{}^{13}CO(1–0) in ETGs

Evidence of boosted Co/CO ratio in early-type galaxies in dense environments


We present observations of CO(1–0) in 17 Combined Array for Research in Millimeter Astronomy (CARMA) ATLAS early-type galaxies (ETGs), obtained simultaneously with CO(1–0) observations. The CO in six ETGs is sufficiently bright to create images. In these 6 sources, we do not detect any significant radial gradient in the CO/CO ratio between the nucleus and the outlying molecular gas. Using the CO channel maps as 3D masks to stack the CO emission, we are able to detect 15/17 galaxies to (and 12/17 to at least ) significance in a spatially integrated manner. Overall, ETGs show a wide distribution of CO/CO ratios, but Virgo cluster and group galaxies preferentially show a CO/CO ratio about 2 times larger than field galaxies, although this could also be due to a mass dependence, or the CO spatial extent (). ETGs whose gas has a morphologically-settled appearance also show boosted CO/CO ratios. We hypothesize that this variation could be caused by (i) the extra enrichment of gas from molecular reprocessing occurring in low-mass stars (boosting the abundance of C to C in the absence of external gas accretion), (ii) much higher pressure being exerted on the midplane gas (by the intracluster medium) in the cluster environment than in isolated galaxies, or (iii) all but the densest molecular gas clumps being stripped as the galaxies fall into the cluster. Further observations of CO in dense environments, particularly of spirals, as well as studies of other isotopologues, should be able to distinguish between these hypotheses.

galaxies: clusters: general – galaxies: elliptical and lenticular, cD – galaxies: ISM

1 Introduction

Figure 1: (Left): Distribution of CO luminosities (gray) among both CO-detected ETGs (green) and this work (blue). The CO comparison sample is from the flux-limited CO survey of Young et al. (2011). The CO sample represents a large percentage of ETGs detected at  L, though misses some of the faintest ETGs detected in CO. (Right): Distribution of all galaxies with measured , including LTGs (gray). The CO-detected ETGs are distinguished (blue) along with galaxies part of this work (green). In this case, while the ETGs fall in the low end of the distribution, it appears that this is due to the fact that ETGs in general have lower (further detailed in Young et al. 2011), rather than the CO-detected ETGs being biased. In both cases, the smallest value of is NGC 404 from Sage (1990), at  L.

While early-type galaxies (ETGs) were originally assumed to be mainly devoid of molecular gas (e.g., Bower et al. 1992), it has become increasingly apparent that these “red and dead” early-type galaxies often contain non-negligible amounts of cold gas (Wiklind & Rydbeck, 1986; Phillips et al., 1987; Wiklind & Henkel, 1989; Sage & Wrobel, 1989; Welch & Sage, 2003; Sage et al., 2007). Large-scale searches for molecular gas in ETGs have indeed shown that a significant proportion of these galaxies do contain a molecular gas reservoirs (e.g., Combes et al. 2007; Young et al. 2011), though many of the early searches targeted relative small unrepresentative samples of ETGs, rather than unbiased, volume-limited samples.

The ATLAS survey (Cappellari et al., 2011a) provides a complete, unbiased sample of morphologically-selected ETGs in a large local volume. All ATLAS galaxies were searched for molecular gas by Young et al. (2011), with a measured detection rate of %. The detection rate of molecular gas did not appear to be dependent on galaxy environment, with a similar fraction of group and cluster ETGs detected as field ETGs. Follow-up efforts included CO imaging (e.g., Young et al. 2008; Crocker et al. 2011; Alatalo et al. 2013). Davis et al. (2011) found that despite the fact that Virgo galaxies are detected at a similar rate to field ETGs, the origin of the gas within Virgo galaxies appears to be exclusively internal, while a large fraction of field ETGs acquire their molecular gas via accretion of gas-rich companions.

Figure 2: Integrated intensity map of the CO(1–0) (blue contours) overlaid upon the CO moment0 map (colourscale) for NGC 3665, NGC 4429, NGC 4435, NGC 4710, NGC 4753 and NGC 5866, the galaxies with strong (see §3.1) CO detections. The CARMA CO(1–0) beam is shown at the bottom-left of each panel, indicating that the galaxies are well resolved, and a scale bar representing 10″ is shown on the bottom-right. Galaxies known to be in the Virgo cluster include NGC 4429, NGC 4435 and NGC 4710 (Cappellari et al., 2011a). NGC 4753 is well-known to be in a group (Steiman-Cameron et al., 1992) and NGC 3665 and NGC 5866 are within a galaxy association (Garcia, 1993). The CO levels are [0.1, 0.3, 0.5, 0.7, 0.9] times the maximum.

CO(1–0) has long been used as an optically thin tracer of CO, and for the most part has been shown to be a faithful tracer of it (Young & Sanders, 1986; Sage, 1990; Sage & Isbell, 1991; Aalto et al., 1995). The range of CO/CO ratios (hereafter ) seen in past studies (mainly in spirals, starbursts and mergers) was 0.06–0.4, with ongoing major mergers populating the lower end of the distribution (Aalto et al., 1991). Sage (1990) studied the CO(1–0) properties of 3 ETGs (NGC 404, NGC 4710 and NGC 5195) with the National Radio Astronomy Observatory (NRAO) 12m single-dish, and concluded that this small sample of S0 galaxies seems to have a similar value of  as spiral galaxies on average. NGC 5195, known to have recently interacted with M51, has a much smaller  than the other ETGs, similar to what is seen in NGC 1266 (Alatalo et al., 2011; Crocker et al., 2012), a peculiar ETG with compact, dense molecular gas (Alatalo et al., 2015) currently hosting an active galactic nucleus (AGN)-driven massive molecular outflow (Davis et al., 2012; Alatalo et al., 2014).

Name Velocity Range
         () (km s) (km s) (Jy km s) (Jy km s)
IC 676 1360 – 1520 20 67.95.8
IC 719 1670 – 2080 40 23.82.9
NGC 2764 2570 – 2840 30 79.74.2
NGC 3607 680 – 1210 40 53.23.2
NGC 3619 1280 – 1770 80 15.11.3
NGC 3626 1320 – 1670 50 40.13.4
NGC 3665 1750 – 2400 50 91.24.4
NGC 4324 1520 – 1770 40 20.02.1
NGC 4429 830 – 1370 30 70.43.2
NGC 4435 560 – 980 30 32.62.6
NGC 4694 1150 – 1210 30 17.00.8
NGC 4710 940 – 1310 30 350.66.0
NGC 4753 919 – 1470 50 70.24.9
NGC 5379 1700 – 1880 30 22.92.3
NGC 5866 500 – 1010 30 250.98.4
PGC 029321 2760 – 2910 30 19.61.7
PGC 058114 1370 – 1640 30 73.43.1

Channel width of the CO cube.
Does not include the 20% absolute millimeter flux calibration uncertainty. Errors were calculated as discussed in §3.2.
30m CO integrated flux densities (Crocker et al., 2012).
CO integrated flux densities from the CO cubes described in Alatalo et al. (2013), but derived with the same velocity resolution (and channels) as the CO cubes to match the CO resolution for our stacking analysis.
The CO map of IC 676 is uniformly weighted in Alatalo et al. (2013), whereas the other CO and CO maps in this work and are naturally weighted
from Crocker et al. (2012). Others values are calculated from IRAS fluxes.
IRAS data were unavailable, so the flux ratio was calculated based on the Herschel ratio (T. Bitsakis, private communication).

Table 1: CO(1–0) observations for the CARMA ATLAS survey

Dense gas (i.e., gas that is traced by HCN and HCO emission) was observed in a much larger sample of ETGs through the SAURON (Bacon et al., 2001) and ATLAS (Cappellari et al., 2011a) surveys, that targeted the strongest CO-detected ATLAS galaxies with the Institut de Radioastronomie Millimétrique (IRAM) 30m single-dish telescope (Krips et al., 2010; Crocker et al., 2012; Davis et al., 2013). In these works, there was some evidence that while ETGs show similar ratios of dense gas as spirals, the ratios may depend on environment, with Virgo galaxies showing boosted  and HCN/CO ratios compared to field galaxies. With the Crocker et al. (2012) sample, statistically significant correlations between  and absolute K-band luminosity, stellar population age, molecular to atomic gas ratio, dust morphology and dust temperature were found. The correlation between  and environment was not however statistically significant, perhaps due to the low numbers of galaxies in dense environments such as the Virgo cluster.

Here we present the CO(1–0) observations from the CARMA array. In §3, we discuss the simultaneous CO(1–0) and CO(1–0) observations, and the techniques that were used to reduce and analyse the CO(1–0) data, including imaging strong detections and stacking all non-detections. In §2, we describe our sample galaxies and their properties. In §4, we discuss the implications of the CO distributions and how our new detections inform upon what affects  in galaxies. In §5, we summarize our conclusions. The ATLAS sample properties are described in detail in Cappellari et al. (2011a)3, and details specific to the CARMA observations are reported in Alatalo et al. (2013). We use the cosmological parameters  km s, and (Spergel et al., 2007) throughout.

2 The Sample

The CARMA sample has been extracted from the ATLAS sample of galaxies, which is a volume-limited sample (D42 Mpc) of massive (M-21.5 mags) morphologically-selected early-type galaxies (Cappellari et al., 2011a). The subsample of galaxies observed with CARMA have been selected among those detected in the CO line by Young et al. (2011), who observed the complete ATLAS sample. CARMA then imaged all previously un-imaged CO-detected ETGs above a flux limit of 18.5 Jy km s (Alatalo et al., 2013). Figure 1a shows the distributions of CO luminosities for CO-detected ETGs from Sage (1990); Crocker et al. (2012) and this work compared to the unbiased, flux-limited survey of Young et al. (2011), indicating that CO-detected ETGs tend to populate the high end of the distribution. Given that CARMA observed the brightest CO ETG detections from Young et al. (2011), the fact that CO detections tend to populate the bright portion of the CO distribution is unsurprising.

Figure 1 shows that when we draw comparisons to the available CO-detected galaxies in the literature (Young & Sanders, 1986; Sage & Isbell, 1991; Aalto et al., 1995), that are either late-type galaxies (LTGs), starbursts or mergers, the CO-detected ETGs fall to the lower end of the distribution of CO luminosities, consistent with what is found in CO-detected ETGs in general (Young et al., 2011). So our CO ETG sample has properties entirely consistent with a population of CO-bright ETGs.

It also appears clear that the galaxies added in this work do not fundamentally differ in their CO properties from the galaxies considered in previous ETG work (Young et al., 2011), and appear to be a robust addition to the CO-detected ETGs. If we look specifically at the representation of Virgo galaxies, we find that 7/26 (268%) are Virgo members within the CO-detected sample, compared to 12/56 (215%) of the CO-detected ETGs from Young et al. (2011). Thus it does not appear that Virgo members are over- or under-represented in our sample. CO-detected ETGs were also selected from this subsample, using identical selection criteria to Virgo and thus are also neither over- nor under-represented in the sample.

Though the 7 Virgo members have comparable average properties to the non-Virgo CO-detected ETGs (within the scatter), including stellar mass ( vs. ) and molecular gas extent ( vs. ; Davis et al. 2013), they do skew more massive with more compact molecular discs. This is likely due to the small number of Virgo galaxies within this sample, but the scatter is large due to small number statistics. Further CO studies of more Virgo members should reduce the scatter.

3 Observations and Results

CO(1-0) observations were obtained for 17 of the 30 original CARMA CO(1–0) galaxies. The galaxies missing CO(1-0) data are those imaged before Spring 2009, when upgrades to the CARMA receivers allowed for simultaneous imaging of both lines. The total available bandwidth from CARMA in each line varies as well depending on the semester the source was observed, due to upgrades to the correlator. Total bandwidths for each object are listed in Table 2 of Alatalo et al. (2013). All data reduction and calibration were done simultaneously with the CO imaging published in Alatalo et al. (2013). Table 1 presents the galaxies observed in CO, along with CO-specific observing parameters (including beam size, bandwidth and channel widths). The calibrators used, and total observing time for each source, can be found in Alatalo et al. (2013).

Figure 3: Stacked CO(1–0) spectra of the 15/17 detected galaxies in the CARMA ATLAS CO survey. The dark blue shaded regions represent uncertainties, using the noise calculation laid out in §3.2, with the light blue representing the integrated signal within the 2D CO(1–0) mask of each channel, compared with the CO(1–0) spectra from Alatalo et al. (2013) in red. The un-shaded CO spectra are those channels where the number of pixels in the CO mask fall to fewer than total number of pixels of the beam. In these channels, the spectrum is calculated by summing pixels in a contiguous region of the channel map the size of the beam area, and can be considered the baseline scatter for the stacked CO spectra, with corresponding error bars in the top left of each panel. The scale on the left of each panel is the CO(1–0) line flux, while the scale on the right of each panel is the CO(1–0) line flux. The velocities are the optically-defined, local standard of rest () velocities of the molecular gas. Total bandwidths from the CARMA observations are listed in Table 2 of Alatalo et al. (2013). In all cases except NGC 3665, the continuum (reported in Alatalo et al. 2013) is insubstantial compared to the stacked CO spectra (and has been removed from all line data). A dashed black line shows the continuum detection level of 8.98 mJy in NGC 3665.

3.1 Strong CO detections: CO maps

The CO(1–0) emission was strong enough (emission was detected with signal-to-noise ratio 3) to map 6 of the 17 observed galaxies. Moment maps were constructed based on the methods described in Alatalo et al. (2013), with a full velocity width determined by the CO(1–0) line. Figure 2 shows the six mapped galaxies: NGC 3665, NGC 4429, NGC 4435, NGC 4710, NGC 4753 and NGC 5866. To test the agreement between the CARMA detections and the single dish detections from Crocker et al. (2012), we convolved the images of the three overlapping galaxies with strong CO detections: NGC 3665, NGC 4710 and NGC 5866, to the size of the IRAM 30m beam, and compared , finding them to be in agreement within errors with  reported in Crocker et al. (2012).

For these six galaxies, we attempted to investigate whether  varies with position. We first regridded both the CO and CO maps so that the area of one square pixel equalled the area of the CO beam, this using the Interactive Data Language (IDL) Astronomer’s Library routine hcongrid4, with bilinear interpolation of all points. We then measured both a nuclear (from the regridded pixel corresponding to the 2 Micron All Sky Survey K band-determined nuclear position; Skrutskie et al. 2006) and a non-nuclear  value (from the remaining CO-detected pixels). To evaluate the uncertainty on the  measurements, we first evaluated the uncertainty on each integrated flux by measuring the root mean square noise (RMS) in the regridded CO and CO integrated intensity maps, respectively, taking the standard deviation of the pixels outside of the region where CO is detected, and multiplying by , where is the number of pixels in each region (one for nuclear regions; the total number of detected off-nuclear pixels for the non-nuclear regions). These  gradient measurements are listed in Table 2 and discussed in Section 4.

NGC 3665
NGC 4429
NGC 4435
NGC 4710
NGC 4753
NGC 5866
Table 2: Resolved  values

3.2 Weak CO detection: CO spectra

To extract CO integrated flux densities for the fainter galaxies, where CO is not detected in individual beams, we utilize our prior knowledge of the molecular gas distribution from the brighter CO emission. This technique is also used to extract integrated flux densities for the strong CO detections. We used a smoothed CO datacube (created to correspond both to the weighting scheme as well as velocity properties of the CO maps) for each galaxy as an priori dataset, and for each channel determined the two-dimensional region over which the CO line is detected with a signal-to-noise ratio above 1.5. This 3D region then represents a CO detection mask (see Schruba et al. 2011 for details of a similar method). For each galaxy, we then sum the CO datacube over its detection mask channel by channel, to obtain an integrated spectrum (shown in Fig. 3) and thereafter an line flux5. (tabulated in Table 1) by summing the channels with emission (shaded blue in Figure 3). The un-shaded portion of the CO spectra in Fig. 3 are channels where the number of pixels in the CO mask fall to fewer than the number of pixels in the beam. In these channels, the channel flux is calculated by summing pixels in a contiguous region of the channel map the size of the beam area, and can be considered the baseline scatter for the stacked CO spectra.

The RMS per channel was determined by calculating the standard deviation of the line flux in an area equivalent to the mask, created by the miriad task restor in “residual” mode, and multiplying by , where is the total size of the CO mask area normalized to the beam area. The RMS on the total line flux was then calculated by multiplying the noise per channel by the width of the channel (shown in Table 1) and multiplying by the square root of the sum of the noise squared in each individual channel (the shaded channels in Fig. 3):

where is the channel number.

There is a further 20% absolute flux calibration uncertainty associated with millimeter observations, due to the time variability of millimeter flux calibration sources. While this 20% uncertainty impacts the total CO line flux measurements, it does not factor into , given that all CO and CO observations were taken simultaneously. A further 30% up-correction was made to the RMS to attempt to account for the oversampling of our interferometric maps. The 30% correction was derived by calculating the noise in the oversampled (1 pixel) maps, and comparing that noise to the RMS derived from maps whose pixel area equals that of the synthesized beam. The RMS per channel for each galaxy is shown in the upper-left corner of each panel.

CO Type N Mean Std. Deviation
All CO 72 0.111 0.046
All Field 62 0.096 0.034
Spirals 32 0.098 0.029
Mergers 10 0.086 0.020
Field galaxies ( 32 0.112 0.036
All ETGs 27 0.128 0.081
Interacting ETGs 8 0.068 0.026
Settled ETGs 19 0.154 0.074
Virgo ETGs 7 0.221 0.064
Field ETGs 20 0.096 0.052
Field ETGs ( 8 0.120 0.066

The field sample includes all galaxies not in the Virgo cluster

Table 3: Average  properties of each subsample of galaxies

Table 1 lists the galaxies that were observed in CO during the CARMA ATLAS survey, with the derived integrated flux densities and  values from the stacked spectra. Of the 17 galaxies, 15 are detected using the a priori masking technique. Figure 3 presents the stacked CO spectra of the 15 detected galaxies. Comparing the line shapes of CO (blue) and CO (red), we see that they are generally in good agreement, supporting our claim that the stacking technique as well as our chosen mask threshold have successfully detected CO emission.

Crocker et al. (2012) published IRAM 30m single-dish spectra of CO(1–0) and CO(2–1) for 18 of the ATLAS galaxies. Our sample overlaps the Crocker et al. (2012) sample in the case of 8 galaxies: IC 676, NGC 2764, NGC 3607, NGC 3665, NGC 4694, NGC 4710, NGC 5866 and PGC 058114. We detect all overlapping galaxies at a signal-to-noise ratio . CARMA recovers more emission than the IRAM 30m in all cases except IC0676, equal within the uncertainty (see Table 1). In many cases, this is likely because the CO emission subtends a larger angular size than the 30m beam (as was often the case for the CO emission published in Alatalo et al. 2013). Because CO and CO were measured simultaneously with CARMA,  does not suffer from flux calibration uncertainties. We therefore elect to use the CO integrated flux densities from the current work for all overlapping galaxies.

Radio continuum pollution leading to enhanced CO fluxes is not likely to be an issue. NGC 3665 and NGC 4710 are the only two galaxies in this sample with detected continuum, at and 4 mJy, respectively, at 110 GHz. The CO data processing includes a continuum subtraction step (detailed in Alatalo et al. 2013).

Table 2 compares  in the nuclear and extended regions of NGC 3665, 4429, 4435, 4710, 4753 and 5866 (see Section 3.1). In the first four,  is larger in both the nuclear and extended region than in the stacked spectrum, and vice versa for NGC 4753 and NGC 5866. The discrepancies between stacking and resolved  measurements can be explained by many effects. Stacking is able to detect fainter CO emission from a larger area than the resolved data. Compared to well-resolved objects, where the brightest CO peaks were able to fill the beam, the average  from stacking would be lower than from the resolved maps. On the other hand, in cases where CARMA has marginally detected CO in the maps, stacking would be able to increase the total CO flux detected, resulting in a higher  for stacking than mapping. Deeper CO observations might be able to fully explain this discrepancy, but are beyond the scope of this paper.

4 Discussion

4.1 Distribution of CO in the galaxies

Figure 2 shows the spatial distribution of the CO emission, compared to the CO  in ATLAS galaxies. Only the galaxies with the brightest CO are detectable in a spatially-resolved way. There are many features worthy of notice in the CO distributions.

NGC 4710 and NGC 5866 have strong, edge-on bars whose kinematic features were detected in the CO emission (see §4.2.2 in Alatalo et al. 2013). In these cases, the CO preferentially traces the bar region, particularly a nuclear disc or ring and an inner ring, the locations of the largest concentrations of molecular gas in barred systems (Athanassoula & Bureau, 1999). NGC 4710 and NGC 5866 were also among the handful of galaxies detected in OH absorption by McBride et al. (2015), confirming high optical depths. The CO detections at the disc edges further confirm the conclusions of Alatalo et al. (2013) and Davis et al. (2013), that these systems are bar+ring systems, as the CO there is likely tracing the portions of the edge-on rings with higher column densities (edge brightening). Topal et al. (in preparation) confirm this and study the spatially-resolved bar signatures using a multitude of high-density molecular tracers in both galaxies.

The Virgo galaxies NGC 4429 and NGC 4435 are robustly detected in CO throughout the extent of the CO detections. In both cases, CO faithfullly traces CO. NGC 4753, known to be at the centre of a developing group (Steiman-Cameron et al., 1992), also seems to have CO that traces the brightest portions of CO, though it is likely that beyond these points the CO emission simply drops below the detection threshold (rather than being absent).

Figure 4: Resolved  in the six strong CO detections amongst the ATLAS galaxies, separating the nucleus (points) and the surrounding gas (open circle). These numbers are likely upper limits for integrated , as only CO integrated flux densities exceeding our detection threshold are represented.

4.2 Radial gradients in

Figure 4 show  in nuclear and surrounding gas in the 6 galaxies mapped, with the points representing the nuclear pixels and the open circles representing the extended emission (with resolved  values shown in Table 2). As discussed in §3.2, ratios measured from the images differ slightly from those measured with the stacked spectra, likely due to the images missing faint CO emission picked up in the stacked spectra. Within the selected pixels, these ratios are thus likely reasonable, but they are not necessarily representative of the entire galaxy, where the stacking method is better able to recover emission below the local detection threshold.

In spiral galaxies, Paglione et al. (2001) found a  gradient, increasing from the nuclear region to the outlying regions on average, though this result had large scatter. Molecular discs in ETGs are much less extended than those in spirals (Davis et al., 2013) in absolute terms, so it is possible that a  variation will be much less pronounced in ETG discs, either due to resolution or because the gas is better mixed (i.e. shorter dynamical times). In the 6 spatially-resolved CARMA detections (shown in Fig. 4), there is no decipherable trend, and in 5 of the 6 objects the nuclear and non-nuclear  values are consistent within the errors. NGC 3665 is the single object in which the trend is larger than the uncertainties, but follows the opposite trend from spirals, and could be due to sensitivity limitations in detecting fainter CO flux. There might be a slight increase in  from the nuclear region to the off-nuclear regions in NGC 4429 and NGC 5866, but  remains flat in NGC 4710 and decreases in NGC 3665, NGC 4435 and NGC 4753. It is however possible that we are missing a non-negligible amount of faint extended emission, due to the sensitivity limitations of the CARMA sample. As mentioned above (§3.2), the disagreement in  between the image measurements and the stacked measurements is evidence that this might be the case.

We therefore do not see any evidence for an increase in  across galactic discs in this subset of ETGs, in contrast to what has been seen in spirals, but we cannot rule out that such a gradient could exist, given the sensitivity as well as resolution limitations of our observations. Deeper, higher resolution follow-up observations should more robustly establish whether ETGs indeed lack a  gradient, whether this is an environmental effect, or whether our result is due to sensitivity issues.

4.3 Variations of  based on galaxy type and environment

Figure 5 presents the distributions of  as a function of galaxy types, in the new CARMA detections of Virgo and non-Virgo ETGs, the Crocker et al. (2012) ETGs, NGC 404 and NGC 5195 from Sage (1990), as well as literature spirals and mergers (Young & Sanders, 1986; Sage & Isbell, 1991; Aalto et al., 1995). Figure 5 also shows the  distribution as a function of environment in the case of ETGs only (as there is no Virgo spiral with CO data currently in the literature). Spirals and mergers tend to have , while ETGs are distributed much more evenly as a function of .

We use the morphological classifications from Alatalo et al. (2013) to differentiate between settled and disturbed gas morphologies, though NGC 1266 was re-classified as disturbed, since the turbulent nature of the molecular gas is visible in the CO and CS(2–1) lines using higher resolution imaging (Alatalo et al., 2011, 2015). The CO in NGC 404 was classified as a ring (Wiklind & Henkel, 1990), and is therefore settled, while that in NGC 5195 would be classified as mildly disrupted (Kohno et al., 2002), and is considered disturbed. The  value within ETGs seems to have a strong dependence on the disruption of the molecular gas, with disrupted gas distributions tending toward the lowest Ẇhen ETGs are split by environment, the Virgo ETGs represent the majority of ETGs with , while field ETGs average less than =0.1. The distributions not only show that environment seems to play a role, but also that disrupted molecular gas distributions tend toward lower  in ETGs.

Figure 5: (Top): Distribution of all galaxies with (gray), compared with the distribution of spirals (orange), mergers (red) and ETGs (turquoise) only. Both spirals and mergers seem to have an average , whereas ETGs have a much flatter distribution. (Bottom): Distribution of all ETGs with (gray), compared to the distribution of CO-disrupted ETGs (pink), field ETGs (blue), and Virgo ETGs (green) only. Disrupted ETGs tend to have the smallest  values, while field ETGs are consistent with spirals. Virgo ETGs account for the majority of large .

Figure 6 compares  with the dust temperature, represented by the 60-to-100 m flux ratio. Young & Sanders (1986) discussed the dependence of  on the dust temperature, with lower values of  corresponding to higher values of , further confirmed by the observations of Sage & Isbell (1991) who increased the sample size. Aalto et al. (1995) drastically increased the number of spirals for which CO observations exist, confirming the trend as well as hypothesizing that turbulence within the molecular gas might influence . Warmer dust temperatures are a reasonable tracer of turbulence in the underlying molecular gas, as an abundance of young stars will provide both UV photons to heat the dust and momentum to stir up the molecular gas. (Ogle et al., 2010). However, the presence of Virgo galaxies at low also having the highest  indicate that environment could also be a factor in setting .

Crocker et al. (2012) found evidence that Virgo galaxies and those galaxies in groups have boosted  values, but due to small number statistics were not able to confirm that result. The CARMA CO detections add many more ETGs to the Crocker et al. (2012) work, including an additional 3 Virgo galaxies. The addition of these ETGs not only strengthens the trend, but appears to also further support the suggestion that turbulence might influence  (Young & Sanders, 1986; Aalto et al., 1995). Magenta outlines in Fig. 6 correspond to ETGs with disturbed molecular gas. The disturbed class of ETGs almost exclusively appears at values of , the only exception being NGC 4694 that also is a member of the Virgo cluster.

In spirals and mergers, there appears to be a ceiling in  of around 0.15 for the majority of sources. Once the ETGs are included, a large proportion of them inhabit the region, with Virgo ETGs exclusively lying above this threshold. NGC 4710 (identified as a Virgo member in Cappellari et al. 2011a) is found to have the lowest  in the Virgo ETG sample, of 0.15, though it is of note that while NGC 4710 has a velocity similar to other Virgo cluster members, it is not formally within the virial radius of Virgo.  in the CARMA sample is on average larger than that in all other samples discussed, due in part to the fact that the CARMA observations include more ETGs with predominantly settled gas distributions (which were not among the brightest CO detections in ATLAS first followed up in CO with the IRAM 30m; Crocker et al. 2012). This inclusion of the ETGs with settled molecular gas seems have increased the population of objects in the region by over a factor of two.

Table 3 gives the  mean and standard deviation of each subsample of galaxies shown in Fig. 5, including isolated galaxies based on environment, morphology and dust temperature. Overall, Virgo ETGs have a larger mean  value than all other subsamples and are distinct (means inconsistent within the standard deviations) from all other subsamples except for cold dust field ETGs (of which there are only 8). While these two populations agree within the errors, the cold dust field ETG population is much more similar to the complete cold dust field sample than to Virgo ETGs. Thus, while we cannot completely rule out that the variation in  that we observe is solely dependent on molecular gas turbulence (as traced by dust temperature), it does appear that  in Virgo is boosted.

Figure 6: Integrated CO/CO ratio for ETGs (green stars) are shown as a function of the 60–to–100m flux ratio (), and compared to spiral galaxies (spirals), and mergers (clovers). Upper limits are shown as downward facing arrows. Literature spiral galaxies and mergers are from Young & Sanders (1986), Sage & Isbell (1991), and Aalto et al. (1995), with ETGs (green) from Sage (1990), Crocker et al. 2012, and this work. The two ETGs from Sage (1990) are NGC 5195 (the minor interaction companion to M 51; Kohno et al. 2002) and NGC 404. The black dotted horizontal line shows the location of =0.15. Virgo ETGs are represented by green stars with a circular black background. All spirals are field spirals, as there is no Virgo spiral with  literature data. While a few ETGs from Crocker et al. (2012) have elevated  ratios, the majority have ratios similar to those of spirals and mergers from previous work. Moreover, these ETGs also follow the trend with dust temperature , with the highest  corresponding to the coldest dust temperatures. ETGs (stars) with a magenta outline represent those that are morphologically classified as either mildly or strongly disrupted in Alatalo et al. (2013). It is clear that ETGs with this classification are much more likely to have a low , except in the case of NGC 4694, which is also a Virgo galaxy. While two of the high dust temperature and low  ETGs from our study contain an AGN (NGC 3607 and PGC 029321), PGC 058114, which also has a high dust temperature (but also a high ), does not appear to (Nyland et al., in preparation).

Figure 7 isolates the ETGs and spirals in the ATLAS parent sample (Cappellari et al., 2011a), performing Spearman rank statistics as was done for the original sample from Crocker et al. (2012). The Spearman rank correlation coefficient and the probability that the null hypothesis (no correlation) is true are given. Low probabilities for the null hypothesis (we define ‘low’ to be less than 0.03) suggest the data are correlated. In most cases, the correlations that were originally seen in Crocker et al. (2012) (including those with (1) dust temperature, (2) stellar mass and (3) environment) persist within the larger sample, and a new correlation is observed with the (4) molecular gas extent ( from Davis et al. 2013), which was only performed on the ETGs. The dependence on (the 3 Mpc number density; a proxy for environment from Cappellari et al. 2011a) has strengthened with the increased number statistics, but so has the dependence on galaxy mass. We are thus unable to determine whether the variation in  is independently correlated with stellar mass, environment, or molecular extent, or if our CO-detected Virgo ETGs are simply more massive and compact than the CO-detected field ETGs (discussed in §2) or spirals, causing the correlation.

When we differentiate between types of molecular gas in ETGs, either using the CO morphological classifications from Alatalo et al. (2013), molecular gas that is disturbed morphologically preferentially has a much lower  than that of settled or aligned discs. In fact, on average morphologically-settled molecular discs have  values almost two times higher than those of disturbed molecular distributions. When we compared the  to the kinematic misalignment between the gas and the stars (; Davis et al. 2011), the probability of a null hypothesis was 24%, meaning that it appears that the disruption of the molecular gas matters far more than whether the gas was externally accreted. Figure 6 outlines in magenta ETGs that have been classified as “disrupted” by Alatalo et al. (2013), and shows that in general those galaxies have lower , (in addition to high CO integrated flux densities, given they were part of the flux-limited dense gas survey from Crocker et al. 2012).

This fits quite well with the work of Aalto et al. (1995), who see similarly low  values in mergers, compared to normal spirals. It is difficult to parse whether the differences in distributions we see are because galaxies in Virgo tend to have much more settled molecular discs, or if there are other effects at play. Further studies of CO in a large number of ETGs with settled molecular discs, in a wide variety of environments, will likely shed light on the relative importance of morphology and environment. No previous CO study includes observations of Virgo spirals.

4.4 Possible causes of the variation in , optical depth or abundance?

In the giant molecular clouds (GMCs) of our Milky Way, typical  values are 0.2 for whole GMCs, and 0.3 for GMC centres (Polk et al., 1988). So an ensemble dominated by cold GMC-like clouds should have a high . Ratios above this can be understood as arising from optically thick and very cold gas, perhaps like a GMC but with its surrounding envelope stripped away. We know that the CO/CO line ratio correlates strongly with the dust colour (or temperature; Young & Sanders 1986; Aalto et al. 1995), as is clearly seen in Figure 6. Alternatively, a high  value could be partially a result of enhanced C over C, although this would still require very optically thick gas. An illustration of the latter is the fact that the C/C abundance ratio decreases strongly with radius in the Milky Way (Milam et al., 2005), but  has a strong opposite trend (Digel et al., 1990; Carpenter et al., 1990; Sodroski, 1991). This is because the CO optical depth in Galactic centre clouds is significantly lower due to both temperature effects and the presence of unbound gas.

Figure 7:  against ISM properties and environment of the CO-detected ETGs from Crocker et al. (2012) and this work in addition to the spiral galaxies that are part of the 871 galaxy parent sample from which Cappellari et al. (2011b) selected the ATLAS ETGs. From left to right: (1) ratio of IRAS fluxes (a dust temperature proxy); (2) M (a stellar mass proxy); (3) , the 3 Mpc galaxy density from Cappellari et al. (2011a) (an environment proxy). (Bottom): (4) , the molecular gas extent from Davis et al. (2013). The Spearman rank correlation coefficient (‘r’) and the probability that the null hypothesis (’P’, no correlation) is true are given at the top of each plot. Field ETGs are red and Virgo ETGs are blue, while spirals are green with gray error bars. There appears to be a correlation in each with each parameter, though the galaxies within Virgo tend to be very massive and to contain more compact molecular discs. Virgo galaxies also tend to have the highest  even within individual mass bins.

For the CO/CO line ratio, optical depth effects may either be caused by temperature/density changes (e.g. Aalto et al. 1997), or by increased line-widths in diffuse, non self-gravitating gas (e.g. Aalto et al. 2010). These do not require a secondary boost in the C isotope compared to C, that has been suggested as taking place during stellar processing, specifically in low-mass stars (Casoli et al., 1991). Crocker et al. (2012) considered abundance, fractionation and optical depth explanations for the trends seen in their smaller sample of ETGs, and concluded that optical depth (mediated by the gas dynamical state) dominates the  ratio changes. Here, with additional sample galaxies and more information about the gas dynamics from the interferometric maps, we return to this question. In addition, we consider a new option, the evaporation of lower-mass GMCs, or “the survival of the densest.”


Carbon-bearing molecules are susceptible to a process called fractionation. This is particularly true for CO, and therefore CO/CO abundance ratios may deviate from the C/C ratio from nucleosynthesis. There are two competing processes. The deviation may be in the form of an enhancement or a reduction depending on which process dominates. The first process is selective photodissociation, which will enhance CO since it is the more abundant isotopic variant and self-shields more effectively (e.g. van Dishoeck & Black 1988). The second process operates when gas kinetic temperatures are low to moderate (below 35 K) and elevates CO through the isotopic charge exchange reaction:

This results in CO/CO abundances above the C/C ratio from nucleosynthesis by up to a factor of five (e.g. Sonnentrucker et al. 2007), without also requiring an enhancement in the C/C abundance.

How can we test if our  ratios are caused by ion molecule exchange reactions? The dust temperatures are low, so if they are related to the gas kinetic temperatures fractionation seems possible. Ritchey et al. (2011) point out that because CO is the most abundant carbon-bearing molecule in the ISM, an enhancement in CO will result in a depletion of the carbon reservoir of C. Other molecules originating from the remaining carbon should therefore have a C/C ratio that behaves in the opposite sense compared to , i.e. we would expect to see smaller C/C ratios in these species for Virgo galaxies. Ritchey et al. (2011) suggest CN and CH as comparison species, and for our purposes CN should work well. CN has the additional advantage that the two adjacent spin groups can be used for an accurate measure of the optical depth. HCN and HCO should also work quite well. We note that, for homogeneous clouds, the enhancement of CO should occur mostly in the outer regions where there is sufficient C. If the clouds have structure the enhancement will leak further into the clouds.

While fractionation is a compelling possibility to create the enhanced  ratios seen, Crocker et al. (2012) did not see a correlation between CO/HCN (compared to CO/HCN) and morphological type, or temperature, potential signs that fractionation is an important driver of the  variations. Observations of dense gas across a larger sample of molecular gas-containing ETGs would further clarify whether or not we can completely rule out fractionation, but current data suggest it is not a dominant effect.

Enhanced C abundances from stellar reprocessing

A possible origin for the distribution of  ratios is that the molecular gas in Virgo galaxies has resided in those galaxies for far longer than that of average field ETGs. Davis et al. (2011) discovered that the angular momentum of ionised and molecular gas in Virgo galaxies is always aligned with the stellar angular momentum, which was posited to mean that the molecular gas within Virgo ETGs is not externally accreted, and either is due to stellar mass loss (Faber & Gallagher, 1976) or remains from aligned molecular gas when the galaxy first entered Virgo, thus supporting this claim.

The observation that some Virgo spirals may have significantly higher O/H abundances within their ionised gas than field galaxies (Shields et al., 1991) also support this claim, although more recent metallicity studies contradict it (Hughes et al., 2013). Casoli et al. (1991); Casoli et al. (1992a, b) discuss low  ratios as likely being an abundance effect, with more primordial (and largely unprocessed) molecular gas containing fewer C atoms, since this isotope is formed via low-mass stellar nucleosynthesis. The interstellar medium (ISM) in Virgo galaxies would have existed sufficiently long for this enrichment to occur. NGC 4710, which is on the outskirts of Virgo, has =0.15, at the junction between most spirals, interacting galaxies, and Virgo galaxies. If NGC 4710 is on its first approach (suggested by its presence outside of the Virgo virial radius), and therefore just losing the ability to accrete new primordial gas, then this enrichment has just started to boost the C abundance and is beginning to impact .

Objects with morphological signs of recent accretion events are the most likely to also have low  values, as is expected if the gas in the objects that recently accreted is more primordial and less enhanced. Though this is the case, kinematic misalignments with stars, consistent with external accretion, did not show such a distinctive difference, especially compared to the disruption seen in the gas (see §4.3). If we assume that on average, the accreted molecular gas has a lower C abundance, then there should be a trend toward lower  from externally acquired gas, not solely morphologically disrupted gas.

Studying isotopic line ratio in other gas species (HCN vs. HCN, for instance) might be able to confirm whether enrichment plays a critical role in the  ratio. Alternatively, studying  at higher excitation, and getting a different value of  in the settled discs, could rule out enrichment (though not necessarily confirm it).

Midplane pressure

Rather than being an abundance effect, the elevated  in Virgo ETGs could also be the result of midplane pressure, that can modify the optical depth of the molecular gas. First posited by Crocker et al. (2012) to describe the correlations seen in the original sample of ATLAS ETGs, it is possible that the hot intracluster medium (ICM) within Virgo produces a larger than normal midplane pressure in galaxies. This excess pressure on the molecular disc would compactiy it, creating a disc with higher optical depth (and thus higher ) than is seen in field galaxies. The fact that the resolved maps of NGC 4429 and NGC 4435 do not appear to show any significant spatial variation in  seems to support the possibility of a high pressure throughout.

However, high optical depths within the discs of these galaxies would likely imply a higher fraction of dense gas, and thus impact their star formation efficiencies (Gao & Solomon, 2004), but Davis (2014) have shown that the star formation efficiencies of Virgo ETGs are similar to those of all ETGs. Since it also appears that the molecular gas distribution plays a role in this ratio for ETGs, the high  in non-Virgo settled ETGs seems to indicate that midplane pressure is not the dominant effect in these systems.

Preferential evaporation of lower density GMCs

It is possible that the observed enhancement of  is nothing more than “survival of the densest” (GMCs). As Virgo galaxies move through the ICM, and the low density clumps are stripped (leaving only the densest clumps, likely with high ), one would expect our  measurements to probe only the highest end of the usual density distribution of GMCs. Pineda et al. (2008) show that the highest  values in local clumps (  0.4), with derived densities of  cm, are indeed comparable to the average  of the Virgo galaxies. The mild correlation of the HCN/CO ratio with seen by Crocker et al. (2012) also supports this hypothesis. In fact, results from the Herschel Virgo Cluster Survey (Davies et al., 2012) suggest that dust and gas densities are higher in Virgo cluster galaxies than in field galaxies, likely requiring denser constituent molecular clouds. Higher resolution studies of GMCs in Virgo galaxies with ALMA would further elucidate whether the properties of the dust transfer directly to the molecular gas properties in these systems.

4.5 Investigating the origins of the  variation

To fully understand the dependence of  on environment, as well as what mechanism drives that dependence, further CO observations of Virgo galaxies, including Virgo spirals, will likely be required, hopefully shedding light on what is causing the large spread in  values observed (including the Virgo–field galaxy dichotomy). If C abundance is the most important factor, then the gas phase metallicity (a proxy for “primordality”) should correlate with . If midplane pressure is the dominant factor for the variation in , then the  ratio in varying environments (including galaxy groups and possibly the Coma Cluster) should correlate. Finally, if entry into the hot X-ray gas halo of the Virgo cluster has stripped all but the densest clumps, then a measure of how long each galaxy has been within the cluster should correlate with . A high resolution census of the GMCs in Virgo (similar to the investigation of Utomo et al. 2015) might also provide evidence for whether “survival of the densest” is able to explain the variations we see in .

5 Conclusions

The CARMA ATLAS CO survey of Alatalo et al. (2013) included 17 galaxies simultaneously observed in CO(1–0), increasing the number of ATLAS ETGs observed in CO by 50%, and including far fainter CO detections (and more objects with settled molecular gas) than the original work of Krips et al. (2010) and Crocker et al. (2012).

We presented here the CO(1–0) resolved maps of 6 ETGs from the CARMA ATLAS survey, those detected with sufficient significance to derive spatial-resolved CO information. An a priori masking analysis based on the method of Schruba et al. (2011) allowed us to detect CO in 15/17 of the observed sources, this in a spatially-integrated manner.

There does not appear to be evidence for spatial variations in  in the 6 ETGs we were able to map, though it is likely we would not have been able to detect a gradient at the level found in Paglione et al. (2001), given the signal-to-noise and resolution of our data.

Combining these new observations with those from Crocker et al. (2012) and other data from the literature, we find that Virgo ETGs have a significantly higher  than both field ETGs, and spirals and interacting galaxies, although ETGs with settled molecular discs also have preferentially higher  than both spirals and ETGs with morphological disruptions. It is also possible that the variations of  are driven by mass or molecular extent, as Virgo ETGs are, on average, more massive and contain more compact molecular discs than field ETGs.

We hypothesize that the higher  ratio in Virgo could be caused by extra low-mass stellar enrichment taking place in Virgo cluster galaxies (boosting the C elemental abundance in the absence of external gas accretion), by increased midplane pressure due to the ICM, or by the survival of only the densest clumps of molecular gas (as the galaxies enter the Virgo ICM). Studies of CO in more Virgo galaxies, particularly spirals, as well as studies of other C and C isotopologues, should be able to distinguish between these hypotheses.


K.A. thanks Erik Rosolowsky, Mark Lacy, Francoise Combes, Adam Leroy, Asunción Fuente and Jeff Kenney for insightful conversations that have improved this work, Theodoros Bitsakis for Herschel related questions and Selcuk Topal for a detailed investigation of the single dish data from NGC 5866. K.A. would also like to thank the anonymous referee for insightful comments and thoughtful recommendations. K.A. is supported by funding through Herschel, a European Space Agency Cornerstone Mission with significant participation by NASA, through an award issued by JPL/Caltech. M.B. is supported by the rolling grants ‘Astrophysics at Oxford’ PP/E001114/1 and ST/H002456/1 from the UK Research Councils. L.M.Y. is supported by the National Science Foundation under Grant No. NSF AST-1109803. Support for CARMA construction was derived from the Gordon and Betty Moore Foundation, the Kenneth T. and Eileen L. Norris Foundation, the James S. McDonnell Foundation, the Associates of the California Institute of Technology, the University of Chicago, the states of California, Illinois, and Maryland, and the National Science Foundation. Ongoing CARMA development and operations are supported by the National Science Foundation under a cooperative agreement, and by the CARMA partner universities. 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.


  1. pagerange: Evidence of boosted CO/CO ratio in early-type galaxies in dense environmentsReferences
  2. pubyear: 2015
  3. http://www.purl.org/atlas3d
  4. http://idlastro.gsfc.nasa.gov/ftp/pro/astrom/hcongrid.pro
  5. Line flux is defined as being the emission detected by the interferometer in the CO masking aperture, summed over the channel range determined by the observer


  1. Aalto S., Beswick R., Jütte E., 2010, A&A, 522, A59
  2. Aalto S., Booth R. S., Black J. H., Johansson L. E. B., 1995, A&A, 300, 369
  3. Aalto S., Johansson L. E. B., Booth R. S., Black J. H., 1991, A&A, 249, 323
  4. Aalto S., Radford S. J. E., Scoville N. Z., Sargent A. I., 1997, ApJ, 475, L107
  5. Alatalo K., et al., 2011, ApJ, 735, 88
  6. Alatalo K., et al., 2013, MNRAS, 432, 1796
  7. Alatalo K., et al., 2014, ApJ, 780, 186
  8. Alatalo K., et al., 2015, ApJ, 798, 31
  9. Athanassoula E., Bureau M., 1999, ApJ, 522, 699
  10. Bacon R., et al., 2001, MNRAS, 326, 23
  11. Bower R. G., Lucey J. R., Ellis R. S., 1992, MNRAS, 254, 601
  12. Cappellari M., et al., 2011a, MNRAS, 413, 813
  13. Cappellari M., et al., 2011b, MNRAS, 416, 1680
  14. Carpenter J. M., Snell R. L., Schloerb F. P., 1990, ApJ, 362, 147
  15. Casoli F., Dupraz C., Combes F., 1992a, A&A, 264, 49
  16. Casoli F., Dupraz C., Combes F., 1992b, A&A, 264, 55
  17. Casoli F., Dupraz C., Combes F., Kazes I., 1991, A&A, 251, 1
  18. Combes F., Young L. M., Bureau M., 2007, MNRAS, 377, 1795
  19. Crocker A., et al., 2012, MNRAS, 421, 1298
  20. Crocker A. F., Bureau M., Young L. M., Combes F., 2011, MNRAS, 410, 1197
  21. Davies J. I., et al., 2012, MNRAS, 419, 3505
  22. Davis T. A., 2014, MNRAS, 445, 2378
  23. Davis T. A., Bureau M., Cappellari M., Sarzi M., Blitz L., 2013, Nature, 494, 328
  24. Davis T. A., et al., 2011, MNRAS, 417, 882
  25. Davis T. A., et al., 2012, MNRAS, 426, 1574
  26. Davis T. A., et al., 2013, MNRAS, 429, 534
  27. Digel S., Thaddeus P., Bally J., 1990, ApJ, 357, L29
  28. Faber S. M., Gallagher J. S., 1976, ApJ, 204, 365
  29. Gao Y., Solomon P. M., 2004, ApJ, 606, 271
  30. Garcia A. M., 1993, A&AS, 100, 47
  31. Hughes T. M., Cortese L., Boselli A., Gavazzi G., Davies J. I., 2013, A&A, 550, A115
  32. Kohno K., et al., 2002, PASJ, 54, 541
  33. Krips M., Crocker A. F., Bureau M., Combes F., Young L. M., 2010, MNRAS, 407, 2261
  34. McBride J., Alatalo K., Nyland K., 2015, MNRAS, 447, 392
  35. Milam S. N., Savage C., Brewster M. A., Ziurys L. M., Wyckoff S., 2005, ApJ, 634, 1126
  36. Ogle P., et al., 2010, ApJ, 724, 1193
  37. Paglione T. A. D., et al., 2001, ApJS, 135, 183
  38. Phillips T. G., et al., 1987, ApJ, 322, L73
  39. Pineda J. E., Caselli P., Goodman A. A., 2008, ApJ, 679, 481
  40. Polk K. S., Knapp G. R., Stark A. A., Wilson R. W., 1988, ApJ, 332, 432
  41. Ritchey A. M., Federman S. R., Lambert D. L., 2011, ApJ, 728, 36
  42. Sage L. J., 1990, A&A, 239, 125
  43. Sage L. J., Isbell D. W., 1991, A&A, 247, 320
  44. Sage L. J., Welch G. A., Young L. M., 2007, ApJ, 657, 232
  45. Sage L. J., Wrobel J. M., 1989, ApJ, 344, 204
  46. Schruba A., et al., 2011, AJ, 142, 37
  47. Shields G. A., Skillman E. D., Kennicutt Jr. R. C., 1991, ApJ, 371, 82
  48. Skrutskie M. F., et al., 2006, AJ, 131, 1163
  49. Sodroski T. J., 1991, ApJ, 366, 95
  50. Sonnentrucker P., Welty D. E., Thorburn J. A., York D. G., 2007, ApJS, 168, 58
  51. Spergel D. N., et al., 2007, ApJS, 170, 377
  52. Steiman-Cameron T. Y., Kormendy J., Durisen R. H., 1992, AJ, 104, 1339
  53. Utomo D., et al., 2015, ApJ in press (arXiv:1502.01017)
  54. van Dishoeck E. F., Black J. H., 1988, ApJ, 334, 771
  55. Welch G. A., Sage L. J., 2003, ApJ, 584, 260
  56. Wiklind T., Henkel C., 1989, A&A, 225, 1
  57. Wiklind T., Henkel C., 1990, A&A, 227, 394
  58. Wiklind T., Rydbeck G., 1986, A&A, 164, L22
  59. Young J. S., Sanders D. B., 1986, ApJ, 302, 680
  60. Young L. M., Bureau M., Cappellari M., 2008, ApJ, 676, 317
  61. Young L. M., et al., 2011, MNRAS, 414, 940
This is a comment super asjknd jkasnjk adsnkj
The feedback cannot be empty
Comments 0
The feedback cannot be empty
Add comment

You’re adding your first comment!
How to quickly get a good reply:
  • Offer a constructive comment on the author work.
  • Add helpful links to code implementation or project page.