1. Introduction
During the hierarchical formation of structure, galaxy clusters grow through constant infall of galaxies from surrounding filaments, the accretion of groups, and merging with other clusters (Sarazin Reference Sarazin, Feretti, Gioia and Giovannini2002). Among these accretion mechanisms, major mergers between clusters of roughly equal mass are the most energetic events (
${\sim 10^{63-64}}$
erg) since the Big Bang (Markevitch et al. Reference Markevitch, Forman, Sarazin and Vikhlinin1998). A significant fraction of galaxy clusters at
$z \lt 1$
are still undergoing mergers (Jones & Forman Reference Jones and Forman1999; Mann & Ebeling Reference Mann and Ebeling2012), raising the key question of how such events affect the constituent galaxies.
In the local universe, we have a good understanding of the link between galaxy environments and galaxy properties such as morphology (Dressler 1980) and star formation (SF) activity (Lewis et al. Reference Lewis2002; von der Linden et al. 2010; Barsanti et al. Reference Barsanti2018). Clusters are predominantly populated by red, passive early-type galaxies, whereas blue, late-type star-forming galaxies (SFGs) are more commonly found in low-density environments. A significant fraction of cluster galaxies may already be partially/fully quenched prior to their infall into cluster environments. This quenching is known as pre-processing and can contribute to observed quenched galaxy fractions out to 5
$R_{200}$
(McGee et al. Reference McGee, Balogh, Bower, Font and McCarthy2009; Haines et al. Reference Haines2015; Rhee et al. Reference Rhee2020; Piraino-Cerda et al. Reference Piraino-Cerda2024). Within
$R_{200}$
, cluster-specific mechanisms can be divided into two main classes of quenching mechanisms, namely gravitational and hydrodynamical interactions.
The gravitational mechanisms, such as galaxy mergers (Toomre & Toomre Reference Toomre and Toomre1972), harassment by nearby galaxies and/or the cluster potential (Moore et al. Reference Moore, Katz, Lake, Dressler and Oemler1996), tidal stripping (Moore et al. Reference Moore1999), and strangulation (Larson, Tinsley, & Caldwell Reference Larson, Tinsley and Caldwell1980) can alter both galaxy morphology and SF activity. The hydrodynamical processes directly act on the gas component of the cluster galaxies through interactions between the hot (
$T \sim 10^7$
K) intracluster medium (ICM) and the interstellar medium (ISM) – ram-pressure stripping (RPS; Gunn & Gott Reference Gunn and Gott1972), viscous stripping (Nulsen Reference Nulsen1982), and thermal evaporation (Cowie & Songaila Reference Cowie and Songaila1977). Simulations indicate that RPS can trigger SF within cluster galaxies by compressing the gas (Fujita & Nagashima Reference Fujita and Nagashima1999; Bekki & Couch Reference Bekki and Couch2003; Bekki, Owers, & Couch Reference Bekki, Owers and Couch2010; Roediger et al. Reference Roediger, Bruggen, Owers, Ebeling and Sun2014) and then quench galaxies by removing the gas that would otherwise be available for future SF. RPS may occur rapidly (Domainko et al. Reference Domainko2006; Kapferer et al. Reference Kapferer, Sluka, Schindler, Ferrari and Ziegler2009) or gradually over a longer timescale (aka. strangulation; Larson et al. Reference Larson, Tinsley and Caldwell1980). Unlike the gravitational interactions, hydrodynamical mechanisms have little effect on the overall stellar distribution (Cortese, Catinella, & Smith Reference Cortese, Catinella and Smith2021).
Compared to relaxed clusters (i.e. virialised), merging clusters harbour more extreme environments. During core passage, the relative speed of the merging haloes can reach up to 4 500
$\,\textrm{km s}^{-1}$
(Sarazin Reference Sarazin, Feretti, Gioia and Giovannini2002; Markevitch et al. Reference Markevitch2002; Markevitch et al. Reference Markevitch2004), which generates high-speed shocks travelling through the ICM. Because the pressure of the ICM in the region of the shock is higher than the overall ICM pressure, RPS could favour SF more easily (Bekki & Couch Reference Bekki and Couch2003; Bekki et al. Reference Bekki, Owers and Couch2010; Roediger et al. Reference Roediger, Bruggen, Owers, Ebeling and Sun2014; Ruggiero et al. Reference Ruggiero2019). Therefore, additional impacts on galaxy properties are expected (post-processing; Vijayaraghavan & Ricker Reference Vijayaraghavan and Ricker2013). That said, contradictory results have been presented by authors on whether cluster mergers promote or quench SF. The most important open problems in the field are discussed in the following paragraphs.
Do cluster mergers trigger star formation? Increased global SF activity in merging clusters has been widely reported. Clusters with multiple components, indicative of ongoing dynamical activity, show higher fractions of SFGs relative to the single-component clusters (Cohen et al. Reference Cohen, Hickox, Wegner, Einasto and Vennik2014; Cohen, Hickox, & Wegner Reference Cohen, Hickox and Wegner2015; Yoon & Im Reference Yoon and Im2020). Similarly, Aldás et al. (Reference Aldás, Gómez, Vega-Martnez, Zenteno and Carrasco2024), based on clusters selected in the Illustris TNG simulation, found that disturbed clusters host more blue and SFGs than their relaxed counterparts. Stroe et al. (Reference Stroe2017) found an overdensity of H
$\alpha$
emitters in clusters hosting radio shocks that are indicators of ongoing merger activity. On smaller scales, mergers appear to enhance SF as well. Cortese et al. (Reference Cortese, Gavazzi, Boselli, Iglesias-Paramo and Carrasco2004) reported three infalling, highly star-forming subgroups in the young binary cluster A1367. Galaxies between merging subclusters have been observed to exhibit SF and nuclear activity (Ferrari et al. Reference Ferrari, Benoist, Maurogordato, Cappi and Slezak2005; Hwang & Lee Reference Hwang and Lee2009). Owers et al. (Reference Owers, Couch, Nulsen and Randall2012) identified multiple jellyfish galaxies near the merging features in A2744 as an indication of enhanced SF activity due to the merger. Stroe et al. (Reference Stroe2015) found unusually high star formation rates (SFR) and fractions of H
$\alpha$
emitters in the region affected by merger-induced shocks in the young post-merger Sausage cluster, where the SFR density is comparable with that seen at cosmic noon (
$z\sim$
2). In the multiple cluster system A901/2, Roman-Oliveira et al. (Reference Roman-Oliveira2019) identified 70 jellyfish galaxy candidates with specific star formation rates (sSFR) indicating an enhancement. Ruggiero et al. (Reference Ruggiero2019) conducted a simulation-based analysis on the same system, which showed jellyfish galaxies are preferentially formed near the leading edges of interacting halos due to the boosted ram pressure. Finally, a recent study by Lourenço et al. (Reference Lourenço2023) reported a mild, yet marginally significant, enhancement in the RPS candidates in interacting clusters in the Wide-Field Nearby Galaxy-Cluster Survey (WINGS; Cava et al. Reference Cava2009) sample.
Do cluster mergers have any effect on star formation? Interestingly, merging clusters are also found to be efficient in suppressing SF activity. Numerical simulations reveal an increase in gas mass loss due to RPS during major mergers, leading to weakening/quenching of SF activity (Fujita et al. Reference Fujita, Takizawa, Nagashima and Enoki1999; Domainko et al. Reference Domainko2006; Kapferer et al. Reference Kapferer, Sluka, Schindler, Ferrari and Ziegler2009). Passive spiral/disc galaxies were found in the centres of merging clusters (i.e. between their brightest cluster galaxies, BCGs) by Pranger et al. (Reference Pranger2014) and Kelkar et al. (Reference Kelkar2020), suggesting merger-driven quenching. Deshev et al. (Reference Deshev2017) found neither recent nor ongoing SF in members in the core of the merger A520, suggesting a scenario of rapid quenching without a starburst episode. In merging Illustris TNG clusters, Li et al. (Reference Li2023) concluded that merger-induced shocks lead to rapid quenching of SF in low-mass satellites, which is also supported by Roberts (Reference Roberts2024) for nine nearby clusters hosting radio relics. There are also studies finding no significant enhancement in SF or quenching in merging systems (Kleiner et al. Reference Kleiner, Pimbblet, Owers, Jones and Stephenson2014; Wittman et al. Reference Wittman, Imani, Olden and Golovich2024; Astudillo et al. Reference Astudillo, Carrasco, Castellón, Zenteno and Cuevas2025).
Across previous studies, a bimodal approach emerges: studies either focusing on a single cluster with well-defined dynamics (Owers et al. Reference Owers, Couch, Nulsen and Randall2012) or on a large cluster sample (Cohen et al. Reference Cohen, Hickox, Wegner, Einasto and Vennik2014), where dynamical activity is less well-characterised. In this paper, we use a sample of four clusters from the SAMI Galaxy Survey (hereafter SAMI-GS; Bryant et al. Reference Bryant2015) where dynamical states are well-characterised. We therefore adopt an approach that is a compromise between the two that have been used so far. This enables a homogeneous analysis of galaxy populations in clusters that have a range of dynamical conditions.
The paper is organised as follows: Sections 2 and 3 introduce the cluster and galaxy samples, and the spectroscopic data used in this paper. In Section 4, we outline the spectroscopic classification scheme to define the galaxy populations across the sample. Section 5 presents our results, and Section 6 interprets them in the context of previous studies and simulations. Section 7 highlights the key findings in this study. Throughout this paper, we adopt a flat
$\Lambda$
CDM cosmology with
$H_0 = 70$
km s
$^{-1}$
Mpc
$^{-1}$
,
$\Omega_{m} = 0.3,$
$\Omega_\Lambda = 0.7$
.
2. Cluster sample
In this study, we use a sample of clusters from the SAMI-GS (Bryant et al. Reference Bryant2015). As outlined in Owers et al. (Reference Owers2017), the sample contains eight low-redshift clusters – namely A85, A119, A168, A2399, A3880, A4038, APMCC 0917, and EDCC 0442- spanning a redshift range of
$0.029\lt z \lt 0.058$
, selected from the catalogue of De Propris et al. (Reference De Propris2002) in the 2
$^\circ$
Field Galaxy Redshift Survey (2dFGRS; Colless et al. Reference Colless2001) fields and the Cluster Infall Regions in the Sloan Digital Sky Survey, herefter SDSS, (CIRS; Rines & Diaferio Reference Rines and Diaferio2006). The clusters were chosen to fall within the mass range of
$\log(M_{200}/M_\odot) = 14.25 - 15.19$
, yet they show differences in their dynamical states. This selection enables us to study how the different merger stages of clusters influence the galaxy properties. Consistent with this scientific motivation, we use a subsample containing four clusters – two merging (A168, A2399) and two relaxed clusters (A3880, EDCC 0442). Section 2.1 briefly summarises the literature on the dynamical states of the individual clusters in the sample.
2.1 Notes on individual clusters
A168
$(z=0.0449)$
is a well-studied, off-axis cluster merger occurring nearly on the plane of the sky (Ulmer, Wirth, & Kowalski Reference Ulmer, Wirth and Kowalski1992; Yang et al. Reference Yang2004a, b; Hallman & Markevitch Reference Hallman and Markevitch2004; Hwang & Lee Reference Hwang and Lee2009; Fogarty et al. Reference Fogarty2014; Owers et al. Reference Owers2017; Dwarakanath et al. Reference Dwarakanath, Parekh, Kale and George2018). As shown in the top left panel of Figure 1, it exhibits elongated X-ray emission with two peaks separated by
$\sim$
700 kpc in projection, aligned along the NW-SE merger axis. The X-ray peaks are centred on the BCG of each subcluster and are offset from the galaxy density peak. Yang et al. (Reference Yang2004a, b) infer a crossing time-scale of
$\sim$
0.6 Gyr based on the spatial separation, indicating an early-stage post-merger with a mass ratio, estimated via simulations, between 1:1 and 1:3. Hallman & Markevitch (Reference Hallman and Markevitch2004) identified a cold front situated ahead of the northern peak, pointing to a more evolved merger stage. Hwang & Lee (Reference Hwang and Lee2009) used a linear two-body model to refine the merger timescale. Combining the observational indicators available – only in X-ray without existing radio observation – at that time, they favoured a scenario where core-passage occurred
$\sim$
5 Gyr ago (for our adopted cosmology) and the subclusters are approaching for a second passage. A recent study by Dwarakanath et al. (Reference Dwarakanath, Parekh, Kale and George2018) identified twin radio relics near the northern X-ray peak, indicating a merger-induced shock, with an estimated merger timescale of
$\sim$
0.5 Gyr.
Figure 1. Legacy Survey imaging (Dey et al. Reference Dey2019) of the cluster sample used in this study, overlaid with X-ray emission (cyan contours) from XMM-Newton for A2399 and EDCC 0442 and Chandra for A168 and A3880. The top and bottom panels highlight the merging and relaxed clusters, respectively, with names and redshifts written in the upper left of each panel. In the top left panel, the gold contours show radio emission (170–231 MHz) from the GLEAM survey (Hurley-Walker et al. Reference Hurley-Walker2017). The lime crosses and gold diamonds denote the BCGs and the adopted centres of each system, respectively. As shown by the arrows in the top left plot, the orientation is North up, East to the left, and this applies to all clusters. It is evident that X-ray and radio properties of merging clusters differ markedly from those of the relaxed clusters, highlighting the ongoing dynamical activity.
A2399
$(z=0.0580)$
is a dynamically young system exhibiting low surface brightness and strongly bimodal X-ray emission with peaks
$\sim$
600 kpc apart (Böhringer et al. Reference Böhringer2007, Reference Böhringer2010; Owers et al. Reference Owers2017) as presented in the top right panel of Figure 1. The X-ray peaks are coincident with the BCGs and the galaxy density peaks (Fogarty et al. Reference Fogarty2014). Moreover, Fogarty et al. (Reference Fogarty2014) identified extended X-ray emission (referred to as the ‘plume’ in Fogarty et al. Reference Fogarty2014) extending northward from the western X-ray peak, indicating ongoing merging activity. Substructure analyses of A2399 reveal contradictory results. Using WINGS data, Flin & Krywult (Reference Flin and Krywult2006), Ramella et al. (Reference Ramella2007), and Moretti et al. (2017) could not detect any substructures in A2399. In contrast, Owers et al. (Reference Owers2017) detected the presence of substructures through multidimensional (i.e. velocity, spatial, and combined) analysis. One of these substructures is coincident with the western X-ray structure and was also detected by Lourenço et al. (Reference Lourenço2020) via 3D (i.e. spatial+velocity) analysis. The X-ray properties of A2399 reveal contact discontinuities in both temperature and brightness profiles, as well as a high central entropy and metallicity (Mitsuishi et al. Reference Mitsuishi2018; Lourenço et al. Reference Lourenço2020). However, the former authors interpret these features as evidence of a cold front, whereas the latter authors consider the feature as a shock front. Regardless, both suggest that A2399 might have undergone a recent head-on collision (i.e. post-core-passage phase) on the plane of the sky along the NE-SW direction. Yet, neither study considers the plume identified in Fogarty et al. (Reference Fogarty2014) in their merger scenario. In this paper, we consider an alternative merger scenario where the plume is a result of RPS of gas from the western substructure. The offset of the plume to the NW indicates that the western clump is travelling towards the SE. This scenario suggests that A2399 is a pre-merger system right before the first core passage.
In contrast, A3880
$(z=0.0578)$
and EDCC 0442
$(z=0.0498)$
show no evidence of ongoing merger activity (Owers et al. Reference Owers2017; Lourenço et al. Reference Lourenço2023), having relatively regular, symmetric X-ray emission detected with Chandra and XMM-Newton, respectively, well centred with the BCGs as illustrated in the bottom row of Figure 1.
3 Data and galaxy sample
3.1 Spectroscopic data
In this study, we exploit the single-fibre spectroscopy from the SAMI Cluster Redshift Survey (hereafter SAMI-CRS; Owers et al. (Reference Owers2017)) and SDSS-III (Ahn et al. Reference Ahn2014; Alam et al. Reference Alam2015). The SAMI-CRS was conducted as a part of the SAMI-GS. The survey gathered spectra using the 2dF/AAOmega multi-object spectrograph mounted at the prime focus of the 3.9-m Anglo-Australian Telescope over seven nights in 2013. The 2dF instrument contains 392 fibres, each with 2 arcsec diameter, spread over a two-degree field of view. These fibres feed light to the AAOmega dual-beam spectrograph (Sharp et al. Reference Sharp, McLean and Iye2006). The SAMI-CRS targets were observed with the low-resolution 580V and 385R gratings for the blue and red arms, respectively, with a 5 700 Å dichroic beam splitter. The setup covers a wavelength range of 3 700–5 850 Å in the blue arm and 5 600–8 800 Å in the red arm, with 3.53 and 5.32 Å full width at half-maximum (FWHM) resolution for the blue and red arms, respectively.
In total, 22 066 spectra were taken during the SAMI-CRS. The details of the redshift measurements, the quality assessment (i.e. accuracy and precision), and duplicate analysis are given in Owers et al. (Reference Owers2017). Based on the quality measurements given in Owers et al. (Reference Owers2017), we only consider the SAMI-CRS spectra with reliable redshift measurements (i.e.
$\textrm{confidence}\gt0.9$
), yielding 7 648 spectra including duplicates for our subsample (out to
$\sim 4.5 R_{200}$
). SDSS provides 492 spectra for A168 and A2399 out to similar cluster-centric radii.
3.2 Galaxy sample
The SAMI-CRS provides a high completeness in redshifts measurements (
$\sim$
94%) within 2
$R_{200}$
for Petrosian r-band magnitudes
$r_\textrm{petro}\leq19.4$
(Owers et al. Reference Owers2017), by combining AAOmega observations and a compilation of pre-existing redshifts from other surveys, such as SDSS (Ahn et al. Reference Ahn2014), 2dFGRS (Colless et al. Reference Colless2001), 6dFGS (Jones et al. Reference Jones2009), WINGS (Cava et al. Reference Cava2009), CAIRNS (Rines et al. Reference Rines, Geller, Kurtz and Diaferio2003), NFPS (Smith et al. Reference Smith2004), ENACS (Katgert et al. Reference Katgert1996), and a catalogue for A85 from Durret et al. (Reference Durret, Felenbok, Lobo and Slezak1998). Cluster membership is determined via an iterative procedure based on these redshifts and the spatial positions of galaxies as described in Owers et al. (Reference Owers2017). The global cluster properties (i.e.
$\sigma_{200}, R_{200}$
) and the dynamical characterisation (e.g. presence of any substructures) are determined considering these confirmed members. This procedure (detailed in Owers et al. Reference Owers2017) results in a membership catalogue containing 1 167 cluster galaxies for A168, A2399, A3880, and EDCC 0442 out to 2
$R_{200}$
.
Based on the member galaxies, we select our galaxy sample for the analysis outlined in Section 4 by applying two main criteria. As a first step, we only include members within 2
$R_{200}$
having either SAMI-CRS spectra with reliable redshifts (i.e.
$\textrm{confidence}\gt0.9$
) or SDSS spectra, removing 27 galaxies. Another key consideration is probing the same stellar mass range across the sample, which required homogeneous photometric measurements for stellar mass estimation. The SAMI-CRS photometry comes from SDSS-DR9 (Ahn et al. Reference Ahn2012) and VST/ATLAS (Shanks et al. Reference Shanks2015), and the homogenisation procedure is outlined in Section 5 of Owers et al. (Reference Owers2017). There are 1103 members with available photometry. For these objects, the stellar masses are estimated using the empirical relation between (g-i) colour and (M/L)
$_i$
given by Taylor et al. (Reference Taylor2011), also outlined in Bryant et al. (Reference Bryant2015) and Owers et al. (Reference Owers2017) for apparent magnitudes. The colour-magnitude diagram shown in Figure 7 of Owers et al. (Reference Owers2017) and colour-mass diagram given in Figure 2 indicate that the low-mass regime is incomplete in terms of red sequence galaxies, especially for EDCC 0442. This could introduce a bias towards blue cloud galaxies. Therefore, we apply a cut on the stellar masses as
$\log(M_\ast/M_\odot)\geq9.25$
. This yields 829 galaxies –
$N_{\text{A168}}=193$
,
$N_{\text{A2399}}=241$
,
$N_{\text{A3880}}=211$
, and
$N_{\text{EDCC 0442}}=184$
.
Figure 2. Colour-mass diagram of members for each cluster. The red dots, cyan stars, and green squares represent the galaxies classified as ‘passive’, ‘star-forming’, and ‘H
$\delta$
-strong’ based on the classification scheme outlined in Section 4. The magenta dashed, the black solid, and the red dashed lines represent the mass limits for log
$(M_\ast/M_\odot)$
= 9.00, 9.25, and 9.50, respectively.
4. Spectroscopic classification
Cluster mergers that trigger SF in their resident galaxies are expected to produce a higher incidence of SFGs (Stroe et al. Reference Stroe2015) with respect to their relaxed counterparts. In the case of a boost in rapid quenching of SF activity, a notable fraction of recently quenched galaxies with strong H
$\delta$
signatures is expected (Couch & Sharples Reference Couch and Sharples1987; Poggianti et al. Reference Poggianti1999; Owers et al. Reference Owers2019) to emerge in these environments. To address this, one needs to identify these populations by examining their spectral features. In this section, we outline the line measurements and the definition of populations through a spectroscopic classification scheme.
4.1 Emission line measurements
We measure emission line properties by employing the penalised PiXel Fitting code (pPXF; Cappellari & Emsellem Reference Cappellari and Emsellem2004; Cappellari Reference Cappellari2017). pPXF allows us to simultaneously fit the emission lines along with the stellar kinematics and stellar populations. In order to measure emission line properties accurately, the underlying stellar continuum must be modelled and subtracted. As recommended in Section 3.6 of Cappellari (Reference Cappellari2017) and implemented by van de Sande et al. (Reference van de Sande2017), Owers et al. (Reference Owers2019), and Oh et al. (Reference Oh2020), this can be achieved by iteratively running pPXF. The steps adopted here are described below:
-
• Initially, the spectrum and noise are fed into pPXF, excluding bad pixels. Instead of masking the emission lines, we also simultaneously fit gas templates to account for the contribution of the emission features to the continuum. We only use an additive polynomial with a degree of 12 (degree=12, mdegree=-1) to minimise mismatches between theoretical MILES templates (Sánchez-Blázquez et al. Reference Sánchez-Blázquez2006; Vazdekis et al. Reference Vazdekis2010; Falcón-Barroso et al. Reference Falcón-Barroso2011) and the spectrum. The original noise is then rescaled based on the residual between the spectrum and the best fit and used in the next step.
-
• The second run involves the removal of
$3\sigma$
outliers, using the CLEAN parameter. The best-fit solution for the stellar kinematics is kept as an initial guess for the third iteration. -
• In the final run, we use a 12th-degree multiplicative polynomial in place of the additive polynomial (degree=-1, mdegree=12).
Throughout the procedure, we assume the same kinematics for all the emission lines.
We measure the equivalent widths (EW) of H
$\alpha$
and [NII]
$\lambda$
6584 using the results from the pPXF procedure outlined above. Firstly, a continuum band is defined on either side of the H
$\alpha$
and [NII]
$\lambda$
6584 lines to determine the continuum level from the best-fit stellar continuum, and then H
$\alpha$
and [NII]
$\lambda$
6584 fluxes are divided by that continuum level to measure their EWs. A similar approach is also adopted for the absorption line measurements discussed in the next section.
4.2 Absorption line measurements
The H
$\delta$
absorption line equivalent width, EW(H
$\delta$
), is measured from the best-fit stellar continuum from pPXF in Section 4.1 using the procedure described in Cardiel et al. (Reference Cardiel, Gorgas, Cenarro and Gonzalez1998). We estimate EW(H
$\delta$
) and the associated uncertainties using a Monte Carlo (MC) routine, performing 1 000 iterations (i.e.
${N}_{\textrm{iteration}} = 1\,000$
) of resampling on the best-fit stellar continuum. For each iteration, we perturb the best-fit continuum by adding a random value sampled from a normal distribution using the noise spectrum as the standard deviation at each spectral pixel. Subsequently, absorption line properties are measured in the observed frame; equivalent width and its uncertainty are defined as the mean and standard deviation of the distribution of EW values measured in each MC iteration.
4.3 Classification scheme
Once both emission and absorption line properties are measured, we are able to conduct a spectral classification to define galaxy populations. For this purpose, we adopt a similar classification scheme to that outlined in Owers et al. (Reference Owers2019).
Emission – We define an emission spectrum as one in which at least two emission lines are detected with SNR
$\gt$
3: one must be a primary line (e.g. H
$\alpha$
or [NII]
$\lambda$
6584) and the second one must be one of H
$\alpha$
, [NII]
$\lambda$
6584, H
$\beta$
, [OIII]
$\lambda$
5007. Here, SNR is the ratio between the flux and its uncertainty (i.e.
$flux/\sigma_{flux}$
). We employ two additional criteria that help to remove spurious low-amplitude emission line detections that occur due to template mismatch during the pPXF fitting process. First, we only include those spectra where EW(H
$\alpha$
or [NII]
$\lambda$
6584)
$\gt$
1 Å. Second, we only include spectra with reliably measured gas kinematics, that is, where
$\sigma_{gas} \lt 500$
$\,\textrm{km s}^{-1}$
and
$|v_{gas}|\lt500$
$\,\textrm{km s}^{-1}$
.
The ionising source of emission line galaxies is classified as follows:
-
• For emission-line galaxies where all of the lines H
$\alpha$
, [NII]
$\lambda$
6584, H
$\beta$
, [OIII]
$\lambda$
5007 are detected, we use the BPT diagram (Baldwin, Phillips, & Terlevich Reference Baldwin, Phillips and Terlevich1981) and the populations are defined based on the demarcation line descriptions from Kewley et al. (Reference Kewley, Dopita, Sutherland, Heisler and Trevena2001) (Ke01) and Kauffmann et al. (Reference Kauffmann2003) (Ka03); the line ratios of SFGs (SF) lie below and left of the dividing line of Ka03, intermediate (INT) galaxies lie between that and the dividing line of Ke01, while non-SFGs (NSF) lie above and right of it. -
• For galaxies without H
$\beta$
and/or [OIII]
$\lambda$
5007, we use the WHAN diagram of Cid Fernandes et al. (Reference Cid Fernandes2010), in which emission line classification is done using the EW(H
$\alpha$
) and the ratio between [NII]
$\lambda$
6584 and H
$\alpha$
. The demarcation lines to separate SF, INT, and NSF galaxies are log([NII]
$\lambda$
6584/H
$\alpha$
) = −0.32 and log([NII]
$\lambda$
6584/H
$\alpha$
) = −0.1, which are approximations to the lines of Ka03 and Ke01, respectively. -
• Galaxies not presenting both primary lines are classified as SF if H
$\alpha$
is present. Otherwise, when only [NII]
$\lambda$
6584 is detected, they are classified as NSF. -
• Depending on the value of EW(H
$\alpha$
), sub-classification follows Cid Fernandes et al. (Reference Cid Fernandes2010). If EW(H
$\alpha$
)
$\lt$
3 Å, the galaxy can be classified as wSF or rINT, or rNSF. NSF spectra with 3 Å
$\lt$
EW(H
$\alpha$
)
$\lt$
6 Å are wNSF, while those with EW(H
$\alpha$
)
$\gt$
6 Å are sNSF. The prefixes ‘r/w/s’ denote ‘retired/weak/strong’.
Absorption – We consider the spectra that do not satisfy the criteria given above for absorption line classification. We classify them based on their absorption line properties, following the criteria from Owers et al. (Reference Owers2019) as follows:
-
• The spectra with
$SNR(H\delta) \gt 3$
, where
$SNR = |EW(H\delta)|/\sigma_{EW}$
, and
$EW(H\delta)\lt-3$
Å are classified as H
$\delta$
-strong (i.e. HDS). -
• If
$EW(H\delta)\gt-3$
Å, they are classified as having passive spectrum.
Passive versus star-forming classes – Combining all spectral information from both emission and absorption line diagnostics, we divide cluster galaxies into passive galaxies (PASG), SFGs, and H
$\delta$
-strong galaxies (HDSG). PASGs include galaxies with spectra classified as passive, r/w/sNSF, rINT, wSF; SFGs are the remaining combination of SF and INT galaxies; HDSGs are those labelled as HDS.
5. Results
5.1 Population fractions
As a first step towards understanding if the merger activity in A168 and A2399 is affecting the star-forming properties, we compare the fractions of different spectroscopic classes to those in the relaxed sample. If mergers are responsible for triggering SF, we expect a higher fraction of SFGs in merger-affected regions (Ferrari et al. Reference Ferrari, Benoist, Maurogordato, Cappi and Slezak2005; Stroe et al. Reference Stroe2015; Hwang & Lee Reference Hwang and Lee2009). Alternatively, if mergers are more effective at rapidly quenching galaxies, we expect a higher fraction of HDS galaxies to be found in merger regions. To that end, we define the fraction of SFGs,
$f_{SFG}$
, in a region as
\begin{equation} f_{SFG} = \frac{N_{SFG}}{N_{PASG} + N_{SFG} + N_{HDSG}},\end{equation}
where
$N_{PASG}$
,
$N_{SFG}$
, and
$N_{HDSG}$
denote the number of passive, star-forming, and H
$\delta$
-strong galaxies, respectively. We measure the fractions within two cluster-centric regions: (i)
${R} \leq {R}_{200}$
, where the impact of the mergers is expected to be greater, and (ii) the cluster outskirts (
$ 1 \lt {R/R}_{200} \leq 2$
), where they are expected to have had less effect. The outskirts may also offer insight into SF activity around merger regions, making them a great benchmark for comparing non-merger-affected regions across the sample.
Figure 3 and Table 1 show the fraction of SFGs within
$R_{200}$
and in the range
$1 \lt R/R_{200}$
$\lt$
2 for our cluster sample. The results for individual clusters are shown as open grey symbols. The blue stars and red squares show the combined results for the merging and relaxed clusters, respectively. Uncertainties represent
$1\sigma$
confidence intervals derived from the recipe of Cameron (Reference Cameron2011).
Figure 3. Star-forming galaxy fractions for different cluster-centric regions –
$R \lt R_{200}$
and
$1 \lt R/R_{200}$
$\lt$
2. The grey open symbols (i.e. pentagon, hexagon, thick cross, and diamond) show each cluster’s fraction (i.e. A168, A2399, A3880, and EDCC 0442, respectively). The blue stars and the red squares represent the stacked samples based on similar cluster dynamics (i.e. merging and relaxed, respectively). The error bars show the uncertainties on the fractions estimated through the approach given in Cameron (Reference Cameron2011).
Table 1. Population table for galaxies for each cluster and stacked cluster sample within different cluster-centric portions. Errors are estimated based on the recipe given by Cameron (Reference Cameron2011).
Figure 4. Spatial distribution of cluster galaxies. Columns represent individual clusters, while rows highlight all, passive (red) and star-forming galaxies (blue) from top to bottom, respectively. The
$16^{\rm th}$
,
$50^{\rm th}$
, and
$84^{\rm th}$
percentile contours are generated from the smoothed kernel density estimate (KDE) for the population highlighted in the lower left of each panel using ks package (Duong, Goud, & Schauer Reference Duong, Goud and Schauer2012) in R. As indicated by the black arrows in the top row, North is up and East is to the left.
When
$R \lt R_{200}$
is considered, the PASGs are the main population regardless of the cluster merger state. PASGs account for
$70.2^{+2.4}_{-2.6}$
% (229/326) of the merging sample, and
$75.2^{+2.7}_{-3.2}$
% (161/214) of the relaxed clusters. SFGs contribute
$29.1^{+2.6}_{-2.4}$
% (95/326) and
$24.3^{+3.1}_{-2.7}$
% (52/214) of the central galaxy population in merging and relaxed clusters, respectively. The
$f_{SFG}$
within
$R_{200}$
is higher in merging clusters (29.1%) compared to relaxed clusters (24.3%), although the difference, as estimated through a two-proportion Z-test, is not significant (
$\sim1.2\sigma$
). Across the whole sample, only 3 galaxies (i.e.
$\lesssim$
1%) present strong H
$\delta$
signatures, making up
$0.6^{+0.8}_{-0.2}$
% (2/326) and
$0.5^{+1.1}_{-0.1}$
% (1/214) of merging (only in A168) and relaxed clusters (only in A3880) for R
$\lt$
$R_{200}$
, respectively. Due to the low count, we hereafter combine HDSGs and PASGs.
In contrast, the trend seen within
$R_{200}$
is reversed at the outskirts (i.e.
$1 \lt R/R_{200}$
$\lt$
2) where the presence of SFGs becomes more pronounced. Merging and relaxed clusters show similar fractions of PASGs and SFGs without any HDSGs detected.
$42.6^{+4.8}_{-4.6}$
% (46/108) and
$40.9^{+3.7}_{-3.5}$
% (74/181) of galaxies are classified as PASGs, whereas SFGs account for
$57.4^{+4.6}_{-4.8}$
% (62/108) and
$59.1^{+3.5}_{-3.7}$
% (107/181) of outskirt galaxies for each sample, respectively.
5.2 The spatial distributions of passive and star-forming galaxies
In Section 5.1, we found hints that there is a higher
$f_{SFG}$
found within
$R_{200}$
for the merging clusters when compared to the relaxed sample. To determine if the higher
$f_{SFG}$
is related to the merger activity, in this section, we investigate the spatial distributions of galaxies across the sample by highlighting each spectral class.
Figure 4 illustrates the spatial distribution of members for each cluster. The red and blue points represent PASGs and SFGs, respectively. The positions of the BCGs are indicated by yellow crosses. From left to right, the columns show the distributions for A168, A2399, A3880, and EDCC 0442, respectively, while from top to bottom each row shows the distribution for one of the galaxy populations, being ‘all galaxies’ (i.e. PASGs+SFGs), ‘PASGs’, and ‘SFGs’. Contours represent the
$16{\rm th}, 50{\rm th}$
, and
$84{\rm th}$
percentiles of the distribution for the corresponding population. The contours are generated from the smoothed kernel density estimate (KDE) using the ks package (Duong Reference Duong2007) in R. Below, we describe the results for each cluster.
A168: The overall galaxy distribution is elongated along the merger axis, running NW to SE. This elongation is more pronounced for the PASGs that dominate the region within
$R_{200}$
. The elongation is likely due to unresolved substructures, whose presence is revealed by the X-ray data. In contrast, the SFGs do not share the NW-SE elongation that is seen in the PASGs. The distribution of SFGs is more centrally concentrated; the majority of the SFGs (
$\sim$
60% – 41/68) are located within
$R_{200}$
. There is a dense clump of SFGs just north of the southern BCG (yellow cross) in the region between substructures, consistent with findings of Hwang & Lee (Reference Hwang and Lee2009). Two galaxies classified as HDSGs are also found to be in the region between substructures.
A2399: The density peak coincides with two central BCGs. The
$50{\rm th}$
percentile KDE contour extends to the NW, indicating a local excess in the projected galaxy density at
$X=0.5, Y=0.5$
. The PASGs show a similar spatial distribution to the overall sample, although the elongation is more pronounced, and there is no evidence for a local excess at
$X=0.5, Y=0.5$
. The elongation of the PASG distribution indicates that there may be an unresolved substructure associated with the third BCG at
$X=0.3, Y=-0.1$
. As shown in Figure 1, this third BCG is associated with the X-ray structure. The SFGs present an asymmetric spatial distribution, with the main surface density peak located at
$(X, Y) =$
$\sim(0.5, 0.5) R_{200}$
.
A3880: As expected for a dynamically relaxed system, A3880 hosts a quite symmetric galaxy distribution. The main galaxy density peak is well-centred around the BCG. There is a secondary peak situated at the outskirts
$(R\sim 2R_{200}$
) to the SW, at
$X=1.25$
,
$Y=-1.25$
, likely associated with an infalling group. PASGs mainly populate the central region. The distribution of the SFGs is more extended when compared with those seen in A168 and A2399; the majority (
$\sim$
70%) of SFGs are found outside
$R_{200}$
. The only HDS galaxy in A3880 is found close to the centre.
EDCC 0442: The surface density distribution is elongated along a NE-SW axis, and the peak is coincident with the BCG. The distribution of passive galaxies is similar to the overall galaxy distribution, although the SW extension is more prominent in PASGs, indicating accretion from the surroundings. In contrast to A168, A2399, and A3880, SFGs are completely absent from the central
$\sim 0.5 R_{200}$
of EDCC 0442. Like A3880, the SFGs are predominantly located on the cluster outskirts.
We can conclude that merging clusters have a more mixed spatial distribution compared to the relaxed systems. PASGs reside in the cluster core regardless of the dynamical state. On the other hand, SFGs in the relaxed clusters have a relatively symmetric spatial configuration, predominantly occupying the outskirts, whereas merging systems host an SFG population that also populates the central
$R_{200}$
and exhibits an asymmetric distribution on the sky.
5.3 Radial distribution
In Section 5.2, we show qualitatively that there appears to be a difference in the radial distribution of SFGs for the merging and relaxed clusters. In order to quantify this result and determine the statistical significance, we look at the distribution of cluster-centric distances per population across the sample.
Figure 5 shows the distribution of galaxies as a function of normalised cluster-centric distances (
$R/R_{200}$
) across the sample. In each panel of Figure 5, the red and blue histograms and vertical lines show the distributions and their mean values calculated via biweight estimation (Beers, Flynn, & Gebhardt Reference Beers, Flynn and Gebhardt1990) for PASGs and SFGs, respectively. The uncertainties are estimated through bootstrapping. Regardless of the dynamical states, PASGs are localised in the central
$R_{200}$
with a tail towards larger cluster-centric distances. The mean
$R/R_{200}$
of the PASGs, shown as a solid red vertical line in each panel of Figure 5, ranges from
$\sim$
0.5 to 0.7
$R_{200}$
(A168 = 0.584, A2399 = 0.496, A3880 = 0.598, and EDCC 0442 = 0.665). The distributions show positive skewness values (A168 = 0.879, A2399 = 1.229, A3880 = 0.715, and EDCC 0442= 0.636). The number of PASGs per bin for
$R \gtrsim 0.5 R_{200}$
is relatively constant in the relaxed clusters A3880 and EDCC 0442, whereas the merging clusters A168 and A2399 present decreasing numbers of PASGs in the same cluster-centric portion.
Figure 5. Normalised cluster-centric distance distributions. The red and blue histograms highlight the distributions for PASG and SFGs, respectively. The vertical lines show the mean values for PASGs and SFGs, respectively, calculated via biweight estimation and given in the upper left of each panel alongside the biweighted dispersion and the skewness. The stripe density plots highlight exact values for each galaxy.
In order to quantify how significantly different the PASG distributions are across the sample, we apply the two-sample Kolmogorov–Smirnov (KS; Smirnov Reference Smirnov1939) and Anderson–Darling (AD; Anderson & Darling Reference Anderson and Darling1952) tests on a cluster-to-cluster basis. Both tests return higher p-values (i.e.
$p\gg0.05$
) for clusters with similar dynamical states (i.e. A168 vs A2399 and A3880 vs EDCC 0442). This means that we cannot reject the null hypothesis that both distributions are drawn from the same parent sample. In contrast, when we compare merging and relaxed clusters, the KS test suggests the opposite situation – p-values are
$\sim$
0.04,
$\sim$
0.09 (AD test returns 0.035) for A168 vs A3880 and A168 vs EDCC 0442, respectively, and
$p\lt0.05$
for both A2399 vs A3880 and A2399 vs EDCC 0442, respectively, indicating that two distributions are likely to be drawn from different parent samples.
In the case of SFGs, the
$R/R_{200}$
distributions notably differ between merging and relaxed clusters. Figure 5 shows that the number of SFGs in relaxed clusters increases with increasing cluster-centric distances. On the other hand, the number of SFGs in merging clusters is approximately constant or decreasing towards the outskirts. In particular, A2399 shows a peak in the number of SFGs around
$\sim 0.6 R_{200}$
. In contrast with that of the PASGs, the mean values of
$R/R_{200}$
distributions of the SFGs appear similar for clusters with similar dynamics;
$\sim$
0.86, and
$\sim$
0.88 for A168 and A2399, and
$\sim$
1.21 and
$\sim$
1.22 for A3880 and EDCC 0442, respectively. Moreover, the skewness values also highlight this difference between dynamical states. A168 and A2399 have positive skewness values as
$\sim$
0.35 and
$\sim$
0.24, respectively, whereas relaxed clusters present negative skewness values -
$\sim$
−0.42 and
$\sim$
−0.01 for A3880 and EDCC 0442, respectively.
To determine the significance of the differences in the
$R/R_{200}$
distributions of SFGs, we again use the KS-test. Comparing the merging clusters A168 and A2399, the KS-test returns a p-value of
$\sim$
0.28. Similarly, the KS-test returns a p-value
$\sim$
0.61 for the comparison of the relaxed clusters A3880 and EDCC 0442. Thus, there is no statistically significant difference between the
$R/R_{200}$
distributions for the two clusters within the relaxed sample, nor the two clusters within the merging sample. In contrast, comparing clusters across the merging and relaxed samples yields significant differences; each of the comparisons A168 vs A3880, A168 vs EDCC 0442, A2399 vs A3880 and A2399 vs EDCC 0442 returns a p-value
$\ll0.05$
. These results indicate that the distributions of
$R/R_{200}$
for SFGs in the merging clusters are significantly different from those in the relaxed clusters in our sample. Moreover, we test the consistency between the means of the
$R/R_{200}$
distributions of SFGs using both biweighted metrics and the original means with standard deviations, applying a two-sample t-test. For each merging–relaxed cluster pair, the difference between means exceeds
$\sim3.5\sigma$
, supporting the results from the KS test.
Lastly, if we stack the clusters based on their dynamical states, the difference in the
$R/R_{200}$
distributions for each galaxy population becomes more significant. The mean
$R/R_{200}$
values of PASGs are 0.52 and 0.52 for merging and relaxed clusters, respectively, yet the KS test returns a p-value of
$\sim$
0.002. Regarding SFGs, the mean values are 0.83 and 1.21 for merging and relaxed clusters, and the p-value from the KS test is
$\sim$
$10^{-7}$
.
5.4 Projected phase-space distribution
The position of a galaxy in projected phase-space (PPS) is determined using the projected cluster-centric distance and the line-of-sight velocity relative to the cluster redshift. The position in PPS is a useful diagnostic of cluster membership and also for segregating populations based on their accretion times (Haines et al. Reference Haines2012; Noble et al. Reference Noble2013; Oman, Hudson, & Behroozi Reference Oman, Hudson and Behroozi2013; Haines et al. Reference Haines2015; Owers et al. Reference Owers2017; Rhee et al. Reference Rhee2017; Owers et al. Reference Owers2019).
Figure 6 shows the projected phase-spaces for each cluster, highlighting each spectral class, where the red and blue points represent PASGs and SFGs, respectively. The positions of the BCGs are indicated by yellow crosses. As in Figure 4, each column represents an individual cluster, and each row shows a different population; the contours are also generated for
$16^{\rm th}$
,
$50^{\rm th}$
, and
$84^{\rm th}$
percentiles with the ks package.
Figure 6. Same as Figure 4, but for projected phase-space distribution.
Focusing on the top row of Figure 6, which shows the PPS for both PASGs and SFGs, there is a clear difference in the PPS distribution of galaxies in clusters with different dynamical states. Considering the
$50^{\rm th}$
percentile contour of the distribution, merging systems host galaxies mostly within
$R_{200}$
; relaxed systems have more extended galaxy distribution up to 2
$R_{200}$
– the distributions follow the trumpet shape of caustics observed in typical virialised systems (Kaiser Reference Kaiser1987; Regos & Geller Reference Regos and Geller1989; Diaferio & Geller Reference Diaferio and Geller1997). To test the significance of the differences across the sample, we use the multivariate two-sample KDE test introduced by Duong et al. (Reference Duong, Goud and Schauer2012). The KDE test is a non-parametric method in which the integrated mean square error is used to quantify the difference between two samples and has already been adopted to analyse PPS in the literature (Lopes, Ribeiro, & Rembold Reference Lopes, Ribeiro and Rembold2017; de Carvalho et al. Reference de Carvalho2017; Owers et al. Reference Owers2019; Costa et al. Reference Costa, Ribeiro, de Carvalho and Benavides2024). The KDE test returns p-values of 0.311, 0.013, 0.193, 0.004, 0.089, and 0.485 for A168 vs A2399, A168 vs A3880, A168 vs EDCC 0442, A2399 vs A3880, A2399 vs EDCC 0442, and A3880 vs EDCC 0442, respectively, also summarised in Table 2. The p-values are higher when clusters with similar dynamical states are compared, while they are lower when relaxed clusters are compared with merging clusters. However, A168 vs EDCC 0442 and A2399 vs EDCC 0442 result in
$p \gt 0.05$
. This is most likely due to a similar density peak (i.e. the contour of
$16^{\rm th}$
percentile) observed in the region between
$\sim 0.3 \lt R/R_{200}$
$\lt 0.7$
and
$\mid v/\sigma \mid \lt 0.5$
. For the stacked merger vs relaxed PPS comparison, the same analysis yields a lower p-value of 0.005.
Table 2. p-values, returned by the KDE test, for cluster-to-cluster comparisons of the projected phase-spaces of all galaxies.
The PPS distribution for PASGs, shown in the middle row of Figure 6, appears similar to the overall distribution shown in the top row. The PASGs mostly dominate smaller cluster-centric distances (and lower
$\mid v/\sigma \mid$
values) consistent with the so-called ‘virialised’ population (Oman et al. Reference Oman, Hudson and Behroozi2013; Rhee et al. Reference Rhee2017). However, the presence of this virialised region is more prominent for relaxed systems having a relatively symmetric velocity distribution around
$v/\sigma \sim 0$
and a clear deficit of SFGs in the same region. Moreover, there are a few PASGs at
$R \gt R_{200}$
for merging systems, while relaxed systems contain a noticeable passive population in the outskirts seen as a secondary peak – as an indication of an infalling population. Cluster-to-cluster comparisons return
$p \gt0.05$
, indicating that PASGs exhibit similar distributions in the PPS. This also holds for the stacked samples, for which the KDE test returns
$p = 0.124$
.
Table 3. Same as Table 2, but for SFGs.
The PPS distribution for the SFGs shown in the bottom row of Figure 6 reveals intriguing differences between the merging and relaxed clusters. For the relaxed systems, the SFGs are found at larger cluster-centric distances, as shown in Section 5.3, and have relatively symmetric distributions about the velocity axis. In contrast, the SFGs in the merging clusters are often found at small cluster-centric distances and show larger asymmetry in velocity. Comparisons between clusters reveal p-values from the KDE test of 0.089, 0.005, and 0.024 for A168 vs A2399, A168 vs A3880, and A168 vs EDCC 0442; 0.036, and 0.291 for A2399 vs A3880, and A2399 vs EDCC 0442; 0.374 for A3880 vs EDCC 0442, respectively (for
$R \lt 2R_{200}$
), also summarised in Table 3. These results highlight that clusters with similar merger states (i.e. A168 vs A2399 and A3880 vs EDCC 0442) demonstrate similarities in their PPS distribution among themselves. A168, as a post-merger cluster, presents an SFG phase-space distribution that is statistically different to A3880 and EDCC 0442. SFGs in A2399 show a similar distribution to EDCC 0442. However, their PPS distribution is inconsistent with the null hypothesis of being drawn from the same parent sample as A3880. In the case of analysis based on the stacked samples, the KDE test returns
$p = 0.014$
, inconsistent with the hypothesis that two samples are drawn from the same parent sample.
5.5 EW(H
$\alpha$
) distributions
In the previous sections, we showed that SFGs in mergers are more likely to be found in the region with ongoing dynamic activity. In this section, we investigate the mode of SF in order to understand if there is evidence for merger-related triggering or quenching of SF. To do this, we examine their distributions of EW(H
$\alpha$
) values, as tracers of ongoing SF, globally (Section 5.5.1) as well as by focusing on merger-specific regions (Section 5.5.2).
5.5.1 Global comparison
Figure 7 shows the distribution of log EW(H
$\alpha$
) for each cluster. Each red violin plot represents the distribution within
$R_{200}$
for a cluster, while the black violin plot shows the distribution for
$1\lt R/R_{200} \lt 2$
. Above each plot, the median and the standard deviation determined from the median absolute deviation (1.4826
$\times$
MAD) of the EW(H
$\alpha$
) distribution are given and coloured to match the violin distributions. The uncertainties on these values are estimated through bootstrapping. The solid and dashed lines indicate the
$50{\rm th}, 16{\rm th}$
and
$84{\rm th}$
percentiles. The stripes represent the individual values for each sample. The median values representing each distribution do not differ significantly across the sample compared to the dispersion.
We apply the two-sample KS and AD tests to test for differences between the distributions of EW(H
$\alpha$
) within
$R_{200}$
and on the outskirts. For
$R\lt R_{200}$
, comparing the EW(H
$\alpha$
) distributions of the merging and relaxed clusters individually, we only find a statistically significant difference between A168 and EDCC 0442, where the p-values are 0.006 (KS) and 0.022 (AD), respectively. If we stack the relaxed clusters, this becomes less significant with
$p_{KS} = 0.076$
and
$p_{AD} = 0.142$
. Comparisons of A168 versus A2399, and A3880 versus EDCC 0442 return p
$_{KS/AD}$
= 0.032/0.054 and 0.081/0.044, indicating marginally significant differences between these samples. As we look at the outskirts (i.e.
$R \gt R_{200}$
), no statistically significant differences (i.e.
$p\gg0.05$
) are found across the sample. We also compare the distributions between galaxies within
$R_{200}$
and the outskirts per cluster and find no statistically significant differences (
$p\gg0.05$
) either.
The KS and AD tests also reveal no significant difference between the stacked merging and relaxed samples for
$R\lt R_{200}$
and
$R\gt R_{200}$
, returning
$p\gg0.05$
. These results indicate that there is no statistically significant difference in the global star-forming properties probed by EW(H
$\alpha$
) between the relaxed and merging clusters.
5.5.2 A closer look into merger regions
As shown in Section 5.5.1, there is no clear evidence for global enhancement or quenching of SF in the merging clusters. To detect any possible merger-induced signal, we now focus on the galaxies situated in the vicinity of merger regions (e.g. between substructures) and features such as radio relics (i.e. shocks) to compare them with the remaining SFGs. We define subsamples of SFGs based on their spatial and/or PPS distribution.
-
• A168:
-
– ‘Clump’ galaxies (C): Galaxies located between substructures
$(N=12)$
. -
– ‘Shock’ galaxies (S): Galaxies identified near the shock/relic
$(N=7)$
. -
– ‘Rest’ galaxies (R): All remaining galaxies not in ‘C’ or ‘S’
$(N=50)$
.
-
-
• A2399:
-
– ‘Near-subcluster’ galaxies (NSC): Galaxies close to the western clump in the PPS distribution
$(N=15)$
. -
– ‘Between XP’ galaxies (B): Galaxies located between substructures (i.e. between X-ray peaks)
$(N=3)$
. -
– ‘Rest’ galaxies (R): All remaining galaxies not in ‘NSC’ or ‘B’
$(N=72)$
. -
– One galaxy (ID: 9239900059) is common to both ‘B’ and ‘NSC’ categories.
-
Figure 8 shows the log-scaled EW(H
$\alpha$
) distribution of SFGs in A168. The dark blue, magenta, and cyan histograms represent the galaxies in the clump, shock, and rest samples. The inset plot on the upper right highlights the spatial distribution of these populations by zooming into the central region; the red and orange contours display the X-ray and radio maps; the inset plot on the middle right shows their distribution in the PPS; the bottom panel displays the stripe density plot for each population with the same colour code. In the upper right corner, median and standard deviation determined from MAD are given per subsample, and the associated uncertainties are estimated in a similar fashion to that described previously. The median values are nearly identical, and the dispersions are similar (the conventional mean returns 1.10 dex, 1.26 dex, and 1.15 dex for the clump, the shock, and the rest). However, despite having a smaller sample size, shock galaxies extend towards the higher end of the distribution;
$\sim$
30% (2/7) are in the top 5% of EW(H
$\alpha$
) values of
$\sim$
69 Å (the highest), and
$\sim$
54 Å (third highest). The probability of 2 or more of the 7 lying in the 5% tail is 0.044. The galaxy with the highest EW(H
$\alpha$
) is located at the SE end of the shock, which suggests SF may be triggered by being overrun by the shock.
Figure 7. The EW(H
$\alpha$
) distributions. The red and black violin plots represent the distribution for
$R\lt R_{200}$
and
$1\lt R/R_{200}$
$\lt$
2, respectively, for each cluster. On top of each violin distribution, the median and the standard deviation estimated via median absolute deviation (1.4826
$\times$
MAD) of EW(H
$\alpha$
) distribution are given and coloured to match the violin distributions. The uncertainties on these values are estimated through bootstrapping. The solid and dashed lines indicate the
$50{\rm th}, 16{\rm th}$
and
$84{\rm th}$
percentiles. The stripes represent the individual values for the given samples.
Figure 8. EW(H
$\alpha$
) distribution of subsamples defined in A168. The dark blue, magenta, and cyan symbols represent ‘Clump’, ‘Shock’, and ‘Rest’ samples, respectively. The vertical lines mark the median of each subsample, which is given in the upper right of the figure with the standard deviation estimated via median absolute deviation (1.4826
$\times$
MAD) and uncertainties estimated via bootstrapping. The plot in the upper inset shows the spatial distribution of these samples close to the central region. The red and gold contours show the X-ray and radio emission from Figure 1. The symbol sizes are proportional to the EW(H
$\alpha$
); the larger the symbol, the higher the value. The lower inset shows the projected phase-space distributions. The stripes shown in the bottom panel represent the individual values for each sample.
To determine the significance of the difference between subsamples, we run internal and external comparisons. The clump and shock galaxies are first compared with the remaining SFGs (after excluding each subsample, respectively) in A168 for
$R\lt R_{200}$
and
$R\gt R_{200}$
and then with each other. For external comparisons, we select SFGs within the same cluster-centric regions for A2399 and the stacked relaxed sample. Through the KS and AD tests, we find no significant difference for any comparison, that is, all
$p\gg0.05$
.
Figure 9 shows the EW(H
$\alpha$
) distributions of the subsamples defined in A2399. The dark blue, magenta, and cyan histograms, vertical lines, and circles represent the ‘NSC’, ‘B’, and the ‘rest’ samples, respectively. The contours shown in the inset plot display the X-ray data. The medians and standard deviation values estimated from MAD are shown at the upper right for each subsample in the same colour and the uncertainties are estimated via bootstrapping. The samples of ‘NSC’ and ‘B’ show slightly higher values (1.30 dex and 1.47 dex) than the rest (1.26 dex). Yet, like A168, allowing for the dispersions, we find no significant differences between subsamples.
Figure 9. Same as Figure 8, but for A2399. The dark blue, magenta, and cyan symbols represent ‘NSC’, ‘B’, and ‘Rest’ samples, respectively. The vertical lines are the medians of each subsample and given in the upper right of the figure with the standard deviation estimated via median absolute deviation (1.4826
$\times$
MAD). The inset in the upper right highlights the central region, showing the spatial distribution of the samples in greater detail. The red contours are for the X-ray emission shown in Figure 1.
We conduct a similar set of comparisons for the subsamples defined in A2399. We only detect a significant difference between A2399’s NSC subsample and SFGs residing within
$R\lt R_{200}$
in A168 with p-values of 0.049 and 0.038 from the KS and AD tests, respectively, which might hint at a difference between their pre- and post-merger nature.
Lastly, we follow up on the significant difference found for A168 and A2399 by comparing the subsamples defined in both clusters. There is no significant difference revealed by either statistical test. Only the AD test yields a marginal difference with
$p = 0.071$
for the comparison between A168’s ‘C’ and A2399’s ‘B’ subsample.
Overall, our analysis of the distributions of EW(H
$\alpha$
) in the SFGs provides no evidence for SF triggered by the merger in the central regions of these clusters.
6. Discussion
The main goal of this study is to understand the interplay between the SF activity of cluster galaxies and their hosts’ assembly histories, in order to disentangle any additional impacts to those seen in the virialised/relaxed cluster environments. SF activity was traced through the
$f_{SFG}$
and the distributions of EW(H
$\alpha$
). While SFGs in merging clusters exhibit spatially and kinematically more mixed distributions, the EW(H
$\alpha$
) properties are similar to those of relaxed systems, leading us to a scenario of dynamical mixing rather than enhancement in SF activity. In this section, we discuss our interpretation in the context of the existing literature and from a simulation point of view, as well as the limitations of the method and data.
6.1 Comparison with the previous studies
Combining the results presented in Sections 5.1, 5.2, and 5.3, we are able to reproduce the well-established radial trend where the cluster centres are mainly dominated by PASGs, while SFGs are more common in the outskirts, agreeing with the previous studies (Dressler 1980; Biviano et al. Reference Biviano1997; Lewis et al. Reference Lewis2002; von der Linden et al. 2010; Barsanti et al. Reference Barsanti2018; Owers et al. Reference Owers2019). This implies that the trend is not significantly affected by the dynamical states of the clusters. A similar result is reported by Astudillo et al. (Reference Astudillo, Carrasco, Castellón, Zenteno and Cuevas2025) that the morphology/SFR-density relation still holds for unrelaxed clusters.
The mild increase found in
$f_{SFG}$
(
$\sim$
20%) for merging clusters is comparable with findings in Cohen et al. (Reference Cohen, Hickox, Wegner, Einasto and Vennik2014, Reference Cohen, Hickox and Wegner2015) and Yoon & Im (Reference Yoon and Im2020). They studied more than 100 nearby (
$z\lt0.2$
) SDSS clusters and found that clusters having substructures exhibit enhancement in the
$f_{SFG}$
(
$\sim$
20–30%). However, our signal is less prominent (
$\sim1.2\sigma$
) compared to theirs (
$\gtrsim2\sigma$
), likely due to the difference in sample sizes. Stroe et al. (Reference Stroe2017) studied 19 intermediate redshift clusters and 3 000 associated H
$\alpha$
emitters, and the merging clusters, traced by radio shocks, were found to host more H
$\alpha$
emitters relative to relaxed clusters. More recently, Aldás et al. (Reference Aldás, Gómez, Vega-Martnez, Zenteno and Carrasco2024) also reached similar conclusions based on Illustris TNG100 simulation. Accordingly, our result on
$f_{SFG}$
is qualitatively in agreement with the earlier works. Furthermore, the cluster outskirts present similar
$f_{SFG}$
, regardless of cluster dynamics, indicating that the outskirts are more resilient to dynamical state. This contradicts Aldás et al. (Reference Aldás, Gómez, Vega-Martnez, Zenteno and Carrasco2024), who found the majority of SFGs in the disturbed clusters reside in the outskirts (
$R\gt R_{200}$
).
Furthermore, we could not find any evidence of a recently quenched population as an indication of ongoing quenching. HDSGs make up a very small fraction of our sample, being represented by only 3 galaxies, which corresponds to
$\sim$
0.36% of the total sample (3/829). This is significantly lower than that found in WINGS clusters by Paccagnella et al. (2017) (i.e.
$\sim$
7.3%). Owers et al. (Reference Owers2019), exploiting the spatially resolved data from SAMI-GS, found 17 HDSGs for the SAMI cluster sample, considering galaxies only with
$\log(M_\ast/M_\odot)\gt10$
within
$R_{200}$
(
$N=579$
galaxies), resulting in a fraction of
$\sim$
3%. They estimated that no more than
$\sim$
0.5% would be recovered as H
$\delta$
-strong in single-fibre spectroscopy, agreeing with our HDSG fraction. Moreover, EDCC 0442, as a relaxed system, is the only cluster exhibiting a clear absence of SFGs within 0.5
$R_{200}$
. Compared to a strong deficit in SFGs found in post-merger A520 by Deshev et al. (Reference Deshev2017), we are not able to see any evidence for elevated quenching in merging systems.
To search for any merger-induced SF activity, we examined the EW(H
$\alpha$
) distributions for both global cluster environments and in the vicinity of merger features. Even though the global EW(H
$\alpha$
) properties show marginal cluster-to-cluster differences, no evident trend as a function of dynamical state is found due to the large scatter in the distributions. Figures 8 and 9 of Stroe et al. (Reference Stroe2017) show that, despite having a higher characteristic H
$\alpha$
density, merging clusters display similar H
$\alpha$
luminosities (proportional to SFR or EW(H
$\alpha$
) in our case) to relaxed clusters, in agreement with our result on global EW(H
$\alpha$
) properties. Moreover, Aldás et al. (Reference Aldás, Gómez, Vega-Martnez, Zenteno and Carrasco2024) show that the sSFR distribution within
$R_{200}$
is similar for relaxed and disturbed clusters, also complementing our result. Close-up investigation of the galaxies near the merger feature does not indicate any strong merger-induced activity for our merging clusters. For A168, Hwang & Lee (Reference Hwang and Lee2009) investigated the angular variance of galaxies between substructures (i.e. the ‘clump’ sample), and projection effects are unlikely to account for this clustering of SFGs, suggesting it is due to merger-induced activity. However, our EW(H
$\alpha$
) comparisons involving the ‘clump’ galaxies indicate no significant difference compared to the remaining SFGs (including the relaxed systems), so that there is no strong evidence for merger-related triggering of SF. On the other hand, galaxies near the shock in A168 present a wider range of EW(H
$\alpha$
). Two of them exhibit high EW(H
$\alpha$
), a hint of merger-induced activity, consistent with Owers et al. (Reference Owers, Couch, Nulsen and Randall2012) and Stroe et al. (Reference Stroe2015, Reference Stroe2017). Due to the low number statistics, this is not conclusive evidence for induced SF activity.
Regarding the origin of the mild enhancement in
$f_{SFG}$
in merging clusters across our sample, we suggest a dynamical mixing scenario (i.e. redistribution of galaxies) where merging clusters represent an intermediate stage of cluster formation, being fed by star-forming rich filaments/groups. This explanation is also preferred by Cohen et al. (Reference Cohen, Hickox and Wegner2015). Figure 10 of Owers et al. (Reference Owers2017) illustrates the probable substructures in A2399 in addition to the western cluster. The peak in SFGs shown in Section 5.2 and the SE extension of the ‘NSC’ subsample coincide with these substructures, indicating multiple ongoing coalescences. Given the fact that A2399 is a dynamically young pre-merger system sitting at the centre of the Aquarius-Cetus Supercluster (Bregman, Dupke, & Miller Reference Bregman, Dupke and Miller2004), it is expected to see the infall of SF-rich galaxies or groups. Similar results are found by Cortese et al. (Reference Cortese, Gavazzi, Boselli, Iglesias-Paramo and Carrasco2004) and Deshev et al. (Reference Deshev2017) for the merging systems A1367 and A520, respectively, where infalling substructures exhibit a higher
$f_{SFG}$
. An earlier study by Fujita et al. (Reference Fujita, Takizawa, Nagashima and Enoki1999) using numerical simulations showed that due to the high relative velocities, blue SFGs can reach the core region of the merging clusters before they quench, supporting the presence of SFGs in the cores of merging clusters.
6.2 Simulation perspective
To validate our interpretation for the enhancement in the incidence of SFGs in merging clusters from a simulation perspective, we utilise the Box2-hr volume from the Magneticum PathfinderFootnote
a
suite of hydrodynamical cosmological simulations. The simulations were run with a modified version of GADGET-2 (Springel Reference Springel2005) with various improvements (Dolag et al. Reference Dolag, Vazza, Brunetti and Tormen2005; Beck et al. Reference Beck2016). The large volume for Box2-hr, of
$(352\mathrm{Mpc}/h)^3$
, includes a significant number of galaxy clusters to compare with. For more details on the simulations, including particle resolution, structure identification, baryonic physics, galaxy cluster populations, and cluster galaxy populations, see Springel et al. (Reference Springel, Yoshida and White2001), Dolag et al. (Reference Dolag, Borgani, Murante and Springel2009), Teklu et al. (Reference Teklu2015), Remus et al. (Reference Remus, Dolag and Hoffmann2017), Lotz et al. (Reference Lotz, Dolag, Remus and Burkert2021).
Figure 10. Magneticum simulation results for relaxed (left) and merging (right) clusters. Top panel: The xy projected phase-space distributions. The inset plots show the 6D phase-space diagram using the radial component of the 3D velocity and the 3D radius. The red hexagonal bins show the passive galaxies; the redder the colour higher the density; the cyan stars represent star-forming galaxies for each plot. The black solid lines define the virialised regions (Rhee et al. Reference Rhee2017) for each panel. Middle panel: The median sSFR as a function of normalised cluster-centric distance. The blue solid line with stars represents the sSFR profile based on the xy projection (2D), while the cyan dashed line with stars is the same for the real (3D) space. The uncertainties are standard errors of medians (i.e.
$1.253 \times \sigma_{bin} / \sqrt{N_{bin}}$
). Note that the central bin of the 3D sSFR profile for relaxsed clusters (left panel) contains only 1 galaxy. Bottom panel: The star-forming galaxy fractions as a function of normalised cluster-centric distance. While the colours are the same as in the middle panel, we use open squares for the symbols to improve visual clarity. The error bars show
$1\sigma$
uncertainties estimated through the recipe given in Cameron (Reference Cameron2011). For each panel, the grey vertical dotted line indicates
$R_{200}$
. The horizontal grey dotted lines shown in the top panels are centred at 0.
We chose snapshot 140 at
$z=0.03$
, comparable to our observed sample, where there are 886 clusters with halo masses of
$M_\mathrm{200c}\geq 10^{14}\,{\rm M}_\odot$
, reaching as high as
$M_\mathrm{200c}\approx 2\times 10^{15}\,{\rm M}_\odot$
. This sample is reduced to the 868 clusters with unbroken assembly histories up to
$z=3$
, equivalent to the sample used by Kimmig et al. (Reference Kimmig2025) to study the link between the distribution of stellar mass in galaxy clusters and their formation histories. We consider the galaxy populations of these clusters with a stellar mass cut of
$M_*\geq 5\times10^{9}\,{\rm M}_\odot$
, for a total number of 45 902 galaxies in the clusters. In order to cover the same mass regime, we also apply a cut on cluster mass, excluding those outside of the SAMI range (i.e.
$14.25\lesssim \log M_{200}\lesssim15.19$
), leaving 340 clusters.
The large number provides a huge range of assembly histories, with formation redshifts (defined as the time when the main progenitor of the cluster first reached 50% of the final
$z=0$
halo mass) between
$z_\mathrm{form}=0.058$
and
$z_\mathrm{form}=1.77$
, with a median formation at
$z_\mathrm{form}\approx0.67$
. We split the sample into early and late forming by the
$1\sigma$
-bounds of the
$z_\mathrm{form}$
distribution of the remaining 340 clusters. The cuts are therefore
$z_\mathrm{form}\lesssim 0.26$
(
$t_\mathrm{lookback}\simeq3$
Gyr) for dynamically active clusters and
$z_\mathrm{form}\gtrsim0.97$
(
$t_\mathrm{lookback}\simeq8$
Gyr) for dynamically relaxed clusters.
$z_\mathrm{form}\lt0.26$
is well-representative of cases where a system experiences a merger within the last
$\sim$
3 Gyr (which is quite comparable with A168). After applying the cuts, the size of the final sample size is reduced to 55 merging and 55 relaxed clusters with 4638 and 3622 members, respectively.
Figure 10 shows the results for the Magneticum clusters. The top panel shows the stacked projected phase-space distributions (in the 2D
$x-y$
projection; normalised
$R_{xy}$
and
$v_{z}$
) of each cluster sample, whereas inset plots are based on the galaxies’ 6D properties (i.e. real distance,
$R_{3D}$
; velocities,
$v_{3D}$
, with respect to the centre). The hexagonal bins represent PASGs; the redder the colour, the higher the phase-space density per bin, and SFGs are plotted as cyan stars. Here,
$v_{circ}$
denotes the circular velocity (
$\sqrt{GM/r}$
) of the halo as a tracer of velocity dispersion, and
$R_{200c}$
is the radius within which the average mass density is 200 times the critical mass density at the given redshift. In the middle panel, the blue stars with a solid line and the cyan stars with a dashed line display the median sSFR as a function of projected and 3D distances, respectively. The uncertainties are the standard errors on the median (i.e.
$1.253\times\frac{\sigma_{bin}}{\sqrt{N_{bin}}}$
). The cyan and blue open squares in the bottom panel highlight projected and 3D
$f_{SFG}$
within R
$\lt$
$R_{200}$
and
$1 \lt R/R_{200}$
$\lt$
2 for each sample, respectively, and are shifted radially in order to avoid overlapping. The error bars in the bottom row indicate 1
$\sigma$
uncertainties estimated through Cameron (Reference Cameron2011). The grey vertical dashed lines denote
$R_{200}$
for all panels. The horizontal dashed lines are centred at
$v/v_{circ} = 0$
for the top panels. The black solid lines in the top row of Figure 10 mark the region defined by Rhee et al. (Reference Rhee2017), primarily enclosing the ancient infaller or so-called virialised population. As merging cluster cores may not be fully virialised, this region may not represent the true virialised population. Rhee et al. (Reference Rhee2017) caution that, in equal-mass mergers, adopting a single cluster centre can introduce PPS scatter. With these caveats, we retain the term ‘virialised’ to refer to the central/core regions of these systems.
The simulated PPSs show striking similarities to our observations. The relaxed clusters host a few SFGs in the virialised region, whereas the presence of SFGs is more pronounced in the same space for merging clusters. In order to rule out possible projection effects, we also show 6D phase spaces. SFGs in relaxed clusters are mainly consistent with an infalling population (mostly first infallers; Rhee et al. Reference Rhee2017), where the majority of SFGs contribute to this population. This aligns with Lotz et al. (Reference Lotz, Remus, Dolag, Biviano and Burkert2019), who found that the majority of SFGs do not survive beyond the first core passage. In contrast, the merging systems host nearly half of their SFGs in the inner regions represented by intermediate and ancient infallers (Rhee et al. Reference Rhee2017). Regardless of the projection, these results support that the merging clusters host a more mixed SFG population and exhibit slightly higher
$f_{SFG}$
within
$R_{200}$
(bottom panel of Figure 10) compared to the relaxed systems.
Consistent with our findings, the simulated merging clusters also seem to have a high incidence of SFGs within
$R_{200}$
, where the impact of the merger is greatest. In order to test whether this is because of possible merger-induced SF activity or not, we now investigate the SF activity of SFGs in both relaxed and merging clusters. We trace SF activity via median sSFR values as a function of cluster-centric distance with a step size of
$0.2R/R_{200c}$
for the simulated clusters. Both samples exhibit a similar trend in median sSFR with a relatively constant radial profile for projected and 3D spaces. The two innermost bins for 3D space for the relaxed clusters are represented by just three galaxies, so are not reliable. Combined with the PPS properties, we can conclude that the constant radial trend in sSFR indicates that cluster mergers have little impact on global SF activity. In other words, there is no clear evidence for triggered SF activity, supporting our interpretation that the differences are due to dynamical mixing rather than enhanced activity. Moreover, given the fact that relaxed clusters, representing the dynamically more evolved, passive, older systems (i.e. higher
$z_\mathrm{form}$
), have been suppressing the SF over longer timescales compared to young merging systems, this may also reflect as the difference in
$f_{SFG}$
. Recently, Kimmig et al. (Reference Kimmig2025) found a positive correlation (shown in their Figure 9), although a weaker relation, between quenched galaxy fractions and formation redshift for different simulations, including Magneticum, which also supports this.
6.3 Caveats and limitations
In cluster environments, RPS mainly regulates SF activity in an outside-in manner as the galactic gravitational potential weakens toward the outskirts (Cortese et al. Reference Cortese, Catinella and Smith2021). It presents observationally by forming stripped star-forming tails, recently quenched regions, and truncated discs (Koopmann & Kenney Reference Koopmann and Kenney2004; Boselli et al. 2006; Chung et al. Reference Chung, van Gorkom, Kenney, Crowl and Vollmer2009; Smith et al. Reference Smith2010; Poggianti et al. Reference Poggianti2016; Owers et al. Reference Owers2019; Bellhouse et al. Reference Bellhouse2022). This means that any enhancement/quenching in SF due to RPS is likely to appear in the outskirts of galaxies. The results presented in this study rely on single-fibre spectroscopy from SAMI-CRS using the 2dF instrument (2 arcsec fibre diameter) and SDSS (3 arcsec). Considering the redshift range (
$0.04 \lt z \lt 0.06$
), the coverage of the spectra only ranges over physical scales of
$\sim$
1.6 and 2.4 kpc. This means that we are only able to probe the very central regions of galaxies and cannot retrieve information from the outskirts, where any signal due to environmental effects might be found. In this regard, the resolved spectroscopy from the SAMI and Hector Galaxy Surveys will allow SF activity to be investigated over larger regions.
The sample studied here only consists of four clusters. Studies based on a single or handful of merging systems present contradictory results (Ferrari et al. Reference Ferrari, Benoist, Maurogordato, Cappi and Slezak2005; Hwang & Lee Reference Hwang and Lee2009; Owers et al. Reference Owers, Couch, Nulsen and Randall2012; Pranger et al. Reference Pranger2014; Deshev et al. Reference Deshev2017), indicating that cluster-to-cluster variations affect the overall interpretation. However, a handful of studies based on larger datasets have provided a hint of merging clusters triggering SF or hosting more SFGs (Cohen et al. Reference Cohen, Hickox, Wegner, Einasto and Vennik2014; Cohen et al. Reference Cohen, Hickox and Wegner2015; Yoon & Im Reference Yoon and Im2020; Lourenço et al. Reference Lourenço2023). Therefore, disentangling the impact of merging clusters requires more complete, statistically robust studies over larger cluster samples, with a well-covered range of dynamical states.
7. Summary and conclusions
Many questions concerning galaxy evolution in extreme merging clusters remain open. This paper addresses some of these using a sample of four nearby SAMI clusters (
$0.04 \lt z \lt0.06$
) with different dynamical activity – A168, A2399, A3880, and EDCC 0442. Utilising spectra from the SAMI Cluster Redshift Survey and Sloan Digital Sky Survey, galaxy populations were classified as passive, star-forming, and H
$\delta$
-strong based on spectral features, through a classification scheme outlined in Section 4. We have investigated their spatial and projected phase-space distributions, and the EW(H
$\alpha$
) properties, as a tracer of SF activity, with a focus on SFGs. Our key findings are the following:
-
1. We found a mild increase (
$\sim$
20%) in
$f_{SFG}$
within
$R_{200}$
for the merging clusters compared to their relaxed counterparts in our sample. Yet, the difference is within a 1
$\sigma$
level. -
2. Spatial distributions and projected phase-space diagrams reveal differences between merging and relaxed clusters. In the relaxed clusters, there is clear segregation between the PASG and SFG populations, while the galaxy populations are better mixed in the merging clusters, with the majority of SFGs found within
$R_{200}$
. -
3. Despite the mixing, the well-defined relation between SF activity and cluster-centric distance (i.e. SF density) is still visible for the merging clusters.
-
4. We found no significant difference in the properties of EW(H
$\alpha$
) between merging and relaxed clusters, both globally and locally, ruling out the merger-induced activity. -
5. Comparison with the Magneticum simulations indicates that the additional SFGs in the central regions of merging clusters are most likely undergoing their first infall into the merged system.
We conclude that the mild enhancement seen in the
$f_{SFG}$
in the merging clusters is probably due to the redistribution (i.e. mixing) of galaxies in dynamically young environments that are in an intermediate stage of formation; whereas the relaxed counterparts, which represent a dynamically older population of clusters, have had much more time to suppress SF in their members.
The contradictory results reported in the literature emphasise the need to study a large sample of clusters with well-characterised dynamical activity. To that end, upcoming surveys and missions such as 4MOST Hemisphere Survey (4HS; Taylor et al. Reference Taylor2023) and NewAthena (Nandra et al. Reference Nandra2013) will provide multi-wavelength data across different environments and enable us to study statistically the impact of varying dynamical states on galaxy properties with a larger sample.
In the future, we will investigate the environmental impact on SF activity using resolved data from SAMI-GS, which will enable us to search for direct evidence for triggering/quenching in SF (Çakır et al. submitted). Moreover, the next-generation Hector Galaxy Survey (Bryant et al. Reference Bryant, Bryant, Motohara and Vernet2024; Oh et al. Reference Oh2025) will provide resolved data for a larger cluster sample, including more massive systems with more diverse dynamical activity, out to 2
$R_{200}$
, where we will be able to address the same questions with spatially resolved galaxy properties.
Acknowledgement
We thank the anonymous referee for their constructive feedback that has helped to improve the clarity of the manuscript. We thank Luca Cortese for his valuable feedback on this manuscript. We also thank Stefania Barsanti for her feedback on the discussion section.
OÇ and GQ are supported by the Australian Government Research Training Program Scholarship (RTP) to conduct this research. LCK acknowledges support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under project nr. 516355818, the COMPLEX project from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme grant agreement ERC-2019-AdG 882679 and by DFG under Germany’s Excellence Strategy – EXC-2096 – 3900783311. This research was partially supported by the Australian Research Council Centre of Excellence for All Sky Astrophysics in 3 Dimensions (ASTRO 3D), through project number CE170100013. This work was supported by the Korea Astronomy and Space Science Institute under the R&D programme (Project No. 2025-1-831-01) supervised by the Ministry of Science and ICT (MSIT). MP acknowledges support from the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2022R1A2C1004025).
The SAMI-CRS made extensive use of the Anglo-Australian Telescope (AAT). We acknowledge the traditional custodians of the land on which the AAT stands, the Gamilaraay people, and pay our respects to elders past and present, and all the technical and observing support of the staff at the AAT throughout the SAMI-CRS.
Funding for SDSS-III has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, and the US Department of Energy Office of Science. The SDSS-III web site is http://www.sdss3.org/.
SDSS-III is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS-III Collaboration including the University of Arizona, the Brazilian Participation Group, Brookhaven National Laboratory, Carnegie Mellon University, University of Florida, the French Participation Group, the German Participation Group, Harvard University, the Instituto de Astrofisica de Canarias, the Michigan State/Notre Dame/JINA Participation Group, Johns Hopkins University, Lawrence Berkeley National Laboratory, Max Planck Institute for Astrophysics, Max Planck Institute for Extraterrestrial Physics, New Mexico State University, New York University, Ohio State University, Pennsylvania State University, University of Portsmouth, Princeton University, the Spanish Participation Group, University of Tokyo, University of Utah, Vanderbilt University, University of Virginia, University of Washington, and Yale University.
This study made extensive use of Python (Van Rossum & Drake Reference Van Rossum and Drake2009) libraries – Astropy (Astropy Collaboration et al. 2013, 2018, 2022), Numpy (Harris et al. Reference Harris2020), Scipy (Virtanen et al. Reference Virtanen2020), statsmodels (Seabold & Perktold Reference Seabold and Perktold2010), Matplotlib (Hunter Reference Hunter2007), Smplotlib (Li Reference Li2023), Pandas (Pandas Development Team Reference Pandas Development Team2020), Seaborn (Waskom Reference Waskom2021), and Jupyter (Kluyver et al. Reference Kluyver2016); and R (R Core Team 2021).
Open access
