2.1. (Predominantly) inactive stellar systems

2.1.2. UCD galaxies and NSCs

A local sample of UCD galaxies and NSCs with directly measured SMBH masses has come from Graham & Spitler (Reference Graham and Spitler2009) and (Graham Reference Graham2020, and references therein).Footnote ^j This sample includes the SMBH and NSC masses for the (stellar-stripped S0 and now) compact elliptical (cE) galaxy M32 (Nguyen et al. Reference Nguyen2018) and the flattened ‘peculiar’Footnote ^k dwarf ETG NGC 205 (Nguyen et al. Reference Nguyen2019), plus an updated black hole mass for NGC 404 (Davis et al. Reference Davis2020)Footnote ^l and NGC 4395 (Brum et al. Reference Brum2019).Footnote ^m The sample is supplemented with the NSCs in NGC 5102, NGC 5206 (Nguyen et al. Reference Nguyen2018; Nguyen et al. Reference Nguyen2019)Footnote ⁿ and NGC 3593 (Nguyen et al. Reference Nguyen2022). UCD736 orbiting within the Virgo Galaxy Cluster (Liu et al. Reference Liu2020; Taylor et al. Reference Taylor2025) has also been added, as have the globular clusters B023-G078 around M31 (Pechetti et al. Reference Pechetti2022) and $\omega$ Centauri (D’Souza & Rix Reference D’Souza and Rix2013; Häberle et al. Reference Häberle2024) for which the Gaia -Sausage/Enceladus host galaxy mass is used (Lane, Bovy, & Mackereth Reference Lane, Bovy and Mackereth2023), as discussed by Limberg (Reference Limberg2024). Finally, the dwarf spheroidal galaxy Leo I (Mateo, Olszewski, & Walker Reference Mateo, Olszewski and Walker2008) has been included, although it may not contain a massive black hole (Pascale et al. Reference Pascale, Nipoti, Calura and Della Croce2024).

The original $M_\mathrm{ bh}$ – $M_\mathrm{ \star,nsc}$ relation, involving nuclear star cluster masses, $M_\mathrm{ \star,nsc}$ , was first published in Graham (Reference Graham, Meiron, Li, Liu and Spurzem2016), having been presented at a 2014 IAU conference in Beijing. The relation was updated by (Graham Reference Graham2020, Equation 6) and is shown in figure 1, along with the galaxy-morphology-specific $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ relations. The former relation stems from the discovery of the $M_\mathrm{ \star,nsc}$ – $M_\mathrm{ \star,sph}$ relation (Balcells et al. Reference Balcells, Graham, Domnguez-Palmero and Peletier2003; Graham & Guzmán Reference Graham and Guzmán2003) coupled with the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ relation. It is important to recognise that it appears to hold until (i) the erosion of the star clusters at high black hole masses due to binary SMBHs (Bekki & Graham Reference Bekki and Graham2010; Gualandris & Merritt Reference Gualandris and Merritt2012) or (ii) the appearance of a nuclear disc 10s-to-100s of parsec in size rather than just an ellipsoidal star cluster. To date, the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,nsc}$ relation has not received anywhere near the attention of the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ relation, yet it holds insight into the coevolution of dense star clusters and massive black holes with important consequences for gravitational wave science from extreme mass ratio inspiral (EMRI) events (e.g. Preto & Amaro-Seoane Reference Preto and Amaro-Seoane2010; Mapelli et al. Reference Mapelli, Ripamonti, Vecchio, Graham and Gualandris2012; Babak et al. Reference Babak2017; Gair et al. Reference Gair, Babak, Sesana, Amaro-Seoane, Barausse, Berry, Berti and Sopuerta2017; Amaro-Seoane et al. Reference Amaro-Seoane2023).

Figure 1.

$M_\mathrm{ bh}$ - $M_\mathrm{ \star,sph}$ diagram and relations. This is an extension of Figure 5 from Graham (Reference Graham2024b), itself an adaption of Figure 6 from Graham & Sahu (Reference Graham and Sahu2023b). From right to left, the lines from existing studies are as follows. The right-most red line represents BCGs, and the second-from-right red line represents non-BCG E galaxies (Graham Reference Graham2024b, Table 2), both primarily built from ‘dry’ major mergers. The green line represents (‘wet’ major merger)-built dust-rich (dust=Y) S0 and the Es,b galaxies (Graham Reference Graham2024b, Table 2), hence the asterisk on the Y in the figure legend. Next, the blue line represents S galaxies (Graham Reference Graham2023b, Table 1), while the orange line represents dust-poor (dust=N) S0 galaxies referred to here as primaeval (Graham Reference Graham2023b, Table 1). Stripping and threshing the stars from these galaxies may produce cE and UCD galaxies, respectively. The left-most solid and dotted lines represent the NSCs and inner component of UCD galaxies (Graham Reference Graham2020, Equation 6). Notes: Spheroid masses of AGN with $4\times10^6 \lesssim M_\mathrm{ bh}/M_\odot \lesssim 5\times10^7$ have likely been overestimated (in these suspected S galaxies, based on their location in figure 2). Without any structural decomposition of the LRDs, their total stellar mass is shown here under the implicit assumption, which we denounce (Section 3.2), that they are spheroidal structures without a disc component. Upper-left legend: Lin $+$ 24 = Lin et al. (Reference Lin2024b); ‘LRD/AGN’ covers the LRDs reported in recent works, as noted in Section 2; Jiang $+$ 11 = Jiang et al. (Reference Jiang, Greene and Ho2011); GS15 = Graham & Scott (Reference Graham and Scott2015). The NSC and UCD data come from (Graham Reference Graham2020, and references therein). Lower-right legend: Galaxies with directly measured SMBH masses (Graham & Sahu Reference Graham and Sahu2023a), with updates noted in Section 2. Cyan squares (and the green peas and grey AGN samples) are additional galaxies not used to derive the relations.

For the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ diagram shown in figure 1, the stellar mass of the inner component of the UCD galaxies, i.e. the dense NSC obtained from decompositions of their light profiles, is displayed. For the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,gal}$ diagram (Figures 2 and 3), the total stellar mass of the UCD galaxies is presented. This latter approach mirrors the plotting of the total stellar mass for the ‘ordinary’ galaxies shown in figures 2 and 3, while their bulge or equivalently spheroid stellar mass is displayed in figure 1.

Figure 2.

$M_\mathrm{ bh}$ - $M_\mathrm{ \star,gal}$ diagram and relations. Modification of figure 1, building on Figure A4 from Graham (Reference Graham2023b). Here, the inner plus outer components of UCD galaxies are used for their galaxy stellar mass. The right-most red line (upper-right) denotes E BCGs, while the longer red line denotes non-BCG E galaxies (Graham Reference Graham2024b, Table 2) and overlaps with the slightly steeper green line representing dust-rich (dust=Y) S0 and Es,b galaxies (Graham Reference Graham2024b, Table 2). The long black line denotes galaxies built from major mergers (Graham Reference Graham2023b, Table 1). The steepest (blue) line represents the S galaxies (Graham Reference Graham2023b, Table 1), which follow a steep ‘blue sequence’ (Savorgnan et al. Reference Savorgnan, Graham, Marconi and Sani2016; Davis et al. Reference Davis, Graham and Cameron2018). The very numerous (in the local Universe) dust-poor S0 galaxies (orange and red squares) with low stellar masses do not follow either of these relations. The LRD/AGN sample shown here comes from the 2023-2024 data sets mentioned in Section 2.2.2.

Figure 3.

Modification of figure 2. The quasi-quadratic (black) line represents galaxies built from major mergers (Graham Reference Graham2023b, Table 1) while the quasi-cubic (blue) line represents S galaxies (Graham Reference Graham2023b, Table 1), known to follow the steeper ‘blue sequence’ discovered by Savorgnan et al. (Reference Savorgnan, Graham, Marconi and Sani2016). The $z\lt 0.055$ AGN (larger grey dots) from Reines & Volonteri (Reference Reines and Volonteri2015) have been added; they support the steep quadratic/cubic $M_\mathrm{ bh}$ - $M_\mathrm{ \star}$ relations; that is, they are not an offset population. The smaller black dots are low-z galaxies with AGN hosting suspected intermediate-mass black holes (Chilingarian et al. Reference Chilingarian, Katkov, Zolotukhin, Grishin, Beletsky, Boutsia and Osip2018). The open black circles are $z\gtrsim6$ AGN with dynamical, rather than stellar, galaxy masses (Izumi et al. Reference Izumi2021). Lower-right legend: RV15 = Reines & Volonteri (Reference Reines and Volonteri2015); Chilin’ $+$ 18 = Chilingarian et al. (Reference Chilingarian, Katkov, Zolotukhin, Grishin, Beletsky, Boutsia and Osip2018); Izumi $+$ 21 = Izumi et al. (Reference Izumi2021); S = spiral galaxies with a directly-measured SMBH mass. ‘Merger-built ETG’ = S0, ES, E and BCG with a directly-measured SMBH mass and known to have been built from a major merger; ‘S0, primaeval $+$ ’ = dust-poor low-mass galaxies with a directly-measured SMBH mass and not known to have experienced a major merger; the $+$ acknowledges that some of these overlap with the distribution of S galaxies and as such are likely to be faded S galaxies rather than faded/preserved S0 galaxies that never sufficiently grew to host a spiral pattern. The LRD/AGN sample shown here is from Rusakov et al. (Reference Rusakov2025) and has only upper limits for the stellar masses.

2.2. AGN with derived black hole masses

3.1. Background briefing

For those new to the evolving field of black hole scaling relations, a recap of select relevant developments over the past decade may be helpful. Other readers may wish to jump to Section 3.2.

As explained in Graham (Reference Graham2012), it is understood why a near-linear $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ relation is inadequate to describe ‘ordinary’ galaxies with ‘classical’ (merger built) bulges (as observed by Laor Reference Laor1998; Laor Reference Laor2001; Wandel Reference Wandel1999; Salucci et al. Reference Salucci, Ratnam, Monaco and Danese2000). It is now recognised that each galaxy type follows a notably steeper than linear distribution (Reference GrahamGraham 2023b ), and recognition of galaxy-morphology-specific relations has enabled considerable breakthroughs in our understanding of galaxy evolution. For example, recognition of the galaxy-morphology-dependent $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ relations has led to the identification of two types of S0 galaxies (those known to have low-metallicities and old ages are referred to as primaeval (Conselice, Gallagher, & Wyse Reference Conselice, Gallagher and Wyse2003; Lisker et al. Reference Lisker, Glatt, Westera and Grebel2006; Sil’chenko Reference Sil’chenko2013; the other is built from wet major mergers Graham Reference Graham2023a). In the past, these primaeval galaxies may have accreted gas and experienced minor mergers in such a way that they morphed into S galaxies, or a substantial collision may have led them to bypass such a transition and become a second-generation major-merger-built S0 galaxy, which can also be made from S galaxy collisions. Graham (Reference Graham2023b) has suggested that these low-mass, dust-free S0 galaxies are preserved, albeit faded, primaeval galaxies. Ram-pressure stripping of cold gas from these galaxies within the hot X-ray emitting gas clouds/haloes of galaxy groups and clusters can act to preserve them by shutting down star formation. The high velocities of the galaxies in the clusters inhibit mergers of the low-mass galaxies and prevent this avenue of evolution. The cluster environment effectively ‘pickles’ these galaxies.Footnote ^t The bulk of this galaxy type, often referred to as dwarf ETG (dETG), are known to be nucleated (e.g. Binggeli et al. Reference Binggeli, Sandage and Tammann1985; Sandage, Binggeli, & Tammann Reference Sandage, Binggeli and Tammann1985), with even more found using the Hubble Space Telescope (e.g. Lotz et al. Reference Lotz, Telford, Ferguson, Miller, Stiavelli and Mack2001; Graham & Guzmán Reference Graham and Guzmán2003; Côté et al. Reference Côté2006). As galaxy mass functions show, they, and dwarf irregular galaxies, are the most abundant galaxy type in the Universe today (e.g. Reaves Reference Reaves1983; Phillipps et al. Reference Phillipps, Disney, Kibblewhite and Cawson1987). If they contain massive black holes, mergers of these systems, likely in the pre-cluster field and group environment, erode the central star cluster and thereby reduce the central surface brightness (Bekki & Graham Reference Bekki and Graham2010).

As early as Romanishin et al. (Reference Romanishin, Strom and Strom1977), some of these dETGs were considered dS0 disc galaxies. This has contributed to replacing the Tuning Fork diagram with an evolutionary scheme, dubbed the ‘Triangal’, linking the galaxy types through accretions and mergers (Graham Reference Graham2023b). It is the objective of this manuscript to begin to place LRDs in this greater context of galaxy evolution. The impact of mergers is widespread and wide-scale, from the erosion of NSCs to a potential ‘merger bias’Footnote ^u affecting measurements of the baryonic acoustic oscillations (BAO) used to probe cosmology (Eisenstein et al. Reference Eisenstein2005; Percival et al. Reference Percival, Cole, Eisenstein, Nichol, Peacock, Pope and Szalay2007).Footnote ^v As we shall see in Section 3.3, these morphology-specific relations are also important for understanding the distribution of AGN in the $M_\mathrm{ bh}$ – $M_\mathrm{ \star}$ diagram. The distribution of the $z=0$ primaeval S0 galaxies is additionally important for checking potential connections with the distant AGN/LRDs, as done in Section 3.2.

Finally, it may be helpful to note that a detected offset in an $M_\mathrm{ \star}$ - $\sigma$ diagram between ETGs with and without directly measured black hole masses, which had potentially implied a bias in the $M_\mathrm{ bh}$ – $M_\mathrm{ \star}$ relations derived from galaxies with directly measured black hole masses (Shankar et al. Reference Shankar2016), was entirely due to a mismatch in the stellar light-to-mass conversion between the data sets (Sahu, Graham, & Hon Reference Sahu, Graham and Hon2023). The $M_\mathrm{ bh}$ – $M_\mathrm{ \star}$ relations should be applied as is to galaxies without directly-measured black hole masses; one simply needs to ensure that the stellar masses of one’s sample are derived consistently with that used to establish the scaling relation.Footnote ^w

3.2. LRDs with AGN

Figure 1 shows that NSCs and the dense inner components of UCD galaxies have stellar masses 2–3 orders of magnitude smaller than local galaxies with the same black hole mass. Curiously, they roughly follow a distribution with a similar slope in the $M_\mathrm{ bh}$ - $M_\mathrm{ \star,sph}$ diagram. Although new NSC and UCD galaxy data have been included (see Section 2.1.2), the line shown on the left-hand side of figure 1 has been taken from Graham (Reference Graham2020) and is essentially that first reported by Graham (Reference Graham, Meiron, Li, Liu and Spurzem2016). In a future paper, an updated analysis of this relation will be presented. The main objective here is to see how the LRDs compare. In figures 2–3, the total (NSC $+$ envelope) stellar mass of the UCD galaxies is displayed.

Many ideas exist as to how SMBHs established themselves early in the Universe (e.g. Inayoshi, Visbal, & Haiman Reference Inayoshi, Visbal and Haiman2020, and references therein). At first glance, the high (2023–2024 and 2025) SMBH masses reported for the LRDs/AGN at high-z may seem to favour heavy black hole seeds over light seeds, and one might even speculate that they could be primordial, contributing to dark matter (e.g. Argyres, Dimopoulos, & March-Russell Reference Argyres, Dimopoulos and March-Russell1998; Bean & Magueijo Reference Bean and Magueijo2002; Dolgov & Postnov Reference Dolgov and Postnov2017). The direct collapse of gas clouds (Doroshkevich, Zel’dovich, & Novikov Reference Doroshkevich, Zel’dovich and Novikov1967; Umemura et al. Reference Umemura, Loeb and Turner1993) and self-gravitating pre-galactic gas disks (Begelman et al. Reference Begelman, Volonteri and Rees2006; Spaans & Silk Reference Spaans and Silk2006) has been proposed for the production of the more massive seed black holes. At the same time, there can still be a contribution from light black hole seeds, including those built from the cascading collision of stars in dense clusters at the nuclei of haloes (Portegies Zwart & McMillan Reference Portegies Zwart and McMillan2002; Omukai, Schneider, & Haiman Reference Omukai, Schneider and Haiman2008; Das et al. Reference Das, Schleicher, Basu and Boekholt2021). In the presence of substantial gas, dynamical friction will help feed these cluster stars inward (Omukai et al. Reference Omukai, Schneider and Haiman2008; Devecchi & Volonteri Reference Devecchi and Volonteri2009; Davies, Miller, & Bellovary Reference Davies, Miller and Bellovary2011). This will contribute to the partial demise of the clusters and the growth of the massive black holes. There is also near-Eddington (Wolf et al. Reference Wolf, Lai, Onken, Amrutha, Bian, Hon, Tisserand and Webster2024) and super-Eddington accretion (Volonteri & Rees Reference Volonteri and Rees2005; Alexander & Natarajan Reference Alexander and Natarajan2014) onto these high-z black holes, perhaps formed from Pop III stars (Ryu et al. Reference Ryu, Tanaka, Perna and Haiman2016; Banik et al. Reference Banik, Tan and Monaco2019; Tan et al. Reference Tan, Singh, Cammelli, Sanati, Petkova, Nandal and Monaco2024). Indeed, Suh et al. Reference Suh2025) recently reported on LID-568, an SMBH at $z\approx4$ , accreting at 40 times the Eddington limit. A combination of light and heavy black holes may have seeded the LRDs.

There may be clues at low-z as to the evolution of LRDs. The cE galaxies at $z\sim0$ are considered former low-mass ETGs stripped of many of their stars (e.g. Zinnecker et al. Reference Zinnecker, Keable, Dunlop, Cannon, Griffiths, Grindlay and Philip1988; Freeman Reference Freeman, Smith and Brodie1993; Khoperskov et al. Reference Khoperskov, Khrapov and Sirotin2023). At the same time, some rare isolated cEs may have simply never ‘grown up’, unless they are all runaway systems that were stripped and then ejected from galaxy groups or clusters (Chilingarian & Zolotukhin Reference Chilingarian and Zolotukhin2015). As illustrated in (Graham & Sahu Reference Graham and Sahu2023b, their Figure 6), stripping of stars can produce galaxies with $M_\mathrm{ bh}/M_\mathrm{ \star,sph}\approx0.05$ (e.g. NGC 4486B around M87, or NGC 4342). The more extreme version of stripping, referred to as ‘threshing’ (e.g. Bekki & Freeman Reference Bekki and Freeman2003; Ideta & Makino Reference Ideta and Makino2004; Chilingarian & Mamon Reference Chilingarian and Mamon2008; Pfeffer & Baumgardt Reference Pfeffer and Baumgardt2013), is thought to produce UCD galaxies. SDSS J124155.33 $+$ 114003.7 (M59cO) may represent a halfway point between cE and UCD galaxies (Chilingarian & Mamon Reference Chilingarian and Mamon2008). Threshing can pare a galaxy back to $M_\mathrm{ bh}/M_\mathrm{ \star,sph}\sim 0.1$ (e.g. UCD1 around M60) or perhaps even $\sim$ 1 if just the NSC remains.

Given that the stripping process should preferentially remove the dark matter (thought to be dominant at larger, and thus less bound, radii),Footnote ^y a key difference between local UCD galaxies formed via stripping and relic LRDs may be their dark matter fraction. The lack of dark matter in UCD galaxies (e.g. Hilker et al. Reference Hilker, Baumgardt, Infante, Drinkwater, Evstigneeva and Gregg2007; Chilingarian, Cayatte, & Bergond Reference Chilingarian, Cayatte and Bergond2008; Chilingarian et al. Reference Chilingarian, Mieske, Hilker and Infante2011; Frank et al. Reference Frank, Hilker, Mieske, Baumgardt, Grebel and Infante2011) would seems to rule out the notion they are relic compact galaxies formed long ago in dark matter haloes, i.e. relic LRDs, as opposed to tidally-stripped galaxies. Furthermore, the first LRDs to form are perhaps unlikely to be today’s UCD galaxies, as these early LRDs probably resided in larger over-densities that eventually became today’s massive ETGs. However, the LRDs that formed later in the Universe, in smaller haloes, might be the ancestors of today’s UCD galaxies, that is, they could be the nuclei of threshed low-mass disc galaxies that started life as LRDs. One may also speculate that all of today’s UCD galaxies should contain a massive black hole if all high-z LRDs did. It is not, however, established that all local UCD galaxies contain a massive black hole, nor if all nucleated dwarf ETGs galaxies – the likely pre-stripped progenitors of UCD galaxies – do. This state of affairs is an issue of spatial resolution and, therefore, probably does not yet provide useful constraints.

Based on the 2023–2024 masses for LRDs, they are not similar to NSCs for which Graham & Spitler (Reference Graham and Spitler2009) quantified their $M_\mathrm{ bh}/M_\mathrm{ \star}$ ratios. Figure 2 reveals that the LRDs tend to have smaller $M_\mathrm{ bh}/M_\mathrm{ \star}$ ratios, with some matching that of UCD galaxies.Footnote ^z Through the LRDs, we may be witnessing the reverse of a stellar stripping process, in which the black hole mass is already in place or quickly forms before the bulk of a galaxy’s stellar mass builds around it (e.g. Kokorev et al. Reference Kokorev2024b, their section 5.2) and (Tripodi et al. Reference Tripodi2024). However, with the revised black hole masses in the distant AGN/LRDs studied by Rusakov et al. (Reference Rusakov2025), figure 3 reveals that they require more definitive stellar masses before firm conclusions can be reached. It is considered unlikely that the LRDs evolve along the steep $M_\mathrm{ bh}$ - $M_\mathrm{ \star,gal}$ relations defined by local galaxies because if their current upper stellar mass estimates (shown in figure 3) are close to their true stellar masses, then their small half-light sizes (tens to a few hundred parsec) would imply a problematically high stellar density. LRDs likely have considerably lower stellar masses than the upper limits shown in figure 3, perhaps overlapping with the green peas or perhaps forming an extension towards the UCD galaxies and NSCs in the $M_\mathrm{ bh}$ - $M_\mathrm{ \star,gal}$ diagram.

Finally, we are open to the possibility that whether UCD galaxies and some LRDs are structurally equivalent could be a case of ‘nave realism’ based on ‘the illusion of information adequacy’ (Gehlbach, Robinson, & Fletcher Reference Gehlbach, Robinson and Fletcher2024), stemming from their congruent location in the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,gal}$ diagram. Additional information may suggest that their overlap in SMBH and stellar masses is a mere coincidence. In this regard, size information of LRDs and age estimation of UCD galaxies (e.g. Chilingarian et al. Reference Chilingarian, Cayatte and Bergond2008) should be valuable. At least some UCD galaxies are measured to be reasonably old, such as the Sombrero galaxy’s SUCD1 at 12.6 $\pm$ 0.9 Gyr (Hau et al. Reference Hau, Spitler, Forbes, Proctor, Strader, Mendel, Brodie and Harris2009) and M60-UCD1 with a formal age of 14.5 $\pm$ 0.5 Gyr (Seth et al. Reference Seth2014), older than the Universe. However, the Virgo cluster’s VUCD3 is reported to have an age of just 11 Gyr with a 9.6-Gyr-old inner component, while M59cO has an age of 11.5 Gyr but with a blue inner component (Chilingarian & Mamon Reference Chilingarian and Mamon2008) dated at just 5.5 Gyr old (Ahn et al. Reference Ahn2017). This may reflect the ability of star clusters to rejuvenate themselves, at least those residing at the bottom of a galaxy’s gravitational potential well, or a late-time creation for some. Detecting elongated LRD host galaxies with light profiles having Sérsic indices $n \lesssim 1$ would suggest they are disc-like structures, whereas a distribution of somewhat spherical shapes would match that of local dETGs (e.g. Binggeli & Popescu Reference Binggeli and Popescu1995). While single Sérsic fits to galaxies with point-like AGN may yield small sizes, when the imaging data permits it, simultaneously fitting a point-source plus a Sérsic model may enable more reliable information on the underlying host galaxy.

In passing, it is noted that the stellar population of LRDs (and red nuggets) would have initially been blue due to hot, massive stars, just as the stars in the discs of today’s low-mass disc-dominated S0 galaxies would have been moreFootnote ^aa blue in the past when they were younger. Ideally, future work will reveal further connections between LRDs and ‘green peas’ (Cardamone et al. Reference Cardamone2009; Lin et al. Reference Lin2025), blue dwarf ETGs (Cairós et al. Reference Cairós, Vlchez, González Pérez, Iglesias-Páramo and Caon2001, Reference Cairós, Caon, Papaderos, Noeske, Vlchez, Garca Lorenzo and Muñoz-Tuñón2003; Driver et al. Reference Driver, Allen, Liske and Graham2007; Cameron et al. Reference Cameron, Driver, Graham and Liske2009; Moffett et al. Reference Moffett2019), and luminous blue compact galaxies (Guzmán et al. Reference Guzmán, Östlin, Kunth, Bershady, Koo and Pahre2003; Shi et al. Reference Shi, Kong, Li and Cheng2005; Bergvall et al. Reference Bergvall, Zackrisson, Andersson, Arnberg, Masegosa and Östlin2006; Pérez-Gallego et al. Reference Pérez-Gallego2011).

3.3. Additional AGN: Toeing the line

As noted in the Introduction, the steep quadratic nature of the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ relations for ETGs built from major mergers naturally place bright QSOs (e.g. Wang et al. Reference Wang2010; Bongiorno et al. Reference Bongiorno2014; Wang et al. Reference Wang2016; Shao et al. Reference Shao2017; Decarli et al. Reference Decarli2018; Venemans et al. Reference Venemans, Walter, Zschaechner, Decarli, De Rosa, Findlay, McMahon and Sutherland2016) above the original near-linear relation. At the same time, the super-quadratic $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ (and cubic $M_\mathrm{ bh}$ – $M_\mathrm{ \star,gal}$ ) relation for S galaxies naturally places faint Seyfert galaxies below the original near-linear relation. Bridging these extremes are the low-luminosity QSOs and regular AGN of intermediate-luminosity (Willott, Bergeron, & Omont Reference Willott, Bergeron and Omont2015; Willott, Bergeron, & Omont Reference Willott, Bergeron and Omont2017; Izumi et al. Reference Izumi2018) that overlap with the original near-linear $M_\mathrm{ bh}$ – $M_\mathrm{ \star}$ relation. Izumi et al. (Reference Izumi2021) reported how the less luminous QSOs beyond $z\sim6$ had $M_\mathrm{ bh}/M_\mathrm{ \star}$ ratios more in line with the original near-linear $M_\mathrm{ bh}$ – $M_\mathrm{ \star}$ relation. Rather than talking in terms of differing and disjoint populations of AGN with over- or under-massive black holes, it makes more sense to recognise that the original proposition of a linear $M_\mathrm{ bh}$ – $M_\mathrm{ \star}$ scaling relation requires adjusting and that a steeper scaling relation unifies many galaxies.

Unfortunately, the lack of morphological information among AGN samples has hampered past understanding of galaxy speciation. However, as briefly noted before, one slight mystery has now been resolved. In figure 1, and the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ diagram presented in Graham & Scott (Reference Graham and Scott2015), some of the AGN were seen to reside to the right of the sample of galaxies with predominantly inactive black holes with directly measured masses. This is evident at $M_\mathrm{ bh}\approx10^7$ M $_\odot$ . From the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,gal}$ diagram (Figure 2), it is apparent that these AGN are probably S galaxies. It is likely that the published $B/T$ ratios for these AGN, which were typically greater than 0.5-0.6, were too high, thereby artificially shifting them to the right in the $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ diagram. S galaxies tend to have $B/T\lt$ 0.1–0.2 (Graham & Worley Reference Graham and Worley2008; Davis et al. Reference Davis, Graham and Cameron2019).

Given the location of the AGN with $M_\mathrm{ bh} \lesssim 10^6$ M $_\odot$ in figures 1 and 2, they appear to be hosted by S galaxies and primaeval S0 galaxies. Again, the term primaeval is used to imply a first-generation galaxy type not altered by substantial accretion or major mergers. These AGN are consistent with the steep morphology-specific $M_\mathrm{ bh}$ – $M_{\star}$ scaling relations, even though they were not used to define them.

A simplified variant of figure 2 is presented in figure 3, such that the local sample of galaxies with directly measured SMBH masses is now separated into just three types. There are those previously identified as primaeval; they are the low-mass, dust-poor S0 disc galaxies that tend to have an old, metal-poor stellar population. Second are the disc galaxies with a spiral pattern, which tend to have ongoing star formation.Footnote ^ab Then there are the galaxies built from major mergers, such as the (often dust-rich) S0 galaxies built from a wet major merger, the E galaxies built from a dry major merger, and the BCG typically built from more than one major merger. This subdivision offers an alternative view of how the local AGN mesh with, rather than deviate from, galaxies with varying formation histories.

Figure 3 also displays new samples of AGN. The AGN with estimated black hole masses from Reines & Volonteri (Reference Reines and Volonteri2015) and Chilingarian et al. (Reference Chilingarian, Katkov, Zolotukhin, Grishin, Beletsky, Boutsia and Osip2018) are included, as are the AGN at $z\gtrsim6$ from Izumi et al. (Reference Izumi2021). However, only dynamical, rather than stellar, galaxy masses are published for this final sample. Therefore, a (not overly prominent) small circle is used to show those systems in figure 3. While (Izumi et al. Reference Izumi2021, their Figure 13) reported that many of these systems have ‘overmassive’ black holes (by up to a factor of $\sim$ 10) relative to the old near-linear $M_\mathrm{ bh}$ - $M_\mathrm{ \star}$ relation, the bulk of them are not outliers relative to the steeper galaxy-morphology-dependent $M_\mathrm{ bh}$ - $M_\mathrm{ \star}$ relation for merger-built S0 galaxies, a point made by Graham (Reference Graham2023a), see also Graham & Sahu (Reference Graham and Sahu2023a). Just 3–5 of these AGN from Izumi et al. (Reference Izumi2021) have $M_\mathrm{ bh}/M_\mathrm{ dyn}$ ratios that are a factor of (only) 2–3 times higher than the distribution seen for local systems with directly measured SMBH masses. This is reconcilable with the sample selection bias of the most luminous QSOs at high-z and the greater inaccuracy of indirect measures of black hole mass in AGN. The subset of luminous $z\gtrsim6$ AGN from Izumi et al. (Reference Izumi2021) shown in figure 3 is likely to be wet-merger-built S0 galaxies.

At $M_\mathrm{ \star,gal} \gt 10^{10}$ M $_\odot$ , in figure 3, the overlap of low-z AGN with the $z\approx0$ sample of SMBHs with directly measured masses reveals that these AGN are predominantly S galaxies or merger-built S0 galaxies. These AGN are not an offset population from ordinary galaxies with predominantly inactive SMBHs. This information is crucial if we are to connect the evolutionary path of high-z AGN/LRDs with other active and inactive galaxies.

Knowledge of galaxy-morphology-specific black hole scaling relations enables an improved means for deriving the virial factor(s) for converting AGN virial products into black hole masses (Peterson Reference Peterson1993; Onken et al. Reference Onken, Ferrarese, Merritt, Peterson, Pogge, Vestergaard and Wandel2004; Bentz et al. Reference Bentz, Peterson, Pogge and Vestergaard2009). To date, these conversion factors have been obtained with little attention to galaxy morphology. A better approach will involve matching AGN virial products with directly measured black hole masses from galaxies of the same morphological type (Graham et al. 2025, in preparation). It can already be seen in the data of (Bentz & Manne-Nicholas Reference Bentz and Manne-Nicholas2018, their Figure 5) that the reverberation-mapped AGN follow a steep $M_\mathrm{ bh}$ – $M_\mathrm{ \star,sph}$ distribution well-matched by the super-quadratic relation quantified by (Scott et al. Reference Scott, Graham and Schombert2013, their Figure 3) for galaxies without depleted stellar cores, i.e. those not built from a dry major merger. Similarly, the AGN data of (Bentz & Manne-Nicholas Reference Bentz and Manne-Nicholas2018, their Figure 6) display a steep trend similar to that of Savorgnan et al. (Reference Savorgnan, Graham, Marconi and Sani2016) for S galaxies. This yet-to-be-applied approach at determining the virial factor(s) will avoid a bias such that the reverberation-mapped AGN sample may be skewed towards a distribution of galaxy type, such as S and dust-rich S0 galaxies, that differs from the bulk of the sample with directly measured black hole masses to which it is compared. Similarly, AGN sample selection of certain galaxy types and not others, for example, S and/or dust-rich S0 but not primaeval dust-poor S0 galaxies or ‘red and dead’ E galaxies, may also explain the tight AGN relations reported by Bennert et al. (Reference Bennert2021). The location of the LRDs and other AGN in figures 1–3 can be refined once improved virial factors, used to calibrate secondary relations, are established through application of the galaxy-morphology-dependent scaling relations (Graham Reference Graham2023b). Better constraints on the stellar masses of the LRDs are also needed to decipher their evolutionary trajectory in the $M_\mathrm{ bh}$ - $M_{\star}$ diagrams.