaining galaxies are optically faint and therefore not likely to be strong X-ray sources. The 14 pointings were done between 2007 June and 2008 March for 30 ks each, except for the central pointing which was observed for 50 ks to achieve a similar contrast for galaxies in the X-ray bright central region. All observations were conducted in VFAINT telemetry mode. The level 1 event files were processed with ciao¹¹ v4.0 and CALDB¹² v3.4.3, filtering out bad pixels and applying the latest gain files. Events with ASCA grades¹³ 1, 5, and 7, and a non-zero status flag were discarded. Finally, background flares were screened for using the routine "lc_clean.sl" in ciao, excluding time periods with background count rates greater than 20% of the quiescent rate. Flaring was not a problem for any of our observations.

Source detection was performed using the "Mexican Hat" wavelet source detection algorithm "wavdetect" in ciao, with scales of 1, 2, 4, 8, 16, and 32 pixels, using soft (0.3–2.0 keV), hard (2.0–7.0 keV), and "full" band (0.3–7 keV) event lists. The detection threshold was set to limit the number of false detections to ≈4 per CCD. As summarized in Table 1, a total of 17 confirmed group members were detected, including the largest elliptical, NGC 2563 itself. This galaxy is coincident with the group center as defined by the peak of the diffuse group X-ray emission. As it is not straightforward to cleanly separate the emission of this galaxy from that of the intragroup medium, the central galaxy is treated as a special case for the remainder of the paper.

For all detected group members, source spectra and associated response files were extracted within circular regions extending to where the galaxy surface brightness becomes consistent with the local background level. Background spectra were extracted in surrounding annuli, and results were fitted in the 0.3–7.0 keV band using xspec v. 12.3 (Arnaud 1996). As many of the sources have few counts, the maximum likelihood-based C-statistic (Cash 1979) was used to determine best-fit parameters. The goodness of fit was verified using the χ² statistic for all sources with sufficient photon statistics to allow meaningful constraints from spectra accumulated in bins of at least 20 net counts.

The X-ray emission in luminous early types is generally dominated by that of hot gas and low-mass X-ray binaries (e.g., Fabbiano 2006). We modeled any hot gas emission in the group members using the mekal optically thin thermal plasma model (Mewe et al. 1985; Liedahl et al. 1995) in xspec, assuming solar abundance ratios. Any X-ray binary/active galactic nucleus (AGN) component was modeled with a power law. In all fits, the absorbing hydrogen column density was fixed at the Galactic value (N_H = 4.0 × 10²⁰ cm⁻²; Kalberla et al. 2005), since the limited photon statistics generally precluded useful constraints on N_H from spectral fitting itself. Following Sun et al. (2007) and Jeltema et al. (2008), the metallicity in the mekal model was fixed at 0.8 Z_☉, the mean value found for galactic hot gas in the large cluster galaxy sample of Sun et al. (2007). This left four free model parameters: Gas temperature T, power-law index Γ, and two normalizations. However, due to limited number of counts, the 3σ uncertainties on T and Γ were unconstrained for most of the galaxies. In such cases we first fixed Γ at 1.7, as is typical for X-ray binary and AGN spectra, and, if necessary, also T at 0.7 keV. Based on the photon statistics and spectral results, the detected galaxies were then divided into three categories, as summarized in Column 12 of Table 1.

• 1.  
  Galaxies for which the normalization of both components was well constrained at the 2σ level (three galaxies in total; well constrained here means that the 2σ uncertainties are finite, and that the normalization is non-zero within those uncertainties). For the central galaxy NGC 2563, all of the four parameters are well constrained, but its emission is not easily deblended from that of the intragroup medium. The other two galaxies are NGC 2557, where Γ was fixed, and IC 2339, where also T was fixed. Best-fit values and 1σ errors for all fit parameters were determined using Markov Chain Monte Carlo (MCMC) simulations in xspec. The values reported were obtained from the median value of the chain. We note for completeness that the standard deviation leading to a parameter being here considered well constrained (which is based on the change in the adopted fit statistic when the parameter is varied, using xspec's standard "error" command) is not necessarily identical to that resulting from the MCMC simulations.
• 2.  
  Galaxies with ≳ 30 counts but for which the normalization of one or both components remained unconstrained within the 2σ uncertainties (six galaxies in total). In the cases where the 2σ lower limit on the thermal luminosity L_{X, th} was consistent with zero, whereas that on the power-law component was not, the spectrum was assumed to be consistent with a power law only. IC 2338 is consistent with displaying both a thermal and a power-law component, but, unlike galaxies in category (1), its 2σ upper limit on L_{X, th} remained unconstrained. Using MCMC, we report the corresponding 1σ upper limit. For the remaining galaxies in this category, the best-fit Γ was consistent with ≈1.7 as expected from X-ray binaries or an AGN, whereas the best-fit temperature was much higher (T > 4 keV) than expected for hot galactic gas. Hence, a power-law component likely dominates their X-ray output. For these sources, we fixed the best-fit power-law model and added a T = 0.7 keV thermal component. The limit on L_{X, th} was determined from the 1σ upper limit of the normalization of the mekal model.
• 3.  
  Galaxies detected with ≲ 30 net counts (eight in total). In these cases, statistics are insufficient to allow a robust determination of the nature of their X-ray emission. A 1σ upper limit to their L_{X, th} was estimated assuming a T = 0.7 keV mekal model, but these galaxies are labeled with "U" (for unknown spectral composition) in Column 12 of Table 1. To help clarify the nature of these sources, we also considered their hardness ratios HR (here the 2–7 keV/0.3–2 keV flux ratio). This ratio is 0.02–0.05 for a mekal model with T = 0.7–1.0 keV, and 0.35–0.47 for a power-law spectrum with Γ = 1.7–2.0. For 0.05 < HR < 0.35, the dominant component is ambiguous, but some contribution from a thermal component is allowed. Although these classifications are only tentative, given the Poisson uncertainties on HR, we do note that all group members spectroscopically confirmed to require only a power law show HR > 0.4, whereas HR ≲ 0.35 for the galaxies with a spectroscopically identified thermal component. In Table 1, a "T" is listed after "U" in Column 12 if HR < 0.35 to indicate that the galaxy may contain thermal emission. The remaining galaxies are labeled with either "TP," meaning that both thermal and power-law components are likely present (0.05 < HR < 0.35), or "P," indicating consistency with a power law only (HR > 0.35).

To further aid in identifying the nature of galactic X-ray emission, we used its spatial extent to help discriminate between nuclear (AGN) and galaxy-wide emission (e.g., hot gas halos, X-ray binaries). For each detected group member, we fitted the 0.5–2 keV brightness profile with a Gaussian, and classified it as extended if the 1σ lower bound of its full width at half-maximum (FWHM) was ⩾10% larger than that of the local point-spread function (PSF) extracted at peak source energy. Overall, most of the detected group members appear extended in X-rays, and these are labeled with an "E" in Column 13 of Table 1, or with a "P" (point-like) otherwise. For most sources with ≲ 30 counts the extent remains poorly constrained, however, and Column 13 then includes a question mark.

For the X-ray-undetected group members, we followed the procedure of Sun et al. (2007) and Jeltema et al. (2008) to estimate luminosity upper limits. Count rates were measured within a circular aperture of R = 3 kpc from the optical galaxy center, in a few cases expanded to include the 90% encircled energy radius of the local PSF at E = 1 keV to account for PSF smearing. The upper limit on L_{X, th} was derived from the Poisson 3σ upper limit on the count rate (Gehrels 1986) assuming a T = 0.7 keV mekal model.

The H i analysis was based on two different observing runs at the VLA in its most compact, 1 km (D-array), configuration. The first run, done in 1999, covered the entire group with a 6 × 6 point mosaic. Individual pointings were separated by 15', fully sampling the primary beam with its FWHM of 30'. To probe as wide a velocity range as possible with sufficient velocity resolution, we used a total bandwidth of 3.125 MHz and four intermediate frequency (IF) channels with two frequency settings and slightly overlapping bands. The total velocity range covered was ∼1100 km s⁻¹ with a resolution of 20 km s⁻¹. The data were calibrated using standard AIPS¹⁴ procedures, and then imported into MIRIAD¹⁵ for mosaicking and joint deconvolution, turning the individual pointings into one mosaicked cube for each IF. The continuum was then subtracted from the u–v data using a fit through the line-free channels identified within the cube. However, inspection of the final cube showed that our velocity coverage was insufficient, with several group members detected close to the edge of the band and their H i velocity range only partly probed.

To remedy this, NGC 2563 was reobserved in 2007, using a 14-point mosaic that covered an area identical to that of our Chandra mosaic. To also cover a larger velocity range, we used a total bandwidth of 6.25 MHz with two IFs and no online Hanning smoothing. Each pointing was observed at two separate frequencies, resulting in a total velocity coverage of 2100 km s⁻¹. Each pointing and velocity setting was calibrated separately, and cubes were made using AIPS. Overlapping frequencies were then combined in the image plane, resulting in a 101-channel cube per pointing. After identifying channels containing H i, the continuum was iteratively subtracted from each cube using a linear fit in the image plane through the line-free channels. The resulting cubes were then CLEANed, combined into one, and the output corrected for the primary beam response; where cubes overlapped, they were averaged with weighting that accounted for the location within the pointing and for the distance from the center of each field. Although interference generated by the VLA rendered data unusable within a narrow velocity range (4331–4374 km s⁻¹), the only group member affected is an Sm dwarf with L_K ≈ 1 × 10⁹ L_☉, so this has little impact on our results.

Instrumental parameters and specifics of the cubes for both data sets are summarized in Table 2, including their velocity ranges (heliocentric, optical definition). Since complete velocity coverage is important here, the 36-point mosaic was used exclusively to search for H i emission, combining it with the central part of the 14-point mosaic. All remaining H i analysis employed the 14-point mosaic only. This was searched for neutral hydrogen using three different methods. First, the 101-channel cube was divided into groups of 25, smoothed spatially to half resolution, and Hanning-smoothed in velocity. This was then used as a mask for the full-resolution cube, blanking all pixels below 2σ in the smoothed cube and summing over 25 channels. Next, we searched the cube by eye, stepping through the velocity channels at different speeds and comparing our detections with the first method. Finally, we plotted spectra integrated over a few beams at the location of the optically identified group members. This did not give any additional detections, so our search was essentially an optically blind one.

Figure 1 shows a gray-scale image of a typical channel in the cube to illustrate the spatial noise distribution and the position of the group members. In the central square degree (R ≲ 600 kpc ≈0.7R_vir) the rms noise (using robust weighting) is remarkably uniform at 0.6 mJy beam⁻¹, equivalent to N_H ≃ 4 × 10¹⁸ cm⁻². Assuming 3σ over five channels, the corresponding sensitivity is $M_{\rm {\rm H}\,{\mathsc{i}}} = 2 \times 10^8$ M_☉. Along the 9' wide outer edges, the noise and sensitivity rise to 3.0 mJy beam⁻¹ and 1 × 10⁹ M_☉, respectively, due to the correction for the primary beam response. This only affects five out of the 64 group members and none of our conclusions. A total noise-free H i image was produced by again using the smoothed cube as a mask and summing each detected galaxy over the narrow velocity range in which H i was detected. The total H i mass for each galaxy was then determined by summing the flux in individual channels in a box centered on the galaxy. The resulting typical uncertainty is 10% for the galaxies with $M_{\rm H\,\mathsc{i}}$ above a few times 10⁹ M_☉, and 2 × 10⁸ M_☉ for the remainder.

Zoom In Zoom Out Reset image size

Table 1 lists all 17 X-ray-detected group members (of the 54 observed), and whether these show evidence for star formation or AGN activity. This is based on their location in a BPT diagram (Baldwin et al. 1981; Kewley et al. 2001; Kauffmann et al. 2003a), as inferred from an emission-line analysis of their SDSS spectra, with optical line fluxes measured using the code described in Tremonti et al. (2004).

Figure 2 shows a 0.5–2 keV mosaic image of the group from our 14 Chandra pointings, adaptively smoothed using the "csmooth" algorithm, with signal significance set between 3σ–5σ. To produce this, spectrally weighted exposure maps were smoothed to the same spatial scales, and the exposure-corrected images combined into one. The impact of edge effects was reduced by removing regions with (smoothed) exposure values below 3% of the maximum, but small enhancements at the boundaries of adjacent pointings still remain. Detected group galaxies are marked with squares in the figure, with interlopers accounting for the remaining X-ray sources.

Zoom In Zoom Out Reset image size

Due to the spatial variation in Chandra effective area, our X-ray detection limits are not uniform across the field, but we reach a 3σ upper limit in the 0.5–2.0 keV band of L_X ≈ 3 × 10³⁹ erg s⁻¹ or better for all of the observed group members. These span almost three orders of magnitude in L_K, with nearly all galaxies above 0.5L* in the g or K band detected in X-rays (7/8 in K, 15/16 in g), whereas the corresponding fraction is less than half (4/9) for those fainter than 0.2L*_K. This is qualitatively consistent with other studies which show a strong correlation between galaxy optical and X-ray luminosity (Fabbiano et al. 1992; Read & Ponman 2001; O'Sullivan et al. 2001, 2003). As noted earlier, the 10 members not observed by Chandra are mostly optically faint, with seven below 0.1L*_K and only one above L*_K.

Among the nine galaxies for which X-ray spectral fitting provides useful constraints (cases 1 and 2 in Section 3.1), four show evidence for a thermal component, including the central NGC 2563 itself. From the hardness ratios of the remaining (case 3) galaxies, two additional objects show evidence for thermal emission for a total of six such galaxies. However, recalling from Section 3.1 that HR < 0.05 for a thermal-only model, we note that no source besides NGC 2563 itself has a hardness ratio HR < 0.2, so a power-law component is likely present (if not dominant) in all the X-ray-detected galaxies. In all group members with more than 40 counts, the emission is inferred to be spatially extended, and their smoothed 0.5–2 keV contours overlaid on SDSS r-band images are shown in Figure 3. Note the offset between the X-ray peak and the optical center in NGC 2560, possibly indicating the presence of an ultra-luminous X-ray source as seen in other nearby galaxies (Miller & Colbert 2004; Roberts 2007).

Zoom In Zoom Out Reset image size

Previous studies have revealed a correlation between L_K and thermal X-ray luminosity for early-type galaxies in all environments (O'Sullivan et al. 2001; Sun et al. 2007; Jeltema et al. 2008; Mulchaey & Jeltema 2010). Taking early types to be Sa or earlier for consistency with these works, then roughly half of the members in NGC 2563 belong to this category. Figure 4(a) compares L_{X, th} for our optically bright group members to the scaling relation derived by Jeltema et al. (2008) for early types in groups. The central galaxy NGC 2563 itself is excluded here for reasons discussed earlier. Our two other early types with evidence for a thermal component (NGC 2557 and NGC 2569) are consistent with the Jeltema et al. (2008) relation, while the majority of the X-ray-undetected galaxies are too optically faint to offer much further insight with the present X-ray detection limits. With the limited statistics, we simply conclude that the detected early types in NGC 2563 do not deviate significantly from the L_X–L_K relation found for other group ellipticals.

Zoom In Zoom Out Reset image size

To further elucidate the potential hot gas contribution to the total L_X of our early types, Figure 4(b) compares the 0.3–8 keV L_X of the X-ray-detected group members to that expected from low-mass X-ray binaries using the scaling relation of Kim & Fabbiano (2004). NGC 2563 itself is again excluded. None of our early types shows evidence of a significant X-ray excess relative to this expectation, suggesting that the majority of their X-ray output is of stellar origin. A few of the more luminous galaxies even lie below this relation, including NGC 2557 which does contain a thermal component.

Of the eight X-ray-detected late-type members, three show evidence for a thermal component. All eight galaxies have L_g > 0.5L*_g but span a relatively larger range in L_K, suggesting their total X-ray output is more directly related to the blue stellar light than to stellar mass. We note the similarity to other studies of late-type galaxies that find their diffuse L_X to correlate more strongly with L_B (and hence star formation rate) than with L_K or stellar mass (Tüllmann et al. 2006; Sun et al. 2007). Our SDSS spectra generally support this, with seven of our eight X-ray-detected late-type members showing optical line emission consistent with star formation. The eighth galaxy (NGC 2558) has line ratios consistent with a LINER. Based on the BPT diagram, there are 12 other actively star-forming late types in the group, but all have L_g  <  0.5L*_g and remain X-ray undetected.

We detect H i in 20 of the 64 group members down to our H i mass limit of ∼2 × 10⁸ M_☉ (recall that our limits are higher for galaxies at the very edge of the VLA mosaic). The results for these 20 objects are summarized in Table 3, along with results for three large spirals that remain undetected in H i but whose H i deficiencies can still be estimated (see below). Quoted H i velocity widths correspond to the velocity range of channels with signal at the ⩾2σ level; note that for UGC 04332 the H i profile extends to the edge of our velocity coverage, so the H i width and mass provided in Table 3 are lower limits. Two of the 20 H i detections are associated with galaxies not previously identified as group members from optical spectroscopy (SDSSJ082044.60+210715.0 and SDSSJ081931.17+203843.9). Both of these represent optically faint irregular galaxies.

Table 3. H i Detections and Deficiencies

Galaxy	Morphology	R	H i width	H i Velocity	Optical Velocity	Log ($M_{\rm H\,\mathsc{i}}$)	Def$_{\rm H\,\mathsc{i}}$
		(kpc)	(km s−1)	(km s−1)	(km s−1)	(M☉)
SDSSJ082044.60+210715.0	Irr	73	42	4406a	...	8.28	...
CGCG119−061	Sab	112	...	...	5235	<8.60b	>+0.56
CGCG119−059	Sc	165	170	4215	4211	8.73	+0.39
UGC 04344	Sc	238	149	5033	5041	9.89	−0.10
SDSSJ081938.81+210353.0	Irr	254	43	4236	4240	8.48	...
UGC 04332	Sa pec	264	>449c	5481	5514	>9.41c	< + 0.04c
CGCG119−053	Sa pec	339	171	4852	4877	9.26	−0.17
UGC 04329	Sc pec	439	254	4109	4099	9.98	+0.10
SDSSJ081905.46+211448.0	Sm?	451	106	4842	4844	8.51	...
IC 2293	SBbc	470	212	4088	4094	8.91	+0.55
CGCG119−043	Sc pec	506	170	4470	4458	8.65	+0.47
CGCG119−051	Sbd	513	191	5033	5028	9.32	−0.02
SDSSJ081931.17+203843.9	Irr	564	85	4980a	...	8.81	...
CGCG119−040	Sa	663	21	4841	4816	7.78	+1.23
UGC04324	Sabd	664	340	4831	4814	9.20	+0.32
SDSSJ081904.25+213521.0	Irr	724	106	4863	4865	8.95	...
NGC 2558	Sb	739	405	4990	4998	10.08	−0.23
IC 2338	SBcdd	848	313e	5413e	5400	9.76e	−0.34e
IC 2339	SBc pec	858	313e	5413e	5420	9.76e	−0.34e
CGCG119−082	SBa	898	...	...	4783	<9.00b	>+0.09
UGC 04386	Sab pec	921	...	...	4640	<9.00b	>+0.61
CGCG119−047	Sabd	923	276	4502	4506	9.73	−0.45
SDSSJ082352.25+212507.2	Sc	964	171	5086	5095	8.85	+0.45

Notes. ^aNew H i group member. ^bH i undetected. ^cIncomplete H i velocity coverage. ^dH i tail. ^eClosely interacting pair. Values apply for the entire system.

Download table as:  ASCIITypeset image

H i is detected only in galaxies of types Sa and later, including in all 15 such galaxies with L_K > 0.1L*_K. Figure 5 shows contours from our noise-free H i mosaic overlaid on the Chandra mosaic of the full group, and Figure 6 displays an H i/optical overlay for each H i-detected galaxy. The latter figure reveals four significant H i extensions in the group associated with six galaxies (IC 2238/IC 2339, UGC 04324/CGCG119-040, CGCG119-047, and CGCG119-051, all labeled with an asterisk in the figure). This implies that ∼30% of our H i-detected group members display H i evidence for being involved in an ongoing interaction. The galaxies IC 2338 and IC 2339 represent a particularly close pair, and the H i quantities quoted in Table 3 therefore apply for the entire system. The absence of detectable H i among the early-type group members is consistent with previous studies finding H i above our detection limits in only a small fraction of early types (Burstein et al. 1987; Morganti et al. 2006; di Serego Alighieri et al. 2007). Since we have performed an optically blind search for H i across the entire group, we can furthermore rule out the existence of any optically dark H i clouds not associated with any galaxy down to our H i mass limit.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To investigate to what extent the group members may be deficient in H i, we estimated H i deficiencies Def$_{\rm H\,\mathsc{i}}$ following the usual prescription,

$$\begin{equation} \mbox{Def}_{\rm H\,\mathsc{i}} = \mbox{log}\,(M_{\rm H\,\mathsc{i}})_{\rm expected} - \mbox{log}\,(M_{\rm H\,\mathsc{i}})_{\rm observed}. \end{equation} \tag{ 1 }$$

The expected H i masses were computed from the best-fit relationship between total H i mass and galaxy optical diameter D derived for galaxies in the field (Solanes et al. 1996). D is here defined as the galaxy major axis measured from the Palomar Observatory Sky Survey blue prints; where not already available from the Uppsala General Catalog of Galaxies (Nilson 1973), we measured D directly from these prints. As the Solanes et al. (1996) field sample was restricted to rather large spirals of types Sa through Sc, only group members of those types and with D > 10 kpc were considered here. Note, as mentioned above, that this includes three galaxies not detected in H i, and these are also listed in Table 3. For the IC 2338/2339 pair, a single deficiency was calculated from the total observed and expected H i masses given the optical sizes of both galaxies.

The resulting deficiencies listed in Table 3 span a wide range, but given the uncertainties in D and optical morphology, as well as the dispersion in the H i masses of field galaxies, objects with observed H i masses within a factor of a few of the expected values cannot be considered deficient. However, there are six galaxies with Def$_{\rm H\,\mathsc{i}} \gtrsim 0.45$, consistent with an H i deficiency of at least a factor of three. One of these is UGC 04386, but Def$_{\rm H\,\mathsc{i}}$ for this object is not well established, in part because it resides at the eastern edge of the H i mosaic where the data are noisier, and in part because it is viewed edge-on and is of uncertain morphological type.

Although our main goal is to study the ISM in individual group members, it is also relevant to establish which galaxies are embedded within detectable intragroup gas. To do so, a radial surface brightness profile was extracted from the unsmoothed exposure-corrected Chandra mosaic, centered on NGC 2563 itself, and with all other bright sources masked out. The background level was estimated from the four outermost pointings, which are all centered at R = 40'–45' from the X-ray peak and show consistent full-chip count rates. The resulting background-subtracted profile, shown in Figure 7, reveals diffuse emission out to at least R ≈ 21' (≈400 kpc in projection). This is a few arcminutes further than the maximum extent determined from ROSAT observations (Mulchaey et al. 2003; Osmond & Ponman 2004), and roughly corresponds to our estimate of R₅₀₀  ≈  420 kpc. Modeling the profile as a sum of two β-models yields β₁ = 0.61^+0.09_−0.03, r_c1 = 1.0 ± 0.1 kpc, and β₂ = 0.32 ± 0.01, r_c2 = 1.1^+1.3_−1.0 kpc for the inner and outer components, respectively. With a reduced χ² = 3.0, the fit is poor however, in part because the profile steepens beyond that of a β-model at large radii, as also seen in other groups and clusters (Vikhlinin et al. 2006; Sun et al. 2009).

Zoom In Zoom Out Reset image size

The relatively high central surface brightness could potentially compromise our ability to detect individual galaxies in the group core. We have attempted to circumvent this issue by requiring a longer central Chandra exposure (50 ks). In addition, the surface brightness has dropped from its central value by three orders of magnitude already at R = 5' ≈ 100 kpc. Apart from the central galaxy, there are no optically luminous group members within the central few arcmin, so our X-ray detection limits are impacted minimally by the presence of intragroup gas. This is corroborated by the fact that we find no systematic radial variation in limiting L_{X, th} for the 54 group members covered by Chandra. A detailed analysis of the intragroup medium in NGC 2563 will be the subject of future work.

The detection of X-ray tails with Chandra has provided direct evidence for ram pressure stripping in both groups (Rasmussen et al. 2006; Jeltema et al. 2008) and clusters (Sun et al. 2007), and similar inferences have been made from the presence of H i tails within these systems (Kenney et al. 2004; Crowl et al. 2005; Chung et al. 2007). In X-rays, however, the detection rate of these features is generally very small (<10%), suggesting that strong stripping of hot ISM is either very rare or proceeds very rapidly. No evidence for such activity in the form of X-ray tails is seen within NGC 2563, but this is not surprising given the low detection rate of thermal coronae within the sample.

In H i, we identify two tails associated with the relatively isolated CGCG119-047 and CGCG119-051, both with H i extending beyond the optical disk in the direction opposite of the group center (Figure 6). Neither of these objects is H i deficient, so any stripping activity must have recently commenced. Their respective radial velocities relative to the group mean of 200 and 320 km s⁻¹ imply Mach numbers of ${\cal M} \gtrsim 0.4$ and ≳ 0.6, respectively, for an ambient gas temperature of T ≈ 1 keV. Although both galaxies are currently beyond the radius to which intragroup gas is detected (R ≈ 500 and ≈900 kpc), their H i morphologies are suggestive of ongoing ram pressure (or viscous) stripping. Multi-wavelength observations and higher-resolution H i data would be required to confirm this (e.g., Murphy et al. 2009). Another relatively isolated spiral, IC 2293 at R ∼ 470 kpc, displays no H i tail but is highly deficient (Def$_{\rm H\,\mathsc{i}} = 0.55$), and is also a candidate for recent ram pressure stripping.

To test for the global importance of ram pressure stripping within the group, we next consider the radial distribution of ISM detections. To put all galaxies on an equal footing, their thermal X-ray luminosities and H i masses were normalized by the galaxy L_K. Excluding again the central galaxy NGC 2563, evidence for hot gas is seen in only five of the group members, and Figure 8 shows that all of these reside relatively far from the group center, with four out of five located beyond the radius R ∼ 400 kpc to which intragroup gas is detected. In contrast, no hot ISM component is found within R ∼ 300 kpc, where several of the optically luminous early types reside. This includes NGC 2562 at R ∼ 90 kpc, the closest bright galaxy to the group center, with an upper limit to L_{X, th} which is an order of magnitude below the value of log (L_{X, th}/L_K) ≃ −4.6 suggested by the L_K–L_{X, th} relation for early types in groups (Jeltema et al. 2008).

Zoom In Zoom Out Reset image size

To further illustrate this point, we note that, given our X-ray detection limits (Section 4.1), the Kim & Fabbiano (2004) relation would suggest that we should detect all early-type members brighter than log (L_K/L_☉) ≈ 10.0, irrespective of their hot gas content. In practice, only 8/14 (57% ± 25%) are detected, even if including NGC 2563 itself and excluding Sa galaxies. The bottom panel in Figure 8 shows the cumulative fraction of group members covered by Chandra above this L_K that is X-ray detected, and the fraction that also contains evidence for a thermal component. While the former remains constant with R, the thermal fraction drops toward the group core. Although subject to large Poisson errors, this result is consistent with hot ISM having been stripped within the dense group core, while the more distant galaxies retaining a thermal component have yet to experience peak ram pressure during their orbit (but see also Balogh et al. 2000).

In contrast, galaxies with a detectable cold ISM component are distributed more evenly with R. This is illustrated in Figure 9(a), which compares our measured H i masses to those of similar galaxies in the field. A handful of galaxies show evidence of a significant shortfall in H i content (i.e., Def$_{\rm H\,\mathsc{i}}\gtrsim 0.5$), but the values scatter broadly around Def$_{\rm H\,\mathsc{i}}=0$ and show no systematic dependence on R. A Kolmogorov test was performed to test whether the inferred deficiencies are, in fact, consistent with being drawn from a Gaussian parent distribution centered at Def$_{\rm H\,\mathsc{i}} = 0$ with some width σ. We find that for σ in the range 0.3–0.5, this probability is at least 85%, regardless of whether the upper/lower limits and the highly deficient CGCG119-040 are included. There is thus no strong evidence from this that the large Sa–Sc group members are globally H i deficient. However, we present evidence in Section 5.3 that the late-type population as a whole is deficient relative to the field.

Zoom In Zoom Out Reset image size

The estimated H i deficiencies depend on assumed galaxy morphology and the somewhat uncertain comparison values for field galaxies. A more robust measure of the relative H i content within the group members might be provided by $M_{\rm H\,\mathsc{i}}/L_K$. The dependence of this quantity on R is shown in Figure 9(b). Note that at fixed R, lower-mass galaxies would be more easily ram pressure stripped, but they are also likely to be more H i-rich to begin with. To take this into account, the galaxies were divided into three bins of comparable L_K. Figure 9(b) confirms that even at "fixed" L_K, there is no systematic radial dependence of H i content within the group, and hence that ram pressure stripping of cold ISM is unlikely to be globally important in this system.

The presence of H i tails associated with close galaxy pairs is commonly taken as a sign of tidal encounters (de Mello et al. 2008; Koopmann et al. 2008). NGC 2563 contains two galaxy pairs with clear H i extensions indicating such encounters. The most prominent of these is associated with IC 2338 and IC 2339, a spiral pair separated by only ∼15 kpc in projection and ∼20 km s⁻¹ in velocity. The SDSS images show evidence for a stellar bridge between the galaxies, supporting a tidal origin for the H i tails. An H i extension is also connecting UGC 04324 with CGCG119-040; the latter represents the most H i deficient group member (deficient by a factor of ∼15), with all its H i associated with the H i bridge. While no corresponding stellar feature is seen in the relatively shallow SDSS images, the H i morphology is strongly suggestive of a tidal encounter.

To explore whether recent removal of cold ISM through tidal interactions can generally explain the observed H i deficiencies within the group, we show in Figure 10(a) the H i deficiencies as a function of projected distance R₁ to the nearest neighbor. If excluding the closely interacting IC 2338/2339 pair and the upper/lower limits, a Kendall correlation test suggests an anticorrelation at the 1.0σ level. Including the upper/lower limits at their nominal values strengthens the correlation significance to 2.2σ. The correlation with R₁ remains significant at the 2σ level when instead considering the normalized H i masses in Figure 10(b). While only indicative, this result is consistent with some H i having been removed in galaxy–galaxy interactions within the group.

Zoom In Zoom Out Reset image size

Since strong tidal interactions require proximity of two galaxies in both position and velocity space, a plot of H i content against R₁ may be subject to large scatter due to some galaxies being close in projection only. For each galaxy, we therefore also consider the minimum value of

$$\begin{equation} \xi = \sqrt{ (\Delta R/\Delta R_{\rm max})^2 + (\Delta v_r/\Delta v_{\rm max})^2}, \end{equation} \tag{ 2 }$$

where ΔR and Δv_r are the separations in projection and radial velocity between any two group members, and ΔR_max ≈ 2.1 Mpc and Δv_max ≈ 1600 km s⁻¹ are the corresponding maximum values between all galaxies in NGC 2563. Strong interactions would require ξ ≪ 1, and, if generally prominent in removing H i, would imply a correlation between H i content and ξ for small values of the latter. For NGC 2563, one might expect interactions to be important up to at least ξ ≈ 0.1, as this would correspond to two galaxies at the same v_r separated by ΔR ≲ 200 kpc, or to a physically "overlapping" pair with Δv ≲ 160 km s⁻¹. As suggested by Figures 10(c) and (d), there is some evidence of such a correlation for ξ ≲ 0.1 among the individually H i-detected galaxies. However, this is significant at less than 2σ, so we can only conclude that our results are at least consistent with some H i removal due to galaxy–galaxy interactions. Repeating this analysis for the X-ray-detected group members reveals no indication of a systematic trend in L_{X, th}/L_K with either R₁ or ξ. However, many of these galaxies have X-ray luminosities only slightly above our completeness limit, so deeper observations of a larger sample would be required to confirm this result within groups in general.

Our comprehensive VLA coverage of NGC 2563 enables a census of the global amount and distribution of H i in the group, and provides further evidence of recent H i mass loss from the group members. Based on measurements from the H i Parkes All Sky Survey, Evoli et al. (2011) have quantified the typical H i mass for late-type galaxies (Sb and later) in the general "field" with M_* > 10⁸ M_☉:

$$\begin{equation} M_{\rm H\,\mathsc{i}} = M_1 \left(\frac{M_\ast }{M_2}\right)^{0.19} \left[1+\left(\frac{M_\ast }{M_2}\right)^{0.76}\right], \end{equation} \tag{ 3 }$$

where M₁ = 3.36 × 10⁹ M_☉ and M₂ = 3.3 × 10¹⁰ M_☉. To compare this to results for NGC 2563, we used estimates of M_* for the group members from SDSS,¹⁶ available for all but six of the 64 members; for those six, we assume a K-band stellar mass-to-light ratio of 0.55, which is the average for the members with both L_K and M_* available. Equation (3) would then suggest a total H i mass of the relevant group members of 6.7 ± 2.0 × 10¹⁰ M_☉, whereas the value inferred from Table 3 is $M_{\rm H\,\mathsc{i}} < 4.4\times 10^{10} \,M_\odot$, when including the average 3σ upper limit of 2 × 10⁸ M_☉ for the undetected members. This suggests that globally the M_*  >  10⁸ M_☉ late types in the group have suffered mild H i mass loss with respect to similar field galaxies. Compared to estimates of Def$_{\rm H\,\mathsc{i}}$, this result may represent a more robust estimate of global H i deficiency within the group, as it also includes smaller galaxies, depends less sensitively on the inferred morphology and optical size of individual group members, and takes the dispersion seen for field galaxies into account.

We also derive a total stellar mass in the group of M_* ≈ 1.1 × 10¹² M_☉ for all the 64 spectroscopically identified members, and a corresponding total range of $M_{\rm H\,\mathsc{i}} = (5.7\hbox{--}6.5)\times 10^{10} \,M_\odot$. This implies $M_{\rm H\,\mathsc{i}}/M_\ast =0.054$–0.062 for the full group (down to our 98% spectroscopic completeness limit of M_r = −17), under the plausible assumption that our VLA observations are not missing any large-scale diffuse H i. Future comparison of this result for an X-ray bright group to other groups and clusters could provide further information on the nature of gas removal in these environments.

On the largest scales, the distribution of H i-detected galaxies in Figure 5 reveals a peculiar lopsidedness, with 16 of the 20 detections occurring on the western side of the group. This is not simply due to asymmetric spatial VLA coverage, as demonstrated in Figure 1, nor due to a larger fraction of H i deficient galaxies in the east. The wide velocity coverage of our VLA data (Δv_r ≈ ±3σ_biwt) further renders it unlikely that a significant fraction of potential group members have been missed on the eastern side of the system. The lopsided H i distribution thus reflects a real asymmetry in the distribution of late-type members, with nearly all the large spirals located in the western half of the group.

The origin of this asymmetry remains unclear. One possibility is that the western spirals are part of an infalling subgroup, but echoing an earlier conclusion by Zabludoff & Mulchaey (1998b), we find no evidence for kinematic substructure in the group using the technique pioneered by Dressler & Shectman (1988). Another possible scenario is that galaxy morphologies have been modified by strong interactions with a similarly asymmetric intragroup medium. This can be ruled out, however, because the intragroup gas distribution in NGC 2563 appears fairly symmetric on large scales and, if anything, seems more strongly extended to the west (see Figure 2 and Mulchaey & Zabludoff 1998). Finally, exploiting the extensive SDSS coverage in this field, we have also examined the overall galaxy distribution within a projected distance of R = 5 Mpc of the group center, and find no evidence for a general large-scale enhancement in galaxy density on the western side of the group. Further studies of other groups on very large scales may help establish how common such morphological segregations are in these systems.