Dwarf Galaxy Discoveries from the KMTNet Supernova Program. II. The NGC 3585 Group and Its Dynamical State^*

To find dwarf galaxy candidates in the two NGC 3585 fields, we visually select dwarf galaxy candidates by identifying diffuse sources (see Park et al. 2017 for details). For dwarf galaxy candidates in the NGC 3585 group, the ∼20 Mpc distance, or (m − M) ∼ 31.5 mag, means that the stellar populations will present as unresolved diffuse features because even the stars on the tip of the red giant branch (TRGB), one of the brightest stellar populations, cannot be resolved. In our stacked image, which has a point-source magnitude limit of about 24 mag, the TRGB stars with M_I = −4.0 mag (Lee et al. 1993) have an apparent magnitude of about 27.5 mag. We use primarily the I-band image to identify galaxies because it is the deepest of the three bands for a given exposure time and because most dwarf galaxies have a red color, (V − I) > 0.5 (see Section 3.4). We use the B- and V-band images, as well as RGB color images made from the three bands, to rule out artifacts.

To enhance faint surface brightness features, we set the dynamical scale range on a display screen (e.g., DS9) such that the lowest and highest intensities are approximately −3σ and 10σ from the mean, where σ refers to the scatter of the sky background. We select any diffuse sources that are approximately larger than 10'' as dwarf galaxy candidates, and currently reject sources with special features (e.g., spiral or bulge) except for those with a point source at the center, which are classified as nucleated dwarf galaxy candidates (see Section 4.2). To minimize biases in the detection and incompleteness, three authors (H.S.P., S.C.K., and Y.L.) searched the images independently and then compared the results. For candidates selected by only one or two classifiers, all three classifiers reviewed the object together and decided whether to retain or reject the candidate basically based on the above criteria. The retained candidates are roughly 20% of all the candidates that were finally accepted.

For the regions near NGC 3585 itself, we use an image where we have subtracted a model for the galaxy constructed with the IRAF/BMODEL task. The model is constructed using the output from the IRAF/ELLIPSE task, using nonlinear steps along the semimajor axis and floating ellipticity and position angle, with the brightness profile measured using median isophotal fluxes after interactively masking bright objects (e.g., foreground stars or background galaxies) and applying two iterations of 3σ clipping. This model subtraction was only performed on the 025 × 025 area centered on NGC 3585 galaxy in the N3585-1 field. From our visual inspection of the subtracted image we conclude that the model fitting is adequate to within 1' from the center of NGC 3585.

Despite the difference in depth between the two observed fields, we recover the same candidates in both fields within the overlapping region. We conclude that our ability to recover dwarfs is not set by the point-source limiting magnitude of each field, but rather by other factors such as scattered light from bright stars, image defects, and a spatially variable background. Because none of these limiting factors are affected by the modest difference in exposure time between the two fields, we treat the two fields in the same way. We also searched for dwarf galaxy candidates in the image stacked within the overlapping region, but did not find any new diffuse sources. In total, we identify 46 such diffuse objects and plot their positions in Figure 1.

We then use the NASA Extragalactic Database (NED) to search for known galaxies in the NGC 3585 KMTNet fields. We find eight galaxies with existing radial velocity measurements that are compatible with membership in the NGC 3585 group as defined by being within ±3σ_v, 1220 < cz < 1650 km s⁻¹. We define this velocity range using the measured group velocity dispersion, σ_υ = 70 km s⁻¹ (Makarov & Karachentsev 2011) and the radial velocity, υ = 1434 km s⁻¹ (NED). The four of these eight galaxies that have an absolute magnitude fainter than −18 mag and approximately exponential brightness profiles are dwarf galaxy group members. These galaxies are, in retrospect, detected in our KMTNet images, but do not belong to the 46 dwarf candidates. They were not included in our candidate list because they have high surface brightnesses and are much more difficult to distinguish from background galaxies. That is, these bright dwarf galaxies have an extended central region (approximately μ ≲ 23 mag arcsec⁻²) that is saturated in our displayed range, which we generally consider to be a signature of background galaxies. As such, there could be additional high surface brightness group members that are among neither the NED galaxies nor our cataloged candidates.

We also examined whether archival images, such as those from the Dark Energy Camera Legacy Survey (DECaLS) and the Hubble Space Telescope (HST), could be used to confirm our candidates and to search for new diffuse sources. Unfortunately, the survey region of the former does not overlap with our NGC 3585 field. The latter, HST, has two overlapping fields (a central region and a region ∼30' away from the NGC 3585 galaxy). However, none of the dwarf galaxy candidates we identified lie within these fields, and we did not find any additional diffuse candidates.

We now add the four NED dwarfs to our dwarf galaxy candidate list for the remainder of this study. We use the brighter four confirmed group galaxies, which are not dwarfs, only when discussing the galaxy LF in Section 4.4. In Figure 2 we present the I-band grayscale images of all the dwarf galaxy candidates and the additional four confirmed dwarfs. They tend to be diffuse with a variety of sizes and morphologies. Some show concentrated nucleated emission.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We estimate the limiting brightness of our I-band images using a completeness test where we create and attempt to recover several hundred artificial galaxies. We create the galaxies and add them to our I-band images of the deep field (N3585-1-Q2) and the shallow field (N3585-2-Q1), including the NGC 3585 galaxy, using the mkobjects task in the IRAF/artdata package. In our modeling, we adopt surface brightness profiles ranging from n = 0.6 to n = 1 and an effective radius distribution that matches the effective radius–magnitude relation of the NGC 3585 dwarf galaxy candidates in Figure 6 (see Section 3.4). Based on reproducing our the visual inspection procedure, we determine that we are 90% complete down to I ≈ 19.8 mag in the deep field (M_V ≈ −11.0 mag at the distance of NGC 3585) and to I ≈ 19.6 mag in the shallow field.

For a galaxy survey that targets galaxies larger than a given radius (r_lim) with a given surface brightness (μ_lim), we can predict detection limits for radius or surface brightness versus total magnitude (Ferguson & Sandage 1988; Müller et al. 2015). For our dwarf galaxy survey, in which we detect galaxies larger than 10'' (r_lim = 5'') down to a μ_lim = 28 mag arcsec⁻², the completeness boundaries are drawn as dotted curves in Figure 6, assuming that the dwarfs have exponential surface brightness profiles. Most of our dwarf galaxy candidates are located on the left side of the boundaries, as expected. The results from this calculation are also consistent with those from the injection of artificial galaxies.

The central region of a group is often contaminated by diffuse intragroup light (Watkins et al. 2014, 2015; Mihos et al. 2017). We now check whether diffuse intragroup light affects our detection of dwarf galaxy candidates. We compare the recovery for the inner (within ∼0.15 Mpc, the area that would suffer most from intragroup light) and the outer regions. The completeness difference is smaller than 5%, which is within our 1σ uncertainty. We therefore conclude that intragroup light does not affect this study.

Finally, our dwarf galaxy candidates might be confused with Galactic cirrus, which is often observed in deep optical images (Miville-Deschênes et al. 2016). Because Galactic cirrus is attributed to dust grains, we examine the WISE 12 μm map of this region to help us distinguish between dwarf galaxy candidates and Galactic cirrus. Our NGC 3585 fields are located outside clear regions of Galactic cirrus that can be seen in the WISE 12 μm map. In terms of color and surface brightness, Galactic cirrus has red colors (1.3 mag < (g − r)₀ < 2.0 mag and 1.5 mag < (B − V)₀ < 2.2 mag; Ludwig et al. 2012), while our dwarf galaxy candidates have relatively blue colors, 0.1 mag < (B − V)₀ < 1.0 mag (see Section 3.4). Last, all of our dwarf galaxy candidates have a brighter central surface brightnesses than the surface brightness of Galactic cirrus (μ_B > 27 mag arcsec⁻²; Cortese et al. 2010). We conclude that most if not all of our dwarf galaxy candidates are not misclassified Galactic cirrus.

We measure the surface brightness profiles of the 46 dwarf galaxy candidates and the 4 previously known dwarf galaxies in B, V, and I using the IRAF/ELLIPSE task. Because many of our candidates have low surface brightness, we performed the photometry in two steps. First, we measure the brightness of the candidates without fixing the parameters (e.g., center, position angle, and ellipticity) using the ELLIPSE task on the I-band image, which is relatively deeper than other band images. We then fix the parameters to be the values measured around the resulting effective radius and measure the surface brightness profile of each candidate on the BVI-band images. We use 12 linear steps along the semimajor axis and measure the mean isophotal flux. The uncertainty in each measurement is determined from the combination of three sources as follows: (1) the mean error of the isophotal flux as given by ELLIPSE, (2) the variation of sky background level, which is estimated from the fluxes in several of the outermost radial bins, and (3) the variation among fitting parameters, which we estimate by calculating several hundred trials while varying the parameters according to the uncertainties of the fixed parameters returned originally by ELLISPE. Finally, we transform the instrumental surface brightness using the zero-points and color terms measured in Section 2.

In Figure 3 we show the surface brightness (upper panels) and color (lower panels) profiles of the NGC 3585 dwarf galaxy candidates. Most candidates have roughly an exponential (n ∼ 1) surface brightness profile and a constant color profile. The faintest surface brightnesses reach approximately 28 mag arcsec⁻². The weighted mean colors ($\langle {\mu }_{(V-I)}\rangle$ and $\langle {\mu }_{(B-V)}\rangle$) are denoted by the blue dot–dashed and green dotted lines, respectively, in the color profile panels. We adopt the mean color as the color, (V − I) and (B − V), of each dwarf galaxy candidate. The color ranges are 0.3 < (B − V) < 1.1 and 0.6 < (V − I) < 1.3.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We use the Sérsic function (μ₀ + 1.0857(r/r₀)^1/n), where μ₀, r₀, and n are the central surface brightness, scale length, and Sérsic curvature index, respectively, to fit the surface brightness profiles. The best fits of the Sérsic functions are presented as solid curves in Figure 3. We adopt the 1σ statistical errors estimated from the nonlinear least-squares fitting procedure (Markwardt 2009) for the uncertainties of the Sérsic parameters. Typical uncertainties for our candidates ($\langle \sigma ({\mu }_{0,I})\rangle \approx 0.12$ mag arcsec⁻², $\langle \sigma ({r}_{0,I})\rangle \approx 0\buildrel{\prime\prime}\over{.} 59$, and $\langle \sigma ({n}_{I})\rangle \approx 0.13$) are plotted in Figure 6. All of the dwarf galaxy candidates have central surface brightness values brighter than μ_0,I ∼ 25.4, μ_0,V ∼ 26.2, and μ_0,B ∼ 27.1 mag arcsec⁻². They have Sérsic scale lengths of 2'' < r_0,I < 24'' and curvature indices of 0.4 < n_I < 2.0 with a median value of ∼0.8. The best fits for dwarf galaxy candidates in the overlap regions of N3585-1 and N3585-2 fields (Figure 1) are comparable, showing consistency in our measurements.

We estimate their total magnitudes and effective radii from the extrapolation of the Sérsic fits (see Section 3.1 of Chiboucas et al. 2009 for the details of the method). The faintest total magnitudes in the B, V, and I bands are 22.4, 21.6, and 20.5 mag, respectively. The smallest effective radius (r_e,I) in the I band is ∼15 (corresponding to ∼150 pc at the distance of NGC 3585). The uncertainties of their total magnitudes and effective radii are derived by measuring several hundred trials while varying according to the uncertainties of the Sérsic parameters. The typical values, $\langle \sigma (I)\rangle \approx 0.14$ mag and $\langle \sigma ({r}_{e,I})\rangle \approx 0\buildrel{\prime\prime}\over{.} 76$, for our dwarf galaxy candidates are also plotted in Figure 6.

In Table 3 we present the photometric results for the 46 dwarf galaxy candidates in the study and for the 4 confirmed dwarfs. The first column contains the name. The second and third columns list the central coordinates. The fourth column presents the total I-band magnitude based on the Sérsic fit. The fifth and sixth columns are the observed (V − I) and (B − V) colors, respectively. The seventh, eigth, ninth, and tenth columns list the central surface brightness (μ_0,I), scale length (r_0,I), curvature index (n_I), and I-band effective radius (r_e,I) derived from the Sérsic fit, respectively. The "N" and "U" flags in the last column indicate whether the galaxy candidate is nucleated ("N") and/or if it can be qualified as an ultra-diffuse galaxy ("U"). See Sections 4.2 and 4.3 for the details of these classifications.

Table 3.  Photometric Catalog for Dwarf Galaxy Candidates in the NGC 3585 Field

ID a	R.A.(J2000)	Decl.(J2000)	I b	(V − I)c	(B − V)c	μ0,Id	r0,Id	${n}_{I}$ d	${r}_{e,I}$ d	Flage
	(hh:mm:ss)	(dd:mm:ss)	(mag)	(mag)	(mag)	(mag arcsec−2)	(arcsec)		(arcsec)
ESO502-G018	11:07:18.1	−25:34:23	14.01	0.98	0.83	22.35 ± 0.04	23.57 ± 0.68	0.70 ± 0.03	25.72
KSP-DW32	11:07:29.4	−25:53:50	18.29	1.00	0.83	23.74 ± 0.08	5.85 ± 0.50	0.78 ± 0.09	7.71
KSP-DW33	11:08:40.8	−25:52:02	19.48	0.93	0.81	23.56 ± 0.15	3.40 ± 0.40	0.66 ± 0.12	3.39
KSP-DW34	11:08:42.3	−25:51:26	18.01	0.75	0.63	22.98 ± 0.06	5.36 ± 0.27	0.57 ± 0.05	3.78
KSP-DW35	11:09:02.3	−25:57:16	19.02	0.85	0.83	23.53 ± 0.11	4.12 ± 0.38	0.66 ± 0.12	4.12
KSP-DW36	11:10:14.2	−26:27:24	20.13	0.90	0.47	23.65 ± 0.08	2.29 ± 0.31	0.85 ± 0.20	3.32
KSP-DW37	11:10:36.1	−26:06:59	18.99	0.88	0.92	23.34 ± 0.29	2.88 ± 0.84	1.03 ± 0.25	4.90	N
KSP-DW38	11:10:42.7	−27:01:15	20.17	0.80	0.82	23.69 ± 0.05	2.74 ± 0.15	0.58 ± 0.09	2.02
KSP-DW39	11:10:46.2	−26:31:31	18.78	1.12	0.73	24.55 ± 0.15	7.65 ± 0.91	0.60 ± 0.22	6.21
KSP-DW40	11:10:48.5	−25:15:32	16.38	0.92	0.63	21.94 ± 0.04	5.48 ± 0.29	0.93 ± 0.04	8.68
GALEXASCJ	11:10:51.5	−25:44:58	14.19	0.98	0.81	19.15 ± 0.15	2.29 ± 0.31	1.49 ± 0.06	4.40
ESO438-G010	11:10:51.8	−27:53:51	13.08	0.88	0.71	20.05 ± 0.02	8.72 ± 0.19	1.13 ± 0.01	15.62	N
KSP-DW41	11:11:24.1	−27:58:09	18.35	0.85	0.73	23.52 ± 0.12	5.34 ± 0.54	0.73 ± 0.11	6.39	N
KSP-DW42	11:11:26.1	−26:32:54	18.14	0.96	0.87	24.89 ± 0.08	12.65 ± 0.63	0.49 ± 0.07	5.24
KSP-DW43	11:11:44.4	−26:48:38	18.60	0.84	0.95	24.83 ± 0.11	9.40 ± 0.77	0.60 ± 0.15	7.63
KSP-DW44	11:11:57.2	−26:54:59	15.16	1.05	0.95	21.99 ± 0.03	9.92 ± 0.33	0.92 ± 0.03	15.58	N
KSP-DW45	11:12:02.7	−26:25:44	20.45	0.92	1.00	24.07 ± 0.78	2.42 ± 1.55	0.84 ± 0.59	3.44
KSP-DW46	11:12:13.3	−27:14:17	18.03	0.83	0.51	23.14 ± 0.16	5.09 ± 0.75	0.76 ± 0.14	6.45
KSP-DW47	11:12:16.3	−26:11:16	16.23	1.05	0.80	23.03 ± 0.04	10.46 ± 0.37	0.84 ± 0.04	14.99
KSP-DW48	11:12:23.4	−25:50:52	17.75	1.01	0.90	23.75 ± 0.07	7.80 ± 0.52	0.74 ± 0.08	9.38
KSP-DW49	11:12:34.8	−28:01:43	18.82	0.99	0.68	23.30 ± 0.12	4.41 ± 0.34	0.51 ± 0.09	2.18
KSP-DW50	11:12:51.6	−27:07:33	18.19	0.98	0.84	23.30 ± 0.06	5.55 ± 0.28	0.63 ± 0.06	4.89
KSP-DW51	11:12:53.2	−26:29:03	18.62	1.06	0.90	23.68 ± 0.10	5.24 ± 0.42	0.68 ± 0.09	5.56
KSP-DW52	11:12:56.2	−27:26:41	20.18	1.16	0.87	24.60 ± 0.18	3.67 ± 0.71	0.78 ± 0.25	4.77
KSP-DW53	11:12:56.8	−26:39:55	19.51	0.82	0.29	25.41 ± 0.30	8.82 ± 1.32	0.40 ± 0.27	1.39
KSP-DW54	11:13:01.0	−26:51:23	18.23	0.91	0.80	25.04 ± 0.20	11.00 ± 1.83	0.78 ± 0.26	14.42	U
KSP-DW55	11:13:09.0	−26:19:56	18.52	1.13	0.64	23.35 ± 0.28	3.33 ± 1.13	1.11 ± 0.30	5.90
KSP-DW56	11:13:27.9	−26:56:07	20.15	0.91	0.67	24.11 ± 0.12	3.29 ± 0.36	0.62 ± 0.16	2.83
KSP-DW57	11:13:32.6	−26:32:48	18.29	1.11	1.09	24.37 ± 0.12	8.40 ± 0.79	0.68 ± 0.14	8.92
KSP-DW58	11:13:36.3	−26:33:49	17.85	0.91	0.71	23.86 ± 0.08	8.16 ± 0.57	0.68 ± 0.07	8.50	N
KSP-DW59	11:13:48.1	−26:07:58	20.38	1.07	0.75	25.15 ± 0.17	5.17 ± 0.66	0.44 ± 0.22	1.38
KSP-DW60	11:13:55.2	−26:22:20	16.98	0.86	0.61	23.52 ± 0.06	8.92 ± 0.54	0.89 ± 0.06	13.56	N,U
ESO503-G001	11:14:02.2	−26:21:56	13.00	1.11	0.90	19.08 ± 0.06	5.87 ± 0.27	1.11 ± 0.03	10.43
KSP-DW61	11:14:13.1	−26:31:38	19.57	0.98	1.05	23.86 ± 0.13	3.47 ± 0.52	0.77 ± 0.18	4.47
KSP-DW62	11:14:14.4	−26:30:47	16.57	1.21	0.71	24.74 ± 0.07	23.93 ± 1.09	0.52 ± 0.07	12.45	U
KSP-DW63	11:14:16.9	−27:04:44	18.44	0.90	0.45	23.87 ± 0.27	5.64 ± 1.29	0.83 ± 0.40	7.90
KSP-DW64	11:14:17.2	−26:29:51	18.69	1.31	0.91	23.72 ± 0.41	3.63 ± 1.67	1.11 ± 0.47	6.44	N
KSP-DW65	11:14:18.9	−27:20:56	18.33	0.86	0.83	24.27 ± 0.85	1.70 ± 2.30	1.95 ± 1.55	3.27
KSP-DW66	11:14:27.9	−26:55:00	15.89	0.86	0.64	23.52 ± 0.05	14.32 ± 0.82	0.92 ± 0.06	22.47	U
KSP-DW67	11:14:29.6	−25:16:10	16.32	1.05	0.77	22.51 ± 0.04	7.79 ± 0.27	0.86 ± 0.03	11.43
KSP-DW68	11:14:36.4	−26:51:15	18.70	0.94	0.70	23.71 ± 0.16	4.98 ± 0.65	0.73 ± 0.16	5.85
KSP-DW69	11:14:54.3	−27:49:00	19.26	0.95	0.85	22.98 ± 0.09	2.71 ± 0.25	0.75 ± 0.10	3.34
KSP-DW70	11:15:24.1	−26:45:14	17.23	1.01	0.82	22.96 ± 0.04	6.95 ± 0.28	0.73 ± 0.04	8.21
KSP-DW71	11:15:58.1	−27:29:57	19.76	0.68	0.25	23.65 ± 0.08	3.24 ± 0.25	0.58 ± 0.12	2.35
KSP-DW72	11:16:21.2	−26:24:26	19.56	0.75	0.47	23.96 ± 0.21	4.10 ± 0.65	0.58 ± 0.18	3.04
KSP-DW73	11:16:47.3	−27:40:17	16.39	1.14	0.78	22.86 ± 0.03	9.52 ± 0.27	0.76 ± 0.03	12.07
KSP-DW74	11:17:55.0	−26:07:35	18.37	1.01	0.75	23.03 ± 0.75	1.99 ± 2.03	1.49 ± 0.82	3.83	N
KSP-DW75	11:18:26.3	−26:41:52	18.93	0.97	0.74	23.29 ± 0.20	3.21 ± 0.62	0.91 ± 0.16	4.97
KSP-DW76	11:18:26.4	−26:11:39	16.46	0.84	0.60	22.83 ± 0.03	9.02 ± 0.30	0.77 ± 0.03	11.62
KSP-DW77	11:18:37.1	−26:55:09	16.53	0.55	0.37	23.14 ± 0.05	9.81 ± 0.57	0.81 ± 0.06	13.45

Notes.

^aKSP-DW** candidates are newly discovered objects in KSP. The full name of GALEXASCJ candidate is GALEXASCJ111051. ^bI is the total I-band magnitude derived from the Sérsic fit. ^c(V − I) and (B − V) are colors without correction for extinction. ^dμ_0,I, r_0,I, n_I, and r_e,I are central surface brightness, scale length, curvature index, and effective radius derived for the I-band Sérsic fits, respectively. ^eN and U flag a nucleated galaxy candidate and UDG candidate, respectively.

Download table as:  ASCIITypeset image

For the purpose of constructing the luminosity function (Section 4.4), we also photometered the four bright group galaxies (see Section 3.1 for details). The resulting photometry of the bright galaxies is as follows: NGC 3585 (M_V = −22.52), UGCA230 (M_V = −19.09), ESO 503-G007 (M_V = −18.70), and ESO 438-G012 (M_V = −18.67). These values are within 0.5 mag of the total magnitudes in the NED.

In Figure 4 we present the dwarf galaxy candidate number density as a function of the projected radius (R) from the center of NGC 3585. The profile declines from the center, followed by flattening in the outermost region (15 ≲ R ≲ 2°). If the flattening represents the distribution of unrelated background galaxies, then the observed number density profile is compatible with a surface density distribution that is either exponential (Figure 4(a)) or follows a power law (Figure 4(b)). For the exponential form, the best-fit function is Σ = e^{−1.95(±0.67)R + 3.31(±0.55)} + 1.80(±0.50) deg⁻², while for the power-law form, it is Σ = 2.90(±0.93)R^{−1.52(±0.29)} + 1.80 deg⁻². Based on the former, we estimate the background contamination to be 10 objects (23% of our total) within a 15 radius of NGC 3585. We believe that the remaining 4 dwarf galaxy candidates located beyond R = 1.5 are most likely background sources. The contamination level we derive from the exponential model, 1.80 deg⁻², is consistent with the contamination value, 1.75 deg⁻², estimated from the background field counts (see Section 3.5 for details).

Zoom In Zoom Out Reset image size

We considered the possibility of using SBF measurement to ascertain membership, but considering our observing conditions (13 seeing in I band), the target distance (20 Mpc), and the exposure time (10 hr). By comparing our results to some work in this area (e.g., Mieske et al. 2003; Carlsten et al. 2019b), we confirmed out expectation that this is not possible.

We present the color–magnitude diagrams for the dwarf galaxy candidates in the NGC 3585 field in Figure 5. The mean (B − V)₀ and (V − I)₀ of the dwarf galaxy candidates corrected for Galactic extinction are 0.69 ± 0.18 and 0.87 ± 0.14, respectively. These mean values are similar to those obtained for other groups: the M106 group, $\langle {(B-V)}_{0}\rangle \approx 0.73$ (Kim et al. 2011), the NGC 2784 group, $\langle {(B-V)}_{0}\rangle \approx 0.67$, $\langle {(V-I)}_{0}\rangle \approx 0.85$ (Park et al. 2017), and the M83 group, $\langle {(B-V)}_{0}\rangle \approx 0.82$ (Müller et al. 2015). When the colors of the dwarf galaxy candidates are overlaid with those of dwarf galaxies in other groups (Figure 5), we find that the color distributions of the NGC 3585 dwarf galaxy candidates are consistent with those of other groups (e.g., the M83 group from Müller et al. 2015 and the NGC 2784 group from Park et al. 2017). In general, the early-type galaxies in galaxy clusters have a color–magnitude relation: the brighter the galaxies, the redder they appear (e.g., the Virgo cluster from Lisker et al. 2008 and the Ursa Major cluster from Pak et al. 2014). The dwarf galaxy candidates in the NGC 3585 also follow the relation, as shown in Figure 5. When we use dwarf galaxy candidates brighter than M_V = −11 mag in the NGC 3585 group from this study and the NGC 2784 group from Park et al. (2017), the dwarf galaxy candidates follow color–magnitude relations: (B − V)₀ = −0.012  M_V + 0.518 and (V − I)₀ = −0.013  M_V + 0.691. These results for the color distribution and relation indicate that many of the dwarf galaxy candidates in the NGC 3585 field may be early-type galaxies.

Zoom In Zoom Out Reset image size

In Figure 6 we compare the structural parameters, V-band central surface brightnesses (μ₀), effective radii (r_e), and Sérsic-n indices of the dwarf galaxy candidates in the NGC 3585 field to those found in other groups. The μ_0,V and M_V values are derived from the estimated μ_0,I and M_I (Section 3.2) using the (V − I) color of the dwarf galaxy candidates assuming that they are at the same distance as NGC 3585. The r_e and n values are those obtained from fits to the I-band images. The values for the structural parameters of dwarf galaxies in other groups are compiled from the literature: Müller et al. (2015) for the M83 group, Chiboucas et al. (2009) for the M81 group, and Park et al. (2017) for the NGC 2784 group. As seen in the M83 group (Müller et al. 2015) and in the NGC 2784 group (Park et al. 2017), the central surface brightnesses and the effective radii of the dwarf galaxy candidates in the NGC 3585 group increase with galaxy luminosity. In the case of the Sérsic curvature index, the median value, n ≈ 0.8 ± 0.1, of the dwarf galaxy candidates in the NGC 3585 field is similar to that of dwarf galaxies in other groups: n ≈ 1.1 ± 0.3 for the M83 group (Müller et al. 2015), n ≈ 0.6 ± 0.2 for the M81 group (Chiboucas et al. 2009), and n ≈ 0.8 ± 0.1 for the NGC 2784 group (Park et al. 2017). The properties of the NGC 3585 dwarf galaxy candidates are consistent with those of dwarf galaxies in other groups.

Zoom In Zoom Out Reset image size

In order to estimate the level of contamination of the dwarf galaxy candidates in NGC 3585 fields identified in Section 3.1 by sources in background, we adopt the KK196 field as a control field. The KK 196 field (R.A._J2000 = 13^h21^m4742, decl._J2000 = −45° 03'462), which is about 30° from that of NGC 3585, has been observed together with the NGC 3585 field in our KSP program, so that both the fields have almost identical integrated exposure times and depth in stacked images. In addition, the KK 196 field belongs to the Centaurus A group at 4 Mpc distance, and this makes the use of the field as a control field more reliable because the dwarf galaxies in this field associated with the Centaurus A group can easily be resolved due to their proximity. Our three classifiers searched for dwarf galaxy candidates in the KK196 field in the same manner as described in Section 3.1 As a result, we found four resolved dwarf galaxy candidates that belong to the Centaurus A group—two of them were previously identified as KK196 and KK203, while the rest two are newly indentified in this study—as well as seven unresolved dwarf galaxy candidadtes that appear to be unassociated with the Centaurus A group. We therefore consider these seven candidates as background dwarf galaxies.

We now carry out the same analysis on these seven candidates as we did on the NGC 3585 dwarf galaxy candidates. Results will be reported in greater detail by H. S. Park et al. (2019, in preparation). We summarize our results with a focus on understanding the background contamination in the NGC 3585 field as follows: (1) the spatial distribution of the seven observed diffuse candidates is compatible with a random distribution from both the 1D and 2D Kolmogorov–Smirnov tests (Press & Teukolsky 1998), (2) assuming that these objects are at the distance of NGC 3585, their distribution in the color–magnitude diagram (mean colors (B − V)₀ = 0.67 ± 0.08 and (V − I)₀ = 0.82 ± 0.10 and −13.6 < M_V < −11.1 mag) and their Sérsic structural parameters (median n ≈ 0.7 ± 0.2,) are consistent with those of the NGC 3585 dwarf galaxy candidates, (3) their projected number density is about 7/(2° × 2°), or 1.75(±0.66) deg⁻², which is very similar to the value estimated from the number density profile of NGC 3585 dwarf galaxy candidates (1.80 deg⁻², see Section 3.3), and (4) none of these candidates have either a nuclear source or are UDGs. While the color and magnitude similarities make it difficult to reject this type of background source, we have a robust estimate of their contribution to the satellite number density profiles and an indication that one way to identify bona fide dwarf galaxies may be if they are either nucleated (see Section 4.2) or ultra diffuse (see Section 4.3).

In clusters and groups, the more massive galaxies tend to be centrally concentrated (Presotto et al. 2012; Roberts et al. 2015). To investigate if dwarf galaxies in the NGC 3585 group conform to this trend, we divide our sample into two: bright (M_V < −12.5 mag) and faint (M_V ≥ −12.5 mag) objects. We show the radial number density profiles for these two subsamples in Figure 7(a). The best-fit exponential functions for the bright and faint populations are ln(Σ) ≃ − 0.58 (±0.41) R + 0.97 (±0.34) and ln(Σ) ≃ − 0.76 (±0.08) R + 1.26 (±0.07), respectively. Given the uncertainties in the fits, both populations of dwarf galaxy candidates show similar density profiles. Thus, mass segregation is not apparent among the dwarf galaxy candidates in the NGC 3585 group. In Figure 7(b) we plot the mean color of the dwarf galaxies in NGC 3585 versus projected radius from the group center. There is no apparent color variation with radius.

Zoom In Zoom Out Reset image size

The mass and color radial behaviors of the dwarf galaxy candidates in the NGC 3585 differ from the significant radial dependencies that we found in the NGC 2784 group (Park et al. 2017). This difference suggests that the NGC 3585 group might be a dynamically younger system where mass segregation has not yet developed.

Some dwarf galaxies have a distinct nucleated source in their central region (Côté et al. 2006; Trentham & Tully 2009; Turner et al. 2012) and the dwarf galaxy candidates in the NGC 3585 group are no exception. In Table 3 we classify 8 dwarf galaxy candidates among the 50 in the NGC 3585 group as nucleated based on the images (Figure 2) and the surface brightness profiles (Figure 3). These classified nucleated candidates have an evident point source that is distinctly brighter (approximately ≳0.5 mag arcsec⁻²) than the central surface brightness estimated by Sérsic fit to the entire galaxy. The incidence of the nucleated dwarf galaxy candidates is 20% (8 out of 50) in the NGC 3585 group, which is consistent with that found in other groups (20% in the LG early-type, Turner et al. 2012, and 10%–30% for several nearby galaxy groups, Trentham & Tully 2009), and 10% in the NGC 2784 group (Park et al. 2017)⁸ ). In contrast, in the Virgo and Fornax clusters, studies have found that about 60%–80% of the dwarf galaxies have nuclei (Côté et al. 2006; Lisker et al. 2007; Turner et al. 2012; Georgiev & Böker 2014). Although selection effects could play a role, cluster samples are biased to brighter samples; even our bright group sample has a small nucleated incidence (∼30%; Figure 8(a)). Unfortunately, the significance of the group results is undermined by small number statistics. Confirmation of the difference between groups and clusters awaits larger samples. Nevertheless, none of the 7 dwarf galaxy candidates in the control field (see Section 3.5) has a nucleus, which indirectly supports a conjecture that the nucleated dwarf galaxy candidates are associated with the NGC 3585 group.

Zoom In Zoom Out Reset image size

In Figure 8 we compare the luminosities (M_V), colors ((B − V)₀), radial distribution from the group center (R), central surface brightness (μ_0,V), effective radii (r_e), and Sérsic-n of the nucleated dwarf galaxy candidates to those of other dwarf galaxy candidates. We overlay the distribution of the number ratio (N_NUC/N_ALL) in each panel. From these plots, we conclude that nucleated dwarf galaxy candidates are more common among brighter (a, d), redder (b), larger (e), and more centrally concentrated (f) dwarf galaxy candidates. We note that due to the relationships among these parameters, a high incidence for redder dwarf galaxy candidates is related in corresponding higher incidence among brighter, more concentrated dwarf galaxy candidates due to the color–magnitude relation and the magnitude–structural parameter trends described in Section 3.4 The correlation between nuclear sources and color is consistent with those obtained for dwarf elliptical galaxies (dEs) in nearby galaxy clusters (Rakos & Schombert 2004; Lisker et al. 2007). However, the incidence of nucleated dwarf galaxy candidates does not vary with distance from the center in the NGC 3585 group (Figure 8(c)), which is in contrast to the trend in clusters. The nucleated dwarfs in the Virgo cluster are located in denser environments. If the radial trend in Virgo is due to mass segregation, with the nucleated dwarfs being the more massive and sinking toward the center, our result might again suggest that the NGC 3585 system is dynamically young.

Numerous UDGs have recently been found in nearby galaxy clusters and groups (van Dokkum et al. 2015; van der Burg et al. 2016; Müller et al. 2018; Zaritsky et al. 2019). We find four UDG candidates in the NGC 3585 group based on central surface brightness (μ_0,V ≳ 23.7 mag arcsec⁻²) and effective radius (r_e,V ≳ 1.5 kpc) criteria (van Dokkum et al. 2015). The UDG candidates in the NGC 3585 group (see Table 3)⁹ have parameters in the range of 24.3 < μ_0,V < 25.9 mag arcsec⁻², 1.6 < r_e,V < 2.0 kpc. 0.55 ≤ (B − V)₀ ≤ 0.75, and n_V ≲ 1.0. The latter two quantities are similar to the average values, (B − V)₀ ≈ 0.69 and n ≈ 0.8, of all the dwarf galaxy candidates in the NGC 3585 group. The absence of UDGs among the control field dwarf candidates (see Section 3.5) indirectly supports the conjecture that the UDG candidates are associated with the NGC 3585 group.

An interesting properties of the four UDGs in this group is that they are centrally concentrated, all within R < 0.15 Mpc (Figure 7(b)). A strong central concentration is in conflict with results from UDGs in galaxy clusters (van Dokkum et al. 2015; van der Burg et al. 2016; Román & Trujillo 2017a) and some galaxy groups (Merritt et al. 2016; Román & Trujillo 2017b; Müller et al. 2018), which instead find a deficit of UDGs in the central region. While the statistical significance of the result for NGC 3585 is again limited due to small numbers, the result again suggests that the NGC 3585 group might be a dynamically younger system in which centrally located UDG candidates have not yet been tidally disrupted.

Finally, we add our measurement to previous measurements to investigate the number abundance of UDGs as a function of group mass. In this instance, we consider a velocity dispersion as a system mass. A number of four UDGs in a system with a velocity dispersion of σ_υ = 70 km s⁻¹ (see Section 3.1) causes a flattening of the relationship between the UDG number abundance and system mass (see Figure 6 by Román & Trujillo 2017b or Figure 10 by Lee et al. 2017). If confirmed for similar low-mass groups, this result may indicate that UDGs form preferentially in groups (Román & Trujillo 2017b) or are less effectively destroyed.

To compare the LF of the galaxies in the NGC 3585 group with those of other galaxy groups, we use cumulative LFs because, in general, the number of galaxies in a group is small (Trentham & Tully 2009). To construct the LF for the NGC 3585 group, we only use galaxies with R < 0.5 Mpc to minimize the effect of background contamination (Section 3.3). The other groups in this comparison include the M81 group (Chiboucas et al. 2009), the M83 group (Müller et al. 2015), the M96 group (Müller et al. 2018), the M101 group (Bennet et al. 2017, 2019; Danieli et al. 2017; Carlsten et al. 2019a), the M106 group (Kim et al. 2011), the NGC 2784 group (Park et al. 2017), and the NGC 5128 group (Tully et al. 2015; Crnojević et al. 2016). The R < 0.5 Mpc condition is satisfied for all of these other groups except for the NGC 5128 group. However, in this case, most of the dwarfs are spectroscopically confirmed members, therefore contamination is not an issue. In Figure 9(a) we compare the cumulative LFs of the NGC 3585 group to those of these seven other groups.

Zoom In Zoom Out Reset image size

The eight groups have roughly similar LF shapes and faint-end slopes. However, we do note that certain groups (M81, M83, NGC 2784, and NGC 3585) have a clear faint-end hump, a steep increase in the cumulative numbers around M_V ≈ −13 mag. These LF humps could be a real structure, with physical meaning that could be exploited, or perhaps they are the result of background contamination. To resolve this ambiguity, either deep spectroscopic surveys to determine membership are required or a larger, homogeneous photometric survey, such as our complete KMTNet dwarf galaxy search project. With a much larger sample it would be possible to establish whether this is a universal feature and if it is always observed at the same apparent magnitude, in which case one would suspect that it is the result of background contamination, or at the same absolute magnitude, in which case one would suspect that it is a physical effect. We find from the background contamination test (Section 3.5) that contaminating systems tend to be fainter than M_V ≈ −13 mag, and so the hump in the case of the NGC 3585 group could at least partially be attributed to contamination.

We fit the observed cumulative LF of each group, for M_V < −10 mag, to a cumulative Schechter function (Schechter 1976) using the technique described in Chiboucas et al. (2009). The best-fit faint-end slope, α = −1.39 ± 0.03, for the cumulative LF of the NGC 3585 group is similar to those of other groups (Figure 9(a)). The results for the cumulative LFs of the other groups except for the M96 group with α = −1.36 ± 0.03 are introduced in detail in Section 4.2 by Park et al. (2017), and the M101 group is excluded from further comparison because the group shows an abnormal LF shape: very flat for M_V < −10 mag and very steep for M_V > −10 mag (see Bennet et al. 2019 for more detail). We find that the groups have a range of slopes, with the LF slope of the LG (α ≈ −1.0; Chiboucas et al. 2009; McConnachie et al. 2009; Kim et al. 2011; Ferrarese et al. 2016; Park et al. 2017) being considerably flatter than any of these. On the other hand, all observed faint-end slopes are much flatter than that of the subhalo mass function in the ΛCDM model (α = −1.8) (Trentham & Tully 2002).

To search for a dependence of the faint-end slope on group properties, we initially investigated a correlation between group mass and LF slope in our previous study (Park et al. 2017). There we suggested that the LF slope flattens as group mass increases. However, the small sample size, the large uncertainties, and the membership problem render this a preliminary result only. Including the new results for the NGC 3585 group and the M96 group in Figure 9 seems to support the previous claim, although we still need more data. To construct this figure, we adopted σ_group as an indicator for group mass. The values of σ_group for groups other than NGC 3585 and M96 are from Table 4 in Park et al. (2017). The values for the NGC 3585 group, σ_group = 70 km s⁻¹, and the M96 group, σ_group = 233 km s⁻¹ listed as M105, are from Makarov & Karachentsev (2011). The relation between α and σ_group can be expressed as α = 0.0003  σ_group −1.353, but a correlation analysis suggests that the confidence with which we can claim a correlation is only at about 1σ. This marginal nature of the result could be the result of systematic differences among group results obtained from diverse surveys and of small sample statistics as well as of the group membership. A large and uniform survey is required to establish this result. The homogeneous samples obtained from the KMTNet survey of dwarf galaxies in ≳30 groups will serve this purpose.

We close this section by noting the apparent discrepancy between our results and those of Zabludoff & Mulchaey (2000) and Balogh et al. (2001), who find steeper faint-end slopes in denser, more massive environments. However, these previous results characterize significantly more luminous galaxies than those characterized here (e.g., −17.8 < M_R < −19.8, for Zabludoff & Mulchaey 2000). If the faint-end LF is not a simple power law, as suggested earlier in this section, then constraining α over different absolute magnitude ranges will lead to apparently conflicting results. Interpreting differences among results from various studies is also complicated by the comparison of LFs derived from spectroscopically confirmed samples, such as that used by Zabludoff & Mulchaey (2000), and those derived from samples that either apply statistical background corrections or argue that they are minor. It is therefore premature to place much weight on the trend seen in Figure 9 until the larger group survey securely establishes the shape of the faint-end LF and the role of background contamination.