A MeerKAT view of pre-processing in the Fornax A group
D. Kleiner, P. Serra, F. M. Maccagni, A. Venhola, K. Morokuma-Matsui, R. Peletier, E. Iodice, M. A. Raj, W. J. G. de Blok, A. Comrie, G. I. G. Józsa, P. Kamphuis, A. Loni, S. I. Loubser, D. Cs. Molnár, S. S. Passmoor, M. Ramatsoku, A. Sivitilli, O. Smirnov, K. Thorat, F. Vitello
AAstronomy & Astrophysics manuscript no. MeerKAT_FornaxA © ESO 2021February 17, 2021
A MeerKAT view of pre-processing in the Fornax A group
D. Kleiner , P. Serra , F. M. Maccagni , A. Venhola , K. Morokuma-Matsui , R. Peletier , E. Iodice , M. A. Raj ,W. J. G. de Blok , , , A. Comrie , G. I. G. Józsa , , , P. Kamphuis , A. Loni , , S. I. Loubser , D. Cs. Molnár ,S. S. Passmoor , M. Ramatsoku , , A. Sivitilli , O. Smirnov , , K. Thorat , and F. Vitello INAF – Osservatorio Astronomico di Cagliari, Via della Scienza 5, 09047 Selargius, CA, Italye-mail: [email protected] University of Oulu, Space physics and astronomy unit, Pentti Kaiteran katu 1, 90014, Oulu, Finland Institute of Astronomy, Graduate School of Science, The University of Tokyo, 2–21–1 Osawa, Mitaka, Tokyo 181–0015, Japan Kapteyn Astronomical Institute, University of Groningen,PO Box 800, 9700 AV Groningen, The Netherlands INAF – Astronomical observatory of Capodimonte, Via Moiariello 16, Naples 80131, Italy Netherlands Institute for Radio Astronomy (ASTRON), Oude Hoogeveensedijk 4, 7991 PD Dwingeloo, the Netherlands Deptarment of Astronomy, Univ. of Cape Town, Private Bag X3, Rondebosch 7701, South Africa Inter-University Institute for Data Intensive Astronomy, University of Cape Town, Cape Town, Western Cape, 7700, South Africa South African Radio Astronomy Observatory, 2 Fir Street, Black River Park, Observatory, Cape Town, 7925, South Africa Department of Physics and Electronics, Rhodes University, PO Box 94, Makhanda, 6140, South Africa Argelander-Institut für Astronomie, Auf dem Hügel 71, D-53121 Bonn, Germany Ruhr University Bochum, Faculty of Physics and Astronomy, Astronomical Institute, 44780 Bochum, Germany Dipartimento di Fisica, Università di Cagliari, Cittadella Universitaria, 09042 Monserrato, Italy Centre for Space Research, North-West University, Potchefstroom 2520, South Africa Department of Physics, University of Pretoria, Private Bag X20, Hatfield 0028, South Africa INAF – Istituto di Radioastronomia, via Gobetti 101, I-40129 Bologna, ItalyReceived 13 November, 2020; accepted 25 January, 2021
ABSTRACT
We present MeerKAT neutral hydrogen (H i ) observations of the Fornax A group, which is likely falling into the Fornax cluster forthe first time. Our H i image is sensitive to 1.4 × atoms cm − over 44.1 km s − , where we detect H i in 10 galaxies and a totalof (1.12 ± × M (cid:12) of H i in the intra-group medium (IGM). We search for signs of pre-processing in the 12 group galaxieswith confirmed optical redshifts that reside within the sensitivity limit of our H i image. There are 9 galaxies that show evidence ofpre-processing and we classify each galaxy into their respective pre-processing category, according to their H i morphology and gas(atomic and molecular) scaling relations. Galaxies that have not yet experienced pre-processing have extended H i discs and a highH i content with a H -to-H i ratio that is an order of magnitude lower than the median for their stellar mass. Galaxies that are currentlybeing pre-processed display H i tails, truncated H i discs with typical gas fractions, and H -to-H i ratios. Galaxies in the advancedstages of pre-processing are the most H i deficient. If there is any H i , they have lost their outer H i disc and e ffi ciently convertedtheir H i to H , resulting in H -to-H i ratios that are an order of magnitude higher than the median for their stellar mass. The central,massive galaxy in our group (NGC 1316) underwent a 10:1 merger ∼ × M (cid:12) of H i , which wedetect as clouds and streams in the IGM, some of which form coherent structures up to ∼
220 kpc in length. We also detect giant( ∼
100 kpc) ionised hydrogen (H α ) filaments in the IGM, likely from cool gas being removed (and subsequently ionised) from anin-falling satellite. The H α filaments are situated within the hot halo of NGC 1316 and there are localised regions that contain H i . Wespeculate that the H α and multiphase gas is supported by magnetic pressure (possibly assisted by the NGC 1316 AGN), such that thehot gas can condense and form H i that survives in the hot halo for cosmological timescales. Key words.
Galaxies: groups: general – galaxies: groups: individual: Fornax A – galaxies: evolution – galaxies: interactions –galaxies: ISM – radio lines: galaxies
1. Introduction
Our current understanding of galaxy formation and evolutionis that secular processes and galaxy environment fundamentallyshape the properties of galaxies (e.g. Baldry et al. 2004; Baloghet al. 2004; Bell et al. 2004; Peng et al. 2010; Driver et al. 2011;Schawinski et al. 2014; Davies et al. 2019). In the local Uni-verse (z ∼
0) up to ∼
50% of galaxies reside in groups (Eke et al.2004; Robotham et al. 2011), making it essential to understandthe group environment in the context of galaxy evolution. While there is no precise definition of a galaxy group, it gen-erally contains 3 – 10 galaxies in a dark matter (DM) halo of10 – 10 M (cid:12) (e.g. Catinella et al. 2013). As the galaxy num-ber density and DM halo mass of groups span a wide range, thereis no dominant transformation mechanism that galaxies are sub-jected to, but rather multiple secular and external mechanismsworking together. The properties of group galaxies appear tocorrelate with group halo mass and virial radius, implying thatquenching paths in groups are di ff erent from those in clusters(Weinmann et al. 2006; Haines et al. 2007; Wetzel et al. 2012;Woo et al. 2013; Haines et al. 2015). Article number, page 1 of 18 a r X i v : . [ a s t r o - ph . GA ] F e b & A proofs: manuscript no. MeerKAT_FornaxA
As galaxies fall towards clusters, there is su ffi cient time forexternal (i.e. environmentally driven, such as tidal and hydro-dynamical) mechanisms to transform and even quench the galax-ies, prior to reaching the cluster (e.g. Porter et al. 2008; Haineset al. 2013, 2015; Bianconi et al. 2018; Fossati et al. 2019;Seth & Raychaudhury 2020). This is called “pre-processing”and refers to the accelerated, non-secular evolution of galaxiesthat occurs prior to entering a cluster. As pre-processing requiresexternal mechanisms to transform the galaxies, this evolutioncommonly occurs in groups, where it is generally thought thatgroup galaxies follow a di ff erent evolutionary path compared togalaxies of the same mass in the field (e.g. Fujita 2004; Mahajan2013; Roberts & Parker 2017; Cluver et al. 2020). In particular,pre-processing is likely to be most e ffi cient in massive ( > . M (cid:12) ) galaxies residing in massive (10 – 10 M (cid:12) ) groups (Don-nari et al. 2020). It has also been shown that pre-processing isresponsible for the decrease in star formation activity for late-type galaxies at distances between 1 and 3 cluster virial radii(e.g Lewis et al. 2002; Gómez et al. 2003; Verdugo et al. 2008;Mahajan et al. 2012; Haines et al. 2015).Neutral hydrogen in the atomic form (H i ) is ideal for tracingtidal and hydro-dynamical processes in galaxies and the intra-group medium (IGM). H i is the main component of the inter-stellar medium (ISM) and can show the e ff ects of ram pressure,viscous and turbulent stripping, thermal heating (e.g. Cowie &McKee 1977; Nulsen 1982; Chung et al. 2007; Rasmussen et al.2008; Chung et al. 2009; Steinhauser et al. 2016; Ramatsokuet al. 2020), and moderate and strong tidal interactions (e.g. Ko-ribalski 2012; de Blok et al. 2018; Kleiner et al. 2019), long be-fore these mechanism can be identified in the stars.In this paper we present a detailed analysis of the For-nax A galaxy group based on H i and ancillary observations.The Fornax A group is an excellent candidate to search for pre-processing signatures as it is likely in-falling into the (low mass– 5 × M (cid:12) ) Fornax cluster (Drinkwater et al. 2001) for thefirst time. The group galaxies span a variety of stellar masses andmorphological types, implying that tidal and hydro-dynamicalinteractions are likely to a ff ect the galaxies gas and stellar con-tent (Raj et al. 2020).Using Meer Karoo Array Telescope (MeerKAT) H i observa-tions, deep optical imaging from the Fornax Deep Survey (FDS:Iodice et al. 2016, 2017; Venhola et al. 2018, 2019; Raj et al.2019, 2020), wide-field H α imaging from the VLT Survey Tele-scope (VST) and molecular gas observations from the AtacamaLarge Millimetre Array (ALMA), we identify galaxies at di ff er-ent stages of pre-processing following various types of interac-tions.This paper is organised as follows: Section 2 describes theFornax A group. Section 3 describes the H i and H α observationsand the data reduction process used to produce our images. Wepresent the results of our H i measurements, H i images, and therelation to stellar and H α emission in Section 4. In Section 5we present the atomic-to-molecular gas ratios and discuss theevidence and timescale of pre-processing in the group. Finally,we summarise our results in Section 6. Throughout this paperwe assume a luminosity distance of 20 Mpc to the most massivegalaxy (NGC 1316) in the Fornax A group (Cantiello et al. 2013;Hatt et al. 2018) and assume all objects in the group are at thesame distance. At this distance, 1 (cid:48) corresponds to 5.8 kpc.
2. The Fornax A group
The Fornax A galaxy group is the brightest group in the Fornaxvolume. It is located on the cluster outskirts at a projected dis- tance of ∼ ∼ × the Fornax cluster virialradius) from the cluster centre and has a mass of 1.6 × M (cid:12) ,which is of the same order of magnitude as the Fornax cluster(M vir ∼ × M (cid:12) ) itself (Maddox et al. 2019). Within thevirial radius of the group ( ∼ × M (cid:12) (Iodice et al.2017). NGC 1316 is a giant radio galaxy (Ekers et al. 1983; Fo-malont et al. 1989; McKinley et al. 2015; Maccagni et al. 2020),known merger remnant, and the brightest galaxy in the Fornaxcluster volume (even brighter than the brightest cluster galaxyNGC 1399). There are a number of extended stellar loops andstreams in NGC 1316 that are a result of a 10:1 merger that oc-curred 1 – 3 Gyr ago, between a massive early-type galaxy anda gas-rich, late-type galaxy (Schweizer 1980; Mackie & Fab-biano 1998; Goudfrooij et al. 2001; Iodice et al. 2017; Serra et al.2019). The majority of the remaining bright ( m B <
16) galaxiesare late types that have stellar mass ranges of 8 < log(M (cid:63) / M (cid:12) ) < i in the Fornax A group. Horellou et al. (2001) and Serraet al. (2019) imaged the central region of the Fornax A groupin H i , where the more recent image of Serra et al. (2019) de-tected NGC 1316, NGC 1317, NGC 1310, and ESO 301-IG 11,along with four clouds at the outskirts of NGC 1316 (EELR,SH2, C N , , and C N , ), and two tails (T N and T S ). The re-maining six galaxies, which have previously been detected, areNGC 1326, NGC 1326A ,and NGC 1326B in the H i Parkes AllSky Survey (HIPASS; Meyer et al. 2004; Koribalski et al. 2004),NGC 1316C with the Nançay telescope (Theureau et al. 1998),FCC 35 with the Australian Telescope Compact Array (ATCA)and the Green Bank Telescope (Putman et al. 1998; Courtois &Tully 2015), and FCC 46 with the ATCA (De Rijcke et al. 2013).Within NGC 1316, H i has been resolved in the centre and corre-lates with massive amounts of molecular gas (Morokuma-Matsuiet al. 2019; Serra et al. 2019). H i has also been detected in theouter stellar halo, within the regions defined by the H α extendedemission line region (EELR; originally discovered by Mackie& Fabbiano 1998), in the southern star cluster complex (SH2;Horellou et al. 2001) and in two northern clouds (C N , and C N , )(Serra et al. 2019). Lastly, ∼ × M (cid:12) of H i was detectedin the IGM, defined as the northern and southern tails (T N andT S ). The tails are ejected H i gas from the NGC 1316 merger andextend up to 150 kpc from the galaxy centre (Serra et al. 2019).The Fornax A group is an ideal system to search for pre-processing. Evidence suggests that the group is in the early stageof assembly (Iodice et al. 2017; Raj et al. 2020) and is locatedat the cluster infall distance where pre-processing is thought tooccur (Lewis et al. 2002; Gómez et al. 2003; Verdugo et al.2008; Mahajan et al. 2012; Haines et al. 2015). The BGG ismassive enough to experience e ffi cient pre-processing (Donnariet al. 2020) and Raj et al. (2020) show that there are signaturesof pre-processing in the group; six of the nine late types have anup-bending (type III) break in their radial light profile. This in-dicates that the star formation may be halting in the outer disc ofgalaxies, although, what is driving the decline in star formationis not yet clear. Article number, page 2 of 18leiner et al.: Pre-processing in the Fornax A group
3. Observations and data reduction
Commissioned in July 2018, MeerKAT is a new radio interfer-ometer and a precursor for the Square Kilometre Array SKA1-MID telescope (Jonas 2016; Mauch et al. 2020). MeerKAT isdesigned to produce highly sensitive radio continuum and H i images with good spatial and spectral resolution in a relativelyshort amount of observing time. The MeerKAT Fornax Survey(MFS; PI: P.Serra) is one of the designated Large Survey Projects(LSPs) of the MeerKAT telescope. The MFS will observe theFornax galaxy cluster in H i over a wide range of environmentdensities, down to a column density of a few × atoms cm − at a resolution of 1 kpc, equivalent to a H i mass limit of 5 × M (cid:12) (Serra et al. 2016).The Fornax A group was observed with MeerKAT in two dif-ferent commissioning observations in June 2018, which di ff er bythe number of antennas (36 and 62, respectively) connected tothe correlator. We present the details of these observations and ofthe H i cube in Table 1. The MeerKAT baselines range between29 m and 7.7 km and for both these observations, the SKARABcorrelator in the 4k mode was used, which consists of 4096 chan-nels in full polarisation in the frequency range 856-1712 MHzwith a resolution of 209 kHz (equivalent to 44.1 km s − for H i atthe distance of the Fornax cluster).The first observation (referred to as Mk-36) used 36 antennasand observed the target for a total of 8 h. Results from this ob-servation are presented both in radio continuum (Maccagni et al.2020) and in H i (Serra et al. 2019); these papers provide a de-tailed description of the data reduction process. In this work, weuse the Mk-36 calibrated measurement set in combination withthat from the second observation (detailed below).The second observation (Mk-62) used 62 antennas and ob-served the target for a total of 7 h. PKS 1934-638 and PKS 0032-403 were observed, where the former was observed for 20 minand used as the bandpass and flux calibrator while the latter wasobserved for 2 min every 10 min and used as the gain calibrator.We used the Containerised Automated Radio Astronomi-cal Calibration ( CARACal ; Józsa et al. 2020) pipeline to re-duce the MeerKAT observations. The pipeline uses Stimela ,which containerises di ff erent open-source radio interferometrysoftware in a Python framework. This makes the pipeline bothflexible and highly customisable and has been used to reduceMeerKAT and other (e.g. Jansky Very Large Array) interfero-metric observations (e.g. see Serra et al. 2019; Maccagni et al.2020; Ramatsoku et al. 2020; Ianjamasimanana et al. 2020).We used
CARACal to reduce the Mk-62 observation end-to-end and include the already reduced Mk-36 observation (Serraet al. 2019; Maccagni et al. 2020) at the spectral line imag-ing step. For the Mk-62 observation, we used 120 (1330 -1450) MHz of bandwidth to ensure adequate continuum imag-ing and calibration. We used 18 (1402 - 1420) MHz, which easilycovers the group volume, for the (joint) spectral line imaging.Our choice of data reduction techniques and steps is outlinedusing
CARACal as follows: First, we flag the radio frequency in-terference (RFI) in the calibrators data based on the Stokes Qvisibilites using
AOflagger (O ff ringa et al. 2012). Then, a time-independent, antenna-based, complex gain solution was derivedfor the bandpass using CASA bandpass and the flux scale wasdetermined with
CASA gaincal . A Frequency-independent,time-dependent, antenna-based complex gains were determined https://caracal.readthedocs.io https://github.com/SpheMakh/Stimela using CASA gaincal . The gain amplitudes were scaled to boot-strap the flux scale with
CASA fluxscale , and the bandpass andcomplex gain solutions were applied to the target visibilities us-ing
CASA applycal . The RFI in the target data was then flaggedbased on the Stokes Q visibilites, using
AOflagger (O ff ringaet al. 2012). We imaged and self-calibrated the continuum emis-sion of the target with WSclean (O ff ringa et al. 2014; O ff ringa &Smirnov 2017) and CUBICAL (Kenyon et al. 2018), respectively.This process was repeated two more times, in which each self-calibration iteration was frequency-independent and solved onlyfor the gain phase, with a solution interval of 2 min. The finalcontinuum model was subtracted from the visibilities using
CASAmsutils . The visibilities from both the Mk-36 and Mk-62 cal-ibrated measurement sets were then Doppler corrected into thebarycentric rest frame using
CASA mstransform . Residual con-tinuum emission in the combined measurement set was removedby fitting and subtracting a second order polynomial to thereal and imaginary visibility spectra with
CASA mstransform .Then, we created a H i cube by imaging the H i emission with WSclean (O ff ringa et al. 2014; O ff ringa & Smirnov 2017) andmade a 3D mask through source finding with the Source FindinaApplication ( SoFiA ; Serra et al. 2015). This was then used as aclean mask to image a new H i cube with higher image fidelity.Finally, we applied the primary beam correction of Mauch et al.(2020) down to a level of 2%, which corrects for the sensitivityresponse pattern of MeerKAT.Our H i cube was imaged using an 18 MHz sub-band (cen-tred on NGC 1316) and the basic properties are presented inTable 1. The root mean square (RMS) noise is 90 µ Jy beam − ,which equates to a 3 σ H i column density of 1.4 × atomscm − over a single channel of 44.1 km s − at the angular res-olution of 33.0 (cid:48)(cid:48) × (cid:48)(cid:48) . Compared to Serra et al. (2019), wepresent an image that is approximately twice as large and morethan twice as sensitive and has comparable spatial and velocityresolutions.We searched for H i sources using SoFiA outside the
CARACal pipeline. To ensure that we properly captured H i emis-sion that is di ff use or far from the pointing centre, we tested dif-ferent combinations of smoothing kernels and detection thresh-olds in the SoFiA smooth + clip algorithm, per-source in-tegrated signal-to-noise ratio (S / N) thresholds, and reliabilitythresholds. Pixels in the H i cube are detected if their value isabove a smooth + clip detection threshold of 3.5 (in abso-lute value and relative to the cube noise) for spatial smooth-ing kernels equal to 1, 2, and 3 times the synthesised beam incombination with velocity smoothing kernels over a single (i.e.no smoothing) and three channels. The mean, sum, and maxi-mum pixel value of each detected source (normalised to the localnoise) create a parameter space that can separate real H i emis-sion from noise peaks (Fig. 1; Serra et al. 2012). The reliabilityof each source (defined as the local ratio of positive-to-negativesource density within this 3D parameter space) as well as the in-tegrated S / N are then used to identify statistically significant, realH i sources. Our catalogue was created by retaining only sourceswith an integrated S / N above 4 and a reliability above 0.65. Asshown in Fig. 1, this selection is purposefully designed to be con-servative, ensuring that detected di ff use H i emission (i.e clouds The deep H i imaging revealed periodic, artefacts caused by the cor-relator during this time of commissioning. The artefacts were apparentat the sky position of bright continuum emission. We were able to re-move the artefacts by excluding baselines less than 50 m in the cube and85 m for the single, worst channel. While short baselines are essentialfor di ff use emission, this equates to 5 and 22 baselines out of 1891.Article number, page 3 of 18 & A proofs: manuscript no. MeerKAT_FornaxA
Table 1.
Observation and H i cube properties. The measurements of the H i cube RMS noise and column density (over a single channel of44.1 km s − ) were taken in the pointing centre and restoring beam was taken from the centre channel. Property Mk-36 observation Mk-62 observationDate 2 June 2018 16 June 2018ID 20180601-0009 20180615-0039Time on target 8 hr 7 hrNumber of antennas 36 62Pointing centre (J2000) 03h 22m 41.7s, -37d 12 (cid:48) (cid:48)(cid:48)
Available bandwidth 856 - 1712 MHzH i cube frequency range 1402 - 1420 MHzH i cube spectral resolution 209 kHz (44.1 km s − at z ∼ i cube pixel size 6.5 (cid:48)(cid:48) H i cube weight robust = (cid:48)(cid:48) taperH i cube RMS noise 90 µ Jy beam − H i cube restoring beam 33.0 (cid:48)(cid:48) × (cid:48)(cid:48) σ H i column density 1.4 × atoms cm − in the IGM) is clearly real emission and does not include noisepeaks.However, we found some real H i emission below thesethresholds that should be included in the detection mask. Wethus operated on the detection mask using the virtual reality(VR) software iDaVIE-v (Sivitilli et al. in press) from the In-stitute for Data Intensive Astronomy (IDIA) Visualisation Lab(Marchetti et al. 2020; Jarrett et al. 2020). This allowed us to usea ray marching renderer (Comrie et al. in prep) to view and in-teract with our H i cube, while making adjustments to the maskwithin a 3D digital immersive setting. We were able to inspectthe mask for any spurious H i emission that was included or iden-tify real H i emission that was missed. This was accomplished byimporting the detection mask from SoFiA , overlaying it with theH i cube in the VR environment, and then adjusting the mask us-ing the motion-tracking hand controllers. As part of this process,we added two sources to the detection mask within the VR envi-ronment by marking zones where emission was clearly present.The two sources added in VR were originally excluded fromthe detection mask because they are below the reliability thresh-old of 0.65 (but above the integrated S / N threshold of 4). Thesesources are deemed real because they either coincide with emis-sion at other wavelengths (see below) or are part of large, coher-ent H i emission. Following these edits to the detection mask inVR, we created H i intensity and velocity maps that are presentedin the next section. α observation To generate the H α -emission images, we used a combinationof H α narrow-band images and r (cid:48) broad-band images both col-lected using the OmegaCAM attached to the VST at CerroParanal, Chile (PID: 0102.B-0780(A)). The OmegaCAM is a 32CCD wide-field camera with a 1deg × (cid:48)(cid:48) . We used the NB 659 H α filter with 10 nmthroughput, bought by Janet Drew for the VST Photometric H α Survey (VPHAS; Drew et al. 2014). The imaging was done us-ing large ≈ r (cid:48) and H α bands, respectively. This strategy allows us to makeaccurate sky background removal by subtracting averaged back-ground models from the science exposures, and it also reducesthe amount of imaging artefacts (such as satellite tracks) in thefinal mosaics because those are averaged out when the imagesare stacked. The total exposure times in the r (cid:48) -band and H α -band l o g ( s u m / n o i s e ) PositiveNegativeCloudsIntegrated SNR Threshold
Fig. 1.
Sum of the pixel values as a function of the mean pixel valuefor all sources detected with
SoFiA . The blue points indicate the posi-tive detections and the red points indicate the negative detections (Serraet al. 2012). Detected H i clouds are shown as black crosses. The dot-ted line shows the per-source integrated S / N of 4. Only positive sourcesabove this threshold and with a reliability > / N of 4 is a conservative threshold asit is closer to area of parameter space occupied by the most statisticallysignificant detections (i.e. the positive sources with a high sum / noise fortheir mean / noise value) and is clearly above the edge of non-statisticallysignificant detections (i.e. where the density of positive sources is ap-proximately the same as the density of negative sources). Owing to thisconservative threshold, the detected H i clouds, while often di ff use, oc-cupy the parameter space of real, reliable H i emission. were 8 250s and 31 140s, respectively. Similar data reduction andcalibration was done for both r (cid:48) -band and H α -images. Details ofthe used reduction steps are given by Venhola et al. (2018).As the H α narrow-band images are sensitive both to H α emission and flux coming from the continuum, we needed tosubtract the continuum flux from the H α images before they canbe used for H α analysis. As the flux in the r (cid:48) band is dominatedby the continuum, we use scaled r (cid:48) -band images to subtract thecontinuum from the H α . The optimal scaling of the r (cid:48) -band im-age was selected by visually determining the scaling factor thatresults in a clean subtraction of the majority of stars and early-type galaxies. Article number, page 4 of 18leiner et al.: Pre-processing in the Fornax A group
However, there are some caveats in this procedure, whichleaves some systematic over- and under-subtraction in the H α images. If the seeing conditions or point spread functions (PSFs)di ff er between the broad- and narrow-band images there will besome residuals in the continuum subtracted image. In addition tothese residuals caused by the inner parts of the PSF ( (cid:46) (cid:48)(cid:48) ), alsothe extended outer parts (see Venhola et al. 2018) and reflectionrings of the PSF may leave some features in the images. In thecase of bright, extended, and peaked galaxies such as NGC 1316,these PSF features are also significant. As the positions of thereflection rings are dependent on the position of the source onthe focal plane they do not overlap precisely in the narrow- andbroad-band images and thus leave some systematic over- andunder- subtractions in the images. These kinds of features areapparent in the reduced H α emission images.The over- and under-subtraction artefacts dominate inand around objects with bright stellar emission. Therefore,NGC 1316 is significantly a ff ected to the extent that the artefactsobscure real H α emission. To rectify this, we select a sub-regionthat includes NGC 1316, NGC 1317, and NGC 1310 and createa model of the background that is ultimately subtracted from theoriginal image.The background model was created by masking the visible,real H α emission, and replaced with the background local me-dian. The masked image is then filtered with a median filter toeliminate sharp features in the image. Lastly, the (masked, fil-tered) background model is subtracted from the original image.We repeat this process using the residual image to create animproved mask, which is then subtracted from the original im-age. We use a conservative approach to mask the H α emission,as the aim is to remove the dominant artefacts and achieve auniform background throughout the image. We present a com-parison of the images and additional detail in Appendix A.
4. H i distribution in the group In Fig. 2, we present the primary beam-corrected H i columndensity map as detected by MeerKAT, overlaid on a gri stackedoptical image from the FDS (Iodice et al. 2016, 2017; Venholaet al. 2018). Our H i image (Fig. 2) is sensitive to a column den-sity of N H i = × atoms cm − in the most sensitive part(pointing centre), equating to a 3 σ H i mass lower detection limit1.7 × M (cid:12) for a point source 100 km s − wide.As a result of the improved sensitivity of our image, inH i we detect 10 galaxies out of the 13 spectroscopically con-firmed galaxies (Maddox et al. 2019), all the previously knownclouds and streams, and a new population of clouds and streamsin the IGM. Eleven of our H i detections (10 galaxies andSH2) have corresponding optical redshifts (Maddox et al. 2019).NGC 1341, FCC 19, and FCC 40 are the 3 galaxies with opticalredshifts in which we do not detect any H i . NGC 1341 is a late-type (SbcII) galaxy with a stellar mass of 5.5 × M (cid:12) (Raj et al.2020), in which H i has beeen previously detected (Courtois &Tully 2015). However, NGC 1341 is outside our H i image fieldof view and we do not include it in our sample. FCC 40 is a lowsurface brightness dwarf (dE4) elliptical (Iodice et al. 2017) andis unlikely to contain massive amounts of H i . It is also locatedin a region of the image in which the sensitivity is 75% worsethan the pointing centre, such that we do not detect H i below 5.6 × atoms cm − . FCC 19 is a dS0 with a stellar mass of 3.4 × M (cid:12) (Iodice et al. 2017; Liu et al. 2019). As it is near thepointing centre (70 kpc in projection from NGC 1316), we wouldexpect to detect H i if there were any. However, no H i is detectedin FCC 19 and we discuss the implications of this in section 5.2. We present the three-colour (constructed using the individual g , r , and i images) FDS (Iodice et al. 2016) optical image cutoutfor each group galaxy in our sample, which has been overlaidwith the H i contours at their respective column density sensitiv-ity (or upper limit) in Fig. 3. The integrated H i flux and massof the H i detections and the basic properties of the group galax-ies within the H i image field of view are presented in Table 2.The velocity field is presented in Fig. 4 and highlights some newlarge-scale coherent H i structures, which extend up to ∼
220 kpcin length. i Our H i image is the widest and deepest interferometric image ofthe Fornax A group to date. Naturally, we detect new H i sources,additional H i in known sources and resolved H i in previouslyunresolved sources. All the sources are presented in Table 2, Fig.2, and 4. As described in Section 2, several sources in the For-nax A group have been previously detected. The new H i sourcesdetected in this work are as follows: resolved H i tails associatedwith FCC 35, NGC 1310, and NGC 1326; an extension of T N inthe form of additional, coherent clouds; an additional componentto T S in the form of a western cloud; and a population of cloudsin the IGM (unlabelled in Fig. 4). i in galaxies We detect H i in ten galaxies, where the H i is well resolved ineight of them (Fig. 3). Out of those, two galaxies have H i thatis confined to the stellar disc, while the remaining 6 have H i emission that extend beyond the stellar disc. The two galaxieswith unresolved H i are NGC 1316C and FCC 46.The two well-resolved galaxies with H i confined within thestellar discs are NGC 1316 and NGC 1317 (Fig. 3). We detected6.8 × M (cid:12) of H i in the centre of NGC 1316, 60% more H i in the centre than previously detected in (Serra et al. 2019). TheH i has complex kinematics (also seen in the molecular gas anddust) beyond a uniformly rotating disc. The H i in NGC 1317 issharply truncated at the boundary of the stellar disc. Given itsstellar mass and morphology, NGC 1317 is H i deficient by atleast an order of magnitude (discussed in detail in section 5.2).There are six galaxies in the group that have extended H i discs. Three galaxies (NGC 1326A / B and ESO 301-IG 11) haveslightly extended and mostly symmetric H i discs, while the otherthree galaxies (FCC 35, NGC 1326, and NGC 1310) have ex-tended H i features that are significantly disturbed and asymmet-ric (Fig. 3).NGC 1326A and B have extended H i discs and although theyoverlap in projection, they are separated by ∼
800 km s − in ve-locity. There is no H i connecting these two galaxies along theline of sight down to a column density of 2.8 × atoms cm − ,which is also confirmed through visual inspection in virtual real-ity. Future, more sensitive data from the MFS (Serra et al. 2016)will unambiguously show whether these galaxies are interactingor not.The collisional ring galaxy ESO 301-IG 11 has a slightly ex-tended H i disc, where the extension is in the south-east direction(away from the group centre). As suggested by its classification,the H i is likely to have been tidally disturbed in the collision thatformed the ring.In the three galaxies with disturbed or asymmetric H i discs(detailed below), strong tidal interactions can be reasonably ex-cluded as the cause, as the deep g -band FDS images show no Article number, page 5 of 18 & A proofs: manuscript no. MeerKAT_FornaxA h m m m m m − ◦ − ◦ − ◦ RA (J2000) D ec ( J ) NGC 1310NGC 1316NGC 1316C NGC 1317NGC 1326NGC 1326ANGC 1326B ESO 301-IG 11 FCC 19FCC 35FCC 40FCC 46 T o c l u s t e r
20 kpc
Fig. 2.
Primary beam-corrected constant H i contours from MeerKAT (blue) overlaid on a FDS (Iodice et al. 2016) gri stacked optical image. Thelowest contour represents the 3 σ column density level of N H i = × atoms cm − over a 44.1 km s − channel, where the contours increaseby a factor of 3 n ( n =
0, 1, 2, ...). The group galaxies are labelled and the galaxies not detected in H i are outlined by a dashed black ellipse. Thegrey circles indicate the sensitivity of the primary beam (Mauch et al. 2020) at 50%, 10%, and 2%. The red dashed circle denotes the 1.05 degree(0.38 Mpc) virial radius of the group as adopted in Drinkwater et al. (2001), where the restoring beam (33.0 (cid:48)(cid:48) × (cid:48)(cid:48) ) is shown in the bottom leftcorner and a scale bar indicating 20 kpc at the distance of Fornax A in the bottom right corner. The direction to the Fornax cluster is shown by theblack arrow. In H i , we detect 10 (out of 12) galaxies, previously known clouds and streams in the IGM and a population of new H i clouds in theIGM. The previously known IGM H i structures are labelled in Fig. 4 for clarity. stellar emission associated with the extended H i down to a sur-face brightness of 30 mag arcsec − . The H i tails and asymme-tries all di ff er in these galaxies, likely because each galaxy isa ff ected by di ff erent processes, such as gentle tidal interactions,ram pressure, and accretion.The dwarf late-type galaxy FCC 35 has a long, asymmetric(kinematically irregular) H i tail pointing away from the groupcentre. The two closest galaxies (spatially with confirmed red-shifts) are NGC 1316C and FCC 46, a dwarf late type and dwarfearly type. These two galaxies have unresolved H i and are moreH i deficient than the majority of the group galaxies. Neither adynamical interaction between these galaxies nor a hydrodynam- ical mechanism (such as ram pressure) can be ruled out as thecause for the long, H i tail of FCC 35.NGC 1326 is a barred spiral galaxy with a ring and hasclumpy, extended, and asymmetric H i emission in the south,pointing towards the group centre. The one-sided H i emissioncould be indicative of a tidal interaction. However, this couldalso be an instrumental e ff ect, as the galaxy is located very farfrom the pointing centre and is subjected to a variable sensitiv-ity response. The southern side (where the H i tail is) is sensitivedown to ∼ × atoms cm − , while the northern side hasa lower sensitivity of ∼ × atoms cm − . As the tails aredi ff use ( < × atoms cm − ), more sensitive observations are Article number, page 6 of 18leiner et al.: Pre-processing in the Fornax A group h m s s s m s − ◦ D ec ( J ) × cm − NGC 1310 h m s m s s s − ◦ × cm − NGC 1316 h m s s m s s − ◦ × cm − NGC 1316C h m s s s s − ◦ D ec ( J ) × cm − NGC 1317 h m s s m s s − ◦ × cm − NGC 1326 h m s s s s m s − ◦ × cm − NGC 1326A h m s s s s − ◦ D ec ( J ) × cm − NGC 1326B h m s m s s s − ◦ × cm − ESO 301-IG 11 h m s s s s s − ◦ < × cm − FCC 19 h m s s s m s − ◦ − ◦ RA (J2000) D ec ( J ) × cm − FCC 35 h m s s s s s − ◦ RA (J2000) < × cm − FCC 40 h m s s s s − ◦ RA (J2000)7.0 × cm − FCC 46
Fig. 3.
Optical three-colour composite of each group galaxy in our sample with overlaid H i contours. The colour image is comprised of the g -, r -, and i -band filters from the FDS (Iodice et al. 2016); the white dashed contour shows the most sensitive, constant column density of N H i = × atoms cm − from Fig. 2 and the blue contours start from the local column density sensitivity (i.e. 1.4 × atoms cm − scaled by theprimary beam response; see top left corner of each cutout) and increase by a factor of 3 n with n =
0, 1, 2, ..., at each step. For non-detections, the3 σ H i column density upper limit over a single channel is shown in red in the top left of the cutout. The restoring beam (33.0 (cid:48)(cid:48) × (cid:48)(cid:48) ) is shownin orange in the bottom left corner and a 5 kpc scale bar is shown in the bottom right corner. Article number, page 7 of 18 & A proofs: manuscript no. MeerKAT_FornaxA
Table 2.
Basic properties of the group galaxies and H i detected sources within the H i image field of view. The primary beam-corrected integratedH i flux, mass, and upper limits are included for all sources while the morphological type, stellar mass, and g – r colour is included for all thegalaxies. The H i mass was calculated using a distance of 20 Mpc and the statistical uncertainty of the flux was measured and propagated to the H i mass. The 3 σ upper limits of the H i flux and mass are calculated for non-detections using the local RMS and a 100 km s − wide integrated fluxfor a point source. All previously known sources are individually identified and the remaining H i IGM detections are summed into the remainingclouds category. The galaxy morphologies are classified in Ferguson (1989), the photometry is used to estimate the stellar mass (with the methodof Taylor et al. 2011), and g - r colours are measured in Raj et al. (2020) for the majority of the galaxies and in Venhola et al. (2018) for FCC 19and FCC 40. The photometry, g - r colour, and stellar mass of NGC 1316 are measured independently in Iodice et al. (2017). Source Integrated flux H i mass Morphological type Stellar mass g – r (Jy km s − ) (10 M (cid:12) ) (10 M (cid:12) ) (mag)NGC 1310 5.13 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± < < ± ± ± ± < < ± ± ± ± N ± ± S ± ± N , ± ± N , ± ± ± ± ± ± ± ± i emission onthe northern side.Finally, the massive late-type galaxy NGC 1310 is sur-rounded by H i extensions and clouds of di ff erent velocities,which is unusual, because it is a relatively (compared toNGC 1317) isolated galaxy, with an undisturbed optical spiralmorphology and a uniformly rotating H i disc. Despite the coarsevelocity resolution, we can determine from our observations thatthe majority of the H i extensions and clouds (except for the ex-tended component of the disc to the south) are not rotating withthe disc (Fig. 4) and cover a broad range ( ∼ − )in velocity, suggesting that it may be anomalous H i gas froman external origin. Future data from the MFS (Serra et al. 2016)with better velocity resolution will clarify this point. i in the intra-group medium We detect a total of (1.12 ± × M (cid:12) of H i in the IGM.All of the previous clouds in Serra et al. (2019) were detected aswell as additional H i in some of these features. We detect newclouds, the majority residing in the north, with some forminglarge, contiguous 3D structures.We searched for any association between the new H i in theIGM and stellar emission. In particular, as more H i has beendetected within the stellar halo of NGC 1316, we checked forany correlation between the H i and known stellar loops (Fig. 5).Overall, there is very little, clear association between the H i inthe NGC 1316 halo and its stellar loops. The major exceptionsare T S and its newly detected cloud, as they are fully containedwithin the SW stellar loop. The H i in SH2 and EELR may po-tentially correlate with the stellar loop L1 and there are some H i cloud (e.g. C N , ) in the north that partially overlap with the stel-lar loop L7. Other than examples above, all the remaining H i inthe IGM shows no association with stellar emission.We detect an extension in T N , e ff ectively doubling its lengthand mass. The extension smoothly connects in velocity with thepreviously known emission and now extends up to ∼
220 kpcfrom NGC 1316 (Fig. 4), which is where the H i originated fromSerra et al. (2019). The clouds that make up T N now contains(2.10 ± × M (cid:12) of H i . The north and south tails contain60% (6.7 ± × M (cid:12) ) of the total IGM H i mass. The remain-ing clouds in the IGM mostly reside to the north of NGC 1316,with the majority of these existing over a narrow (90 km s − )velocity range. It is possible some of these clouds form largecoherent H i structures, although it is not clear compared to thecase of T N and T S . While T N and T S originate from a single peri-centric passage of the NGC 1316 merger (Serra et al. 2019), theremaining clouds in the IGM are more likely to be the remnantsof recently accreted satellites onto NGC 1316, which is consis-tent with Iodice et al. (2017).The clouds immediately to the north-west of NGC 1317 maybe a remnant of its outer disc. These clouds are within a pro-jected distance of 10 kpc from NGC 1317 and the cloud and thegalaxy have the same velocity. The H i -to-stellar mass ratio of thegalaxy is low by at least an order of magnitude (see below) andthese clouds alone are not enough to explain the H i deficiency.However, these are the only clouds that show potential evidencethat they originated from NGC 1317.All the H i in the IGM located north of the group centre(NGC 1316) and the clouds to the south-east of ESO 301-IG 11appear to be decoupled from the stars. The H i in the south (SH2,T S ) has stellar emission associated with it. Additionally, there are Article number, page 8 of 18leiner et al.: Pre-processing in the Fornax A group h m m m m m − ◦ − ◦ − ◦ RA (J2000) D ec ( J ) NGC 1310NGC 1316NGC 1316C NGC 1317NGC 1326NGC 1326ANGC 1326B ESO 301-IG 11 FCC 19FCC 35FCC 40FCC 46
EELRSH2 C N , C N , T N T S T o c l u s t e r
20 kpc V e l o c i t y ( k m s − ) Fig. 4. H i velocity field, where the known galaxies and previously detected clouds and tails in the IGM are labelled. As in Fig. 2, the two galaxiesnot detected in H i are outlined by black, dashed ellipses and the direction to the Fornax cluster is shown by the black arrow. The velocity colour baris centred on the systemic velocity of the BGG (NGC 1316) at 1760 km s − . The grey circles indicate the sensitivity of the primary beam (Mauchet al. 2020) at 50%, 10%, and 2%. The red dashed circle denotes the 1.05 degree (0.38 Mpc) virial radius of the group as adopted in Drinkwateret al. (2001), where the restoring (33.0 (cid:48)(cid:48) × (cid:48)(cid:48) ) beam and scale bar are shown in the bottom corners. The clouds that make up T N have a new,extended component, e ff ectively doubling the size compared to its original discovery in Serra et al. (2019). a few H i clouds near to the group centre that contain multiphasegas. In Figure 6, we show the ionized H α gas emission detected in thevicinity of NGC 1316 (i.e. the group centre). H α is detected inNGC 1316, NGC 1317, and NGC 1310. However, the most strik-ing features are the H α complexes detected in the IGM.There are giant filaments of H α in the IGM stretching be-tween galaxies of the group. H i is directly associated with someof the ionised gas, showing the coexistence of multiphase gas inthe IGM. These occur in EELR, C N , , the cloud directly below C N , and in five newly detected clouds containing H i that welabel MP in Fig. 6. Additionally, we detect the “Ant” (or ALF;Ant-like feature) first detected as a depolarising feature in Fo-malont et al. (1989) and later in H α by Bland-Hawthorn et al.(1995). The H α emission is thought to provide the interveningturbulent magneto-ionic medium required to depolarize the radiocontinuum emission Fomalont et al. (1989). There is no opticalcontinuum emission nor any H i emission currently associatedwith the Ant.While there are a number of multiphase gas clouds in theIGM, the brightest case is EELR. It is clear that EELR has acomplex multiphase nature, with H i , H α , and dust all previouslydetected in it (Mackie & Fabbiano 1998; Horellou et al. 2001; Article number, page 9 of 18 & A proofs: manuscript no. MeerKAT_FornaxA h m m m m m − ◦ RA (J2000) D ec ( J ) T S C N, C N, EELRSH2
Fig. 5.
Low surface brightness (star removed) image of NGC 1316in g -band, observed with the VST (Iodice et al. 2017). The known(Schweizer 1980; Richtler et al. 2014; Iodice et al. 2017) stellar loopsare labelled and outlined by the dashed green lines. The H i is shown bythe solid blue contours and the previously known H i clouds are labelled.The clouds that make up T S (including the new western H i cloud) over-lap with the stellar SW loop. There is some overlap with some H i cloudsin the north (e.g. C N , and the clouds to the west) and the optical loopL7. Overall, there is no consistent correlation between the stellar loopsand the distribution of H i clouds. Serra et al. 2019). We detect 50% more H i than the previousstudy (Serra et al. 2019) and H i is only present in the region ofthe bright, more ordered ionised gas morphology. Given that ourH i image is sensitive to a column density of 1.4 × atomscm − , it is unlikely that there is any H i in the less ordered (andlikely turbulent) part of EELR. Currently, the origin of EELRis unclear, and we will present a detailed analysis of it and itsmultiphase gas in future work.
5. Pre-processing in the group
The Fornax A group is at a projected distance of ∼ i throughout the Fornax A groupboth in the galaxies and the IGM. While the galaxies range frombeing H i rich to extremely H i deficient, the majority of thegalaxies contain a regular amount of H i for their stellar mass.This is consistent with the group being in an early phase of as-sembly, as the majority of galaxies would be H i deficient for agroup in the advanced assembly stage. The H i detections showevidence of pre-processing in the form of (2.8 ± × M (cid:12) of H i in the IGM, H i deficient galaxies, truncated H i discs, H i tails, and asymmetries. The diversity of galaxy H i morphologiessuggest that we are observing galaxies at di ff erent stages of pre-processing, as we detail below. The most obvious case of pre-processing in the group isNGC 1316, the BGG. It is a peculiar early type that is the bright-est galaxy in the entire Fornax cluster volume and the resultof a 10:1 merger that occurred 1 – 3 Gyr ago between a mas-sive early-type galaxy and a gas-rich late-type galaxy (Schweizer1980; Mackie & Fabbiano 1998; Goudfrooij et al. 2001; Iodiceet al. 2017; Serra et al. 2019). There are large stellar loops andstreams, an anomalous amount of dust and molecular gas (2 × and 6 × M (cid:12) , respectively) in the centre, as well as H i inthe centre and in the form of long tails (Draine et al. 2007; Lanzet al. 2010; Galametz et al. 2012; Morokuma-Matsui et al. 2019;Serra et al. 2019).The H i mass budget for a 10:1 merger to produce the fea-tures observed in NGC 1316 requires the progenitor to contain ∼ × M (cid:12) of H i (Lanz et al. 2010; Serra et al. 2019). Recently,Serra et al. (2019) detected 4.3 × M (cid:12) of H i in the centre ofNGC 1316, overlapping with the dust and molecular gas, and atotal H i mass of 7 × M (cid:12) when including the tails and nearbyH i clouds. While these authors detected an order of magnitudemore H i than previous studies, this is a factor of ∼ i mass in the centre of (6.8 ± × M (cid:12) and a total H i mass 0.9 – 1.2 × M (cid:12) asso-ciated with NGC 1316 in the form of streams and clouds. Thisbrings the observed H i mass budget even closer to the expectedvalue under the 10:1 lenticular + spiral merger hypothesis – justwithin a factor 1.7 - 2.2, which is well within the uncertainties.Since the merger 1 – 3 Gyr ago, NGC 1316 has been ac-creting small satellites (Iodice et al. 2017). The satellites mayhave contributed to the build up of H i , however, we do not ob-serve any H i correlated with dwarf galaxies within 150 kpc ofNGC 1316. Any contributed H i is second order compared to theinitial merger, which is supported by the H i mass of NGC 1316being dominated by the tails. Tidal forces from the initial mergerejected 6.6 × M (cid:12) of H i into the IGM in the T N and T S tailsalone. The remaining H i in the IGM is likely to be a combina-tion of gas decoupled from stars in the initial merger and gasfrom more recently accreted satellites. H i tidal tails that spanhundreds of kpc in galaxy groups have been shown to survivein the IGM for the same timescale (1 – 3 Gyr) from when thismerger took place (Hess et al. 2017). In this section, we identify galaxies at di ff erent stages of pre-processing according to their H i morphology and cool gas (H i The lower limit was determined by only including the same H i sources as Serra et al. (2019) and the T N extension, while the upperlimit includes the remaining H i clouds in the IGMArticle number, page 10 of 18leiner et al.: Pre-processing in the Fornax A group h m m m − ◦ RA (J2000) D ec ( J ) NGC 1310NGC 1317NGC 1316 EELRSH2 C N , MP MP MP MP MP Ant
20 kpc − . − . − . . . . . H α S u r f a ce B r i g h t n e ss ( − e r g s c m − s − a r c s ec − ) Fig. 6.
OmegaCAM H α emission showing the ionised gas in the vicinity of NGC 1316. The blue contour shows the majority of the western lobeof NGC 1316 in radio continuum at a (conservative) level of 1.3 mJy beam − from Maccagni et al. (2020). The white contours show the 3 σ H i column density of 1.4 × atoms cm − (over 44.1 km s − ) from this work. Known sources (i.e. galaxies and IGM H i ) and multiphase (MP) gasclouds that contain H α and H i as well as the Ant-like feature from Fomalont et al. (1989) are labelled. This image reveals long filaments of ionisedgas in the IGM. and H ) ratios. The categories are as follows: i) early, wherea galaxy has yet to experience significant pre-processing; ii)ongoing, for galaxies that currently show signatures of pre-processing; and iii) advanced, for galaxies that have already ex-perienced significant pre-processing.There are a total of 12 galaxies in the sample, which are allthe spectroscopically confirmed galaxies within the H i imagefield of view. In our sample, 10 galaxies have H i detections and2 galaxies (FCC 19 and FCC 40) have H i upper limits (Fig. 3).There are 7 galaxies that have been observed with ALMA. The5 galaxies that were not observed are ESO 301-IG 11, FCC 19,FCC 35, FCC 40, and FCC 46 (Morokuma-Matsui et al. 2019,Morokuma-Matsui et al. in prep). We measure the molecular gasmass of the observed galaxies using the standard Milky Way CO-to-H conversion factor of 4.36 (M (cid:12) K km s − pc − ) − (Bolattoet al. 2013) as well as estimated stellar masses (Table 2) fromRaj et al. (2020) and Venhola et al. (2018), which are derivedfrom the g and i photometric relation in Taylor et al. (2011). Weremove the helium contribution from our molecular gas massesso that we are measuring the molecular-to-atomic hydrogen gasmass (except in the total gas fraction, shown below) and can di-rectly compare our findings to Catinella et al. (2018).We present the H i and H scaling ratios in Fig. 7. We mea-sure the H i gas fraction F HI ≡ log(M HI / M (cid:63) ), the total gas fractionFgas ≡ log(1.3(M HI + M H2 ) / M (cid:63) ) where the 1.3 accounts for thehelium contribution, the molecular-to-atomic gas mass ratio R mol ≡ log(M H2 / M HI ), and the H gas fraction F H2 ≡ log(M H2 / M (cid:63) ).We compare the H i fraction of our galaxies to those in the Her-schel Reference Survey (HRS; Boselli et al. 2010, 2014) and theVoid Galaxy Survey (VGS; Kreckel et al. 2012), which span acomparable stellar mass range of our galaxies. We also compareF HI to the median trend of the extended GALEX Arecibo SDSS Survey (xGASS; Catinella et al. 2018). Furthermore, we com-pare our molecular gas scaling relations to the median trends ofxGASS-CO (Fig. 7), which are xGASS galaxies with CO detec-tions (Catinella et al. 2018). The xGASS and xGASS-CO trendsprovide a good reference for the H i and H scaling relations inthe local Universe as the median F HI trend was derived from1179 galaxies selected with 10 < M (cid:63) (M (cid:12) ) < . and 0.01 < z < mass and scaling relations derived us-ing a subset 477 galaxies from the parent sample that have COdetections.The two galaxies that show no signatures (i.e. in the earlyphase) of pre-processing are NGC 1326A and NGC 1326B. Theyare H i rich galaxies with typical extended H i discs and alow molecular gas content. Both galaxies were observed withALMA (Morokuma-Matsui et al. 2019, Morokuma-Matsui et al.in prep), although no CO was detected, placing upper limits onthe H mass. They have the highest H i fraction and lowest H -to-H i ratios given their stellar mass (Fig. 7). The galaxies arejust within the virial radius of the group, making them furthestfrom the group centre in projected distance. This increases thelikelihood that the galaxies have not undergone pre-processingyet.The galaxies that show current signatures of pre-processing(i.e. the ongoing category) are FCC 35, ESO 301-IG 11,NGC 1310, NGC 1316, and NGC 1326. In general, these galax-ies have H i tails or asymmetric extended H i emission, typicalH i and H ratios (for the galaxies with H observations) that fol-low the median xGASS trends in Fig. 7. The exception to thisis NGC 1316. As this galaxy is the BGG, it has a unique for-mation and evolution history (discussed in section 5.1) that dis-plays both an ongoing (e.g. tidal tails) and advanced state (giantelliptical with a lack of H i contained in the stellar body) of pre- Article number, page 11 of 18 & A proofs: manuscript no. MeerKAT_FornaxA log(Stellar Mass) (M (cid:12) ) − − − − F H I N 1310N 1317ESO 301 N 1326N 1316CFCC 35 N 1326AN 1326BFCC 46 N 1316FCC 19FCC 40EarlyOngoingAdvancedUnclassified . . . . . . . log(Stellar Mass) (M (cid:12) ) − − − F g a s N 1310 N 1317N 1326N 1316CN 1326AN 1326B N 1316 . . . . . . . log(Stellar Mass) (M (cid:12) ) − . − . − . − . . . R m o l N 1310 N 1317N 1326N 1316CN 1326AN 1326B N 1316 − − − − F HI − . − . − . − . − . − . − . F H . . . . N 1310N 1317 N 1326N 1316C N 1326AN 1326BN 1316
Fig. 7.
Atomic and molecular gas scaling ratios. In all figures, the early, ongoing, and advanced pre-processing categories are shown as blue circles,green squares, and red diamonds, respectively and H upper limits are depicted by arrows. Solid markers indicate H i detections and open markersare non-detections. FCC 40 is not assigned to any pre-processing category and is shown as the open black star. Top left panel : The H i gas fractioncompared to galaxies from the HRS (Boselli et al. 2010, 2014) and VGS (Kreckel et al. 2012) (grey points) that show the typical scatter in F HI .The orange shaded region indicates the median trend from xGASS (Catinella et al. 2018). Top right panel : The total gas fraction of our galaxiescompared to the median xGASS-CO trend (Catinella et al. 2018) (orange shaded region).
Bottom left panel:
The molecular-to-atomic-gas ratio ofour galaxies compared to the median xGASS-CO trend (Catinella et al. 2018) (orange shaded region).
Bottom right panel : The H gas fraction asa function of H i gas fraction, showing constant ratios of 100%, 30%, 10%, and 3%. Overall, the galaxies in the early category are H i rich, thegalaxies in the ongoing category typically follow the xGASS and xGASS-CO median scaling relations (Catinella et al. 2018), while galaxies inthe advanced category have no H i or are H i -deficient with irregularly high H -to-H i ratios. processing. In this work, we include this galaxy in the ongoingcategory, although the H i mass range calculated in section 5.1reflects that it could also be part of the advanced category.FCC 35 is the bluest galaxy (Fig. 3 and Table 2) in the group(Raj et al. 2020) and has extremely strong and narrow opticalemission lines that classify it as either a blue compact dwarf oran active star-burst H II galaxy (Putman et al. 1998). Previousstudies (i.e. Putman et al. 1998; Schröder et al. 2001) detecteda H i cloud associated with FCC 35 and suggested it may be aresult of a tidal interaction with the nearest (projected separa-tion of 50 kpc) neighbour NGC 1316C. This is a plausible sce-nario as FCC 35 has an up-bending (Type-III) break in the stel-lar radial profile, and a bluer outer stellar disc (Raj et al. 2020), which could be tidally induced star formation. However, the starformation could also be compression / shock induced (Raj et al.2020). We detect the H i cloud of FCC 35 as part of a long tailpointing away from the group centre, making it the most likelygalaxy to show evidence of ram pressure stripping. The lowerIGM (compared to the ICM) density means that ram pressurestripping is less prevalent in groups. Despite the observationalchallenges, a few cases have been reported (e.g. Westmeier et al.2011; Rasmussen et al. 2012; Vulcani et al. 2018; Elagali et al.2019) and ram pressure is thought to play an important role inthe pre-processing of galaxies in groups. FCC 35 is not H i defi-cient (Fig. 7), implying that the gas has recently been displaced, Article number, page 12 of 18leiner et al.: Pre-processing in the Fornax A group similar to other galaxies showing early signs of gas removal (e.g.Ramatsoku et al. 2020; Moretti et al. 2020).ESO 301-IG 11 is a collisional ring galaxy with a H i gasfraction below the median trend, although it is not the most H i deficient galaxy for its stellar mass. There is clear evidence ofa tidal interaction in the form of irregular optical morphology,an up-bending (Type-III) break in the stellar radial profile anda slightly extended and asymmetric H i disc. The galaxy is bluein colour, although the outer stellar disc is redder than the in-ner disc (Raj et al. 2020), implying that the tidal interaction mayhave restarted star formation in the centre.The asymmetric H i tail of NGC 1326 is di ff use ( < × atoms cm − ) and only detected on one side of the galaxy. Thesensitivity of the opposing side prevents us from detecting H i that di ff use, and we are therefore unable to distinguish whetherthe extended H i is part of a regular extended H i disc or a sig-nature of pre-processing. With the current H i content, it followsthe same H i and H trends as the other galaxies in the ongoingcategory.The optical morphology and gas scaling relations ofNGC 1310 suggest that it is not being pre-processed. The stel-lar spiral structure is completely intact (Fig. 3), ruling out strongtidal interactions and the H i gas fraction and molecular-to-atomic gas ratios are close to the median trends. However, the H i morphology appears complex and incoherent, with many asym-metric extensions and nearby clouds at di ff erent velocities. It isclear that the anomalous H i clouds and extensions are not ro-tating with the main H i disc (Fig. 4), suggesting external ori-gins. The H i extension in the north-west may be emission froma dwarf satellite galaxy, although a spectroscopic redshift wouldbe required to confirm this. Given the presence of the H α fila-ments in the vicinity of NGC 1310, the remaining clouds maybe a result of hot gas, cooling in the IGM (and hot halo ofNGC 1316) and being captured or accreted onto this galaxy.Finally, the galaxies that are in the advanced stage of pre-processing are NGC 1316C, NGC 1317, FCC 19, and FCC 46.There is no H i detected in FCC 19, and the other three galaxieshave truncated H i discs and are H i deficient as their F HI is morethan 3 σ from the xGASS median trend (Fig. 7).NGC 1316C and NGC 1317 have a low H i mass fraction andregular H mass fraction. The total gas fraction of these galaxiesis low and is driven by the lack of H i . Hence, they have signif-icantly more H than H i and a molecular-to-atomic fraction anorder of magnitude higher (the highest in our sample) than themedian trend (Fig. 7). Both these galaxies have no break (Type-I)in their stellar radial profile (Raj et al. 2020), showing no sign ofdisruption to their stellar body and their H i confined to the stel-lar disc, implying that the outer H i disc has been removed. Rampressure or gentle tidal interactions are likely to be responsiblefor removing the outer H i disc of these galaxies. The less dense(compared to the ICM) IGM combined with the group poten-tial allows galaxies to hold on to their gas more e ff ectively thanin clusters (Seth & Raychaudhury 2020). The retained atomicgas within the stellar body can then be converted into molecu-lar gas. This scenario is consistent with the findings of the GAsStripping Phenomena in galaxies with MUSE (GASP; Morettiet al. 2020) project, where pre-processed galaxies in groups (andclusters) have their outer H i removed (via ram pressure) and theremaining H i is e ffi ciently converted into H . These galaxies inthe advanced stage of pre-processing with truncated H i discs andregular amounts of H are similar to some galaxies in the Virgo(Cortese et al. 2010) and Fornax cluster (Loni et al. 2021). Thissuggests that late-type galaxies that have been su ffi ciently pro- cessed lose their outer H i disc and end up with more H thanH i . Despite the similarities between NGC 1316C andNGC 1317, these galaxies have likely been pre-processedon di ff erent timescales. The stellar mass of NGC 1316C is morethan an order of magnitude lower than that of NGC 1317 andaccording to Raj et al. (2020), NGC 1316C only recently ( < < (cid:48) ) disc of NGC 1317 (Raj et al.2020) and even though there is only a projected separation of ∼
50 kpc between NGC 1316 and NGC 1317, a strong tidalinteraction can be reasonably excluded due to the intact spiralstructure of NGC 1317 (Richtler et al. 2014; Iodice et al. 2017).The outer H i disc has been removed and possibly lost to theIGM (i.e. potentially identified as the adjacent clouds at thesame velocity) as a result of gentle tidal or hydrodynamical in-teractions. Alternatively, the outer disc may have been convertedto other gaseous phases on short timescales ( < i disc of NGC 1317, it is evident that thegalaxy has not had access to cold gas over long timescales.Out of all the galaxies with H i , FCC 46 is the most H i defi-cient given its stellar mass. It is a dwarf elliptical with a recentstar formation event and H i was first detected as a polar ring or-biting around the optical minor axis by De Rijcke et al. (2013).As the H i is kinematically decoupled from stellar body, the gaswas likely accreted from an external source (De Rijcke et al.2013). Our measured H i mass (Table 2) is consistent with thatfrom De Rijcke et al. (2013), although, as a result of our sensi-tivity at that position, we do not detect the di ff use H i componentthat shows the minor axis rotation. A minor merger event (e.g.with a dwarf late type) is consistent with the morphology and ∼ M (cid:12) of H i found in FCC 46.FCC 19 is a dwarf lenticular galaxy (Fig. 3) with a stellarmass of 3.4 × M (cid:12) (Liu et al. 2019). It has a g – r colourof 0.58 (Iodice et al. 2017), which is similar to the colour ofNGC 1310, NGC 1326, NGC 1326A, and ESO 301-IG 11 (Table2), which have regular H i fractions and are likely forming stars.However, no H i is detected in FCC 19 and we measure a 3 σ F HI upper limit of -2.3 (Fig. 7) assuming a 100 km s − line width.FCC 19 is situated in the most sensitive part of the image, mean-ing that the galaxy truly does not contain H i . Being so close(70 kpc in projection) to NGC 1316, the tidal field and hot haloof NGC 1316 are likely to have played significant roles in re-moving the H i from FCC 19. The H i has likely been strippedfrom the galaxy and lost to the IGM. The stripped H i may alsobe potentially heated and prevented from cooling.Lastly, we refrain from assigning a category to FCC 40 be-cause we are unable to ascertain whether the galaxy propertiesare a result of secular evolution or have been influenced by pre-processing. This galaxy is a low surface brightness (Fig. 3), low-mass (M (cid:63) = × M (cid:12) ) blue dwarf elliptical (Table 2) withno H i detected. We place an upper limit on the H i mass (andH i fraction), although it is currently unknown if galaxies of thismass, colour, and morphology are expected to contain H i .We show the spatial distribution of each group galaxy andtheir pre-processing status in Fig. 8. The distribution shows a va-riety of pre-processing stages mixed throughout the group, withno clear radial dependence. The majority of on-going and ad-vanced pre-processing are < > ∼ Article number, page 13 of 18 & A proofs: manuscript no. MeerKAT_FornaxA h m m m m m − ◦ − ◦ − ◦ RA (J2000) D ec ( J ) N 1310N 1316N 1316C N 1317N 1326N 1326AN 1326B ESO 301 FCC 19FCC 35FCC 40FCC 46 T o c l u s t e r
20 kpc
Fig. 8.
Pre-processing map of the Fornax A group. The backgroundimage shows the 1.44 GHz MeerKAT radio continuum emission(Maccagni et al. 2020) and the position of each group galaxy are over-laid with the same markers as Fig. 7. The filled markers representH i detections, the open markers indicate H i non-detections, where theearly, ongoing, advanced, and unclassified pre-processing categories areshown as blue circles, green squares, red diamonds, and black stars, re-spectively. The red dashed circle denotes the 1.05 degree (0.38 Mpc)virial radius of the group as adopted in Drinkwater et al. (2001). A20 kpc scale bar is shown in the bottom right corner and the direction tothe Fornax cluster is shown by the black arrow. There is no consistenttrend between projected position and pre-processing status, although themajority of group galaxies show evidence of pre-processing. The extentof the NGC 1316 AGN lobes show that it may be playing a role in thepre-processing of neighbouring galaxies and the magnetic field couldhelp the containment of multiphase gas. (cluster) virial radii from the Fornax cluster, the Fornax groupis located at the distance where pre-processing is thought to bethe most e ffi cient (Lewis et al. 2002; Gómez et al. 2003; Ver-dugo et al. 2008; Mahajan et al. 2012; Haines et al. 2015). Ingeneral, it is not clear whether the pre-processing at this infalldistance is driven by the group interacting with the cluster, orby local (e.g. tidal and hydrodynamical) interactions within thegroup. In this instance, it appears that pre-processing is drivenby local interactions within the Fornax A group for the follow-ing reasons: i) The massive, central galaxy is at least one orderof magnitude more massive (Table 2) than the satellite galaxies.ii) This central galaxy underwent a merger 1 – 3 Gyr ago (dis-cussed in Section 5.1). iii) The majority of galaxies close to thegroup centre ( < / B) clos-est to the Fornax cluster (and furthest from the group centre)show no evidence of pre-processing. In addition to these points,there are four galaxies (NGC 1310, NGC 1317, ESO 301-IG 11,and FCC 19) that spatially overlap (in projection) with the radiolobes of NGC 1316 (Fig. 8) and therefore may be influenced bythe AGN (e.g Johnson et al. 2015).
The H i tails and clouds in the IGM are a direct result of galaxieshaving their H i removed through hydrodynamical and tidal in-teractions over the past few Giga-years. As described in sections4.3 and 5.1, the majority (if not all) of the H i in the IGM is dueto the Fornax A merger and the recent accretion of satellites.The amount (1.12 ± × M (cid:12) ) of detected H i in theIGM is not enough to account for all of the missing H i in theH i deficient group galaxies. However, the outer parts of the im-age are subject to a large primary beam attenuation and some ofthe IGM H i may be hiding in the noise. We estimate the amountof H i potentially missed by assuming that we detect all H i inthe IGM in the inner 0.1 deg (primary beam response > i in this area is representative of the IGMthroughout the entire group both in terms of amount of H i perunit area and H i column density distribution. Under this assump-tion, the primary beam attenuation reduces the detected H i by afactor of ∼ ∼ × M (cid:12) of H i in the IGM. All the (including the missed) H i inIGM is still not enough to explain all the H i deficient galaxiesin the group and clearly gas exists in other phases (i.e. H andH α ). Some of the H i in the galaxies has been converted into H ,which explains why the more advanced pre-processed galaxiesthat have H i , display high molecular-to-atomic gas ratios, andthere is H α in galaxies and the IGM.Currently, the origin of giant ionised gas filaments in theIGM is not well understood. However, they are typically ob-served in high-mass groups or low-mass clusters (e.g. halomasses > . M (cid:12) ), for example the Virgo cluster (Kenneyet al. 2008; Boselli et al. 2018b,a; Longobardi et al. 2020) andthe Blue Infalling Group (Cortese et al. 2006; Fossati et al.2019); see Yagi et al. (2017) for a list of clusters that containlong ionised gas filaments. A likely scenario is that cool gasis stripped from an in-falling galaxy, and subsequently ionised,possibly from ionising photons originating from star-formingregions (Poggianti et al. 2018; Fossati et al. 2019) or throughnon-photo-ionisation mechanisms such as shocks, heat conduc-tion, and magneto-hydrodynamic waves (Boselli et al. 2016). Weuse the relation in Barger et al. (2013) to estimate the total H α mass in the IGM (i.e. EELR, SH2, and the filaments) from ourH α photometry (Fig. 6). Assuming a typical H α temperature of10 K and electron density of 1 cm − , we estimate the total H α mass in the IGM to be ∼ × M (cid:12) , which does not signifi-cantly contribute to the total gas budget in the IGM.Simulations show that ∼ K (i.e. relatively cool) gasclouds can survive in hot haloes (such as NGC 1316) for cos-mological timescales (Nelson et al. 2020). The clouds originatefrom satellite mergers, and are not in thermal equilibrium, butrather magnetically dominated. Cooling is triggered by the ther-mal instability and the cool gas is surrounded by an interfaceof intermediate temperature gas (Nelson et al. 2020). These in-gredients can explain how multiphase gas clouds are present inthe hot halo of NGC 1316 (Fig. 6), such that the H α filamentsare a result of satellite accretion and the H i has rapidly cooledfrom these structures, with the ability to survive in the IGM forcosmological timescales.Recently, Müller et al. (2020) suggest that magnetic fields ofthe order of 2 – 4 µ G can shield H α and H i in the ICM / IGMsuch that the gas clouds do not dissipate. As the H α filamentsand multiphase gas clouds are within the radio lobes (in projec-tion) of NGC 1316, the magnetic field of the lobes (measuredto be ∼ µ G by McKinley et al. 2015; Anderson et al. 2018;Maccagni et al. 2020) may be providing additional stability for
Article number, page 14 of 18leiner et al.: Pre-processing in the Fornax A group the H α and H i to survive. Indeed, the Ant detected by Foma-lont et al. (1989) and Bland-Hawthorn et al. (1995) is a smallportion of the giant H α filaments in the IGM. Even though thereis currently no H i associated with the Ant, other sections of theH α filaments show that neutral and ionised gas can coexist insome regions of the IGM, possibly transform into one another,and accrete onto group galaxies (e.g. NGC 1310).
6. Conclusions
We present results from MeerKAT H i commissioning observa-tions of the Fornax A group. Our observations are reduced withthe CARACal pipeline and our H i image is sensitive to a columndensity of 1.4 × atoms cm − in the field centre. Out of 13spectroscopically confirmed group members, we detect H i in 10and report an H i mass upper limit for 2 (the remaining galaxyis outside the field of view of our observation). We also detectH i in the IGM, in the form of clouds, some distributed alongcoherent structures up to 220 kpc in length. The H i in the IGMis the result of a major merger occurring in the massive, cen-tral galaxy NGC 1316, 1 – 3 Gyr ago, combined with H i beingremoved from satellite galaxies as they are pre-processed.We find that 9 out of the 12 galaxies show some evidenceof pre-processing in the form of H i deficient galaxies, truncatedH i discs, H i tails, and asymmetries. Using the H i morphologyand the molecular-to-atomic gas ratios of the galaxy, we classifywhether each galaxy is in the early, ongoing, or advanced stageof pre-processing.Finally, we show that there are giant H α filaments in theIGM, within the hot halo of NGC 1316. The filaments are likelya result of molecular gas being removed from a satellite galaxyand then ionised. We observe a number of H i clouds associatedwith the ionised H α filament, indicating the presence of multi-phase gas. Simulations show that hot gas can condense into coolgas within hot haloes and survive for long periods of time ona cosmological timescale, which is consistent with the cool gasclouds we detect within the hot halo of NGC 1316. The multi-phase gas is supported by magnetic pressure, implying that themagnetic field in the lobes of the NGC 1316 AGN might be play-ing an important role in maintaining these multiphase gas clouds.The cycle of AGN activity and cooling gas in the IGM could ul-timately result in the cool gas clouds falling back onto the centralgalaxy. We summarise our main findings as follows:1. We present new, resolved H i in FCC 35, NGC 1310, andNGC 1326.2. There is a total of(1.12 ± × M (cid:12) of H i in the IGM,which is dominated by T N and T S (combined H i mass of 6.6 × M (cid:12) ). We detect additional components in both tails, anextension in T N , e ff ectively doubling its length, and a cloudin T S that shows coherence with the stellar south-west loop.3. The H i in the IGM is decoupled from the stars, other than inT S and SH2.4. We measure 0.9 – 1.2 × M (cid:12) of H i associated withNGC 1316, bringing the observed H i mass budget within afactor of ∼ + spiral merger occurring ∼ i disk, high H i content, and molecular-to-atomicgas ratios at least an order of magnitude below the mediantrend for their stellar mass. Galaxies that are currently be-ing pre-processed typically display H i tails or asymmetricextended disks, while containing regular amounts of H i andH . Galaxies in the advanced stage of pre-processing have noH i or have lost their outer H i and are e ffi ciently convertingtheir remaining H i to H .7. We detect the Ant first observed by Fomalont et al. (1989) asa depolarising feature and later in H α by Bland-Hawthornet al. (1995), which turns out to be a small part of long,ionised H α filaments in the IGM. Localised cooling (po-tentially assisted by the magnetic field in the lobes of theNGC 1316 AGN) can occur in the H α filaments to condenseand form H i .In this work, our deep MeerKAT H i image shows many ex-amples of pre-processing in the Fornax A group, such as galax-ies with a variety of atypical morphologies and massive amountsof H i in the IGM. The improved sensitivity and resolution ofthe MFS (Serra et al. 2016) will likely reveal more H i through-out the group and provide kinematic information for the H i ingalaxies and the IGM. Acknowledgements.
The MeerKAT telescope is operated by the South AfricanRadio Astronomy Observatory, which is a facility of the National ResearchFoundation, an agency of the Department of Science and Innovation. We aregrateful to the full MeerKAT team at SARAO for their work on buildingand commissioning MeerKAT. This paper makes use of the following ALMAdata: ADS / JAO.ALMA / NRAO, and NAOJ. This work also made use of theInter-University Institute for Data Intensive Astronomy (IDIA) visualisation lab( https://vislab.idia.ac.za ). IDIA is a partnership of the University ofCape Town, the University of Pretoria and the University of Western Cape. Thisproject has received funding from the European Research Council (ERC) un-der the European Union’s Horizon 2020 research and innovation programme(grant agreement no. 679627; project name FORNAX). The research of OS issupported by the South African Research Chairs Initiative of the Departmentof Science and Innovation and the National Research Foundation. KT acknowl-edges support from IDIA. The work of KMM is supported by JSPS KAKENHIGrant Number of 19J40004. RFP acknowledges financial support from the Eu-ropean Union’s Horizon 2020 research and innovation program under the MarieSkłodowska-Curie grant agreement No. 721463 to the SUNDIAL ITN network.AV acknowledges the funding from the Emil Aaltonen foundation. PK is par-tially supported by the BMBF project 05A17PC2 for D-MeerKAT. AS acknowl-edges funding from the National Research Foundation under the Research CareerAdvancement and South African Research Chair Initiative programs (SARChI),respectively. FV acknowledges financial support from the Italian Ministry of For-eign A ff airs and International Cooperation (MAECI Grant Number ZA18GR02)and the South African NRF (Grant Number 113121) as part of the ISARPRAIOSKY2020 Joint Research Scheme. References
Anderson, C. S., Gaensler, B. M., Heald, G. H., et al. 2018, ApJ, 855, 41Baldry, I. K., Glazebrook, K., Brinkmann, J., et al. 2004, ApJ, 600, 681Balogh, M., Eke, V., Miller, C., et al. 2004, MNRAS, 348, 1355Barger, K. A., Ha ff ner, L. M., & Bland-Hawthorn, J. 2013, ApJ, 771, 132Bell, E. F., Wolf, C., Meisenheimer, K., et al. 2004, ApJ, 608, 752Bianconi, M., Smith, G. P., Haines, C. P., et al. 2018, MNRAS, 473, L79Bland-Hawthorn, J., Ekers, R. D., van Breugel, W., Koekemoer, A., & Taylor, K.1995, ApJ, 447, L77Bolatto, A. D., Wolfire, M., & Leroy, A. K. 2013, ARA&A, 51, 207Boselli, A., Cortese, L., Boquien, M., et al. 2014, A&A, 564, A66Boselli, A., Cuillandre, J. C., Fossati, M., et al. 2016, A&A, 587, A68Boselli, A., Eales, S., Cortese, L., et al. 2010, PASP, 122, 261Boselli, A., Fossati, M., Consolandi, G., et al. 2018a, A&A, 620, A164Boselli, A., Fossati, M., Ferrarese, L., et al. 2018b, A&A, 614, A56 Article number, page 15 of 18 & A proofs: manuscript no. MeerKAT_FornaxA
Calcáneo-Roldán, C., Moore, B., Bland -Hawthorn, J., Malin, D., & Sadler,E. M. 2000, MNRAS, 314, 324Cantiello, M., Grado, A., Blakeslee, J. P., et al. 2013, A&A, 552, A106Catinella, B., Saintonge, A., Janowiecki, S., et al. 2018, MNRAS, 476, 875Catinella, B., Schiminovich, D., Cortese, L., et al. 2013, MNRAS, 436, 34Chung, A., van Gorkom, J. H., Kenney, J. D. P., & Vollmer, B. 2007, ApJ, 659,L115Chung, S. M., Gonzalez, A. H., Clowe, D., et al. 2009, ApJ, 691, 963Cluver, M. E., Jarrett, T. H., Taylor, E. N., et al. 2020, ApJ, 898, 20Cortese, L., Davies, J. I., Pohlen, M., et al. 2010, A&A, 518, L49Cortese, L., Gavazzi, G., Boselli, A., et al. 2006, A&A, 453, 847Courtois, H. M. & Tully, R. B. 2015, MNRAS, 447, 1531Cowie, L. L. & McKee, C. F. 1977, ApJ, 211, 135Davies, L. J. M., Robotham, A. S. G., Lagos, C. d. P., et al. 2019, MNRAS, 483,5444de Blok, W. J. G., Walter, F., Ferguson, A. M. N., et al. 2018, ApJ, 865, 26De Rijcke, S., Buyle, P., & Koleva, M. 2013, ApJ, 770, L26Donnari, M., Pillepich, A., Joshi, G. D., et al. 2020, MN-RAS[ arXiv:2008.00005 ]Draine, B. T., Dale, D. A., Bendo, G., et al. 2007, ApJ, 663, 866Drew, J. E., Gonzalez-Solares, E., Greimel, R., et al. 2014, MNRAS, 440, 2036Drinkwater, M. J., Gregg, M. D., Holman, B. A., & Brown, M. J. I. 2001, MN-RAS, 326, 1076Driver, S. P., Hill, D. T., Kelvin, L. S., et al. 2011, MNRAS, 413, 971Eke, V. R., Baugh, C. M., Cole, S., et al. 2004, MNRAS, 348, 866Ekers, R. D., Goss, W. M., Wellington, K. J., et al. 1983, A&A, 127, 361Elagali, A., Staveley-Smith, L., Rhee, J., et al. 2019, MNRAS, 487, 2797Ferguson, H. C. 1989, AJ, 98, 367Fomalont, E. B., Ebneter, K. A., van Breugel, W. J. M., & Ekers, R. D. 1989,ApJ, 346, L17Fossati, M., Fumagalli, M., Gavazzi, G., et al. 2019, MNRAS, 484, 2212Fujita, Y. 2004, PASJ, 56, 29Galametz, M., Kennicutt, R. C., Albrecht, M., et al. 2012, MNRAS, 425, 763Gómez, P. L., Nichol, R. C., Miller, C. J., et al. 2003, ApJ, 584, 210Goudfrooij, P., Alonso, M. V., Maraston, C., & Minniti, D. 2001, MNRAS, 328,237Haines, C. P., Gargiulo, A., La Barbera, F., et al. 2007, MNRAS, 381, 7Haines, C. P., Pereira, M. J., Smith, G. P., et al. 2015, ApJ, 806, 101Haines, C. P., Pereira, M. J., Smith, G. P., et al. 2013, ApJ, 775, 126Hatt, D., Freedman, W. L., Madore, B. F., et al. 2018, ApJ, 866, 145Hess, K. M., Cluver, M. E., Yahya, S., et al. 2017, MNRAS, 464, 957Horellou, C., Black, J. H., van Gorkom, J. H., et al. 2001, A&A, 376, 837Ianjamasimanana, R., Namumba, B., Ramaila, A. J. T., et al. 2020, MNRAS,497, 4795Iodice, E., Capaccioli, M., Grado, A., et al. 2016, ApJ, 820, 42Iodice, E., Spavone, M., Capaccioli, M., et al. 2017, ApJ, 839, 21Jarrett, T. H., Comrie, A., Marchetti, L., et al. 2020, arXiv e-prints,arXiv:2012.10342Johnson, M. C., Kamphuis, P., Koribalski, B. S., et al. 2015, MNRAS, 451, 3192Jonas, J. 2016, Proceedings of MeerKAT Science: On the Pathway to the SKAJózsa, G. I. G., White, S. V., Thorat, K., et al. 2020, in ASP Conf. Ser., Vol. 527,ADASS XXIX, ed. R. Pizzo, E. Deul, J.-D. Mol, J. de Plaa, & H. Verkouter,San Francisco, 635–638Kenney, J. D. P., Tal, T., Crowl, H. H., Feldmeier, J., & Jacoby, G. H. 2008, ApJ,687, L69Kenyon, J. S., Smirnov, O. M., Grobler, T. L., & Perkins, S. J. 2018, MonthlyNotices of the Royal Astronomical Society, 478, 2399Kleiner, D., Koribalski, B. S., Serra, P., et al. 2019, MNRAS, 488, 5352Koribalski, B. S. 2012, PASA, 29, 359Koribalski, B. S., Staveley-Smith, L., Kilborn, V. A., et al. 2004, AJ, 128, 16Kreckel, K., Platen, E., Aragón-Calvo, M. A., et al. 2012, AJ, 144, 16Lanz, L., Jones, C., Forman, W. R., et al. 2010, ApJ, 721, 1702Lewis, I., Balogh, M., De Propris, R., et al. 2002, MNRAS, 334, 673Liu, Y., Peng, E. W., Jordán, A., et al. 2019, ApJ, 875, 156Longobardi, A., Boselli, A., Fossati, M., et al. 2020, A&A, 644, A161Loni, A., Serra, P., Kleiner, D., et al. 2021, arXiv e-prints, arXiv:2102.01185Maccagni, F. M., Murgia, M., Serra, P., et al. 2020, A&A, 634, A9Mackie, G. & Fabbiano, G. 1998, AJ, 115, 514Maddox, N., Serra, P., Venhola, A., et al. 2019, MNRAS, 490, 1666Mahajan, S. 2013, MNRAS, 431, L117Mahajan, S., Raychaudhury, S., & Pimbblet, K. A. 2012, MNRAS, 427, 1252Marchetti, L., Jarrett, T. H., Comrie, A., et al. 2020, arXiv e-prints,arXiv:2012.11553Mauch, T., Cotton, W. D., Condon, J. J., et al. 2020, ApJ, 888, 61McKinley, B., Yang, R., López-Caniego, M., et al. 2015, MNRAS, 446, 3478Meyer, M. J., Zwaan, M. A., Webster, R. L., et al. 2004, MNRAS, 350, 1195Moretti, A., Paladino, R., Poggianti, B. M., et al. 2020, ApJ, 897, L30Morokuma-Matsui, K., Serra, P., Maccagni, F. M., et al. 2019, PASJ, 71, 85Müller, A., Poggianti, B. M., Pfrommer, C., et al. 2020, Nature Astronomy[ arXiv:2009.13287 ] Nelson, D., Sharma, P., Pillepich, A., et al. 2020, MNRAS, 498, 2391Nulsen, P. E. J. 1982, MNRAS, 198, 1007O ff ringa, A. R., McKinley, B., Hurley-Walker, N., et al. 2014, MNRAS, 444,606O ff ringa, A. R. & Smirnov, O. 2017, MNRAS, 471, 301O ff ringa, A. R., van de Gronde, J. J., & Roerdink, J. B. T. M. 2012, A&A, 539,A95Peng, Y.-j., Lilly, S. J., Kovaˇc, K., et al. 2010, ApJ, 721, 193Poggianti, B. M., Moretti, A., Gullieuszik, M., et al. 2018, ApJ, 853, 200Porter, S. C., Raychaudhury, S., Pimbblet, K. A., & Drinkwater, M. J. 2008,MNRAS, 388, 1152Putman, M. E., Bureau, M., Mould, J. R., Staveley-Smith, L., & Freeman, K. C.1998, AJ, 115, 2345Raj, M. A., Iodice, E., Napolitano, N. R., et al. 2020, A&A, 640, A137Raj, M. A., Iodice, E., Napolitano, N. R., et al. 2019, A&A, 628, A4Ramatsoku, M., Murgia, M., Vacca, V., et al. 2020, A&A, 636, L1Rasmussen, J., Bai, X.-N., Mulchaey, J. S., et al. 2012, ApJ, 747, 31Rasmussen, J., Ponman, T. J., Verdes-Montenegro, L., Yun, M. S., & Borthakur,S. 2008, MNRAS, 388, 1245Richtler, T., Hilker, M., Kumar, B., et al. 2014, A&A, 569, A41Roberts, I. D. & Parker, L. C. 2017, MNRAS, 467, 3268Robotham, A. S. G., Norberg, P., Driver, S. P., et al. 2011, MNRAS, 416, 2640Schawinski, K., Urry, C. M., Simmons, B. D., et al. 2014, MNRAS, 440, 889Schröder, A., Drinkwater, M. J., & Richter, O. G. 2001, A&A, 376, 98Schweizer, F. 1980, ApJ, 237, 303Serra, P., de Blok, W. J. G., Bryan, G. L., et al. 2016, in MeerKAT Science: Onthe Pathway to the SKA, 8Serra, P., Jurek, R., & Flöer, L. 2012, PASA, 29, 296Serra, P., Maccagni, F. M., Kleiner, D., et al. 2019, A&A, 628, A122Serra, P., Westmeier, T., Giese, N., et al. 2015, MNRAS, 448, 1922Seth, R. & Raychaudhury, S. 2020, MNRAS, 497, 466Steinhauser, D., Schindler, S., & Springel, V. 2016, A&A, 591, A51Taylor, E. N., Hopkins, A. M., Baldry, I. K., et al. 2011, MNRAS, 418, 1587Theureau, G., Bottinelli, L., Coudreau-Durand, N., et al. 1998, A&AS, 130, 333Venhola, A., Peletier, R., Laurikainen, E., et al. 2018, A&A, 620, A165Venhola, A., Peletier, R., Laurikainen, E., et al. 2019, A&A, 625, A143Verdugo, M., Ziegler, B. L., & Gerken, B. 2008, A&A, 486, 9Vulcani, B., Poggianti, B. M., Ja ff é, Y. L., et al. 2018, MNRAS, 480, 3152Weinmann, S. M., van den Bosch, F. C., Yang, X., & Mo, H. J. 2006, MNRAS,366, 2Westmeier, T., Braun, R., & Koribalski, B. S. 2011, MNRAS, 410, 2217Wetzel, A. R., Tinker, J. L., & Conroy, C. 2012, MNRAS, 424, 232Woo, J., Dekel, A., Faber, S. M., et al. 2013, MNRAS, 428, 3306Yagi, M., Yoshida, M., Gavazzi, G., et al. 2017, ApJ, 839, 65 Article number, page 16 of 18leiner et al.: Pre-processing in the Fornax A group
Appendix A: H α image comparison In Fig.A.1, we show the H α image after the standard reductionand the image we used that modelled and subtracted the back-ground of the original image using a median filter. The giantH α filaments can be seen in the original image, however, it isdominated by the over- and under-subtracted artefacts and has avariable background.The success of our median smoothing and model backgroundsubtraction is dependent on how well the real H α is masked.Anything that is included in the mask, by definition, is includedin the final image. This is especially challenging for di ff use H α emission located in areas with high background noise. The con-verse is also true: if spurious H α emission is included in themask, it will also be in the final image.To mitigate these issues as best as possible, we used a con-servative approach to carefully mask the real H α that was clearlyvisible in the original image. It is particularly di ffi cult to maskreal H α emission in areas with a highly variable background andwhere the background is significantly under subtracted. The re-sult is that some of the di ff use H α emission is lost and not repro-duced in the final image. As this is an iterative process, we wereable to recover H α emission in the most over-subtracted regionsof the image. Even though we cannot conserve 100% of the H α emission in this process, the purpose of this is to present the un-derlying structure of the new, giant H α filaments detected in theIGM. Article number, page 17 of 18 & A proofs: manuscript no. MeerKAT_FornaxA . . . . . . . . . . . . . . . . . . h m s m s s m s s − ◦ RA (J2000) D ec ( J ) − . − . − . . . . . H α S u r f a ce B r i g h t n e ss ( − e r g s c m − s − a r c s ec − ) h m m m − ◦ RA (J2000) D ec ( J ) − . − . − . . . . . H α S u r f a ce B r i g h t n e ss ( − e r g s c m − s − a r c s ec − ) Fig. A.1.
Comparison of the original and filtered H α images. Top image : H α image after the standard data reduction process. Bottom image :H α image we present in our work that iteratively modelled and subtracted (described in section 3.2) the background of the original image. Bothimages are presented on the same scale. The original image is clearly dominated by over- and under-subtracted artefacts, while the new image hasa smooth and uniform background, which retains the majority of the real H α emission. Some di ff use H α emission is lost in this process, however,the new image is a significant improvement that shows the underlying structure of the giant H αα