MASSIVE RELIC GALAXIES CHALLENGE THE CO-EVOLUTION OF SUPER-MASSIVE BLACK HOLES AND THEIR HOST GALAXIES

It is now well established that the masses of super massive black holes (SMBHs; ${M}_{\bullet }\gt {10}^{6}\;{M}_{\odot }$) strongly correlate with some of their host galaxies' properties, such as their luminosities (e.g., Kormendy & Richstone 1995; Gültekin et al. 2009; Schulze & Gebhardt 2011; Kormendy & Ho 2013; McConnell & Ma 2013), their masses (e.g., Magorrian et al. 1998; Häring and Rix 2004; Beifiori et al. 2012), or their velocity dispersions (σ; e.g., Ferrarese & Merritt 2000; Gebhardt et al. 2000; Gültekin et al. 2009; Graham et al. 2011). This supports the idea that the host galaxies and their SMBHs form and grow in a coordinated way by a common physical mechanism, which could be active galactic nucleus (AGN) feedback, mergers, or secular evolution (e.g., Fabian 1999; Hopkins et al. 2006; Menci et al. 2008; Somerville et al. 2008; Jahnke & Macciò 2011). The correlation between SMBH mass and velocity dispersion is extremely tight, following a log-normal distribution whose steepness increases as the galaxy sample is enlarged and more accurate measurements of the SMBH masses are used (see e.g., the most recent compilations of McConnell & Ma 2013 or Kormendy & Ho 2013). This correlation implies that the most massive galaxies ($\sigma \;\sim$ 300–400 $\mathrm{km}\;{{\rm{s}}}^{-1}$) should have SMBHs with ${M}_{\bullet }\;\sim$ 10¹⁰${M}_{\odot }$, which in turn corresponds to high stellar masses (M ${}_{\mathrm{bulge}}\;\sim$ 10¹² ${M}_{\odot }$) and high luminosities (L ${}_{{\rm{V}}}\;\gtrsim$ 10¹¹ ${L}_{\odot }$).

However, in the last few years, an increasing number of galaxies have been reported to strongly deviate from the above scaling relations. These galaxies show larger SMBHs than what is expected according to their velocity dispersions or bulge masses. This is the case of, e.g., the deeply analyzed NGC 1277 (van den Bosch et al. 2012; Emsellem 2013; Fabian et al. 2013; Yıldırım et al. 2015), NGC 4291 and NGC 4342 (Bogdán et al. 2012), NGC 1332 (Rusli et al. 2011), and SDSS J151741.75–004217.6 (alias b19, Läsker et al. 2013). All of these extreme outliers lie far beyond the intrinsic scatter of the relations, challenging the supposed co-evolution between SMBHs and their host galaxy (≳3σ). The Hobby–Eberly Telescope Massive Galaxy Survey (hetmgs; van den Bosch et al. 2012, 2015) is a new large galaxy compilation aimed at finding nearby galaxies in which black hole masses can be directly measured. As a by-product, they confirmed the existence of two other enormous SMBHs in NGC 1271 (Walsh et al. 2015) and MRK 1216 (Yıldırım et al. 2015) and there are a few dozen more candidates in their sample. Following Läsker et al. (2013) notation, these types of extreme SMBHs have been named über-massive SMBHs (ÜMBHs), though some critics have emerged pointing to an overestimation of the real mass. This overestimation could be due to the methodology employed (e.g., long-slit spectroscopy instead of integral-field to derive proper dynamical models; e.g., Emsellem 2013; Yıldırım et al. 2015) or to variations in the initial mass function (which would vary the measured stellar mass of the host galaxy; e.g., Emsellem 2013; Läsker et al. 2013). Nonetheless, even accounting for these uncertainties, the galaxies hosting ÜMBHs are still extreme outliers of the scaling relations. Consequently, it is worth addressing the physical processes that cause these objects to be outliers. One possibility is that they are galaxies that did form the corresponding stellar mass, but then lost it by, e.g., tidal stripping. However, the recent work from Bogdán et al. (2012), who studied two outliers in the low-mass regime, claimed that if that was the case, then the dark matter halo surrounding the galaxy should have also been stripped. Their results, instead, showed that both galaxies reside in extended dark matter halos.

In this paper, we propose and probe an alternative scenario. We hypothesize that galaxies hosting ÜMBHs form them in an early phase of the galaxy evolution (z ≳ 2) and once the SMBHs are formed, the galaxies remain structurally untouched, which prevents them from reaching the present-day scaling relations, leaving them as outliers. This early formation is supported by the evolution of the ${M}_{\bullet }$ − M_bulge scaling relation at high redshift (e.g., Walter et al. 2004; Jahnke et al. 2009; Greene et al. 2010, Petri et al. 2012). Therefore, in our scenario, massive galaxies hosting ÜMBHs are outliers in the black hole mass scaling relations because they did not follow the expected two-phase growth channel, where merger accretion increases the size (and to a lesser extent, the mass) of the host galaxies after the central massive core is formed (e.g., Naab et al. 2009; Oser et al. 2010; Hilz et al. 2013). Therefore, we would expect the host of ÜMBHs to be galaxies showing the structural properties of the massive galaxy population at z ∼ 2. In other words, they should be the "relics" of the early universe, with old ($\gtrsim 10$ Gyr) stellar populations and compact (R ${}_{{\rm{e}}}\;\lesssim 2$ kpc) sizes, similar to those found at high-z (e.g., Trujillo et al. 2007; Buitrago et al. 2008; Carrasco et al. 2010). The existence of a small number of such galaxies without merging accretion since z ∼ 2 is a natural prediction of the ΛCDM model (i.e., Quilis & Trujillo 2013), but, to date, only one candidate has been robustly confirmed (Trujillo et al. 2014).

To probe such predictions, we explore the star formation histories (SFHs) for a sample of galaxies with ÜMBHs. If this scenario is realistic, then we should be able to pose a lower limit for the age of the SMBH formation, which corresponds to the age of the host galaxy. The layout of this paper is the following. Section 2 describes the data used from the different compilations and our spectroscopic sample. Section 3 analyzes the derived SFHs of the sample in order to determine whether a galaxy is a relic candidate or not. Section 4 shows the results and discusses their implications on the theories of the co-evolution (or not) of the SMBHs and the host galaxy. A final summary is presented in Section 5. Throughout this work, we adopt a standard cosmological model with the following parameters: H₀ = 70 km s⁻¹ Mpc⁻¹, ${{\rm{\Omega }}}_{m}$ = 0.3, and ${{\rm{\Omega }}}_{{\rm{\Lambda }}}=0.7$.

We first collect the largest possible sample of galaxies with SMBH mass measurements. There are three main recently updated compilations of nearby galaxies: McConnell & Ma (2013, McM13), Kormendy & Ho (2013, KH13), and Graham & Scott (2013, GS13). They are all based on the initial sample of Gültekin et al. (2009), but also contain new galaxies and improved measurements on the SMBH masses, galaxy distances, and bulge mass determinations. All McM13 and GS13 galaxies (except for nine new ones in the latter) are included in KH13. In addition, KH13 presents 13 new galaxies. This literature compilation corresponds to a total of 95 galaxies.

A number of caveats are worth mentioning. First, galaxies found in common among the three compilations (McM13, KH13, and GS13) do not always have the same reported values for some of the properties because different techniques were employed. Second, galaxies are morphologically classified following different criteria in each compilation. McM13 discriminate between elliptical galaxies and bulges, but do not differentiate between classical and pseudobulges, while KH13 make this last separation on the bulges, and GS13 only separate by elliptical and disk galaxies. In addition, the authors of each compilation do not include the same galaxies in their fits: some galaxies excluded in one fit may be included in another. Consequently, we use here the entire compilation to illustrate the correlations (Figure 1), but we are not deriving new fits for the correlations. In this exercise, we use the fits corresponding to McM13 for all the galaxies with dynamical mass estimates, with α = 8.32 ± 0.05 and β = 5.64 ± 0.32 (for the ${M}_{\bullet }$ - σ; Figure 1(a)) and α = 8.46 ± 0.08 and β = 1.05 ± 0.11 (for ${M}_{\bullet }$ − M_bulge; Figure 1(b)).

Zoom In Zoom Out Reset image size

Together with the above compilation, we have expanded our sample using the new compilation of galaxies from the hetmgs survey. This survey was especially designed to find nearby galaxies that were suitable for black hole mass measurements. It is composed of 1022 galaxies, and hence is an ideal place to find galaxies with very large SMBHs, which will be the basis of our analysis for ÜMBH candidates, as explained in the next section.

2.1. Defining the ÜMBH Candidates

To identify ÜMBH host candidates, we select galaxies with SMBHs that are significantly larger than what is expected according to the mass of the bulge of their host galaxy. In other words, we identify those galaxies that strongly deviate from the M ${}_{\bullet }$− M_bulge scaling relation. We use the following operative definition to select ÜMBH host candidates: a ÜMBH host candidate is a galaxy where the mass of its SMBH is located at a distance of 3σ above the M ${}_{\bullet }$− M_bulge scaling relation given by McM13 (see Figure 1). With this definition, it is very unlikely that the mass of its SMBH can be explained as simply produced by the uncertainty in the measurement of the mass of its SMBH. To have a significant number of ÜMBH host candidates, we use the hetmgs sample. In this paper, we are interested in the most massive galaxies, hence we will concentrate our efforts on finding ÜMBH candidates in the upper right corner of the ${M}_{\bullet }$− M_bulge scaling relation.

From the hetmgs sample, we first select those galaxies that are good candidates to host ÜMBHs by imposing the following criteria.

• 1.  
  A large dynamical enclosed mass inside the central $R=3\prime\prime$ aperture of the hetmgs, which indicates a very high central mass concentration (from either stars or a black hole). We use the virial mass estimator ${M}_{\mathrm{vir}}={\sigma }_{{\rm{c}}}^{2}\;R\;{G}^{-1}\gg {10}^{9.7}$ ${M}_{\odot }$, where G is the gravitational constant and ${\sigma }_{{\rm{c}}}$ the central velocity dispersion. This mass does not discriminate between baryonic mass and black hole mass. As such, these virial masses are considered upper limits for the black hole masses (see below).
• 2.  
  Near enough so that the putative black hole can be resolved with high spatial resolution facilities; i.e., the sphere of influence ${\theta }_{{\rm{i}}}=G\;{\sigma }_{{\rm{c}}}^{2.5}\;{D}^{-1}$ (being D the Hubble flow distance) to be larger than 006 (see van den Bosch et al. 2015).

With the above criteria, the sample of candidates from the hetmgs consists of 173 galaxies. These galaxies are, as expected, average massive galaxies with sizes of ∼4 kpc. The vast majority of them do not qualify as ÜMBH host candidates (i.e., they are not 3σ outliers of the ${M}_{\bullet }$ − M_bulge scaling relation). With the 3σ criteria, our sample is reduced to 30 ÜMBH candidates. Interestingly, this last criteria dramatically changes the structural properties of the selected galaxies: they now have an average size of ∼1.5 kpc. In other words, the ÜMBH host candidates are massive compact galaxies. The sizes of the hetmgs galaxies have been retrieved from the SDSS database as the half-light effective radii in the r-band (r_e) and then circularized (R_e).

We need to make a last selecting criteria, which is to have pre-existing SDSS⁷ spectroscopy with high quality (S/N ≥ 20). This is crucial to robustly derive the SFHs of the galaxies with the full-spectral-fitting technique employed in Section 3. This reduces our sample of ÜMBH candidates to 10 galaxies. Two of them (NGC 0426 and NGC 2522) were rejected because their spectra were classified by SDSS as AGN broadline. Having an AGN would not only make the determination of the stellar population properties of the galaxies uncertain but also prevent a proper estimation of the structural properties of the galaxies. In fact, the presence of a bright nucleus could bias the measurement of the effective radius of the galaxies toward smaller sizes. In order to avoid this potential source of contamination from our sample, we have ended with a final sample of eight ÜMBH candidates.

We use the central enclosed virial mass estimate from the hetmgs as an estimation of the ${M}_{\bullet }$ of the host galaxies. This means that both the black hole masses and their locus in the SMBH-host galaxy planes should be considered as upper limits. Efforts toward obtaining more accurate estimates using orbit-based dynamical models are underway. For example, the actual black hole mass for NGC 1277 has been lowered to $12\times \;{10}^{9}\;{M}_{\odot }$ (Yıldırım et al. 2015), which is a factor of 1.5 lower than the upper limits. Therefore, if we apply this factor to the rest of the ÜMBH candidates, we have a lower, more conservative, estimate of their black hole masses. Nonetheless, in the absence of high-resolution spectroscopy and orbit-based models for our sample of candidates, we will use the upper limit estimates for our analysis. However, it is worth noting that, even considering the lower estimates, the ÜMBH candidates are still extreme outliers in the local ${M}_{\bullet }$− M_bulge scaling relation (Figure 1(b); ∼3–5σ deviations). It is worth noting that, when placed in the M ${}_{\bullet }$–σ scaling relation, our ÜMBH host candidates are not that extreme (Figure 1(a); ∼1–3σ deviations). The reason for such differences between the two planes is further discussed in Section 4. The details of our complete spectroscopic sample are described in Table 1.

Table 1.  The Spectroscopic Sample

Galaxy	Spectra	${M}_{\bullet }(+,-)$	${M}_{\mathrm{bulge}}(+,-)$	Vel. Disp.	Re	Deviation $(+,-)$	Deviation $(+,-)$
		(109 ${M}_{\odot }$)	(1011 ${M}_{\odot }$)	(km$\;{{\rm{s}}}^{-1}$)	(kpc)	Mbulge	Vel. Disp.
ÜMBH Host Candidates from hetmgs
NGC 1270	SDSS	<12	3.3 (2.7,1.3)	353 ± 4	1.92 ± 0.03	$3.2\sigma$ (−, 1.6)	$1.0\sigma$ (−, 0.7)
NGC 1271	SDSS	<8	1.5 (0.8,0.6)	293 ± 6	1.31 ± 0.02	$4.0\sigma$ (−, 2.0)	$2.0\sigma$ (−, 1.9)
NGC 1277	SDSS	<17	2.1 (1.9,0.7)	385 ± 4	1.64 ± 0.03	$4.2\sigma$ (−, 1.4)	$0.8\sigma$ (−, 0.7)
NGC 1281	SDSS	<5	1.2 (1.0,0.4)	258 ± 4	1.46 ± 0.03	$3.4\sigma$ (−, 0.6)	$2.0\sigma$ (−, 0.9)
NGC 2767	SDSS	<5	1.4 (1.1,0.6)	239 ± 3	1.95 ± 0.03	$3.2\sigma$ (−, 1.7)	$2.5\sigma$ (−, 1.3)
PGC 012557	SDSS	<7	0.6 (0.6,0.3)	283 ± 4	0.71 ± 0.02	$4.7\sigma$ (−, 1.0)	$1.8\sigma$ (−, 0.8)
PGC 012562	SDSS	<6	0.4 (0.3,0.2)	259 ± 4	0.70 ± 0.02	$5.0\sigma$ (−, 1.2)	$2.2\sigma$ (−, 0.7)
PGC 032873	SDSS	<14	2.3 (1.2,0.9)	356 ± 3	1.76 ± 0.03	$3.7\sigma$ (−, 1.2)	$1.0\sigma$ (−, 0.8)
Control Sample from McM13
NGC 3379	Y06	0.4 (0.1, 0.1)	0.7 (0.5, 0.3)	206 ± 10	3.3 ± 0.4	$1.0\sigma$ (1.1, 1.1)	$0.6\sigma$ (0.6, 0.8)
NGC 3842	SDSS	9.7 (3.0, 2.5)	16.0 (14.0, 7.0)	270 ± 14	14.0 ± 1.3	$0.8\sigma$ (1.1, 1.2)	$2.4\sigma$ (0.7, 0.9)
NGC 4261	SDSS	0.5 (0.1, 0.1)	8.3 (7.0, 4.0)	315 ± 15	7.4 ± 0.7	$2.1\sigma$ (1.1, 1.1)	$1.9\sigma$ (0.5, 0.8)
NGC 4472	Y06	2.5 (0.6, 0.1)	9.0 (7.0, 4.0)	300 ± 15	8.2 ± 0.4	$0.2\sigma$ (0.8, 1.1)	$0.2\sigma$ (0.6, 0.7)
NGC 4473	Y06	0.1 (0.0, 0.0)	1.6 (2.0, 0.6)	190 ± 9	4.3 ± 0.3	$2.1\sigma$ (0.9, 1.6)	$0.6\sigma$ (0.7, 1.0)
NGC 4697	Y06	0.2 (0.0, 0.0)	1.3 (0.7, 0.6)	177 ± 8	6.4 ± 0.4	$0.8\sigma$ (1.3, 1.5)	$0.7\sigma$ (0.8, 1.5)
NGC 4889	SDSS	21.0 (16.0, 15.5)	18.0 (14, 8)	347 ± 17	12.6 ± 1.4	$1.6\sigma$ (1.6, 2.5)	$1.7\sigma$ (1.0, 2.6)

Notes.

Column 1: our spectroscopic sample from the compilation of hetmgs (top) and McM13 (bottom). Column 2: origin of the spectra—SDSS or Y06 (from Yamada et al. 2006). Columns 3 and 4: the masses of the SMBHs and the masses of the bulge of the galaxy from their dynamical estimates (from hetmgs and McM13). Column 5: velocity dispersions (derived with STARLIGHT and from McM13). Column 6: circularized effective radii (from the SDSS r-band, ATLAS3D, and Hyperleda). Columns 7 and 8: deviations from the local SMBH mass scaling relations. Note that for the ÜMBH host candidates the black hole masses are an upper limit, thus marked with an <.

Download table as:  ASCIITypeset image

2.2. The Control Sample

As mentioned before, under the proposed scenario present-day massive galaxies hosting ÜMBHs should have the characteristics of the typical population of galaxies at z ∼ 2 after passively evolving since that epoch. In other words, they should be massive (M* ≳ 10¹¹ ${M}_{\odot }$), compact (R ${}_{{\rm{e}}}\;\lesssim 2$ kpc), and have old stellar populations ($\gtrsim 10$ Gyr). Galaxies that fulfill these criteria in the local universe are named relic galaxies and are known to be extreme outliers from the local mass–size relation (Trujillo et al. 2009). Therefore, we have quantified the compactness of our galaxies by measuring how much their size deviates from the expected value according to the present-day local mass–size relation of Shen et al. (2003). This well-known relation was derived using a complete sample of 140,000 galaxies from the SDSS. The stellar masses used in that paper were taken from Kauffmann et al. (2003), calculated by multiplying the SDSS Petrosian luminosity with the model-derived mass-to-light ratio. The radii considered were the Sérsic half-light radius ${R}_{50,S}$ in the z-band. For galaxies with $n\;\gt$ 2.5, the local mass–size relation is

$$\begin{eqnarray}&&\bar{R}(\mathrm{kpc})=b{(M/{M}_{\odot })}^{a}\end{eqnarray} \tag{ 1 }$$

where $\bar{R}$ is the mean radius, M is the stellar mass, and a and b are fitting parameters given by a least-squares fit to the data, a = 0.56 and b = 2.88 × 10⁻⁶. This relation is shown as the solid green line in Figure 2. This figure indicates the location of our spectroscopic sample in the mass–size relation plane. It is clear that our ÜMBH candidates largely deviate from the local relation.

Zoom In Zoom Out Reset image size

We consider here that a galaxy is compact if the ratio R_e/R ${}_{{\rm{e}},\mathrm{shen}03}\;\lesssim 0.33$ (see Table 2). This corresponds to ∼1/3 of the normal size that massive galaxies show at z = 0 at a given stellar mass. The location in the stellar mass–size plane of our ÜMBH host candidates is equivalent to the locus occupied by the massive compact galaxies at high-z.

Table 2.  Fitting and Massive Galaxy Relic Criteria

Galaxy	[${\chi }^{2}$]	Adev(%)	${R}_{{\rm{e}}}/{R}_{{\rm{e}},\mathrm{shen}03}$	$\%M(t\;\lt \;8$ Gyr)	${\rm{\Omega }}(\%)$	Relic?
ÜMBH host candidates
NGC 1270	1.6	1.7	0.24	0	7	✓
NGC 1271	1.7	1.6	0.25	0	10	✓
NGC 1277	1.9	1.5	0.26	0	0	✓
NGC 1281	1.2	1.6	0.33	0	0	✓
NGC 2767	0.7	1.8	0.37	0	0	✓
PGC 012557	1.3	1.7	0.22	0	0	✓
PGC 012562	1.3	1.3	0.23	0	14	∼
PGC 032873	1.0	1.0	0.23	0	0	✓
Control
NGC 3379	0.1	1.9	0.92	5	4	X
NGC 3842	0.8	2.0	0.73	1	14	X
NGC 4261	0.7	1.8	0.55	0	5	X
NGC 4472	1.5	1.7	0.55	7	5	X
NGC 4473	1.1	1.3	0.83	1	14	X
NGC 4697	1.2	1.4	1.26	21	7	X
NGC 4889	0.8	1.8	0.61	0	10	X

Notes.

The table summarizes the different criteria by which a galaxy can be considered a relic candidate. Columns 2 and 3 are measurements of the quality of the fit from STARLIGHT, (see the appendix). Column 4 measures the compactness of a galaxy, by showing how much the galaxy size deviates from the present-day mass–size relation of Shen et al. (2003) for $n\;\gt$ 2.5 galaxies. Column 5 and Column 6 quote the fraction of stellar populations with $t\lt 8$ Gyr, and the fraction of mass with 8 < t < 10 Gyr, Ω. Column 7 states the verdict on whether the galaxy is considered a relic candidate or not.

Download table as:  ASCIITypeset image

To determine the stellar populations of both the outliers and the control sample, we derive the SFHs of our galaxies, applying a full-spectral-fitting approach. This technique has several advantages compared to the classical line strength approach. One of them is that the results are not limited to luminosity-weighted estimates but also give the output in terms of mass, allowing one to recover the true fossil imprints of the epoch when the stars were created. The code employed here is STARLIGHT (Cid Fernandes et al. 2005; Cid Fernandes & González Delgado 2010), which creates a combination of single stellar populations (SSPs) and model predictions that best resemble the observed galaxy spectrum while minimizing the ${\chi }^{2}$. The reader is referred to the source code papers for a deeper description of the methodology, or to the appendix for a brief summary of this technique. The SSP models used are MIUSCAT (Ricciardelli et al. 2012; Vazdekis et al. 2012), which cover a wide range of both ages (0.1–17.8 Gyr) and metallicities (−1.71–0.22 dex). For this exercise, in particular, 46 ages and 4 different metallicites were considered, making a total of 184 SSP templates. These SSP models also allow for variations on the IMF slope and shape. The IMF has recently been reported to be non-universal (e.g., Cappellari et al. 2012; Conroy & van Dokkum 2012; Spiniello et al. 2012; La Barbera et al. 2013; Ferreras et al. 2013) and to have a strong impact on the derived stellar population properties (Ferré-Mateu et al. 2013). However, Martín-Navarro et al. (2015) have recently shown that for the compact massive relic galaxy NGC 1277, changing the IMF does not modify the derived age properties. This indicates that it is still unclear which parameter really drives these IMF variations (e.g., velocity dispersion, metallicity, or α-enhancement). Therefore, here we will focus on the standard assumption of a Kroupa universal IMF.

Figure 3 illustrates the derived SFH from the spectral fitting for one of the galaxies (NGC 1271). The derived SFHs for the rest of our spectroscopic sample can be found in the appendix. The right panels in Figure 3 show the derived SFHs as the percentage of mass created at a given age, considering both the contribution of each individual metallicity from the models and of the total metallicity. It is worth emphasizing here an important caveat of the form in which the derived SFHs are shown. As mentioned by Cid Fernandes et al. (2005), the reader should not take the form of each single burst at face value. This is partly because the SSP models do not uniformly cover the entire age parameter space. For example, there is always a gap at 13 Gyr because there is no SSP model corresponding to that age in the MIUSCAT models. Therefore, the derived SFHs should be better interpreted as clearly differentiated epochs of formation, i.e., young, intermediate, and old. These considerations are fundamental for the purpose of our exercise because one of our criteria to consider a galaxy as a relic of the z ∼ 2 universe, is that all of their stellar populations must be old (>10 Gyr). When looking at the age distributions, it can be seen that sometimes there is a small contribution at 8–9 Gyr, which can be considered to be the tail of the distribution that peaks at ∼10 Gyr. For this reason, in order to be more quantitative in our statement of whether a galaxy has (or not) entirely old stellar populations, we measure the fraction of mass Ω which is within the interval 8 < t < 10 Gyr. We thus consider a galaxy compatible with having all its stellar mass formed at $z\;\gtrsim$ 2 if there are no stellar populations younger than 8 Gyr and Ω is below 10% of the total mass.

Zoom In Zoom Out Reset image size

We summarize in Table 2 the quality parameters from the fit and whether the outliers satisfy our criteria to be considered relic galaxies or not. We find that seven out of the eight objects undoubtedly qualify as relic galaxies, while only one is dubious. As these compact relic galaxies are also outliers in the present-day mass–size relation (Figure 2), finding these ÜMBH host candidates could constitute a new way of detecting these rather elusive relic galaxies in the nearby universe (e.g., Trujillo et al. 2009; Taylor et al. 2010; Ferré-Mateu et al. 2012; Trujillo et al. 2012, Poggianti et al. 2013). Regarding our control sample, none of the galaxies can be considered to be massive relics of the high-z universe because of their large sizes. However, a couple of them are as old as our relic galaxies, indicating that at least the stellar populations of their most central parts were formed at very early epochs as well. We would like to emphasize that our criteria based on the Ω value, in order to be fully consistent with the statement that the galaxy is a relic, should be shown to be true along the entire structure of the galaxy. As we are employing SDSS data, the coverage of the fiber is ∼0.3–0.8R_e for the compact galaxies and ∼0.01–0.03R_e for the control sample. Consequently, the SDSS data does not allow us to perform such a radial study. However, this issue has been assessed for the compact candidates in Trujillo et al. (2014) for NGC 1277 and in the companion paper of A. Ferré-Mateu (2015b, in preparation) for PGC 032873. These works both use high-quality spectroscopy from the William Herschel Telescope to the derive SFHs and stellar population properties out to several effective radii, showing no gradients and supporting our hypothesis of this type of galaxies being relics.

The tight correlations that the mass of the SMBHs and their host galaxy properties follow have been considered to be proof of the co-evolution between galactic bulges and the SMBHs they host. According to this paradigm, they should be coupled by a common physical mechanism, e.g., feedback, mergers, or secular evolution. However, a small sample of galaxies hosting ÜMBHs now challenges this assumed universal co-evolution because they clearly fail to follow such relations. In this paper, we have proposed that the nature of these outliers hosting ÜMBHs is connected to the unusual evolutionary path followed by these massive galaxies (without growing in mass or size). It is generally assumed from galaxy formation models that massive galaxies form in a two-phase mechanism, where the central massive part of the galaxy is created in a fast and violent event at high-z, and then the galaxy grows by posterior merger activity (e.g., Naab et al. 2009; Oser et al. 2010; Hilz et al. 2013). However, because mergers are stochastic events, it is expected that some galaxies remain untouched over cosmic time, remaining compact and entirely old (see Quilis & Trujillo 2013). Therefore, if the ÜMBH-outliers are those relics, as our results suggest, then the nature of the deviations could be explained by the lack of merging activity and its consequent lack of galaxy growth. Under our hypothesis, these galaxies should occupy in the SMBH mass scaling relations the position of the population of galaxies at z ∼ 2. Consequently, present-day most massive galaxies should start at those locations ∼10 Gyr ago and their growth in mass and size should be able to move them to the current scaling relations. If this is true, the expected mass and σ increase of the massive galaxies since z ∼ 2 should be enough to locate our outliers in agreement with the local relations.

We explore this possibility in what follows. The arrows in Figure 1 show the estimated growth in velocity dispersion and stellar mass since z ∼ 2 for individual massive galaxies. From semi-analytical models, if an individual galaxy grows approximately seven times in size, it can increase its mass by almost a factor of five (e.g., Oser et al. 2010; Trujillo et al. 2011). This could account for the missing stellar mass, placing the ÜMBH-outliers closer to the central distribution in the ${M}_{\bullet }$−M_bulge relation. As commented above, our sample is not that extreme when placed in the ${M}_{\bullet }$– σ relation, hence not much variation in the velocity dispersion should be expected, if any. This is in agreement with the model predictions (e.g., Cenarro & Trujillo 2009; Oser et al. 2012; Oogi & Habe 2013; Wellons et al. 2015), where the velocity dispersion varies very slightly since z ∼ 2. Depending on the dominant merger channel, individual galaxies can increase their velocity dispersion very mildly by a factor of ∼1.1 (Hilz et al. 2012; Tapia et al. 2015, submitted). It is worth noting that in our scenario, the arrows move horizontally because an increase in the SMBH is expected to be almost negligible during the second phase. It is typically assumed that mergers that occur since z ∼ 2 have a mass ratio of 1:3–1:5 (Oser et al. 2010, 2012). Here we assume that the massive galaxies, once formed at z ∼ 2, have a much larger SMBH than typical for their stellar masses. Then, they merge with smaller satellites, whose mass–size evolution is milder over cosmic time (Trujillo et al. 2004) and therefore are expected to be located already in, or very close to, the local relation having smaller black holes. This type of encounter would produce a negligible increase in the mass of the SMBH, as seen in cosmological simulations of SMBH growth (e.g., Yoo et al. 2007; Kulier et al. 2015; Wellons et al. 2015). Therefore, it is a valid assumption to account for evolution only in the M_bulge and σ directions for the relic galaxies and neglect evolution in the SMBH mass direction. As a final comment, considering the lower black hole estimates would further support our scenario: the ÜMBH candidates would still be extreme outliers in the ${M}_{\bullet }$−M_bulge relation but they would be located even closer to the local scaling relations after accounting for the missing mass.

In a nutshell, our scenario is schematically depicted in Figure 4. The left panel shows the normal evolution that a massive galaxy follows over cosmic time to become a large elliptical galaxy today, with the two-phase of formation (Oser et al. 2010). The right panel illustrates the expected evolutionary track for the relic/outlier galaxies. We remind the reader that this is a cartoon to illustrate the proposed scenario, where a number of assumptions have been simplified, as many questions in the (co-)evolution of SMBHs and their host galaxies remain unsolved. For example, the exact position in the early universe phase. It is not yet clear if the SMBH and the host started forming at the same time, but several observational and theoretical works suggest that the SMBH started forming first, as inferred from the mild deviations of these objects in the ${M}_{\bullet }$–σ (e.g., Walter et al. 2004; Jahnke et al. 2009, Merloni et al. 2010; Reines et al. 2011; Khandai et al. 2012; Petri et al. 2012). Our scenario also seems to support this case, as by the end of the $z\gtrsim 2$ phase the SMBH should be almost fully in place, being larger than expected from the galaxy stellar mass. Although we are not in the position to confirm this from our derived SFHs, we can pose a lower limit for the SMBHs growth at ∼10 Gyr, when the star formation activity in the host galaxy is halted. At this point, one would expect the massive compact galaxy to move into the second phase of formation, balancing out the deviations with its SMBH, which roughly evolves, by growing largely in size and mildly in mass through mergers. This will eventually place the massive galaxy in the local ${M}_{\bullet }$− M_bulge relation.

Zoom In Zoom Out Reset image size

We are aware of the limited sample employed here, but so is the published sample of galaxies with detected SMBHs and accurate measurements or the sample of relic galaxy candidates. In a recent effort, Saulder et al. (2015) have revisited the number of relic candidates in the SDSS, enlarging the one from Trujillo et al. (2009) to 76 candidates. The next step is to find out how many of these galaxies host an SMBH and if so, if they are ÜMBHs. This will be achieved with the upcoming era of the largest telescopes, when SMBH detections and measurements will be improved. In the meantime, surveys such as the hetmgs are ideal compilations for new SMBH detections. A deeper understanding on the formation of these extreme outliers (both in the upper and the low-mass end) is crucial to test the current theories of galaxy formation and evolution and cosmological models.

In this work, we have studied a sample of eight compact massive galaxies selected from the hetmgs that are candidates for hosting ÜMBHs and that are therefore considered extreme outliers from the correlation that the SMBHs and the mass of their host galaxies follow. We propose that they deviate because the host galaxies did not structurally evolve in the way expected for massive galaxies due to their relic nature. Our main results are briefly summarized here.

• 1.  
  We find that seven out of the eight ÜMBH host candidates fulfill the criteria to be considered relic galaxies (i.e., ${R}_{{\rm{e}}}\lt 2$ kpc and t > 10 Gyr), while the other one is a dubious case. Therefore, selecting galaxies that are extreme outliers in the SMBH mass scaling relations could represent a new way to find the elusive relic galaxies in the nearby universe.
• 2.  
  When we plot the ${M}_{\bullet }$–σ and ${M}_{\bullet }$–M_bulge relations, the loci of the galaxies hosting ÜMBHs should represent the position of the massive galaxy population at z ∼ 2. As cosmic time progresses, galaxies following normal growth paths should evolve toward galaxies with larger host masses in those planes.
• 3.  
  We can pose a lower limit to the age of the SMBH growth at ∼10 Gyr, based on the derived SFHs.

This work was supported by the Japan Society for the Promotion of Science (JSPS) Grant-in-Aid for Scientific Research (KAKENHI) Number 23224005 and by the Spanish Ministerio de Economía y Competitividad (MINECO; grants AYA2009-11137, AYA2011-25527, and AYA2013-48226-C3-1-P).

We briefly summarize here how the full-spectral-fitting code STARLIGHT works, but the reader is referred to the source code papers for a more detailed explanation (Cid Fernandes et al. 2005, Cid Fernandes & González Delgado 2010). The code models the extinction as due to foreground dust, and different reddening-laws can be selected to correct from Galactic extinction. Then, it finds the fraction x_j that a given jth SSP contributes to the total flux of the galaxy (normalized to a certain wavelength ${\lambda }_{0}$), creating a synthetic spectrum. In other words, it creates a combination of SSPs that best resemble the observed galaxy spectrum and that minimize the ${\chi }^{2}$. The first two parameters presented in Table 2 are indicative of the quality of the fit. [${\chi }^{2}$] represents the total ${\chi }^{2}$ divided by the number of λʼs used in the fit. The second parameter, adev, is a proxy for the mean deviation over all fitted pixels. Good values for a fit are those with adev below 2%–3%. A visual inspection of the fits and their residuals confirms the quality of the results obtained (see Figures 5 and 6).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

From each x_j, we can derive its contribution in mass, ${\gamma }_{j}$, directly from the mass-to-light ratio of each individual SSP (from the models), to obtain the desired mass-weighted estimates such as the star formation episodes and mean stellar populations parameters, which are derived as

$$\begin{eqnarray}&&{\langle {t}_{\star }\rangle }_{M}=\displaystyle \sum _{j=1}^{{N}_{\star }}{\gamma }_{j}\;{t}_{j};\;\mathrm{for}\;\mathrm{the}\;\mathrm{age}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{\langle {Z}_{\star }\rangle }_{M}=\displaystyle \sum _{j=1}^{{N}_{\star }}{\gamma }_{j}{Z}_{j};\;\mathrm{for}\;\mathrm{the}\;\mathrm{metallicity}.\end{eqnarray} \tag{ 3 }$$

We can therefore reconstruct the SFH of the galaxy by summing up all of the fractions of stellar mass created at a given epoch, as shown in Figures 5 and 6. They present the whole spectroscopic sample data and their derived SFHs. As in Figure 3, the left panels show the galaxy spectrum with fit from STARLIGHT and the right side ones, the derived SFHs in terms of stellar mass.