The 50 Mpc Galaxy Catalog (50 MGC): Consistent and Homogeneous Masses, Distances, Colors, and Morphologies

Our catalog is formed from three base sources, the Local Volume Galaxy (LVG) catalog (Karachentsev et al. 2013), HyperLeda (Makarov et al. 2014), and the NASA-Sloan Atlas (NSA). ⁹ All three contain a wide variety of information, including redshifts and photometry. The strength of the NSA catalog is its uniform Sloan Digital Sky Survey (SDSS) photometry, which extends to quite faint sources, however, this photometry only extends over about one-third of the sky. The LVG and HyperLeda catalogs combine hetereogenous data sets and are continuously maintained and updated. The LVG catalog contains nearly complete measurements of galaxies brighter than M_B ∼ − 12 within 10 Mpc, but it lacks optical color information for the galaxies. Both NSA and HyperLeda extend to greater distances and have optical color information. Combined, these catalogs can provide the data needed to obtain distances, masses, and Hubble types for galaxies out to a maximum target distance of 50 Mpc.

From HyperLeda, we initially began with a subsample of 18726 objects designated as galaxies and with radial velocity <3500 km s⁻¹. We chose to remove the ∼20% of objects that were missing absolute B-band magnitude values in the column mabs from our sample, because the HyperLeda calculation of the absolute magnitude relied upon apparent B-band magnitude, distance modulus, and extinction. Removal of these objects from the sample ensured the inclusion of the most essential galaxy measurements. This cut the initial HyperLeda sample down to 14941 objects.

For the LVG subsample, we combined the tables from the "Catalog of Nearby Galaxies" and "Global Parameters of the Nearby Galaxies" that are available from the LVG catalog website. ¹⁰ The combined tables totaled 1240 galaxies. ¹¹

Finally, for the NSA data, we used the nsa_v1_0_1 catalog and found 8242 objects with cz < 3500 km s⁻¹. We also added available SDSS spectral line flux data to these objects, matched by plate and fiber IDs. We note that the NSA catalog includes not just objects with SDSS redshift information, but also galaxies with redshifts measured from other sources (ALFALFA, NED, 6dF, 2dF, and ZCAT).

2.1.1. Combining the Sample

The three subsamples contain both duplicate objects and contaminants. Here, we describe how we combined the catalogs while eliminating duplicates and contaminants. We note that the presence of a galaxy in each input catalog is given by flags in our catalog: hl_obj, lvg_obj, and nsa_obj. If a galaxy is present in that catalog, the flag has a value of 1, while if it is not, it has a 0 value.

The HyperLeda sample included some duplicate galaxy entries. To remove them, we crossmatched each entry to its nearest neighbor within the sample. We followed up on galaxies with nearby neighbors within a 2' separation radius for galaxies within 20 Mpc, and a 1' separation for galaxies between 20 and 50 Mpc. For the 580 potential duplicate galaxies, we examined available data and flagged them as follows: 0: Individual galaxy with the best available data. 1: Duplicated galaxy with fewer available data. 2: Off-nuclear position in duplicate galaxy. 3: Image shows a bright or star-forming region within a larger galaxy. We include only objects flagged with 0, which removed 183/580 galaxies. This cut resulted in an initial HyperLeda sample of 14758 galaxies.

We first created a combined LVG and HyperLeda sample. We used astropy.coordinates.SkyCoord to conduct a coordinate-based crossmatch with a maximum separation distance of 30''. Of the 1240 galaxies in the LVG, 744 were found in HyperLeda. The 496 unmatched LVG galaxies almost all have a very low luminosity. This explains their exclusion, as these galaxies are likely to lack the photometric and redshift measurements required to have been included in our initial HyperLeda sample. When combining sources between these catalogs, we prioritize the LVG galaxy properties in almost all cases, i.e., radial velocities, B-band magnitudes, and distances (see Section 3).

We then crossmatched NSA catalog objects to our combined HyperLeda and LVG samples using a maximum separation distance of 30''. Of the 8242 NSA objects, 6966 matched the combined HyperLeda and LVG catalog. For the the 1276 unmatched objects, we found that many were contaminants. To distinguish these contaminants from real galaxies, we visually inspected the SDSS imaging of each galaxy and compared redshifts to those listed in the NED. At the lowest redshifts, the contaminants included foreground stars on top of image artifacts or distant galaxies. At larger distances, bright portions of galaxies with separate spectroscopic measurements from SDSS were often included as separate catalog entries. Some contaminants were also just image artifacts, and a small number contained no obvious image source whatsoever. Figure 1 shows examples of NSA contaminant objects. We flagged each of the 1276 unmatched NSA objects, based on our findings: 1: Image shows a bright or star-forming region within a larger galaxy. 2: Image shows no object, an imaging artifact, a star, or a foreground star over a high-redshift galaxy. 3: Duplicates of otherwise good galaxies, of which we flagged the entry with the most reliable redshift source to keep, in order, SDSS, NED, and ZCAT. 4: Good imaging of a galaxy with a redshift matching the redshift in NED. These qualtity flags are given for all 1276 galaxies in Table 1. Only objects in category 4 are considered as possible unique new NSA galaxies to add to the catalog.

Zoom In Zoom Out Reset image size

Table 1. NSA Objects that Do Not Match Galaxies in the Hyperleda and LVG Catalogs

NSAID	R.A.	Decl.	Quality_FLAG	MATCH_RATIO
30129	343.8028213	−0.350540038	2	191.944
34438	13.23646298	−1.174127269	2	225.389
36914	24.69818421	1.071791959	2	371.004
38419	32.6382837	0.738166461	4	nan
41803	8.850357094	14.79454207	2	122.785
42629	14.93267592	14.56776754	2	331.128
46179	116.1573335	40.88379846	4	151.729
50066	129.4311915	51.64175054	1	0.752
51666	134.7746823	53.76023531	1	0.707
52302	138.5671796	57.04441951	4	318.963

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

Finally, to minimize the potential for duplicate galaxies, we wished to ensure that none of the unique NSA objects in category 4 were observations of the edge of a galaxy that was already included in our sample. To do this, we first needed uniform angular diameter measurements for each sample galaxy.

HyperLeda gives galaxy diameter measurements as the major axis using a standard 25 mag arcsec⁻² isophote (d25), whereas LVG uses a diameter measurement corresponding to the Holmsberg isophote (∼26.5 mag arcsec⁻²; which we call d26). We compared d25 HyperLeda diameters to d26 LVG diameters for galaxies found in both source catalogs to find their typical ratio, and then multiplied the LVG diameters by ∼0.838 to approximate d25 HyperLeda values for all the nearest-neighbor galaxies.

For each NSA object, we then divided the angular separation from its nearest-neighbor's angular diameter to create a match separation ratio, δ_sep/d25. Of the galaxies that were confirmed to be included in our sample, we only included those with a ratio δ ≥ 0.75 (i.e., lying beyond 1.5 galaxy radii). Of the 208 objects in category 4, 170 galaxies passed this cut and were added to our final sample. These galaxies have a median distance of 29.9 Mpc and a median mass of 8.08 × 10⁷ M_⊙ ($\mathrm{log}(M)\sim$ 8), and thus they add significantly to our sample.

We note that many studies (e.g., Geha et al. 2012; Reines et al. 2013) have used the NSA catalog as a sample of nearby galaxies without removing contaminants. We find that ∼13% of the nearby NSA galaxies (1106/8242) are contaminants. Most contaminants show a low-mass estimate from NSA redshift distances and photometry, with a median mass of 2.6 × 10⁷ M_⊙. We therefore include the full list of NSA objects that do not match HyperLeda and LVG and the criteria we used to cut these down in Table 1 for future users.

2.1.2. Galaxy Coordinates

The luminosities and colors of our galaxies are two crucial parts of our catalog. In addition to being intrinsically useful, the colors are also important for calculating stellar masses (Section 4) and for assessing the galaxy type (early versus late; Section 5). To obtain these estimates, we use four different sources for colors and associated luminosities: (1) HyperLeda, (2) NSA, (3) the Siena Galaxy Atlas (SGA), and (4) the NASA/IPAC Extragalactic Database (NED). All luminosity calculations use the best distances described in Section 3.

Extinction Correction: For our foreground-extinction corrections, we uniformly apply the Schlafly & Finkbeiner (2011) corrections; all our sources have Schlegel et al. (1998) extinction estimates available, and we multiply these by 0.86 to convert them into Schlafly & Finkbeiner (2011) values. ¹² For HyperLeda, we used B and V magnitudes, and scaled their A_B as described above. To determine A_V values, we assumed an A_V /A_B = 0.769 (Cardelli et al. 1989). For NSA, we used the provided Schlegel et al. (1998) extinction values for each band and scaled them by 0.86. We retrieved a table of location-based extinction values for our entire catalog from the IPAC Galactic DUST web service and added the table's Schlafly & Finkbeiner (2011) E(B − V) values to the catalog as EBV_irsa. For SGA, we then created band-specific extinction arrays by multiplying EBV_irsa by the proper A/E(B − V) values: g = 3.303, r = 2.285. Finally, for NED, we used B- and R-band magnitudes from the same source for consistency—these primarily come from the southern ESO Uppsala survey (Lauberts & Valentijn 1989); we used the HyperLeda B-band extinction, and assumed an A_R /A_B = 0.628 based on the extinction curves of Cardelli et al. (1989) and O'Donnell (1994), taken from the Padova CMD website. ¹³

We did not apply any internal extinction corrections, thus the luminosities and colors presented here are only corrected for foreground extinction. Internal extinction values are available from both HyperLeda and LVG. However, we found no correlation between these two.

Luminosities and Colors from HyperLeda ((B − V)₀ , L_V , L_B ): HyperLeda provides two different B − V colors: (1) The total asymptotic corrected color, bvtc, which is corrected for both internal and foreground extinction, and (2) the color measured within the effective aperture in which half the total B flux is emitted, bve, which is not corrected for extinction. Various HyperLeda galaxies have one or both measurements available. Our first step to obtain consistent values was to undo the HyperLeda extinction corrections on the bvtc column, and apply the Schlafly & Finkbeiner (2011) foreground-extinction correction to both the total and effective colors—we call these extinction-corrected quantities (B − V)_t,0 and (B − V)_e,0 respectively. Because we are using the colors to calculate total masses, we also attempt to correct the small offset between the (B − V)_t,0 and (B − V)_e,0—the difference between these colors for galaxies that have both values is shown in Figure 2. The offset is clearly color dependent, and the line in that figure shows the color-dependent correction applied to the (B − V)_e,0 to obtain our final (B − V)₀ values for galaxies that only had bve data available. We flagged these data using the bvcolor_f value in our table.

Zoom In Zoom Out Reset image size

To calculate luminosities and total magnitudes in B and V bands, we found that the given bt and vt columns were not consistent with the given colors (after correcting for the different extinction corrections applied)—this is likely due to the heterogeneity of the data sources used. Given the ubiquity of btc values, we first calculate the B total magnitude by removing the extinction corrections and applying Schlafly & Finkbeiner (2011) foreground-extinction corrections to obtain the total B_t,0 values. We then derived the total V-band magnitude by using our B_t,0 and subtracted the (B − V)_t,0 value to obtain a foreground-extinction V-band magnitude, V_t,0. We note the use of B_t,0 values throughout this paper when we plot galaxy luminosities. These are transformed into luminosities using the Vega-based absolute magnitudes of the Sun from http://mips.as.arizona.educnaw/Sun.html.

Luminosities and Colors from NSA ((g − i)₀ , L_i , L_B ): NSA provides Petrosian flux and extinction values for each FNugriz band. We calculate the Pogson g-, r-, and i-band magnitudes using the SDSS relation ¹⁴ : $m=22.5\mbox{--}2.5{\mathrm{log}}_{10}f$. We then used extinction correction for all individual magnitudes as described above. To calculate galaxy masses (Section 4), we use the g₀ and i₀ magnitudes to calculate an extinction-corrected (g − i)₀ color. We also calculate an i-band luminosity for the galaxies (L_i ) using the solar AB magnitude from http://mips.as.arizona.edu/cnaw/Sun.html. To analyze a single luminosity distribution for our entire sample, we also estimate the B-band luminosity L_B from the SDSS magnitudes using the stellar transformation B₀ = g₀ + 0.39*(g − r)₀ + 0.21 from Jester et al. (2005).

Luminosities and Colors from SGA ((g − r)₀ , L_r , L_B ): A complementary source of photometry for primarily southern galaxies is the Siena Galaxy Atlas (SGA ¹⁵ ). We matched our catalog to the SGA catalog after removing sources with known redshifts above 3500 km s⁻¹ using a maximum separation distance of 30''. In cases of multiple galaxies within this threshold, we matched the galaxy with the smallest angular separation. This gave us (g − r)₀ data for 7681 galaxies, 1616 of which otherwise lacked color.

The SGA catalog contains a number of different magnitude measurements. To ensure consistency, we compared overlapping sources with NSA and found that [G,R,Z]_MAG_SB26 matched the NSA magnitudes best. We then calculated reddenings and extinction using the IRSA online portal ¹⁶ to obtain Schlafly & Finkbeiner (2011) extinction values. We corrected for these values to calculate the (g − r)₀ color, L_r , and estimate the B-band luminosity L_B using the same Jester et al. (2005) equation as we applied to NSA photometry.

Luminosities and Colors from NED ((B − R)₀ , L_R , L_B ): To expand our mass estimations for southern galaxies, we used astroquery to request NED photometry for each sample object. We added available B and R data to the sample, specifically, objects with an observed passband of B (Cousins) (B_25) and R (Cousins) (R_25). If multiple magnitude measurements were available for a single object, we included a mean value. Queries returned a uniform B and R magnitude error of 0.09 for all objects. We used these B- and R-band magnitudes, corrected for foreground extinction, to calculate the (B − R)₀ color, L_R , and L_B where available.

Combined Color and Luminosity Estimates: In Section 5 we describe color transformations between (g − i)₀ and other colors. We use these to derive an estimate of the (g − i)₀ color for all galaxies with colors. This is contained in column gi_color. For galaxies with multiple available colors, we give precedence to source colors in order: NSA (g − i)₀, and then synthesized (g − i)₀ values based on (g − r)₀, (B − V)₀, and (B − R)₀. We compile the L_B values from all sources in the column B_lum; most of these are taken directly from HyperLeda, while a handful are synthesized from other sources as described above, with the precedence order being HyperLeda, NED, NSA, and SGA. A vast majority of L_B estimates come directly from Hyperleda. The final distribution of L_B estimates, separated by preferred photometric source, is shown in Figure 3.

Zoom In Zoom Out Reset image size

Due to a number of objects with both NSA and HyperLeda data available, we wished to test which source provided more reliable photometry when both were available. We compared g versus B magnitudes, due to their similar wavelength, and investigated galaxies with significantly differing magnitudes to available NED values. Generally, by comparing to available photometry compiled in NED, we find that NSA photometry is more reliable in cases where it is discrepant from HyperLeda, and we therefore typically prioritize NSA data in estimating galaxy masses. An exception to this is for a selection of galaxies with NSA g > 16 and B − g > 1; for these galaxies, we preferentially use Hyperleda values, as the HyperLeda values matched other measurements from NED more closely. Based on this analysis, we created the mag_flag column to signify which source provided the more accurate measurement at a given magnitude—0: The difference between B and g are within one magnitude, so NSA values are used for calculating masses. 1: B and g are discrepant; NSA photometry is more reliable. 2: B and g are discrepant; HyperLeda photometry is more reliable, and is used to estimate masses.

There are 446 additional objects for which the NSA photometry is not used. These include objects with flags indicating unreliable photometry, specifically those with the fifth or sixth bitwise flag set in the NSA DFLAGS field (M. Blanton, private communication). We also do not use NSA photometry for 20 objects with g or i petrosian flux ≤0 nanomaggies. Of these 446 galaxies, 306 had other colors and luminosities available to calculate masses.

Due to the peculiar motions of galaxies, redshift-based distances are not always accurate in the nearby Universe. However, this accuracy can be improved by using models of the local mass density based on the measured peculiar motions of galaxies (Courtois et al. 2012). We chose to use the Cosmicflows-3 Distance–Velocity calculator (Kourkchi et al. 2020) because it covered the full distance range of our sample using an up-to-date local mass density model (Graziani et al. 2019). We entered heliocentric velocities into the calculator to retrieve distance estimates that were corrected for the local mass density. Velocity measurements are given precedence in the order NSA, LVG, and HyperLeda where available.

To estimate the errors on the redshift-based distances, we first estimated a typical peculiar velocity for galaxies in the Local Volume. We based this on galaxies with redshift-independent measurements (see the next subsection) with errors <10%. We then used the CMB-relative velocity measurements from HyperLeda to calculate the peculiar velocity of each galaxy using v_pec = v_rad − H₀ d assuming H₀ = 70. We then measured the width of this distribution using the 16th and 84th percentiles, taking half the width to obtain a typical peculiar velocity of ∼400 km s⁻¹.

Our redshift-based distance errors were then calculated by adding and subtracting this typical peculiar velocity from the heliocentric velocity and recalculating the distance using the Cosmicflows-3 distance–velocity calculator. The final redshift-dependent distance error uses a symmetrization of the difference between upper- and lower-limit corrected distances. These distance errors are typically ∼65.3% at 10 Mpc and ∼17.4% at 40 Mpc.

We use redshift-independent distances from two sources: (1) the updated LVG catalog distances, and (2) the compilation of redshift-independent distances in the NED distance database (NED-D; Steer et al. 2017). We chose to give the highest priority to the LVG catalog distances, as these distances typically represent the best available measurements and are kept up to date. We manually examined distance estimates where the values from LVG conflicted with those NED-D and NED-MED, and found LVG to typically be superior; for instance, the distances to the Maffei group galaxies were based on the recently updated TRGB distances presented in Anand et al. (2019).

For galaxies in NED-D, we chose our distances using the following list of primary indicators, in order of precedence based on the order given in Steer (2020): Cepheids, TRGB, RR Lyrae, Red Clump, supernova (SN) Ia, FGLR, Horizontal Branch, SBF, GCLF, CMD, Type II Cepheids, Miras, PNLF, asymmetric giant branch (AGB), carbon stars, and SNIa SDSS. We use the single most recent preferred distance estimate of the preferred estimate, except if two preferred measurements were from the same year, in which case, we apply an average of these preferred distances and errors. If NED-D provided more than two older measurements from the best primary indicator, we ensured that the newest distance was not greater than two standard deviations from the averaged older distances. For galaxies with large spreads in their preferred distances, we use the mean of all preferred distances with the best primary indicator. We encode this information in the field 'dist_nedd_flag'—0: uses the single best NED-D measurement, 1: best NED-D distance averaged from multiple measurements, and 2: NED-D best indicator mean used due to discrepancy, as described previously.

Due to the nearby Virgo Cluster's large velocity dispersion, redshift-based distances to Virgo are much more inaccurate than for galaxies in the rest of our sample. Because of this, we treated Virgo as a special case for distance purposes. We matched and compared our sample to galaxies from Mei et al. (2007) and the Extended Virgo Cluster Catalog (EVCC; Kim et al. 2014)—we included certain members from their catalog as special cases to have their distances revised. These galaxies have velocities that lie within the range of galaxies that are gravitationally bound to Virgo at a given projected radius.

For these galaxies, we first combined the separate redshift-based and redshift-independent distance lists, using whichever available option had the lowest fractional error. Galaxies with surface brightness fluctuation distances from Mei et al. (2007) are assigned that distance and error. Other Virgo galaxies with only redshift-based distances available are assigned the average distance of 16.5 ± 1.1 Mpc given by Mei et al. (2007). For galaxies with redshift-independent distances, we found more divergent values than expected, including some that did not agree with the Virgo distances. After careful evaluation of individual cases, we assigned the 16.5 ± 1.1 Mpc distance for most objects, keeping the redshift-independent distance only if ∣D − 16.5∣ > 3σ, where σ is the error on the distance; this is a total of 862 objects.

For galaxies with multiple distance sources, our aim was to identify the more reliable and precise distance source. Our distance sources included Virgo distances, LVG distances, NED-D distances with the lowest error, and redshift-based distances from the Cosmicflows-3 calculator. We prioritized Virgo distances and otherwise chose the distances with the lowest error between the NED-D and CF3 z-distances.

Of the 55 objects that had no assigned distance from any of our sources, 29 are assigned the given distances and errors from HyperLeda. The remaining 26 are NSA-only objects for which velocity measurements returned a Cosmicflows-3 corrected distance D ≤ 0; we do not supply a distance in this case.

As well as best distances (the bestdist column) and errors (bestdist_error), each galaxy in the catalog has a field for a specific distance indicator, if available. We have also included the field bestdist_source. The values of this are–0: Virgo, 1: LVG, 2: NED-D, 3: redshift based, and 4: HyperLeda, "nan": No Distance. The final distribution of galaxy distances, separated using these classifications, is shown in Figure 4.

Zoom In Zoom Out Reset image size

Our mass estimates are based on those of Taylor et al. (2011) and thus should be consistent with the widely used GAMA mass functions (Baldry et al. 2012; Kelvin et al. 2014; Wright et al. 2017; Driver et al. 2022), which provide the deepest mass function estimates for galaxies in the local Universe. The Taylor et al. (2011) mass estimates are based on the Bruzual & Charlot (2003) stellar population models and assume a Chabrier (2003) mass function and the dust extinction law from Calzetti et al. (2000). There is an extensive comparison of how this color–M/L relation compares to earlier color–M/L relations in Taylor et al. (2011). In particular, their Figure 13 shows that their relation is ∼0.2 dex below the Bell et al. (2003) relation (due to differences in the initial mass function; IMF) and a similar normalization but significantly different slopes from the Zibetti et al. (2009) relation due to differences in the weighting of stellar population models. Here we compare our derived masses against dynamical mass estimates from ATLAS^3D in Cappellari et al. (2011) and IR-based mass estimates from the S4G survey (Sheth et al. 2010).

4.1.1. ATLAS^3D Dynamical Mass Comparison

ATLAS^3D is a volume-limited sample of 260 nearby early-type galaxies in Cappellari et al. (2011) within 42 Mpc. Dynamical masses were derived using Jeans anisotropic modeling of the integral field kinematics in Cappellari et al. (2013a) and Cappellari et al. (2013b). We compare our masses to those derived from the stellar-only dynamical (M/L)_stars estimates from Cappellari et al. (2013b), shown in Figure 7. The ATLAS^3D masses are mostly higher than our masses, with a median offset of .34 dex. The offset is near zero in the lowest-mass galaxies (log(M_⋆/M_⊙)<10) and is highest (nearly 0.5 dex) at the highest masses. These offsets are consistent with the findings of Cappellari et al. (2012), who found that the vast majority of ATLAS^3D galaxies had dynamical mass estimates that were higher than stellar population model estimates based on a Chabrier (2003) IMF, with the typical offset roughly a factor of 2 (0.3 dex) higher. They found that this offset strongly depends on the observed velocity dispersion, with the highest dispersion and mass galaxies having the largest offsets, just as observed here. This means that overall, this comparison suggests that our mass estimates are accurate for their assumption of a Chabrier IMF, but this assumption may be a poor one especially for the most massive early-type galaxies, and our stellar mass estimates may significantly underestimate the masses of these galaxies.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.1.2. S4G IR-based Mass Comparison

Many science applications benefit from complete, volume-limited samples of galaxies. To gauge the completeness of our sample, we compare our galaxy catalog to mock galaxy samples created from galaxy stellar mass functions. Above, we have derived masses for our galaxies based on the color–stellar mass relations used in the GAMA survey (Baldry et al. 2010; Taylor et al. 2011); we therefore use the GAMA galaxy stellar mass functions to evaluate the completeness of our catalog.

For the full sample of galaxies, we use the most recent stellar mass function published by GAMA from Driver et al. (2022), which extends down to M_⋆ ∼ 10⁷ M_⊙. We simulate a galaxy population with a volume of 4/3π(50 Mpc)³ based on the Driver et al. (2022) mass function. We then compare the number of galaxies with available masses in our catalog to this mock galaxy sample in 0.2 dex mass bins to calculate the fraction of the expected number of galaxies that we detect. We then translate this fraction to estimate the volume-completeness radius, which we define as the radius out to which we would have a complete sample based on the total number of galaxies in our sample. For instance, if we detect 50% of the number of expected galaxies, this translates into a volume-completeness radius of ∼40 Mpc. Figure 8 shows that the catalog is >50% complete at M_⋆ ≳ 10^8.5 M_⊙, which corresponds to an effective volume with radius R ∼ 40 Mpc. At lower masses, the completeness drops significantly. At the highest masses, the sample also appears to become less complete. We return to this point below.

Zoom In Zoom Out Reset image size

The Driver et al. (2022) paper does not have a comparable morphology or color separation to what we have done here, so we use an earlier GAMA paper by Baldry et al. (2012), which is more directly comparable to our color-based morphology separation (i.e., the color_type in our sample). We note that the color separation used here and in Baldry et al. (2012) are similar, but not identical. Figure 8 shows that for the late-type galaxies, the volume-completeness radius rises to the expected 50 Mpc at stellar masses of ∼10¹⁰ M_⊙. The early types achieve a similar completeness at much lower masses, but then the completeness declines at the highest masses. Values greater than one in this figure could be due to (1) a real difference in the number density of galaxies in the local Universe than in the regions used for the Baldry et al. (2012) mass function, or (2) differences in the color separation between our study and theirs. Note that a direct comparison using identical color separations is not possible due to their use of u − r colors in separating their populations and the lack of u − r colors for many galaxies in our sample. The bottom panel of Figure 8 shows the fraction of early-type galaxies as a function of stellar mass in our sample compared to the Baldry et al. (2012) mass function. This shows that the fraction of red galaxies is higher than expected in the local Universe at low masses, but lower at higher masses.

The lack of high-mass galaxies overall, and specifically, high-mass early-type galaxies, is a surprising result because we expect to be complete at the highest masses. This therefore implies that these galaxies are underdense in the local Universe relative to the volumes probed in the GAMA survey. We repeat that our stellar mass estimates should be consistent with GAMA, thus making the difference more likely to represent a real deficit of massive early-type galaxies locally relative to the GAMA volume.

Many galaxies in the nearby Universe have been morphologically classified using the Hubble and de Vaucouleurs classification systems. Galaxies can be separated into two broad categories: Early-type galaxies are bulge-dominated, elliptical, or lenticular in shape, and host little to no current star formation. Late-type galaxies have a spiral-disk shape with smaller central bulges and undergo current star formation. The numerical Hubble T classification ranges from −6 to 10, with negative values corresponding to early-type galaxies and positive values to late types. In addition, early- and late-type galaxies are often separated using galaxy color–magnitude or color-mass diagrams. At a given stellar mass, early-type galaxies are typically associated with redder colors, while late types are associated with bluer colors. In this section, we discuss the source of our morphological type information. We then also describe how we translate this into a color-based separation of early- and late-type galaxies. We then evaluate the X-ray active fraction as a function stellar mass and morphology/color in Section 6.

We use the numerical T-type translation of the morphological classifications in this work. Both HyperLeda and LVG provide numerical T-types for 13744 of the 15611 (89%) galaxies in our catalog. Of the remaining galaxies, 603 sample galaxies have available color information. Furthermore, as shown in the left panel of Figure 9, the early- and late-type galaxies (also known as the red sequence and blue cloud) show considerable overlap, especially for low-mass (<10⁹ M_⊙) galaxies. In order to give all sample objects an appropriate morphology, we worked to minimize the potential for morphological misclassification and calculated a color-based type for all galaxies with available color information.

Zoom In Zoom Out Reset image size

To separate galaxies into early and late types using the color-mass diagram, we first calculated a morphological separation line by minimizing the misclassification of galaxies with both morphological T-types and NSA colors. We separated galaxies into early and late type based on their T-type (with T > 0 being late type), and we fit a line to the color-mass diagram that minimizes the number of misclassifications, i.e., late types redder than the line, and early types bluer than the line. To fit this line, we first divided the data into 0.25 dex mass bins and calculated the fraction of misclassifications as a function of (g − i)₀ color. We then plotted the (g − i)₀ value with the minimum misclassification in each bin and found the best-fit line to establish our color-based morphological separation line. We note that we only fit the line between log(M_⋆/M_⊙) of 8.5 to 10.5, where galaxy mass estimates are more robust. The equation for this line is

$$\begin{eqnarray}&&{(g-i)}_{0}=0.133* \mathrm{log}({M}_{\star }/{M}_{\odot })-0.294.\end{eqnarray} \tag{ 5 }$$

This equation appears as the upper dashed line in the right panel of Figure 9.

Visual inspection of SDSS imaging of the misclassified galaxies using our morphological separation line suggested that while most red late-type galaxies showed distinct signatures of being dusty late-type galaxies (i.e., spiral arms or clumpy regions of active star formation), many blue early-type galaxies were in fact not properly classified and also showed these signatures of ongoing star formation. This was especially true for early-type galaxies that had much bluer colors than our morphological separation line. To better separate early- and late-type galaxies, we decided to reclassify very blue early-type galaxies. We created a second reclassification cutoff line below which we reclassified these galaxies in our best_type column (as described below). To create this line, we sorted the early-type contaminants by color in 0.25 dex mass bins. For each mass bin, we visually classified these contaminants as early or late type and then calculated the threshold color such that legitimate misclassification was minimized within the cutoff. Then, as above, we fit the reclassification cutoff line to the colors determined in each bin to obtain

$$\begin{eqnarray}&&{(g-i)}_{0}=0.124\,* \,\mathrm{log}({M}_{\star }/{M}_{\odot })-0.344.\end{eqnarray} \tag{ 6 }$$

This equation appears as the lower dotted line in the right panel of Figure 9.

To apply our color-based reclassification across our full sample, we needed to transform our morphological separation and reclassification cutoff lines into (g − r)₀, (B − V)₀, and (B − R)₀. Note that to derive the (B − R)₀, we added (B − R)₀ data from Cook et al. (2014) for the galaxies in our sample. Using overlapping galaxies and using RANSAC as described above, we used the following color transformations:

$$\begin{eqnarray}&&{(g-r)}_{0}=0.562\,* \,{(g-i)}_{0}+0.109,\end{eqnarray} \tag{ 7 }$$

$$\begin{eqnarray}&&{(B-V)}_{0}=0.625\,* \,{(g-i)}_{0}+0.131,\end{eqnarray} \tag{ 8 }$$

$$\begin{eqnarray}&&{(B-R)}_{0}=0.898\,* \,{(g-i)}_{0}+0.133.\end{eqnarray} \tag{ 9 }$$

Applying these color-transformations, we obtain the following morphological separation lines:

$$\begin{eqnarray}&&{(g-r)}_{0}=0.075\,* \,\mathrm{log}({M}_{\star }/{M}_{\odot })-0.056,\end{eqnarray} \tag{ 10 }$$

$$\begin{eqnarray}&&{(B-V)}_{0}=0.083\,* \,\mathrm{log}({M}_{\star }/{M}_{\odot })-0.052,\end{eqnarray} \tag{ 11 }$$

$$\begin{eqnarray}&&{(B-R)}_{0}=0.119\,* \,\mathrm{log}({M}_{\star }/{M}_{\odot })-0.131,\end{eqnarray} \tag{ 12 }$$

and the reclassification lines

$$\begin{eqnarray}&&{(g-r)}_{0}=0.070\,* \,\mathrm{log}({M}_{\star }/{M}_{\odot })-0.126,\end{eqnarray} \tag{ 13 }$$

$$\begin{eqnarray}&&{(B-V)}_{0}=0.062\,* \,\mathrm{log}({M}_{\star }/{M}_{\odot })-0.098,\end{eqnarray} \tag{ 14 }$$

$$\begin{eqnarray}&&{(B-R)}_{0}=0.111\,* \,\mathrm{log}({M}_{\star }/{M}_{\odot })-0.242.\end{eqnarray} \tag{ 15 }$$

For our final catalog, we provide several morphological types. In addition to our catalog columns for numerical T-types and their source, we created two type columns that broadly categorize galaxies into "early" or "late." Our color_type classifies a galaxy as early if its color lies above the morphological separation line and as late if it lies below the line. Our recommended best_typecolumn follows this process:

• (1)  
  Most galaxies are classified based on their T-type from HyperLeda or Karachentsev et al. (2013) with t ≤ 0 = "early" and t > 0 = "late."
• (2)  
  The 387 early-type galaxies with bluer colors than the reclassification cutoff are reclassified as late-type galaxies.
• (3)  
  603 galaxies without a morphology are assigned their color-based type. We reclassify each galaxy based on the color used to estimate the mass in the final catalog.

In total, we have best_type values for 14347 out of 15611 galaxies and color_type values for 11740 galaxies. We show the masses of galaxies separated into early and late type in the left panel of Figure 10, and we also include the masses of all galaxies with T-types in the right panel.

The galaxies in the local Universe span a wide range of environments, from dense clusters such as the Virgo galaxies to galaxies in voids. One common way of identifying galaxies in dense environments is through group catalogs. To enable easy identification of galaxies in groups in our survey, we use the Lambert et al. (2020) group catalog based on the 2MRS survey. This group catalog identifies 3022 groups from the 2MRS catalog (which extends well beyond our 50 Mpc limit) using a friends-of-friends algorithm.

To match our galaxy catalog, we first used Table 3 from Lambert et al. (2020), which gives individual galaxies that are part of a group or subgroup using the friends-of-friends algorithm described in the paper. We crossmatched our sample to galaxies with v_cmb < 3867 km s⁻¹ (based on our highest-redshift galaxy), resulting in 2463/2864 unique matches within 30''. We wished to ensure that we also matched low-mass sample galaxies that might not be included in the Lambert et al. (2020) catalog due to the magnitude limit of 2MRS. To do this, we used Table 1 from Lambert et al. (2020), which includes group positions, comoving distances, velocity dispersion, and radius/R₂₀₀ estimates. We crossmatched our entire sample to find the nearest group for each galaxy, considering sample galaxies to be part of a group when they fulfilled two requirements: (i) the velocity for our sample galaxy must be within twice the velocity dispersion of the group, and (ii) the angular separation between between the galaxy and its nearest group matches was smaller than the angular radius of that group, (d ≤ R₂₀₀). Using this three-dimensional approximation, we flagged an additional 790 sample galaxies as likely group members. As expected, the median log(M_⋆/M_⊙) for the distinct group-method matches is ∼8.4, compared to ∼10 for the direct galaxy matches.

We assigned Lambert et al. (2020) group IDs to 3253 sample galaxies, giving precedence to those with direct galaxy matches. This simultaneously flags galaxies as being part of a group, and easily allows users to seek further group information (including group radius and velocity dispersion) by cross correlating our catalog with the catalogs available from Lambert et al. (2020) on the publisher's website.

The Virgo cluster is indicated in the Lambert et al. (2020) catalog as cluster 2987. Of the 347 galaxies with this group identification, we find that 276 of them (79%) are included in Virgo based on the distance sources in Section 3. This is a small fraction of the 907 total galaxies that we identify as being part of Virgo in our catalog because the Lambert et al. (2020) catalog is based on 2MASS and thus misses many fainter Virgo galaxies.

The initial motivation for creating this sample was to quantify the demographics of central black holes in nearby lower-mass galaxies. These demographics can provide constraints on the formation mechanisms of central massive black holes (see the recent review by Greene et al. 2020). X-ray measurements of AGN in nearby galaxies ≲50 Mpc can provide deep and clean enough observations to constrain the occupation fraction of black holes in galaxies below 10¹⁰ M_⊙. Results on large samples of early-type galaxies (Miller et al. 2015; Gallo & Sesana 2019) suggest a high occupation fraction for 10⁹–10¹⁰ M_⊙ galaxies, in agreement with dynamical measurements in a handful of nearby early-type galaxies (Nguyen et al. 2018, 2019). However, a majority of galaxies below 10¹⁰ M_⊙ are late-type galaxies (Baldry et al. 2012), and thus far, no suitable galaxy sample exists to compare the X-ray active fraction (and thus occupation fraction) of late-type galaxies to the studies of early-type galaxies at masses below M_⋆ ∼ 10¹⁰ M_⊙. Recent compilations of archival X-ray data in nearby galaxy samples cut out most low-mass nearby galaxies by requiring redshift-independent distance measurements (She et al. 2017a, 2017b), or by including only brighter samples dominated by massive galaxies (Bi et al. 2020). The X-ray active fractions of more distant dwarfs (between z of 0.08 and 2.4) have also been studied by Mezcua et al. (2018) in the COSMOS field. Our new 50 Mpc Galaxy Catalog provides the best starting point for constraining the local X-ray active fraction and the occupation fraction of central black holes. We specifically focus on trying to obtain a sample that is as large and complete as possible down to lower masses (i.e., 10⁸ M_⊙), with stellar mass estimates and Hubble-type measurements based on color index. Below, we quantify the X-ray active fraction using our new catalog.

The detection of a nuclear X-ray source above 10⁴² erg s⁻¹ (the "classical" AGN threshold) is invariably associated with accretion-powered emission from a massive (i.e., nonstellar) black hole. Below this threshold, and even more so, below 10⁴⁰ erg s⁻¹, emission from bright X-ray binaries could contribute to the detected signal. The expected luminosity from low-mass X-ray binaries, in particular, scales with stellar mass (Lehmer et al. 2019), which makes spatial resolution the key for discriminating between massive black holse with a low Eddington ratio versus contaminants. Chandra is thus preferable to XMM-Newton for searching and identifying massive black holes with luminosities lower than traditional AGN because its resolution is higher by a factor of 4.

We present the Chandra-based X-ray active measurements of our galaxy catalog here, without an attempt to model the distribution or account for contamination from X-ray binaries. However, in a subsequent paper, we will use these data to constrain the L_X–M_⋆ relation and the occupation fraction as a function of galaxy stellar mass, including estimates of contamination from X-ray binaries (E. Gallo et al. 2023, in preparation).

We determined which galaxies in our catalog have available Chandra X-ray data using the Chandra Source Catalog 2.0 (Evans et al. 2010, and I. N. Evans 2023, in preparation), which includes Chandra observations made public before the end of 2014. We searched for X-ray sources coincident with the nuclei of our galaxies using their catalog ra and dec. Note that our choice of coordinates has been discussed in Section 2.1.2, and the effect of uncertainties on these central coordinates is considered in more detail in Section 6.2.2. Using Chandra's CSCView, ¹⁸ we retrieved two tables from a crossmatch with our catalog: (1) a table of limiting sensitivities for all 1506 objects with Chandra data, and (2) a table of detections within a given angular separation (1''–3'') of the galaxy ra and dec. The second table of detections contains 291 unique galaxies with matching detections when using a 1'' search radius (note that 20 objects returned multiple sources). We used our distance estimates (best_dist) to calculate X-ray luminosities from the 0.5 to 7 keV fluxes. These have a median log(L_X/erg s⁻¹) of 39.21 (hereafter, we just refer to this as log(L_X)), with a full range of log(L_X) from 33.19 to 42.26. While some of these sources are likely contaminants from X-ray binaries and not accreting massive black holes, we attempted to minimize this contamination in two ways: (1) We restricted our X-ray activity analysis below to the 249 (85%) of sources that have available log(M_⋆/M_⊙) and log(L_X)> 38.3 (as in Miller et al. 2015; Gallo & Sesana 2019), and (2) we used a 1'' match as our default matching radius. We note that previous work on X-ray binary contamination in similar Chandra observations has shown that only a small fraction (∼10%) of nuclear X-ray detections with log(L_X)> 38.3 are likely to be contaminants (e.g., Gallo et al. 2010; Miller et al. 2012; Foord et al. 2017). We include the columns chandra_observation and chandra_detection to flag matched catalog galaxies using our default 1'' matching, and we also include a chandra_detection_3arcsec for galaxies with sources within 3'' (see Section 6.2.2); the X-ray luminosity is given in the log_lx column.

Of the 249 galaxies we used for our active fraction analysis, 159 have available SDSS imaging. To visualize the X-ray sources in our sample, we show 42 of these galaxies in Figure 11. In the first three rows of the figure, we show the 18 lowest-mass SDSS galaxies, and we present a sampling of galaxies at higher masses in the bottom half. The 18 lowest-mass galaxies have log(M_⋆/M_⊙) from 7.99 to 9.61; an additional 4 galaxies in this mass range lack available SDSS imaging.

Zoom In Zoom Out Reset image size

6.2.1. Local Active Fraction as a Function of Galaxy Mass and Type

We calculate the X-ray active fraction by dividing the number of galaxies with Chandra detections by the number of galaxies with Chandra observations. We use a 1'' matching radius to limit the impact of X-ray binary contamination and take advantage of Chandra's excellent resolution (e.g., Gallo et al. 2010), however, we examine the results of larger matching radii in the next subsection. The results of this X-ray active fraction binned by mass and separated by morphological type are shown in Figure 12. Although some detections may be caused by nuclear X-ray binaries, this fraction serves as an upper limit of AGN occupation for our catalog.

Zoom In Zoom Out Reset image size

We observe an apparent difference in detection fraction between early- and late-type galaxies, particularly within the mass range $9\lesssim \mathrm{log}({M}_{\star }/{M}_{\odot })\lesssim 10.5$. We examine the significance of this difference following the methods of Hoyer et al. (2021). We calculate p-values under the null hypothesis that both early- and late-type galaxies have the same X-ray active fraction. We split the data into 0.25 dex bins from $7.75\leqslant \mathrm{log}({M}_{\star }/{M}_{\odot })\leqslant 11.25$. We use the total X-ray active fraction in each bin as an estimator of the binomial success probability. Using this fraction, we draw n samples from a binomial distribution, where n is the number of galaxies per bin in the early- and late-type subsamples. We repeat this exercise 10⁶ times and use the fraction of estimates that match or exceed our observed difference between early- and late-type galaxies to estimate a p-value for each bin. Using the Fisher method (Fisher 1992), we combine all p-values into a single parameter under the assumption that the null hypothesis is true and that the data in each mass bin are independent. The final p-value is 0.0004 over the entire mass range, thus suggesting that the observed enhancement of early-type galaxies is significant at a level >3σ. This significance increases, for instance, when we consider just the galaxies between $9\leqslant \mathrm{log}({M}_{\star }/{M}_{\odot })\leqslant 10.5$. Thus, it appears there is a significant enhancement of X-ray detections in early-type versus late-type galaxies. We examine belwo whether this result is due to a higher AGN fraction in early-type galaxies.

6.2.2. Local Active Fraction Uncertainties and Systematics

One significant potential uncertainty is the locations of galaxy centers, and whether our catalog coordinates accurately reflect the true center of each galaxy. To gauge the impact of this uncertainty, we conducted the same analysis using a 2'' and 3'' separation constraint for Chandra detection matching. As shown in the left panel of Figure 13, larger matching radii increase the X-ray active fraction for both types overall. The difference between the early- and late-type active fraction depicted in Figure 12 still exists for larger separation constraints, but the statistical significance decreases to p = 0.0019 for 2'' and p = 0.0104 for 3''.

Zoom In Zoom Out Reset image size

The decreasing difference in X-ray active fractions between early and late types with larger matching radii could be explained if the centers of late-type galaxies were less certain due to their lower surface brightness and more complicated morphology. The central surface brightness does depend on galaxy mass and luminosity (e.g., Blanton et al. 2003; Ferrarese et al. 2006), with decreasing surface brightness in lower-mass galaxies below roughly the luminosity and mass of the Milky Way. Late-type galaxies generally also have a lower surface brightness than early-type galaxies (e.g., Blanton et al. 2003). In addition, many of our late-type galaxies are irregulars; these dominate our late-type galaxy number counts below log(M_⋆/M_⊙) = 8.0. Finally, distance plays a role in our ability to accurately locate the angular center of our galaxies, with more distant galaxies likely having more accurate angular coordinates.

As noted in Section 2.1.1, we see significant discrepancies between our catalog positions (mostly from HyperLeda) and the best available positions from NED. These offsets can give us a sense of how uncertain the nuclear positions of each galaxy are. In the right panel of Figure 13, we show the fraction of galaxies with offsets larger than 1'', 2'', and 3'' as a function of galaxy type (early versus late) and stellar mass. The fraction of galaxies with significant position differences between NED and our catalog is noticeably larger for late-type galaxies than for early-type galaxies in the mass range $7.5\,\lesssim \mathrm{log}({M}_{\star }/{M}_{\odot })\lesssim 9.5$. For a 1'' position match, the differences are large enough (>20% at some masses between early and late types) to explain the lower active fractions we see in late-type galaxies. These larger positional uncertainties are likely tied to the lower surface brightness and irregular morphology among our late-type galaxies. However, at higher masses ($9.5\,\lesssim \mathrm{log}({M}_{\star }/{M}_{\odot })\lesssim 10.5$), the positional uncertainty differences become smaller than the active fraction differences. To further examine the possibility that uncertain nuclear coordinates impact our active fraction measurements, we visually examined all 17 galaxies (16 of which are late-type galaxies) with $\mathrm{log}({M}_{\star }/{M}_{\odot })\lesssim 9.5$ that had X-ray sources within 3'' of their centers, but not within 1''. In 11 of these 17 galaxies, the X-ray sources are clearly offset from the galaxy centers. In another 5 galaxies, it was ambiguous whether the X-ray source was nuclear or not, while in only one galaxy (NGC 598/M33), it was clear that the X-ray source was nuclear, and our nuclear position is offset by ∼1'' from the true center of the galaxy (this galaxy has a known nuclear star cluster that is visible in SDSS imaging). This shows that accurate nuclear positions are an important factor to consider when examining the presence of AGN in nearby galaxies, especially in lower-mass late-type galaxies. However, errors in nuclear positions apparently do not fully account for the difference we see between early- and late-type galaxy active fractions.

6.2.3. Local Active Fraction as a Function of Galaxy Mass and Color