THE JAMES CLERK MAXWELL TELESCOPE NEARBY GALAXIES LEGACY SURVEY. I. STAR-FORMING MOLECULAR GAS IN VIRGO CLUSTER SPIRAL GALAXIES

The JCMT NGLS¹⁹ focuses on the molecular gas and dust content of galaxies within 25 Mpc using three primary samples: (1) galaxies from the SINGS (Kennicutt et al. 2003) sample (see B. E. Warren et al. 2009, in preparation, for first results from the NGLS for additional SINGS galaxies); (2) an H i flux limited sample of galaxies from the Virgo Cluster; and (3) an H i flux limited sample of galaxies in the field. The first phase of the NGLS consists of CO J = 3–2 observations out to D₂₅/2 for 21 galaxies from the SINGS sample as well as the complete Virgo sample; this phase of the survey is currently complete. The second phase consists of CO J = 3–2 observations for the entire sample and began in 2009. The third phase will consist of SCUBA-2 observations of the entire sample and should begin in late 2009.

Specific science goals for the NGLS include searching for evidence of cold dust and measuring its mass fraction in galaxies of different types; measuring the amount of warm, dense molecular gas associated with star formation using the CO J = 3–2 line; comparing galaxies with similar morphologies and luminosities in the field and in the Virgo Cluster to determine the effects of cluster membership; using CO rotation curves to trace the dark matter distribution and the frequency of occurrence of nuclear gas concentrations which may feed central starbursts or black holes; and measuring the local submillimeter luminosity function and dust mass function to luminosities up to 100 times fainter than previous studies (e.g., Dunne et al. 2000).

Four large spiral galaxies (NGC 4254, NGC 4321, NGC 4569, and NGC 4579) are the only Virgo Cluster galaxies included in the SINGS survey (Kennicutt et al. 2003) and there is a large collection of information on their dust and ISM properties. Some basic properties of the galaxies are given in Table 1. The global 1–850 μm spectral energy distributions (SEDs) for these four galaxies are very similar (Dale et al. 2005), although the α parameter in the SED fit ranges from a high 3.8 in NGC 4579 to a low of 2.4 in NGC 4254. Draine et al. (2007) calculate the dust mass and gas-to-dust ratio for 65 galaxies in the SINGS sample. Of the four Virgo spirals discussed in our paper, only NGC 4579 stands out as having a factor of 3 larger gas-to-dust ratio; this is the second highest gas-to-dust ratio in the entire sample analyzed by Draine et al. (2007). This high gas-to-dust ratio may be a by-product of the difficulty in determining the 850 μm flux from the relatively low signal-to-noise SCUBA map (G. Bendo 2008, private communication). Morphologically, of the four Virgo spirals, NGC 4569 is the most centrally concentrated at 24 μm, while NGC 4254 is both the least centrally concentrated and the most asymmetric (Bendo et al. 2007). The star formation distribution based on Hα emission is classified as normal for NGC 4254 but truncated/normal for the other three galaxies (Koopmann & Kenney 2004a). Below we give some important details on each of the individual galaxies.

Recent H i observations of NGC4254 (M99) show a long tidal tail extending to the north; its lopsided morphology, high H i content, and high relative velocity suggest that this galaxy is undergoing its first encounter with the center of the Virgo Cluster (Haynes et al. 2007). The galaxy has a normal molecular gas to total gas fraction (Nakanishi et al. 2006).

NGC 4321 (M100) is classified as H ii/LINER (Ho et al. 1997). High-resolution CO J = 1–0 observations show two spiral arms and gas in the nucleus as well as faint emission bridging the arms and the nucleus (García-Burillo et al. 2005). The H i emission is truncated near the edge of the optical disk (Cayatte et al. 1990; Braun et al. 2007). NGC 4321 interacts with NGC 4322, as shown by the bridge between them that is visible in optical light but not in H i (Knapen et al. 1993). It also has a rather prominent circumnuclear ring with much enhanced star formation (Knapen et al. 1995a, 1995b).

NGC 4569 (M90) has been called an anemic spiral because of the small extent of its H i emission relative to its optical size (Sakamoto et al. 1999), although Koopmann & Kenney (2004a) prefer a designation of truncated/normal because its star formation rate per unit area is relatively normal. Bendo et al. (2007) note that there is almost no 24 μm emission from the extended disk. The CO J = 1–0 emission shows a compact bar oriented roughly north–south (Sakamoto et al. 1999) with complex kinematics indicating noncircular motions in the central arcminute (Jogee et al. 2005). This galaxy has a higher molecular gas to total gas fraction than NGC 4254, perhaps due to ram-pressure stripping of H i even within the stellar disk or a higher external pressure due to the intracluster medium (Nakanishi et al. 2006). However, the fact that the only remaining gas is in the inner disk, which tends to have a high molecular gas fraction, could also explain this result. Koopmann & Kenney (2004a) find that the galaxy has some H ii regions that appear to be outside the plane. They also note that rotation plus forward motion relative to the ICM can create an asymmetric ram pressure, which in simulations can produce a dominant extraplanar gas arm emerging from a truncated gas disk. Vollmer et al. (2004) also present evidence for ram-pressure stripping by comparing the kinematically distinct western arm with dynamical simulations. Boselli et al. (2006) argue that active gas stripping, rather than simply stopping gas inflow, is required to explain the star formation history in NGC 4569. Jogee et al. (2005) suggest that NGC 4569 is in the early stages of bar or tidally driven gas inflow.

NGC 4579 (M58) is classified as LINER 1.9/Seyfert 1.9 (García-Burillo et al. 2005). High-resolution CO J = 2–1 observations show two spiral arms approaching to within a few arcseconds of the active galactic nucleus (AGN; García-Burillo et al. 2005) and weak CO emission at the AGN itself (García-Burillo et al. 2008). At lower resolution, weak CO emission is detected throughout the disk (Nakanishi et al. 2006). Despite the weak emission, this galaxy has a higher molecular gas to total gas fraction than NGC 4254, and so may also have been affected by ram-pressure stripping or a higher external pressure (Nakanishi et al. 2006).

In summary, this small sample of four spiral galaxies probes a large fraction of the characteristics seen for luminous spirals in the Virgo Cluster such as H i deficiency, star formation rate, and the presence of nuclear activity. Thus, a detailed study of the molecular gas and star formation properties of these galaxies should provide insight into the effects on the ISM in galaxies and galaxy structure and evolution that can be addressed using the complete Virgo sample from the NGLS.

The CO J = 3–2 observations were obtained as part of the JCMT NGLS. We used the 16-pixel array receiver HARP-B with the Auto-Correlation Spectrometer Imaging System (ACSIS) configured to have a bandwidth of 1 GHz and a resolution of 0.488 MHz (0.43 km s⁻¹ at the frequency of the CO J = 3–2 transition). The galaxies were observed in raster mapping mode to cover a rectangular area corresponding to D₂₅/2 with a 1σ sensitivity of better than 19 mK (T*_A) at a spectral resolution of 20 km s⁻¹. The angular resolution of the JCMT at this frequency is 145 and the images were processed using 72761 pixels (the recommended pixel size in the reconstruction of HARP raster maps). Details of the observations are given in Table 2.

Table 2. Observational Parameters

Galaxy	ΔXa ('')	ΔYa ('')	P.A.a (o)	Dateb	$\overline{T_{\rm sys}}$ (K)	$\overline{\tau }$c (225 GHz)	tintd (s)	ΔTe (mK)	Resolution ('')
NGC 4254	162	138	0	0214, 0228, 0427	472	0.10–0.16	90	14	14.5
NGC 4321	228	186	30	0211, 0213	318	0.04–0.12	30	15	14.5
NGC 4569	312	138	23	0212, 0213	380	0.07–0.12	40	15	14.5
NGC 4579	168	132	95	0228, 0301	373	0.06–0.12	60	13	14.5

Notes. ^aSize and orientation of mapped region. ^bMonth, day; all observations done in 2008. ^cRange of atmospheric optical depth at 225 GHz on dates when data were obtained. ^dTypical integration time per point in the map. ^eThe rms noise in spectra at 20 km s⁻¹ resolution on T*_A scale; divide by 0.67 for T_MB.

Download table as:  ASCIITypeset image

Details of the reduction of the CO J = 3–2 data are given in B. E. Warren et al. (2009, in preparation) and so we give only a brief summary here. The individual raw data files were flagged to remove data from any of the 16 individual receptors with bad baselines and then the scans were combined into a data cube using a sinc(πx) sinc(kπx) kernel as the weighting function to determine the contribution of individual receptors to each pixel in the final map. A mask was created to identify line-free regions of the data cube and a third-order baseline was fit to those line-free regions. The clumpfind algorithm (Williams et al. 1994) implemented as part of the Starlink/cupid²⁰ task findclumps was used to identify regions with emission above three times the rms noise in a data cube that had been boxcar smoothed by 3 pixels and 25 velocity channels. Moment maps were created from the original data cube using the mask created by findclumps. The data were converted to the main beam temperature scale by dividing by η_MB = 0.67. Maps of the CO J = 3–2 integrated intensity are given in Figure 1 and are overlaid on images from the Digitized Sky Survey in Figure 2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The CO J = 1–0 first moment maps from Kuno et al. (2007) were downloaded from their survey Web site. These CO J = 1–0 data are already in the T_mb temperature scale. The angular resolution of the JCMT in the CO J = 3–2 line closely matches that of the Nobeyama 45 m telescope in the CO J = 1–0 line. Thus, no additional smoothing was applied.

For comparison with the maps from Kuno et al. (2007), the CO J = 1–0 first moment maps from Helfer et al. (2003) were downloaded from the NCSA Astronomy Digital Image Library; there was no CO J = 1–0 image for NGC 4254. The image for NGC 4579 did not include any single-dish (short-spacing) data (Helfer et al. 2003) and so must be considered a lower limit to the total CO J = 1–0 intensity. Each image was convolved with the appropriate two-dimensional Gaussian to give a smoothed image with a 145 circular Gaussian beam to match the JCMT resolution and converted from Jy beam⁻¹ km s⁻¹ to K km s⁻¹ using a conversion factor of 0.438 K Jy⁻¹.

The 24 μm data from the SINGS survey were reprocessed using an updated version of the data reduction description given in the SINGS Fourth Data Delivery User's Guide (SINGS Team 2006) with three modifications. First, a background offset related to the scan mirror position has been measured and removed from the data. Second, asteroids have been identified and removed from the data. Third, the updated flux calibration terms given by Engelbracht et al. (2007) have been applied to the data. To compare the 24 μm images to the JCMT data, we convolved the 24 μm images with special convolution kernels that match the profiles of the point-spread function (PSF) to the 145 Gaussian function that describes the JCMT beam. These convolution kernels were created using the empirical 24 μm PSF described by Young et al. (2009) and the prescription given by Gordon et al. (2008).

The Hα images were downloaded from the SINGS Web site²¹ (Kennicutt et al. 2003); there was no image for NGC 4569, so we used the image from the GOLD Mine Database (Gavazzi et al. 2003). Although the images were background subtracted, a weakly varying background across the image affected the flux levels once the images were convolved to match the JCMT resolution. We therefore binned the image by 100 pixels and clipped any data above 2σ to remove real emission, and then fitted a two-dimensional first-order polynomial to the resulting image. The data were calibrated as described in the SINGS release notes. The images were corrected for contamination by the two [N ii] lines using the [N ii]/Hα ratios from Prescott et al. (2007). Finally, we convolved the images with a 145 Gaussian to match the resolution of the JCMT.

All ancillary data images were aligned with JCMT image orientation and pixel size using the kappa command wcsalign. We calculated an image of the star formation rate surface density for each galaxy by combining the Hα and 24 μm images using the formula in Calzetti et al. (2007). Maps of the star formation rate for each galaxy are given in Figure 3. For NGC 4579, the central AGN dominates the emission at both Hα and 24 μm. Thus, Figure 3 does not provide an accurate measure of the star formation rate in the central regions of NGC 4579.

Zoom In Zoom Out Reset image size

All calibrations to date of the conversion from CO luminosity to molecular gas mass have been for the ground-state J = 1–0 transition (Strong et al. 1988). Thus, the CO J = 3–2/J = 1–0 line ratio is an important quantity to measure in determining molecular gas masses. In addition, the CO J = 3–2/J = 2–1 line ratio has been suggested previously to be a good indicator of temperature in giant molecular clouds (GMCs), with larger ratios corresponding to larger temperatures (Wilson et al. 1997). If the CO J = 2–1/1–0 line ratio is relatively uniform (Leroy et al. 2008), then the CO J = 3–2/J = 1–0 line ratio would also be a good temperature indicator. Alternatively, the line ratio can be increased if the average density of the gas is increased (see, e.g., Petitpas & Wilson 2000, for a discussion of this effect in M82).

High-resolution CO J = 1–0 images for NGC 4321 and NGC 4569 have been published by Helfer et al. (2003) and Kuno et al. (2007). The total CO J = 1–0 fluxes measured directly from these publicly available integrated intensity maps agree to better than 5%. However, a comparison of the two images pixel by pixel shows systematic differences between the two data sets. In particular, the images from Helfer et al. (2003), which are produced by combining interferometric and single-dish data, yield higher intensities around bright and more compact structures compared to the images from Kuno et al. (2007). As such systematic differences could have been produced in the combination of the single-dish and interferometric data (which is a tricky and nonlinear process), we have chosen to calculate the CO J = 3–2/J = 1–0 line ratios using the data from Kuno et al. (2007). Images of the line ratios are shown in Figure 4. In calculating the average values for each galaxy and for regions within each galaxy, we limited our analysis to pixels with a signal-to-noise in the CO J = 3–2 line greater than 3. Global and central values for the line ratios as well as line ratios for particular emission peaks within the disk are given in Table 3.

Zoom In Zoom Out Reset image size

NGC 4254 has a remarkably uniform CO J = 3–2/J = 1–0 ratio of 0.33. NGC 4321 has a similar disk-averaged CO J = 3–2/J = 1–0 ratio of 0.36; however, the average value in the central 9 pixels (22'' diameter) is significantly larger at 0.79. NGC 4569 shows a significant gradient from north to south in its line ratio, from 0.53 at the northern CO J = 3–2 peak to just 0.06 at the southern peak. Its disk-averaged value of 0.25 is somewhat lower than the other two galaxies. Finally, NGC 4579 is a difficult case. It is detected weakly by Kuno et al. (2007) but is only marginally detected at CO J = 3–2 (Figure 1). Our marginal detection near the star formation peak in the disk is 0.9 ± 0.3 K km s⁻¹, while Kuno et al. (2007) detect CO J = 1–0 emission of 6 K km s⁻¹. Combining these two values suggests that the CO J = 3–2/J = 1–0 line ratio in the disk of NGC 4579 is similar to the low line ratio seen in the southern portion of NGC 4569.

Assuming a 15% absolute calibration uncertainty for both CO data sets (Section 2), the mean line ratios in the three galaxies with good CO J = 3–2 detections agree at the 1σ level. However, the southern half of NGC 4569 has a significantly lower line ratio while the central region of NGC 4321 and the northern emission peak of NGC 4569 have significantly higher line ratios. These line ratios suggest that the gas may be on average somewhat warmer and/or denser in the center of NGC 4321 and north of NGC 4569, while the southern disk of NGC 4569 may be somewhat cooler or less dense than typical disk regions. An intriguing possibility is that the gradient in gas properties in NGC 4569 is related to its interaction with the ICM; we discuss this possibility in Section 5.

The average CO J = 3–2/J = 1–0 line ratios in these three galaxies are somewhat smaller than the range of values (0.4–0.8) seen in individual GMCs in M33 (Thornley & Wilson 1994; Wilson et al. 1997). The line ratios do fall within the range of ratios (0.2–0.7) measured with similar angular resolution by Mauersberger et al. (1999), albeit at the low end of the range. However, Mauersberger et al. (1999) observed only the central 21'' of each galaxy in their sample. If many galaxies have enhanced line ratios in the central region, as we see in NGC 4321 and possibly in NGC 4569, this focus on the central regions could explain their somewhat higher line ratios. Our line ratio for NGC 4569 is in reasonable agreement with the lower-resolution results from Hafok & Stutzki (2003), who measured a line ratio of 0.23 ± 0.04. However, our line ratios for the other two galaxies are substantially larger than the values measured by them: 0.21 ± 0.04 for NGC 4254 and 0.24 ± 0.02 for NGC 4321. Applying a 3σ clip to the CO J = 3–2 data has had the effect of limiting our line ratio analysis in these galaxies to brighter regions; for example, the interarm regions of NGC 4321 are not detected at this level. If the interarm gas has a substantially lower line ratio, this could explain the difference between our values and those obtained by Hafok & Stutzki (2003).

The CO J = 3–2 luminosities and molecular hydrogen gas masses are given in Table 4. Values for the center of NGC 4579 are calculated using a 3σ upper limit and a Gaussian line with a full width at half-maximum of 200 km s⁻¹. We give two different estimates for the molecular gas mass. The first value is calculated directly from the CO J = 3–2 luminosity assuming a CO J = 3–2/J = 1–0 line ratio of 0.6 and adopting a CO to H₂ conversion factor of 2 × 10²⁰ cm⁻² (K km s⁻¹)⁻¹ (Strong et al. 1988). The CO J = 3–2/J = 1–0 line ratio of 0.6 is adopted as a typical ratio appropriate to the gas in GMCs. It is based on observations of an average CO J = 2–1/J1–0 ratio of 0.8 in spiral disks (Leroy et al. 2008) combined with the average CO J = 3–2/J = 2–1 ratio of 0.76 ± 0.19 measured in the Galactic molecular cloud M17 (Wilson et al. 1999). A similar line ratio (0.60 ± 0.13) is observed in the central regions of 15 nearby galaxies with modest starbursts (Israel 2008; F. P. Israel 2009, in preparation). Adopting a line ratio specifically appropriate for molecular clouds essentially uses the CO J = 3–2 emission to focus on the star-forming molecular gas; the CO J = 1–0 line has been shown to include emission from lower opacity, more diffuse molecular gas (Wilson & Walker 1994; Rosolowsky & Blitz 2005) which likely does not participate directly in massive star formation.

Table 4. Molecular Gas Masses

Galaxy	SFRa (M☉ yr−1)	LCO(3–2) (108 K km s−1 pc2)	$M_{{\rm H}_2}$b (R = 0.6) (109 M☉)	$M_{{\rm H}_2}$c (R = 0.34) (109 M☉)	$M_{{\rm H}_2}$d (Kuno) (109 M☉)	$M_{{\rm H}_2}$e (Helfer) (109 M☉)
NGC 4254	4.8	5.6	3.0	5.3	6.5	⋅⋅⋅
NGC 4321	2.7	3.7	2.0	3.5	3.1	6.50
NGC 4569	1.1	1.3	0.67	1.2	3.0	2.40
NGC 4579	<0.9	<0.19	<0.10	<0.18	1.4	⋅⋅⋅

Notes. ^aTotal star formation rate derived from Hα and 24 μm images shown in Figure 3 and described in Section 2. Note that the true central star formation rate may be overestimated if nuclear activity contributes substantially to the 24 μm flux; this is certainly the case in NGC 4579. ^bMass of molecular hydrogen gas calculated from the CO J = 3–2 luminosity and assuming a CO J = 3–2/J = 1–0 line ratio of R = 0.6, see Section 3 for details. ^cMass of molecular hydrogen gas calculated from the CO J = 3–2 luminosity using the measured average CO J = 3–2/J = 1–0 line ratio of R = 0.34, see Section 3 for details. ^dMass of molecular hydrogen gas measured from the CO J = 1–0 images from Kuno et al. (2007) and using the CO-to-H₂ conversion factor (2 × 10²⁰ cm⁻² (K km s⁻¹)⁻¹) and distance (16.7 Mpc) adopted in this paper. ^eMass of molecular hydrogen gas from Helfer et al. (2003) measured from the CO J = 1–0 line using the NRAO 12 m telescope, adjusted to a distance of 16.7 Mpc.

Download table as:  ASCIITypeset image

Alternatively, we could adopt the average CO J = 3–2/J = 1–0 line ratio measured in our sample, which is 0.34 averaged over the three galaxies in our sample. This mass estimate essentially measures the total molecular hydrogen content and should be directly comparable (modulo issues of area coverage and sensitivity) to previous estimates of the molecular gas mass derived directly from CO J = 1–0 maps. Table 4 also gives masses measured directly from the data of Kuno et al. (2007) as well as the masses estimated by Helfer et al. (2003) from their single-dish maps. For NGC 4254 and NGC 4321, the CO J = 1–0 masses from Kuno et al. (2007) are in general agreement with the masses calculated from the CO J = 3–2 line when the line ratio of 0.34 is adopted, as expected given the quite uniform line ratio seen in these galaxies. However, for NGC 4569, the CO J = 1–0 mass is almost a factor of 3 times larger than the CO J = 3–2 mass. This larger mass is partly a result of the fact that the CO J = 3–2/J = 1–0 ratio in this galaxy is the lowest of the three (accounting for a factor of 1.4 in mass) and also that the southern region of the galaxy is quite bright in CO J = 1–0 but almost invisible in CO J = 3–2. For NGC 4579, the CO J = 1–0 mass is a factor of 10 larger than the upper limit to the CO J = 3–2 mass, again consistent with the very weak CO J = 3–2 emission in this galaxy as a whole.

How are we to understand the variety of gas masses given in Table 4? The most straightforward interpretation is that the gas mass calculated from the CO J = 3–2 luminosity using a CO J = 3–2/J = 1–0 line ratio of 0.6 (as seen in Galactic GMCs) traces the dense molecular gas, while the mass calculated directly from the CO J = 1–0 luminosity traces the total molecular gas mass. With this interpretation, the dense gas fraction in the galaxies in our sample is 46% for NGC 4254, 65% for NGC 4321, 22% for NGC 4569, and <7% for NGC 4579. However, this simple picture is muddied by the disagreement between the CO J = 1–0 data of Helfer et al. (2003) and Kuno et al. (2007). Although the total fluxes in the published CO J = 1–0 integrated intensity images for NGC 4321 and NGC 4569 are very similar, the total mass from the single-dish maps of Helfer et al. (2003) is a factor of 2 larger in NGC 4321 and a factor of 0.8 smaller in NGC 4569, such that the dense gas fraction in the two galaxies becomes quite similar if we use the masses from Helfer et al. (2003). While some of these variations may be attributed to such factors as calibration uncertainties (to explain NGC 4569) and larger mapping area (for NGC 4321), the fact remains that considerable uncertainties, perhaps as large as a factor of 2, continue to plague these gas mass estimates.

How do the variations in the gas and star formation properties of these four galaxies relate to their internal structure and larger environment? NGC 4254 has been suggested to be on its first encounter with the central regions of the cluster and, indeed, has a normal ratio of molecular to atomic gas and a normal extent of star formation. It is also the only galaxy in our sample that does not show evidence of a bar (Laine et al. 2002). Its smaller gas depletion time translates into a higher current star formation efficiency; this higher efficiency may be due to the fact that the galaxy has only begun to be disturbed by the cluster (e.g., Haynes et al. 2007).

NGC 4321 is classified as weakly barred (SAB(s)bc; Buta et al. 2007); high-resolution CO J = 1–0 observations show two spiral arms and gas in the nucleus as well as faint emission bridging the arms and the nucleus (García-Burillo et al. 2005). There is also a prominent circumnuclear ring with enhanced star formation (Knapen et al. 1995a, 1995b). Thus, the similar gas depletion times calculated for the central region and the spiral arms suggest that increased star formation is balanced by an increased molecular gas content in these regions.

For NGC 4569, the somewhat lower CO J = 3–2/J = 1–0 ratio of 0.25 suggests that the average gas in NGC 4569 may be somewhat cooler and/or less dense than the gas in NGC 4254 or 4321. However, this potential difference in average gas properties does not seem to have a strong effect on the gas depletion time (or star formation efficiency) estimated using the CO J = 3–2 line. Given the nearly edge-on orientation of NGC 4569, it is possible that the contribution from CO J = 1–0 emission in the outer disk of the galaxy contributes significantly to the measured line ratio, while the CO J = 3–2 is confined to the inner disk regions along with the star formation. NGC 4569 is the largest of the Virgo spirals in our sample, yet its CO J = 3–2 emission is the most compact. This morphology suggests that whatever processes are acting to strip the outer portions of the galaxy of gas are also affecting the distribution of the dense molecular gas, perhaps by stripping it directly or by removing the reservoir (atomic hydrogen or diffuse molecular hydrogen) from which it forms. The southern portion of the galaxy contains much less dense molecular gas and star formation than the northern half. Given the geometry of the galaxy (Vollmer et al. 2004), the northern portion seems likely to be more affected by interaction with the ICM as its rotational motion and motion through the cluster are in the same sense. The higher density or warmer gas in the northern half of NGC 4569 could possibly be due to gas compression in this region where ram pressure is active.

NGC 4579 appears to be the most unusual galaxy of our small sample. Its disk has the smallest gas depletion time (highest star formation efficiency) of any of the four galaxies in our sample. The relatively weaker star formation in this galaxy as well as its more limited spatial extent (Figure 3) suggests a galaxy which may be exhausting its fuel for star formation. Its relatively early morphological type (SAB(rs)ab; Buta et al. 2007) is consistent with this picture, although NGC 4569 has an identical classification. However, the higher inclination of NGC 4569 means that a given surface density of molecular gas is more easily detected than in NGC 4579. Indeed, the relatively weak CO J = 3–2 emission in the southern half of NGC 4569 would be undetectable if it were at the same inclination as NGC 4579. In addition, the broader lines (full width at zero intensity 400–500 km s⁻¹) seen in the central region of NGC 4579 by García-Burillo et al. (2008) also imply that its central CO emission is harder to detect than the central emission in NGC 4569. Thus, NGC 4579 may not be such an unusual galaxy for its type, but simply observed at a less favorable inclination angle than NGC 4569.

In Section 4, we calculated gas depletion times by directly combining the CO J = 3–2 intensity with the star formation rate. If instead we use the CO J = 1–0 intensity from Kuno et al. (2007), and study exactly the same regions in each galaxy as in Section 4, the picture becomes more complicated. Values for the gas depletion time calculated using the CO J = 1–0 line are also given in Table 5.

For NGC 4254, the gas depletion times remain very uniform when the CO J = 1–0 line is used instead of the J = 3–2 line to trace the molecular gas. The average gas depletion time shows a larger dispersion, which hints at more local variations than are seen using the CO J = 3–2 line; however, this could also be due to low signal-to-noise CO J = 1–0 spectra being included in our analysis.

For NGC 4321, when the CO J = 1–0 line is used, the line ratio is also quite uniform overall but the gas depletion time for the central region is now roughly twice as small as that of the disk. The gas depletion times in the CO peaks associated with the spiral arms are intermediate between these two extremes. The disk-averaged gas depletion time remains about 50% larger than that of NGC 4254 whether the CO J = 1–0 or J = 3–2 transition is used to trace the molecular gas.

For NGC 4569, the gas depletion times now show significant variations, ranging from 9.0 Gyr at the southern CO J = 3–2 peak to 1.8 Gyr for the central region. The disk-averaged value of 5.4 Gyr, with its larger relative dispersion, also suggests more variation in the gas depletion time derived from the CO J = 1–0 line, and is now almost 3 times larger than the gas depletion time in NGC 4254.

Thus, for these three galaxies, using the CO J = 1–0 line to calculate the gas depletion time instead of the J = 3–2 line results in a greater spread in gas depletion times both within galaxies and from one galaxy to another. If we assume that the star formation efficiency (and hence gas depletion time) should not vary significantly from one normal spiral galaxy to another, then this analysis suggests that the CO J = 3–2 line may be a more direct tracer of the dense, star-forming molecular gas than is the CO J = 1–0 line. This result is consistent with the analysis of Iono et al. (2009), who find a nearly linear correlation between the CO J = 3–2 luminosity and the far-infrared luminosity (as a tracer of star formation) over 5 orders of magnitude in galaxy luminosity. The good correlation between the star formation rate surface density and the CO J = 3–2 emission suggests that the CO J = 3–2 transition probes sufficiently high gas densities to be a good tracer of the dense cores within molecular clouds that are forming stars with a high efficiency. In contrast, the CO J = 1–0 line traces gas with a wider range of densities, including more diffuse material that is not directly involved in star formation. In this interpretation, using the CO J = 3–2 line results in a more accurate measurement of the instantaneous gas depletion time.

An alternative interpretation would be that the star-forming molecular gas (and the gas depletion time) is in fact well traced by the CO J = 1–0 line, and the variations seen both within and between galaxies are real. In this case, regions with higher star formation efficiencies (or smaller gas depletion times) would be those with a higher rate of energy injected due to recent star formation per unit mass of molecular gas. This higher rate of energy injection could warm the molecular gas, producing brighter CO J = 3–2 emission and leading to a more uniform apparent gas depletion times when the CO J = 3–2 emission line is used, as observed. While such heating processes are undoubtedly at work in the ISM, we argue that the interpretation of the CO J = 3–2 line as a direct tracer of dense star-forming gas is the more straightforward explanation.