The Next Generation Fornax Survey (NGFS). III. Revealing the Spatial Substructure of the Dwarf Galaxy Population Inside Half of Fornax's Virial Radius

Large numbers of faint, low surface brightness dwarf galaxies are rapidly being discovered in different environments throughout the local universe (e.g., van Dokkum et al. 2015; Muñoz et al. 2015; Müller et al. 2017; Wittmann et al. 2017). The rise of large detector arrays in present-day observatories such as the Dark Energy Camera (DECam; Flaugher et al. 2015) enables us to survey large areas of the sky down to ultra low surface brightness levels, providing the exciting opportunity to search for undiscovered faint dwarf galaxies in massive clusters, galaxy groups, and in the field (e.g., Muñoz et al. 2015; Ferrarese et al. 2016; Müller et al. 2017; Venhola et al. 2017). Dwarf galaxies are found in galaxy group and cluster environments, which are numerically dominated by early-type dwarf galaxies with characteristically smooth morphologies, exponential surface brightness profiles, and stellar populations consistent with red-sequence galaxies (Sandage & Binggeli 1984; Binggeli et al. 1985; Conselice et al. 2003; Misgeld et al. 2008; den Brok et al. 2011; Ordenes-Briceño et al. 2016; Roediger et al. 2017; Eigenthaler et al. 2018). Early-type dwarf galaxies have typically been classified as dwarf ellipticals (dE), but are also known as dwarf spheroidals (dSph) at fainter magnitudes (Grebel et al. 2003). They exhibit absolute B-band magnitudes fainter than M_B ≃ −16, corresponding to $\mathrm{log}({{ \mathcal M }}_{\star }/{M}_{\odot })\lesssim 9$, and effective radii smaller than ∼1 kpc (cf. Figure 8 in Eigenthaler et al. 2018). A further morphological distinction among dwarf galaxies is whether or not they host a central nuclear star cluster. Recent findings show that dwarf nucleation probability is strongly dependent on its spheroid luminosity (Muñoz et al. 2015; Ordenes-Briceño et al. 2018), with the fraction of nucleated dwarfs systematically increasing toward brighter magnitudes.

Large populations of low-mass dwarf galaxies are ideal for studying the dependence of galaxy formation and evolution processes in the transition zones between field and cluster environments, especially in rich galaxy clusters. Statistically significant samples allow us to study their clustering properties on large scales (Muñoz et al. 2015) and potentially probe the dark matter (DM) fine-structure within the cluster halo. This distribution in return serves as an ideal laboratory for comparisons with predictions from structure formation models (e.g., Bovill et al. 2016).

Recently, observations have revealed that low-mass dwarf galaxies appear to form surprisingly thin planes in the local universe (Pawlowski et al. 2012; Ibata et al. 2013; Tully et al. 2015). The observations of such planes challenge current ΛCDM models of hierarchical structure formation (Kroupa et al. 2005; Pawlowski et al. 2014). The frequency of the occurrence of such anisotropic distributions in dense environments will inform an updated view of structure formation in modern galaxy formation models. These findings show that deep homogeneous surveys are necessary to highlight the spatial distribution of the faint dwarf galaxies in nearby groups and clusters of galaxies without completeness.

In the present work, we attempt to lay the foundation for addressing these issues by investigating the faint dwarf galaxy population in the Fornax galaxy cluster out to half of its virial radius (r_vir), using data obtained as part of the Next Generation Fornax Survey (NGFS). The most prominent survey covering the galaxy cluster outskirts (Ferguson 1989), while seminal, is showing signs of age and suffers from shallow detection limits (m_B ≲ 20 mag for point sources and μ_B ≲ 24 mag arcsec⁻² in surface-brightness sensitivity) compared to the potential of modern instrumentation. Given the faint surface brightness of dwarf galaxies recently detected in various galaxy aggregates, the goal of this paper is to update the known population of Fornax dwarf galaxies out to ≲r_vir/2, where one might expect to witness the transition from the central galaxy population to those residing in the cluster outskirts. The Fornax cluster is the nearest high-density region in the southern hemisphere (m–M = 31.51 mag or D_L = 20.0 Mpc, Blakeslee et al. 2009). Given that its proximity is twice the central galaxy density and its early-type galaxy (ETG) fraction is larger than its northern hemisphere counterpart, the Virgo Cluster, Fornax is an important nearby laboratory to investigate the dependence of galaxy evolution on the dynamical state of the environment.

The data presented in this paper is part of the observed NGFS (Muñoz et al. 2015), an ongoing, panchromatic ∼30 deg² survey of the Fornax galaxy cluster using the Dark Energy Camera (DECam; Flaugher et al. 2015) mounted on the 4 m Blanco telescope at Cerro Tololo Interamerican Observatory (CTIO). Figure 1 illustrates the Fornax cluster region that is covered by our inner NGFS footprint, which consists of seven tiles centered on the dominant NGC 1399 galaxy, and homogeneously maps the cluster out to ∼50% of its virial radius (r_vir ≃ 1.4 Mpc; Drinkwater et al. 2001). The dwarf galaxy population in the central tile corresponding to r ≤ r_vir/4 was already studied in Muñoz et al. (2015) and Eigenthaler et al. (2018). All tiles are observed in three optical bands reaching point-source detections with S/N ≳ 5 at 26.5, 26.1, and 25.3 AB mag in the u'-, g'-, and i'-band, respectively.

Zoom In Zoom Out Reset image size

Initial image processing is carried out by the CTIO Community Pipeline (CP; Valdes et al. 2014), focusing mainly on instrumental signature removal (e.g., bias subtraction, flat-fielding, cross-talk correction). After the CP-processing, we apply further processing using the Astromatic¹⁵ software suite to astrometrically calibrate and stack the individual frames and to conduct basic photometry (scamp, swarp, source extractor, hereafter SE; Bertin & Arnouts 1996; Bertin et al. 2002; Bertin 2006). Astrometric and photometric calibration has been performed using the 2MASS astrometric Point Source Catalog (Skrutskie et al. 2006) and SDSS u'g'i' stripe 82 standard stars, respectively. To ensure accurate photometric calibration, we cross-verified it with the globular cluster catalog from Kim et al. (2013) in the Fornax area, which was compiled using U-, B-, V-, and I-band photometry taken with MOSAIC-II camera on CTIO/Blanco. We use the empirical transformation equations from Jordi et al. (2006) for the comparison, finding good agreement within the uncertainties. Finally, we proceed to the final image stacking and source detection.

In previous contributions (Muñoz et al. 2015; Eigenthaler et al. 2018), we presented the NGFS results, focusing on the inner ∼3 deg² (≲r_vir/4) region of the Fornax cluster. We presented optical colors and structural parameters for ≳250 dwarf galaxy candidates, reaching out to a projected distance of ∼350 kpc from NGC 1399; its spatial distribution suggested a rich and substructured dwarf galaxy population extending well beyond these limits.

In this paper, we expand upon these initial results by using the surrounding DECam pointings (NGFS tiles 2–7; see Figure 1) to identify 271 new dwarf galaxy candidates, of which 39 are nucleated. In the following, we refer to the surrounding NGFS tiles (2–7) as the outer footprint. Together with the 121 previously found NGFS dwarfs from the central tile (Muñoz et al. 2015; Eigenthaler et al. 2018), there is a total of 392 new dwarfs, of which 56 are nucleated, that were discovered in the NGFS data. Our by-eye dwarf detection strategy used for the central tile is again utilized, where several members of the NGFS team (KAM, KXR, MAT, PE, THP, YOB) independently compiled dwarf candidate lists for tiles 2–7 using RGB full-color image stacks, constructed from the individual u'g'i' frames. By using cross-matching lists and setting a minimum threshold of three independent detections, we yielded a new robust list of dwarf galaxy candidates projected within r_vir/2 of NGC 1399. While all frames are fully reduced, we defer a full-color and stellar population analysis to a future paper and limit the scope of the present work to a monochromatic i'-band presentation of magnitudes, structural parameters, and spatial distribution characteristics.

We complement our final catalog with the known dwarf galaxy population in Fornax, using the likely members in the Fornax Cluster Catalog (FCC; Ferguson 1989). In the outer NGFS footprint (0.25 < R/R_vir ≤ 0.5), we find a total of 114 FCC galaxies (29.6%) and in the so-far searched NGFS survey area, a total of 251 literature galaxies (39%; Ferguson 1989; Mieske et al. 2007). Taking into account the dwarfs from existing catalogs and the new NGFS dwarfs, the total Fornax dwarf galaxy population reported in this work consists of 643 dwarfs of which 462 are nonnucleated and 181 are nucleated.

The surface brightness profiles for the dwarfs are studied with GALFIT (v3.0.5 Peng et al. 2002) using a Sŕsic profile (Sérsic 1963; Caon et al. 1993). Our procedure has been described in detail in Muñoz et al. (2015) and Eigenthaler et al. (2018). It is an iterative process where the light profile is approximated with a one-component fit to the two-dimensional (2D) galaxy surface brightness distribution. We run GALFIT on cutout images of 105'' × 105'' in size ($\stackrel{\wedge}{=}10.2$ kpc × 10.2 kpc) using object masks created from SE segmentation maps and PSF models created with PSFex (Bertin 2011). For nucleated dwarfs, we iterate the method described above several times and improve the object mask in each iteration step until the nucleus is completely masked (see Eigenthaler et al. 2018; Ordenes-Briceño et al. 2018, for more details). The final fitting profile considers only the spheroid light component of the dwarf galaxy, leaving a residual image (original–spheroid model) with the nuclear cluster in the galaxy center. The analysis of the nuclear star clusters will be presented in an upcoming NGFS contribution. In the following, we focus on the structural parameters of the spheroid sample. The dwarf candidates from the outer footprint have absolute i'-band magnitudes in the range −18.80  ≤M_i' ≤ −8.78 with photometric errors <0.1 mag, effective radii between 18 and 228 (r_eff,i' = 0.18–2.22 kpc at the Fornax cluster distance of D_L = 20.0 Mpc), a mean Sérsic index of $\langle n{\rangle }_{i}=0.81$, and an average axis ratio of $\langle b/a{\rangle }_{i^{\prime} }=0.69$.

Comparing the central and outer regions in terms of mean structural parameters and nucleation fraction, we see some interesting differences mostly in the mean magnitudes and effective radii. Nonnucleated dwarfs in the central region are brighter and larger with $\langle {M}_{i^{\prime} }\rangle =-11.99\pm 0.12$ mag and $\langle {r}_{\mathrm{eff},i^{\prime} }\rangle =0.61\pm 0.03\,\mathrm{kpc}$, relative to nonnucleated dwarfs in the outer region, which have $\langle {M}_{i^{\prime} }\rangle =-11.65\pm 0.10$ mag and $\langle {r}_{\mathrm{eff},i^{\prime} }\rangle =0.55\pm 0.02\,\mathrm{kpc}$. Nucleated dwarfs in the central region have significantly fainter average luminosities but are similar in average size compared to their nucleated counterparts in the outer footprint, $\langle {M}_{i^{\prime} }\rangle =-12.43\pm 0.21$ mag, $\langle {r}_{\mathrm{eff},i^{\prime} }\rangle \,=0.91\pm 0.04\,\mathrm{kpc}$ (central) and $\langle {M}_{i^{\prime} }\rangle =-13.87\pm 0.2$ mag and $\langle {r}_{\mathrm{eff},i^{\prime} }\rangle =0.95\pm 0.04\,\mathrm{kpc}$ (outer). These differences are more pronounced when comparing the nucleated and nonnucleated dwarf population, i.e., the nonnucleated dwarf population is fainter than nucleated dwarfs by ${\rm{\Delta }}\langle {M}_{i^{\prime} }\rangle =0.44$ mag in the central region and ${\rm{\Delta }}\langle {M}_{i^{\prime} }\rangle =2.25$ mag in the outer region. In addition, the nonnucleated dwarf population is, on average, smaller than the nucleated dwarf population by ${\rm{\Delta }}\langle {r}_{\mathrm{eff},i^{\prime} }\rangle \,=0.3\,\mathrm{kpc}$ and ${\rm{\Delta }}\langle {r}_{\mathrm{eff},i^{\prime} }\rangle =0.4\,\mathrm{kpc}$, in the central and outer regions, respectively.

Nonnucleated dwarfs have similar mean sizes and luminosities independent of local environmental density, i.e., central versus outer region. However, central nucleated dwarfs are on average about 1.4 mag fainter than nucleated dwarfs in the outer regions. In the central regions, nucleated and nonnucleated have similar magnitudes but different mean sizes. However, in the outer region, nucleated dwarfs are on average brighter and larger than the nonnucleated dwarfs. Table 1 lists the IDs, coordinates, and structural parameters for the dwarf candidates at cluster–centric radii r_{NGC 1399} > r_vir/4, complementing the sample from Eigenthaler et al. (2018) for the inner region (r_{NGC 1399} < r_vir/4).

4.1. Size–Luminosity Relation

Scaling relations are useful tools to gain insight into the link of the formation processes between different astronomical objects. We illustrate in Figure 2 the size–luminosity relation in terms of the effective radius and absolute i'-band magnitude for the entire NGFS dwarf sample. A total of 452 nonnucleated and 178 nucleated dwarfs are shown. Thirteen dwarfs do not have structural parameter information due to very low surface brightness and/or complicated contamination in their nearby environment (e.g., bright star spikes, detector blemishes, crowded field). To map a large luminosity range, we overplot different galaxy samples, including Local Group dwarfs (McConnachie 2012), ultra-diffuse galaxies from the Coma and Virgo clusters (Mihos et al. 2015; van Dokkum et al. 2015), and giant ellipticals from Fornax, Virgo and the Carnegie-Irvine catalog (Ferrarese et al. 2006; Ho et al. 2011). The NGFS dwarf sample by itself covers a range in absolute magnitude of −18.80 ≤ M_i' ≤ −8.78 mag and effective radii r_eff,i' = 0.11–2.72 kpc, comprising an effective surface brightness from $\langle {\mu }_{i^{\prime} }{\rangle }_{e}=20\mbox{--}28$ mag arcsec⁻². The sequence of giant ellipticals stretching from the upper-to-center-right of the diagram is connected to that of the NGFS dwarfs and dSphs (center to lower-left) by an intermediate bridge of galaxies. The bridge spans the −20 < M_i' < −15 mag, 0.6 ≤ r_eff,i'/kpc ≤ 2 parameter space and blends the brightest NGFS dwarfs with the faint regime of ETGs. We note here that the dwarfs in this group consist primarily of nucleated candidates such that 44/63 of these dwarfs show clear nuclei.

Zoom In Zoom Out Reset image size

UDGs seem to follow their own sequence, roughly along constant effective surface brightness, avoiding the bridge between dwarf and giant galaxies. Although UDGs seem to have similar magnitudes to the brightest dwarf galaxies, they are much more extended. UDGs have been detected in multiple environments (e.g., Koda et al. 2015; Mihos et al. 2015; Muñoz et al. 2015; van Dokkum et al. 2015; Janssens et al. 2017; Lee et al. 2017; Venhola et al. 2017). They show signatures of massive DM halos (e.g., Beasley et al. 2016; van Dokkum et al. 2016), and their population size scales with the mass of the central halo (van der Burg et al. 2016, 2017; Janssens et al. 2017). Together with predictions from theoretical studies, this suggests that UDGs are a consequence of rapidly spinning, massive halos ($\gtrsim {10}^{10}\,{M}_{\odot }$) that recently fell into denser environments (e.g., Rong et al. 2017; Amorisco 2018). In our NGFS footprint, six UDG candidates are found in the central region and one in the outer region. Their magnitudes and effective radii are in the range $-15.62\leqslant {\text{}}{M}_{i^{\prime} }\leqslant -13.85$ mag and r_eff,i' = 1.79 − 2.72 kpc, respectively. Of the seven candidates, two harbor a nuclear star cluster. Their properties make them very similar to UDGs found in other galaxy cluster environments, such as Coma and Virgo (Mihos et al. 2015; van Dokkum et al. 2015). A curious Local Group counterpart is the currently disrupting Sagittarius dwarf galaxy (M_V = −13.5 mag and r_eff = 2.6 kpc, McConnachie 2012) with its nucleus and the central star cluster M54 (Bellazzini et al. 2008; Mucciarelli et al. 2017).

Figure 2 shows that the nucleation fraction of the NGFS dwarf sample decreases strongly with luminosity. The overall nucleation fraction is 28% for the entire luminosity range of the NGFS dwarfs. Nonetheless, the nucleation fraction reaches ∼90% at bright magnitudes, i.e., M_i' ≲ −17 mag and drops to zero at the faintest galaxy luminosities, i.e., M_i' ≳ −9.56 mag. This limit marks the currently faintest nucleated galaxy in the NGFS dwarf galaxy sample (see also Muñoz et al. 2015; Ordenes-Briceño et al. 2018).

4.2. Spatial Distribution

Figure 3 shows the spatial distribution of the dwarf galaxy candidates in the NGFS survey region, with dashed black circles indicating NGC 1399-centric radii of r_vir/4 and r_vir/2 that correspond to ∼350 and 700 kpc at the distance of Fornax. The dwarfs and giant galaxies are distributed throughout the field with the projected dwarf galaxy surface number density profile, ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$, shown by the color shading, computed along a 15 × 15 bins grid—corresponding to physical bin sizes of 152 × 146 kpc²—and show the resulting 2D histogram smoothed with Lanczos interpolation. We also estimate ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$ by a nonparametric kernel density estimate using a Gaussian kernel of 025 bandwidth and show resulting curves of iso-density contours by white-scaled thin solid lines. We point out that variations in surface brightness limits due to bright galaxy haloes do not affect the results, as the typical size of a galaxy halo is negligible compared to the area studied here and the structures found therein.

Zoom In Zoom Out Reset image size

Both ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$ estimates show a general concentration of dwarfs in the core regions of Fornax within r ≲ 350 kpc. NGC 1399 itself can be seen to occupy an apparent saddle-point between two main dwarf galaxy over-densities toward the East and West (see Figures 1 and 3, and also Muñoz et al. 2015). The projected distribution of dwarfs in the western over-density generally follows that of the giant galaxies, which may suggest a physical association. While the current data cannot confirm such a connection, we note the contrast with the ∼200 kpc scale over-density to the east. This group shows a more regular morphology and lies in between to the projected positions of only two bright galaxies, complicating the notion of physical origins with nearby giant hosts. Weaker density contrasts are found for a third "orphan" group of dwarf candidates located near (α, δ) ≈ (03^h33^m, −3575).

The top two left-hand panels of Figure 4 show the ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$ profile calculated with respect to NGC 1399. To guard against potential biases introduced in choosing arbitrary bin sizes, we resample the dwarf galaxies using both a constant and an adaptive binning strategy. For the former, we choose constant bins between 5' and 25', in steps of 1'. At each step, we calculate ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$ in each of the corresponding annuli and show the aggregated data as dots. We have done this procedure for nonnucleated and nucleated dwarfs and the entire dwarf sample. Similarly, the middle panel of Figure 4 shows the data based on adaptive bin sizes with the three samples described above. Here, bins are chosen such that each of them contains exactly N dwarf candidates with $5\leqslant N\leqslant 25$. The constant binning produces a much smoother variation in ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$, while the adaptive binning shows a sensitivity to local dwarf over-densities resulting in artificially small annuli and corresponding spikes in the ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$ profile. We approximate the dwarf ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$ profiles by a power-law model, following other radially dependent projected densities (e.g., ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$ ∝ exp(A R^α); Sérsic 1963; Einasto 1965) by Markov-Chain Monte Carlo (MCMC) sampling. We use MCMC with a normal prior to sample the posterior probability of (A, α) 10⁵ times, which allows sufficient burn-in to skip over stochastic or unreliable results from early iterations so that each chain converges to consistent, well-defined peaks in the marginalized parameter estimate distributions. The resulting Σ_N(r) estimates are shown in the two top left panels of Figure 4 as solid curves. The text displays the means of the posterior probability density functions for α alongside the corresponding 95% confidence limits. The projected radial surface number density for the nonnucleated dwarfs has a flat distribution up to ∼350 kpc and slowly declines beyond that radial distance. On the other hand, nucleated dwarfs have a steeper Σ_N(r) distribution than nonnucleated dwarfs, being more concentrated in the inner regions near the cD galaxy NGC 1399 and decreasing rapidly outwards. This is similar to the results of Lisker et al. (2007) who found, based on SDSS photometry (M_B ≲ − 13 mag), that the bright nucleated dwarf galaxy population in the Virgo galaxy cluster is more centrally concentrated than the nonnucleated dwarf population.

Zoom In Zoom Out Reset image size

The bottom left-panel of Figure 4 shows the results of a two-point angular correlation function ($\hat{w}(\theta );$ Landy & Szalay 1993) analysis of the dwarf galaxy candidates, using the same three samples (nonnucleated, nucleated, and all dwarfs). The two-point angular correlation function quantifies the excess probability of finding galaxy pairs at a given angular separation over a random distribution without a restriction to Gaussianity (e.g., Connolly et al. 2002; Sato et al. 2009) and is typically used to constrain cosmological parameters and structure formation models (e.g., Bernardeau et al. 2002; Cooray & Sheth 2002; Tegmark et al. 2002, 2004; Dolney et al. 2006). Noting the steepening of the ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$ profile at ∼350 kpc and the Σ_N(r) distribution, we split the total sample into those dwarf galaxies inside and outside of this radius and estimate $\hat{w}(\theta )$ for each subpopulation. The solid lines along with ±1σ bounds show the results for the three samples, in addition to all outer and all inner dwarfs, with correlation lengths θ binned in steps of 10' for the total, nonnucleated, and nucleated samples, and 5' for the two subsamples. Estimating $\hat{w}(\theta )$ using different bin sizes reveals mild deviations but with the overall behaviors unchanged.

Given that the likelihood of finding two points separated by an angular distance θ compared to a purely uniform distribution is encapsulated by $\hat{w}(\theta )$, we find particularly strong evidence for dwarf clustering on sub-100 kpc scales for the overall and outer dwarf population. In particular, the outer dwarf galaxies appear more likely to be clustering on scales approaching ∼50 kpc with a notable decrease on scales ≳100 kpc. Overall, the apparent smaller clustering scale appears superimposed on the ∼350 kpc scale over-density shown in Figure 3. This larger clustering scale is reflected by the almost flat ${{\rm{\Sigma }}}_{N}(\alpha ,\delta )$ profile within ∼350 kpc, outside of which a general steepening of the profile is seen, modulo local, projected dwarf clustering and sub-groups. Conversely, the dwarf population inside of r ≈ 350 kpc does not appear to show evidence as strong for clustering at any scale, but we note that the limited spatial region will tend to mute $\hat{w}(\theta )$. In any case, we check against an underlying uniform distribution of dwarf galaxies by creating a large artificial set of three-dimensional dwarf galaxy coordinates, uniformly distributed within a 1 Mpc radius sphere centered on NGC 1399. We limit this population to those lying within the coordinate limits of our observed sample and extract the 2D projected distances from NGC 1399. Two-sample Kolmogorov–Smirnov tests comparing the simulated NGC 1399-centric projected separations and position angles to the new dwarfs rule out a flat surface density distribution of dwarf galaxies within the NGFS field of view at a very high confidence level (i.e., p = 0.004 for separations and p ≪ 0.001 for galaxy position angles).

The right-hand panel of Figure 4 shows an alternate view of the spatial distribution. Here, we show the projected number density as a function of azimuthal angle (Φ; degrees east of north) and radial distance from NGC 1399. Symbols are the same as in Figure 3 with dashed lines indicating r_vir/2 and r_vir/4, which corresponds to 350 and 700 kpc from NGC 1399. We apply $({\rm{\Delta }}{\rm{\Phi }}=36^\circ ,{\rm{\Delta }}{r}_{\mathrm{NGC}1399})$ binning and smooth the 2D histogram with Lanczos interpolation, which serves to highlight dwarf groupings aligning along "lines-of-sight" to NGC 1399, which do not appear as obvious as in the projected (α, δ) space (see Figure 3). Nevertheless, similar ≲100 kpc scale over-densities are apparent at all radii, further supporting the nonuniformity of the Fornax dwarf galaxy population, in particular the E–W bimodality shown near r_{NGC 1399} = r_vir/4 ≈ 350 kpc. Spectroscopic observations are required to assess the phase-space coherence of the found dwarf galaxy over-densities.

Taken together, the evidence of clustering at ≲100 kpc scales within the central cluster–centric radius of ∼1 Mpc of Fornax broadly concurs with the growing observational evidence for the common occurrence, and importance, of dwarf galaxy pairs and groups in low-mass galaxy evolution and transformation in the local universe (e.g., Martínez-Delgado et al. 2012; Annibali et al. 2016; Ordenes-Briceño et al. 2016; Stierwalt et al. 2017). Theoretical considerations predict that close associations between dwarf galaxies should be common in that as many as 50% of ${10}^{6}\,{M}_{\odot }$ scale dwarf galaxies might be expected to have companions within ∼50 kpc (Wetzel et al. 2015; Wheeler et al. 2015). Given the isolated natures of many of the recently discovered dwarf pairs/groups, combined with the locations of other purported dwarf groups ranging out to ∼100 kpc from giant galaxy hosts, the current findings may indicate that at least some of these groups are truly interacting in the halos of Fornax giant galaxy members or that they may have origins in the primordial universe and are falling into the Fornax cluster environment for the first time.

Y.O.-B. acknowledges financial support via CONICYT-Chile (grant CONICYT-PCHA/Doctorado Nacional/2014-21140651) and the DAAD through the PUC-HD Graduate Exchange Fellowship. P.E. acknowledges support from the Chinese Academy of Sciences (CAS) through CAS-CONICYT Postdoctoral Fellowship CAS150023 administered by the CAS South America Center for Astronomy (CASSACA) in Santiago, Chile. M.A.T. is supported by the Gemini Observatory, which is operated by the Association of Universities for Research in Astronomy, Inc., on behalf of the international Gemini partnership of Argentina, Brazil, Canada, Chile, and the United States of America. T.H.P. acknowledges the support through the FONDECYT Regular Project No. 1161817 and the BASAL Center for Astrophysics and Associated Technologies (PFB-06). T.H.P. and A.L. gratefully acknowledge ECOS-Sud/CONICYT project C15U02. This project used data obtained with the Dark Energy Camera (DECam), which was constructed by the Dark Energy Survey (DES) collaboration.

Facility: CTIO:Blanco/DECam. -

Software: astropy (Astropy Collaboration et al. 2013), matplotlib (Hunter 2007), scikit-learn, (Pedregosa et al. 2012), astroML, (VanderPlas et al. 2012) SCAMP (Bertin et al. 2002), SWARP (Bertin 2006), Source Extractor (Bertin & Arnouts 1996), Galfit (Peng et al. 2002).