UNVEILING A RICH SYSTEM OF FAINT DWARF GALAXIES IN THE NEXT GENERATION FORNAX SURVEY

The Fornax cluster is the most nearby southern galaxy overdensity and an excellent laboratory for studying the formation and evolution of galaxies and other dense stellar systems, such as globular clusters (GCs) and dwarf galaxies. In comparison to its northern counterpart, the Virgo galaxy cluster, the Fornax cluster has twice the central galaxy density, half the velocity dispersion, and a distinctly lower mass (7 ± 2 × 10¹³ M_⊙; e.g., Schuberth et al. 2010). Furthermore, the core of Fornax is dynamically more evolved (Churazov et al. 2008) and its early-type galaxy fraction is significantly larger than that of Virgo. Overall, these properties make the Fornax cluster an excellent target for studies of galaxy evolution processes in dense cluster environments.

Probing deeper into the galaxy luminosity function allows us to put more stringent constraints on galaxy formation models (e.g., Bovill & Ricotti 2009, 2011a, 2011b; Brooks et al. 2011; Pfeffer et al. 2014; Vogelsberger et al. 2014; Burkert 2015; Mistani et al. 2015). In addition to the rich populations of low surface brightness dwarf galaxies in the Local Volume (McConnachie 2012), there have been recent and surprising discoveries of ultra-diffuse galaxies (UDGs) in the Coma and Virgo galaxy clusters (Koda et al. 2015; Mihos et al. 2015; van Dokkum et al. 2015). The extremely low stellar masses of UDGs (∼6 × 10⁷ M_⊙) combined with their large radii (1.5–4.6 kpc) translate into very low surface brightness values in the range μ_0,V ≈ 26.0–28.5 mag arcsec⁻², making them extremely difficult to detect. The existence of this mysterious new dwarf galaxy class in dense galaxy cluster environments prompts the obvious question of whether there might be similar populations in other galaxy clusters. In this Letter, we use multi-passband, wide-field observations conducted by us with the Dark Energy Camera (DECam; Flaugher et al. 2015) at the 4-m Blanco telescope at Cerro Tololo Interamerican Observatory as part of The Next Generation Fornax Survey (NGFS) to search for faint dwarf galaxies in the Fornax galaxy cluster central regions (R_NGC1399 ≲ 350 kpc) and measure their structural parameters and spatial distribution.

Throughout this work, we adopt a Fornax distance modulus of 31.51 mag (20.0 Mpc) based on the surface brightness fluctuation method from Blakeslee et al. (2009).

The NGFS is an ongoing multi-wavelength survey of the central 30 deg² of the Fornax galaxy cluster (Ferguson 1989). We are using Blanco/DECam for near-ultraviolet and optical photometry as well as VISTA/VIRCam (Sutherland et al. 2015) for near-infrared photometry of the same field. The NGFS survey is aimed at detecting point sources at S/N = 5 over the PSF area with u = 26.5, g = 26.1, i = 25.3, and K_s = 23.3 AB mag. Below, we offer a summary of the NGFS features most salient for this Letter and defer a detailed description of the characteristics and data reduction to a subsequent paper.

The current NGFS mosaic consists of nine contiguous tiles, and the present work is based on the optical imaging of the central tile. Our dithering pattern is based on the Elixir-LSB technique developed by Jean-Charles Cuillandre for the Next Generation Virgo Survey (see Ferrarese et al. 2012), which is designed to detect low surface brightness structures. The raw images were processed by the DECam Community Pipeline (v2.5.0), including bias calibration, crosstalk correction, linearity correction, flat fielding, and gain calibration. We apply our own sky background subtraction and stacking recipes via an iterative procedure of object masking and sky modeling using the Astromatic software¹¹ (Sextractor v2.19.5, Scamp v2.0.1, and Swarp v2.38.0). Our astrometric and photometric calibrations are based on 2MASS Point Source Catalog (Skrutskie et al. 2006) reference stars and Sloan Digital Sky Survey stripe 82 standard star frames taken under photometric conditions, respectively.

In this Letter, we present the analysis of the central 3 deg² (Figure 1), covering the Fornax cluster core out to ∼350 kpc radius from NGC 1399, corresponding to 25% of the Fornax cluster virial radius (1.4 Mpc; Drinkwater et al. 2001). We created an RGB image from the processed images, which was visually inspected by several of us (RPM, PE, THP, YO, KAM, KXR) to independently identify low surface brightness dwarf galaxy candidates. Table 1 includes 284 dwarf galaxy candidates that were identified by every person on the search team, 264 of which we derived structural parameters for by analyzing their surface brightness profiles (see Section 3).

Zoom In Zoom Out Reset image size

Table 1.  Dwarf Galaxies in the Core of the Fornax Cluster

ID	α2000	δ2000	mi	Mia	nb	reff	reffa	Typec	Reference
			(mag)	(mag)		(arcsec)	(kpc)
NGFS033301-353635	03:33:00.83	−35:36:34.59	17.70	−13.81	0.89	5.397	0.523	$\circ$	⋯
NGFS033309-352349	03:33:08.63	−35:23:49.02	19.18	−12.33	0.81	7.088	0.687	$\circ$	FCC114
NGFS033311-353956	03:33:10.93	−35:39:56.17	20.56	−10.95	0.63	8.950	0.868	$\circ$	⋯
NGFS033316-353551	03:33:15.98	−35:35:50.90	21.00	−10.51	0.69	2.554	0.248	$\circ$	⋯
NGFS033322-353620	03:33:22.18	−35:36:20.20	18.70	−12.81	0.88	4.439	0.430	⊙	⋯
NGFS033331-352654	03:33:31.41	−35:26:54.00	22.74	−8.77	0.47	2.427	0.235	$\circ$	⋯
NGFS033332-353942	03:33:32.11	−35:39:42.21	19.86	−11.65	1.00	8.258	0.801	⊙	⋯

Notes.

^aAssuming a distance modulus of (m − M)₀ = 31.51 mag (Blakeslee et al. 2009). ^bSérsic index (Sérsic 1968; Caon et al. 1993). ^cMorphological galaxy type classification: ⊙ = nucleated, $\circ$ = non-nucleated dwarf galaxy.

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

We compared our sample with objects flagged as "likely members" in the FCC catalog (Ferguson 1989), finding 92 matches with a maximum separation of 12''. Out of these galaxies, there are 47 matched non-nucleated and 45 nucleated dwarfs. We also find 27 non-nucleated and 7 nucleated dwarfs not present in the FCC catalog, but flagged as likely members in the Mieske et al. (2007) catalog. We fail to recover six objects from the Mieske et al. catalog, which we visually confirm from Figure 1 to be background galaxies.

We estimate the structural parameters of each dwarf candidate in Table 1 using the software package galfit (v3.0.5; Peng et al. 2010). We approximate one-component fits to the two-dimensional galaxy surface brightness distribution assuming a Sérsic profile (Sérsic 1968; Caon et al. 1993). We split the fitting procedure into several steps. We create individual postage stamp images for each dwarf candidate and construct their segmentation maps using SExtractor (Bertin & Arnouts 1996). These segmentation maps are then used to create bad pixel masks for each dwarf, masking any non-dwarf sources above a 3σ threshold, that produce dwarf-only images on which we perform the model fits. We then characterize the dwarfs as either nucleated or non-nucleated.

The non-nucleated dwarfs are fit by a single-Sérsic profile, while the nucleated dwarfs are approximated by two components. The robustness of the fitting procedure is estimated from a few systematically differing initial guesses for all model parameters. For 88 non-nucleated dwarfs, all varying initial guesses immediately converged to the same solution and are thus considered to be robust fits. On the other hand, for 114 mostly fainter objects, galfit could not find a stable solution, so we refine the fitting by adopting the following iterative procedure. First, we estimate the total galaxy luminosity from SExtractor MAG_AUTO values using the corresponding segmentation maps. These magnitudes are then used as initial guesses for the galfit Sérsic models and kept fixed during the fit. In this way, the fits are stabilized so that the remaining model parameters can be derived. Next, the newly estimated parameters are kept fixed and the galaxy luminosity is recomputed with galfit. In the final step of the procedure, the newly determined magnitude is again fixed to recompute the other parameters. In this way, all non-nucleated dwarfs were successfully fit, and each galaxy model and residual image were visually inspected to ensure the derived parameters are statistically robust and provide reliable approximations.

To achieve consistent and meaningful fits for the nucleated dwarfs, we modified the procedure to properly account for the excess light in the nuclei. We first attempted to fit a galaxy model with two components, e.g., a Nuker or King profile for the nucleus and a Sérsic profile for the extended, diffuse galaxy spheroid. However, very few stable solutions were found for the nuclei due to their small sizes. As a solution to this problem, we instead mask the nuclei and then fit a single Sérsic to the outer diffuse galaxy spheroid. From the residual images (including the galaxy nuclei), we generate improved masks for the nuclei and any other sources contaminating the diffuse components of the dwarfs. This procedure resulted in stable fits for the diffuse spheroids of all nucleated dwarf galaxy candidates. Again, the residual and model images for each galaxy were visually inspected to validate the robustness of the computed model parameters.

For 20 galaxies in our sample we could not obtain stable fits mostly due to their locations near bright stars or simply being too faint. All remaining 264 dwarfs can be reliably fit with a Sérsic model with 0 ≲ n ≲ 2 and show a well-defined peak around n ≈ 0.7. This is consistent with previous measurements (e.g., Côté et al. 2007) that show that the main bodies of galaxies are well defined by a Sérsic profile with Sérsic indices that vary smoothly from n ≈ 4 (deVaucouleurs r^1/4 profile) in the most massive systems to n ≈ 1 (exponential profiles) in the lowest-mass galaxies.

In this section, we explore various parameter spaces of the newly expanded Fornax dwarf galaxy population. We first compare them to other galactic systems in the context of their respective size–luminosity relations and then provide an overall view of their spatial distributions with respect to both each other and the population of giant elliptical galaxies in the Fornax cluster core.

4.1. Size–Luminosity Relation

Figure 2 shows the effective radii versus absolute i-band magnitudes for our NGFS sample dwarf galaxies in relation to other stellar systems in the nearby universe. The detected Fornax dwarfs range mainly between values of ${\langle {\mu }_{i}\rangle }_{e}$ ≈ 22 and 28 mag arcsec⁻², with absolute luminosities down to approximately −8 mag—similar to the Virgo ultra-faint dwarf recently detected by Jang & Lee (2014)—and half-light radii from r_e ≃ 90 pc up to ∼2.5 kpc. For the 92 galaxies in common between our NGFS and the FCC catalog, we find good agreement between the magnitudes and effective radii, confirming the robustness of our measurements. The measured luminosities and structural parameters are provided in Table 1.

Zoom In Zoom Out Reset image size

At the bright end of the NGFS sequence in Figure 2, we find dwarf galaxies that resemble UDGs identified by Mihos et al. (2015) in Virgo and van Dokkum et al. (2015) in the Coma cluster. We therefore confirm that UDGs also exist in the central regions of the Fornax galaxy cluster and are not confined to massive galaxy clusters alone. However, at a given luminosity, the Fornax dwarfs are significantly smaller than the Coma and Virgo UDGs. The existence of such galaxies has been predicted and observed by Dalcanton et al. (1997). At the faint end of the NGFS dwarf sequence, the lowest surface brightness dwarfs find analogs in Local Group galaxies such as the Draco, Ursa Major, and Andromeda XXVII dwarf spheroidals (McConnachie 2012). Overall, the NGFS dwarf sequence appears very similar to the size–luminosity relation of Local Group dwarf spheroidal galaxies. However, within the photometric completeness limits of our data, we detect a small offset toward a higher average surface brightness of NGFS dwarfs compared with Local Group dwarfs. This might be due to either (i) sample statistics of the Fornax dwarf population, (ii) the harsher ram-pressure stripping and tidal environment in Fornax that at a given total luminosity can render galaxies more compact (Mayer et al. 2006), and/or (iii) different star formation histories and stellar population properties of NGFS dwarfs (e.g., Guérou et al. 2015). We will discuss these relations in detail in a subsequent paper.

4.2. Nucleation Fractions

4.3. Spatial Distribution

A comprehensive view of the new dwarf candidate spatial distribution is illustrated in Figure 3, where the dwarfs have been limited to those within 60' (∼350 kpc) of NGC 1399 to ensure uniform angular representation. The result of a two-point angular correlation function (Landy & Szalay 1993) analysis of the NGFS dwarf sample is shown in the top left panel along with 1σ bounds. On length scales below ∼100 kpc, an intriguing excess appears, indicating the clustering of dwarf galaxies. This may point to (i) dwarf–giant galaxy clustering and/or (ii) bound dwarf galaxy groups that survive in the galaxy cluster environment on those or smaller spatial scales. Interestingly, such dwarf galaxy clustering has been reported in high-resolution numerical simulations (Kravtsov et al. 2004; Weinberg et al. 2008; van Daalen et al. 2014; Wheeler et al. 2015). We will investigate this result and its interpretation in a dedicated future paper.

Zoom In Zoom Out Reset image size

The bottom left panel of Figure 3 shows the projected radial surface number density (Σ_N) profiles for the non-nucleated dwarfs (red), nucleated fraction (blue), and total population (brown). The Σ_N values show the number of dwarfs per arcmin² within adaptively sized annuli containing 20 galaxies each for the total sample, and appropriately scaled for the two subsamples such that similar total data points are obtained, i.e., 15 and 5 galaxies per bin for the non-nucleated and nucleated samples, respectively. Cluster-centric distances corresponding to annuli mid-points are shown along the top and bottom axes in physical and angular units. The non-nucleated sample and total population are both consistent with having flat Σ_N distributions of ∼0.01–0.02 arcmin⁻² out to ∼350 kpc, whereas the nucleated dwarf population (noting the much smaller sample) appears to decline outside of ∼200 kpc. However, this result requires confirmation with a larger sample from the full NGFS survey footprint.

The distributions in azimuthal angles are shown in the right-hand panel of Figure 3, where the color opacity has been decreased to show where the samples overlap. The smooth curves indicate non-parametric Epanechnikov-kernel probability density estimates based on the discrete data, and each curve exhibits several distinct peaks. For all curves, there is a noteworthy overdensity along the Φ ≃ 100° direction, which coincides with the inter-cluster medium between the giant galaxies NGC 1427 and NGC 1399 (see Figure 1). The non-nucleated population appears much more stochastic, but with a notable broad distribution in the range of 200° ≲ Φ ≲ 300°, with individual peaks that closely follow the distributions of the other giant galaxies in the Fornax core, e.g., NGC 1386 (Φ ≃ 220°), NGC 1387 and 1379 (Φ ≃ 250°), and NGC 1374 (Φ ≃ 275°). While the total number of nucleated dwarf candidates prohibit making definitive statements, their distribution tends to follow the same general east–west bimodality, but without the discrete overdensities shown by the populous samples. Thus, while they are likely to trace the interactions between the dominant galaxies of Fornax, their broader distributions limit what can be concluded regarding specific associations at this point. The full NGFS survey data set will expand this result to a larger-scale analysis.

This project is supported by FONDECYT Postdoctoral Fellowship Project No. 3130750, FONDECYT Regular Project No. 1121005, and BASAL Center for Astrophysics and Associated Technologies (PFB-06). P.E. acknowledges support from FONDECYT Postdoctoral Fellowship Project No. 3130485. K.A.M. acknowledges support from FONDECYT Postdoctoral Fellowship Project No. 3150599. M.A.T. acknowledges the financial support through an excellence grant from the "Vicerrectori´a de Investigación" and the Institute of Astrophysics Graduate School Fund at Pontificia Universidad Católica de Chile and the European Southern Observatory Graduate Student Fellowship program. G.G. acknowledges support from FONDECYT Regular Project No. 1120195.

This project used data obtained with the Dark Energy Camera (DECam), which was constructed by the Dark Energy Survey (DES) collaboration.

This research has made use of the VizieR catalog access tool and the Aladin plot tool at CDS, Strasbourg, France. We are grateful to Roberto González, Francisco Valdes, and David James for helpful discussions.

Facility: CTIO (4-m Blanco/DECam) - .