THE ENVIRONMENTS OF ACTIVE GALACTIC NUCLEI WITHIN THE zCOSMOS DENSITY FIELD

The local environment of galaxies harboring active galactic nuclei (AGNs) and QSOs has long been thought to play a potential role in triggering mass accretion onto supermassive black holes (SMBHs). With many of the properties of galaxies (e.g., morphology, color, star formation rate (SFR)) clearly dependent on environment (e.g., Kauffmann et al. 2004; Baldry et al. 2006; Cooper et al. 2006; Cucciati et al. 2009, in preparation) and the possibility of a common history of mass assembly for SMBHs and their host bulges (e.g., Granato et al. 2004; Bower et al. 2006; Hopkins et al. 2008; Somerville et al. 2008), we expect that AGNs may prefer to reside in specific environments most nurturing for their growth. Identifying environmental factors might allow us to determine the physical mechanism(s) responsible for driving accretion such as major mergers of galaxies that has been demonstrated through numerical simulations to be able to remove angular momentum from rotationally supported gas thus transferring mass to the nuclear regions (e.g. Barnes & Hernquist 1996; Mihos & Hernquist 1996). For example, there is evidence that AGNs reside in dark matter halos with masses M_halo ∼ 10^12–13   M_☉ (Porciani et al. 2004; Hopkins et al. 2007; Mandelbaum et al. 2008; Pasquali et al. 2009; Bonoli et al. 2008), a mass regime comparable to the group-scale environments thought to be fertile ground for galaxies to coalesce. As well, high-density regions such as massive clusters of galaxies are expected to be inhospital environments for AGN (Dressler et al. 1985) given the strong empirical association between AGN activity and concurrent star formation (e.g., Kauffmann et al. 2003; Jahnke et al. 2004; Silverman et al. 2009) although counterevidence exists for the radio-loud population (Hill & Lilly 1991; Best et al. 2007).

Environmental studies to date have presented seemingly disparate results most likely due to varying selection methods and physical scales used to characterize environment. Using a large sample of narrow-line AGN from the Sloan Digital Sky Survey (SDSS), Miller et al. (2003) find that there is no environmental dependence on AGN activity using a magnitude-limited sample of galaxies and meaning environment on scales of around a few megaparsecs. Upon further investigation, Kauffmann et al. (2004) find that an environmental dependence, similar to star formation, emerges when implementing a selection based on stellar mass and considering the luminosity of the AGN. Supportive of this scenario, Coil et al. (2007) find that quasars in the DEEP2 survey fields have environments similar to blue galaxies. On the other hand, luminous quasars from the SDSS have an overabundance of galaxies within their vicinity on smaller scales (less than 0.1 Mpc; Serber et al. 2006) that is in agreement with clustering analysis based on quasar pairs (Hennawi et al. 2006) but may still have significant biases in the employed methods (Padmanabhan et al. 2008). Furthermore, the enhancement in star formation attributed to galaxy mergers on scales of 0.01–0.1 kpc is not reflected in AGN accretion (Li et al. 2008) for which the authors conclude may be due to varying timescales with star formation preceding AGN activity (e.g., Schawinski et al. 2007) occurring within a more relaxed host galaxy.

X-ray-selected surveys with both Chandra and XMM-Newton now enable the study of the environment of AGNs including the obscured population at higher redshifts (z ≳ 0.3) where optical selection of narrow-line AGN is difficult. Grogin et al. (2005) first looked into the environments of X-ray-selected sources detected in the Chandra Deep Fields and found that the near-neighbor counts were identical to the galaxies without AGN. This led the authors to conclude that mergers were not the physical mechanism triggering mass accretion especially since the AGN host galaxies were no more asymmetric than the average galaxy of equivalent luminosity (see Gabor et al. 2009, for a morphological study of AGN hosts in COSMOS). While Georgakakis et al. (2007) claim based on a large spectroscopic sample of galaxies from DEEP2 that X-ray-selected AGNs at z ∼ 1 prefer higher-density environments, their results are consistent with those of galaxies having similar host properties (i.e., absolute magnitude, rest-frame color). From a complementary perspective, Martini et al. (2007) have measured the AGN content of galaxy clusters and found no significant difference with the fraction of field galaxies hosting AGN. Recently, Gilli et al. (2009) found clustering lengths of AGN in the COSMOS field comparable to massive galaxies and concluded that the clustering signal is reflective of SMBHs preferring to reside in galaxies with M_* > 3 × 10¹⁰  M_☉. To date, an environmental analysis of X-ray-selected AGN based on both a large spectroscopic survey of galaxies and careful consideration of selection to disentangle the degeneracy between mass and environment seen in galaxy studies (e.g., Baldry et al. 2006; van der Wel 2008; Cucciati et al. 2009, in preparation) has not yet been attempted.

Here, we utilize the rich multiwavelength observations of the COSMOS field (Scoville et al. 2007) to determine the role of environment in triggering AGN at 0.1 < z < 1.0. The COSMOS survey is roughly a 2 deg² region of the sky selected to have full coverage with all major observatories both from the ground (i.e., Subaru, Very Large Telescope (VLT)) and space (e.g., Hubble Space Telescope (HST), Spitzer, XMM-Newton). The zCOSMOS survey (Lilly et al. 2007; S. J. Lilly et al. 2009, in preparation) targets objects for optical spectroscopy with the VLT in two separate observing programs. A "bright" sample (i_ACS < 22.5) is observed with a red grism to provide a wavelength coverage of 5500–9500 Å ideal to identify galaxies (L_*) up to z ∼ 1.2. A deeper program, not utilized in the present study, targets faint galaxies (B < 25), selected to be in the redshift range 1.5 ≲ z ≲ 2.5 using a blue grism for an effective wavelength coverage of 3600 < λ < 6700 Å. For the present study, we select galaxies based on reliable spectroscopic redshifts from the zCOSMOS "bright" program and their stellar mass estimates based on broadband photometry. Those that host AGNs are identified by their X-ray emission as detected by XMM-Newton (Hasinger et al. 2007; Cappelluti et al. 2007, 2009). We use the nearest-neighbor approach to determine the projected local galaxy density using the three-dimensional galaxy distribution as fully described in Kovač et al. (2009). In addition, the zCOSMOS optically selected group catalog (Knobel et al. 2009) offers a complementary perspective on the role of environment. The low optical luminosity and obscured AGN further enable us to determine if the host galaxies of AGN exhibit a trend similar to the color–density or SFR–density relations found for nonactive galaxies. Finally, we refer the reader to Silverman et al. (2009) that presents the properties (i.e., stellar mass, SFR, rest-frame color) of the hosts of X-ray-selected AGN in the COSMOS field.

Throughout this work, we assume H₀ = 70 km s⁻¹ Mpc⁻¹, Ω_Λ = 0.75, Ω_M = 0.25 and use AB magnitudes.

2.1. Parent Galaxy Sample

In order to compare the environments of AGN hosts with those of normal galaxies, we use for the latter the well-defined "parent sample" of galaxies up to z ∼ 1 from the zCOSMOS 10k spectroscopic "bright" catalog, as done in Silverman et al. (2009). Specifically, we select 7234 galaxies with an apparent magnitude i_ACS < 22.5, and a redshift 0.1 < z < 1.0 having a spectroscopic redshift quality flag greater than or equal to 2.0 (see Lilly et al. 2007) that amounts to a confidence of ∼99% for the overall sample. The spatial sampling as shown in Figure 1 is fairly uniform across the central 1 deg² while the completed zCOSMOS 20k catalog will fill in the gaps mainly located along the perimeter of the COSMOS field. Further details on the spatial sampling and quality assurance can be found in Lilly et al. (2007) and S. J. Lilly et al. (2009, in preparation) as well as a complete description of target assignments, data acquisition, and the subsequent reduction procedure based on the VIMOS Interactive Pipeline and Graphical Interface package (VIPGI; Scodeggio et al. 2005).

Zoom In Zoom Out Reset image size

Stellar masses, including rest-frame absolute magnitudes (M_U, M_V) in the AB system, are derived from fitting stellar population synthesis models from the library of Bruzual & Charlot (2003) to both the broadband optical (CFHT: u, i, K_s; Subaru: B, V, g, r, i, z; Capak et al. 2007) and near-infrared (Spitzer/IRAC: 3.6  μ, 4.5  μ; Sanders et al. 2007) photometry using a chi-square minimization for each galaxy. The measurement of stellar mass (M_*) includes (1) the assumption of a Chabrier initial mass function²⁰, (2) a star formation history with both a constant rate and an additional exponentially declining component covering a range of time scales (0.1 < τ < 30  Gyr), (3) extinction (0 < A_V < 3) following Calzetti et al. (2000), and (4) solar metallicities. In Figure 2a, we show the distribution of stellar mass versus redshift for our sample. Further details on mass measurements of zCOSMOS galaxies can be found in M. Bolzonella et al. (2009, in preparation), Meneux et al. (2009), and L. Pozzetti (2009, in preparation).

Zoom In Zoom Out Reset image size

To minimize any selection biases, we determine a minimum mass threshold that all galaxies must satisfy. The mass limit is set to ensure a fairly complete representation of both blue and red galaxies at all redshifts considered. Meneux et al. (2009) estimate based on a series of mock catalogs from the millennium simulation that the zCOSMOS "Bright sample" is essentially complete for galaxies with log M_* ≈ 10.6 at z = 0.8 while the completeness drops to ∼50% at z = 1. We impose a slightly lower-mass limit of log M_* > 10.4 (units of   M_☉) to provide a fair representation of galaxies covering the full range of rest-frame color U − V up to z ∼ 1 and ensure an adequate sample that host AGN (see below). In Figure 2b, it is evident that this mass limit is essentially imposed by the red galaxy population due to the initial selection on apparent magnitude. Above this mass limit, we have a sample of 2457 galaxies, over the redshift range 0.1 < z < 1.0 (Table 1).

Table 1. Sample Statistics-Density Field Analysis

Sample	Massa Range	Redshift Range	No. of Galaxies	No. of AGNs (L0.5–10 keV)	Use
All	...	0.1 < z < 1.0	7234	147 (>42)	Total sample
A	>10.4	0.1 < z < 1.0	2457	63 (42.48–43.7); 88 (>42.48)	AGN fraction
B	10.4–11	0.1 < z < 1.0	1971	48 (42.48–43.7)	AGN fraction
C	>11	0.1 < z < 1.0	486	14 (42.48–43.7); 20 (>42.48)	AGN fraction
D	>10.2	0.1 < z < 0.8	2482	77 (42–43.7)	Color–density relation

Notes. ^aUnits of log M_☉.

Download table as:  ASCIITypeset image

Further spectroscopic measurements such as emission and absorption line strengths, and continuum indices are performed through an automated pipeline ("platefit_vimos"; Lamareille et al. 2008) similar to that performed with the SDSS (Tremonti et al. 2004). For the present study, we specifically use the [O ii]λ3727 emission-line luminosity, corrected for slit loss and an AGN contribution if present, to determine the mass-weighted (specific) SFR (sSFR) using the empirical relation given in Moustakas et al. (2006) thus allowing us to investigate the environmental dependence of star formation for galaxies hosting AGN. We refer the reader to Maier et al. (2009) and Silverman et al. (2009) for more details regarding spectral measurements of zCOSMOS galaxies and the removal of AGN emission based on the observed or inferred [O iii]λ5007 line flux (M. Bolzonella et al. 2009, in preparation; B. Meneux et al. 2009, in preparation).

2.2. AGN Identification

2.3. zCOSMOS Galaxy Density Field

A major aim of the zCOSMOS survey is to reconstruct the three-dimensional density field using the "bright" sample to characterize the environment up to z ∼ 1 and discern its impact on galaxy evolution. Here, we briefly outline the methodology and refer the reader to Kovač et al. (2009) for full details regarding the galaxy density estimates in the zCOSMOS field. The procedure is based on the algorithm "Zurich Adaptive Density Estimator" (ZADE) that utilizes spectroscopic (10k) and photometric redshifts (30k) for accurate distances to practically all galaxies with i_ACS < 22.5. For galaxies without spectroscopic redshifts, their photometric redshift likelihood functions are adjusted using galaxies in the spectroscopic catalog that are closely along the line of sight and within the effective aperture. This procedure effectively accounts for the incomplete sampling (∼30%) of the current 10k catalog thus improving density estimates throughout the zCOSMOS volume. We use density estimates based on a flux-limited catalog and adaptive apertures (based on nearest neighbors) in order to maintain a reasonable sample of AGNs over the full redshift range (0.1 < z < 1.0) at the expense of having a redshift-dependent smoothing scale (see Figure 6 of Kovač et al. 2009). To minimize redshift-space distortions induced by the intrinsic velocity dispersion of groups and clusters, a projection of ±1000 km s⁻¹ in redshift space is implemented. Finally, we express the quantitative measure of the environment as an overdensity (δ; where 1 + δ = ρ〈ρ(z)〉⁻¹) at the position of each galaxy relative to the mean density (〈ρ(z)〉) at a given redshift. Error estimates are obtained by comparison with reconstructed overdensities from mock catalogs (Kitzbichler & White 2007) extracted from the Millennium simulation (Springel et al. 2005) and tailored to match the current zCOSMOS sample. Typical errors on log(1 +  δ) are between 0.1 and 0.15 over a wide range of overdensity (see Figure 4 of Kovač et al. 2009).

It is important to consider if any difference in spatial sampling exists between the AGN and randomly selected galaxies. Such an effect may be inherently induced by the VIMOS automated slit assignment software when designating sources as "compulsory," as done for a subset of the X-ray sources. We do find that the number of randomly observed galaxies in the vicinity of "compulsory" targets is lower than that of the random sample. This is slightly noticeable in Figure 1 where the surface density of AGNs is not as concentrated toward the center of the field as the galaxies. This effect is both small and compensated by the "ZADE" approach thus we do not expect any significant bias in our density estimates.

In Figure 3, we show the zCOSMOS density field in comoving coordinates with the location of galaxies and those hosting AGN as marked. For display purposes, the overdensity distribution, estimated on a grid (Δα = Δδ = 1 arcmin, Δz = 0.002) has been interpolated to a comoving scale of 1.4 Mpc. It is clearly evident that the COSMOS field encompasses a wide range of environments from voids to dense structures. The survey area (∼2 deg²; ∼80 comoving Mpc at z ∼ 1) is still narrow enough that sheets, filaments, and walls appear to cut across the field.

Zoom In Zoom Out Reset image size

2.4. zCOSMOS Galaxy Groups

In Figure 4, we present the distribution of zCOSMOS galaxies and those hosting X-ray-selected AGN (Table 1; sample A) as a function of their environment (i.e., overdensity). The results are shown using the distance to the nth nearest neighbor (5th, 10th, and 20th) separately that probes a range of comoving scales (∼1–4 Mpc). In all panels (Figures 4(a)–(c)), we generally see that AGNs reside in a broad range of environments (e.g., panel (a); 0 ≲ log 1 + δ ≲ 1.5) remarkably similar to the parent galaxy population of equivalent stellar mass (log M_* > 10.4) as substantiated by the low probability (less than 20% in each case), based on a Kolmogorov–Smirnov (K–S) test, that we can reject the null hypothesis (i.e., two distributions are drawn from the same population). To avoid any possible luminosity-dependent systematics, we perform additional K–S tests in fine redshift bins (Δz = 0.1) over the redshift range 0.3 < z < 0.8; the probability (P = 0.25, 0.49, 0.54, 0.39, and 0.03) that the two distributions (overdensities based on the 5th nearest neighbor) are different conclusively supports this finding.

Zoom In Zoom Out Reset image size

To further investigate if any environment influences are present, we measure the fraction of galaxies that host AGN as a function of overdensity. We follow the technique discussed in Section 3.1 of Lehmer et al. (2007) to determine the AGN fraction for our parent population of galaxies. This method properly accounts for the spatially varying sensitivity limits of the XMM observations of the COSMOS field (Figure 1). The necessity of this approach is demonstrated in Figure 1 of Silverman et al. (2009) that shows the limiting X-ray luminosity as a function of redshift for the entire galaxy sample and the measured X-ray luminosities of those galaxies harboring AGN. The sensitivity of the XMM coverage is remarkably uniform as shown by the relatively narrow distribution of the X-ray limits at each redshift. To properly account for the luminosity–redshift relation, we determine the contribution of each AGN separately to the total fraction. The AGN fraction (f; see Equation (1) below) is determined by summing over the full sample of AGN (N) with N_gal,i representing the number of galaxies in which we could have detected an AGN with X-ray luminosity Lⁱ_X. The sampling rate of the random galaxies (S_gal, 29.8%) and AGN (i.e., X-ray sources, S_x; 71.9%) are incorporated since these differ due to the fact that 54% of the X-ray sources are designated as "compulsory targets" when designing masks for VIMOS. We estimate the associated 1σ error (Equation 2) using binomial statistics where N^eff_AGN is the number of AGNs that would be detected if all galaxies have the same limiting X-ray sensitivity and the sample of AGN was randomly selected. Here, we only consider AGN with 42.48 < log L_{0.5–10 keV} < 43.7. The lower limit ensures that we have a statistically significant sample of parent galaxies (≳700) that could host each AGN while the upper limit restricts the sample to low-to-moderate luminosities thus securing the accuracy of their host-galaxy masses. We refer the reader to Silverman et al. (2008a) for further details and results employing this method:

In Figures 4(a)–(c), we give the results of this exercise with the fraction of galaxies hosting AGN as measured in three bins of overdensity with widths set to have equal numbers of galaxies therein. In agreement with the previous analysis, we find that the AGN fraction is not strongly enhanced at any particular value of overdensity. The AGN fraction is low (∼2%–4%) compared to similar studies (e.g., Kauffmann et al. 2004) and due to our selection in a narrow range of X-ray luminosity. We finally remark that an underlying dependence on environment for a subset of our sample is realized as demonstrated in the next section and possibly seen here by the slight trend in the AGN fraction (Figures 4(b) and (c)).

We further test whether an environmental influence is present for AGN of a specific X-ray luminosity or hardness ratio (HR). The HR is a means of easily characterizing the X-ray spectrum (i.e., level of absorption) by measuring the ratio of X-ray counts in the hard band (H; 2.0–10 keV) relative to the soft band (S; 0.5–2.0 keV): HR= (H−S)/(H+S). In Figure 5, we plot local overdensity versus X-ray luminosity while highlighting hard (HR > − 0.2) sources. We find that there is no correlation between galaxy environment and the intrinsic AGN emissivity or any absorbing material.

Zoom In Zoom Out Reset image size

3.1. High-Mass Galaxies

We investigate whether there exists a relation between AGN activity and environment for galaxies of a specific stellar mass. Our motivation is based on the mass-dependent environmental relationship seen in SDSS studies (Kauffmann et al. 2004). In contrast to our analysis in the previous section, we include more X-ray luminous AGN (log L_{0.5–10 keV} > 42.48) that have optical spectra dominated by their host galaxy as determined by visual inspection thus slightly improving our statistics for the highest mass galaxies. In Figure 6, we show plots equivalent to those presented in the previous section and split into two mass bins (panels (a)–(c): 10.4 < log M_* < 11; panels (d)–(f): log M_* > 11) with the statistics given in Table 1 (Samples B, C). For the lower-mass interval (Figures 6(a)–(c)), we find that the results are consistent with our previous findings and the AGN fraction is remarkably similar over the full range of overdensities. On the other hand, the overdensity distribution (Figures 6(d)–(f)) of galaxies hosting AGN is dissimilar to the underlying massive galaxy population due to the lack of AGNs in higher-density environments log(1 + δ>0.5). The difference in these distributions is significant at the 2.3–2.5σ level based on the probabilities (0.99, 0.98) from K–S tests using either the 5th or 10th nearest neighbor. We note that these results are still apparent when restricting the analysis to lower-luminosity AGN although at a slightly reduced level of significance (2σ). Therefore, the fraction of galaxies hosting AGN is higher in underdense regions (log(1 + δ < 0.5) since the offset between the distributions is substantially greater than typical errors in overdensity. These results may further explain the discrepancy between the environmental studies based on the SDSS (Miller et al. 2003; Kauffmann et al. 2004). We remark that our current sample of massive galaxies (log M_* > 11) in zCOSMOS with AGN is small (20) and will improve with the full zCOSMOS 20k catalog and deeper Chandra observations.

Zoom In Zoom Out Reset image size

3.2. Dependence on Stellar Content of AGN Hosts

The result that AGN prefers to reside in lower-density environments for those having massive host galaxies is reminiscent of the well-known color–density relation for galaxies in general. To explore this aspect further, we plot the distribution of rest-frame color U − V versus overdensity in Figure 7. For the following analysis, we decrease our mass limit log M_* > 10.2 in order to improve upon the numbers of AGN hosts with blue colors thus requiring us to reduce the maximum redshift to z = 0.8 to not induce a color bias (Table 1; Sample D). In panel (a), galaxies exhibit an environmental influence with red galaxies (U − V > 1.6) having a prominent extension of their overdensity distribution toward higher values log(1 + δ₅ ≳ 1.0). The well-known trend of increasing density for redder galaxies (i.e. color–density relation) is clearly seen in panel (b) where we measure the mean overdensity for galaxies in bins of rest-frame color (black points).

Zoom In Zoom Out Reset image size

We test whether galaxies hosting AGN exhibit a similar color–density relation as described above. AGN hosts are marked appropriately in Figure 7. From their mean overdensities (panel (b)), AGNs appear to follow a similar relation to the nonactive galaxies although the errors are substantial with a significance of 1.7σ in the difference (see Georgakakis et al. 2007, for an equivalent analysis and conclusion). In support of the color–density relation for AGN hosts, we investigate the connection between ongoing star formation and environment. In Figure 7(c), we plot the mass-weighted "specific" SFR (sSFR), a quantity based on [O ii]λ3727 luminosity, versus galaxy overdensity. The mean overdensity in bins of sSFR is calculated and shown by the large black dots in panel (d). We find a similar result and uncertainty as above with the environment being related to the stellar content of the hosts of AGN. It is worth highlighting that a clear dichotomy is present between the importance of both mass and environment in determining the likelihood of a galaxy's properties (see Cucciati et al. 2009, in preparation for further analysis based on zCOSMOS galaxies).

Coupled with the results of our companion study (Silverman et al. 2009), AGN prefers to reside in galaxies undergoing star formation irrespective of their environment. We illustrate this conclusion in Figure 8 where the fraction of galaxies hosting AGN is shown as a function of rest-frame color U − V and overdensity. Galaxies bluer (U − V ≲ 1.6) than those that have stopped forming stars (i.e., along the red sequence) have higher levels of AGN activity. We further find here that this result holds for AGN hosts residing in either lower- or higher-density environments.

Zoom In Zoom Out Reset image size

By looking in detail at the most massive galaxies (log M_* > 11; Figure 9), we can further understand the relation of AGNs to their environment and the presence of star formation. The locus of AGNs in the color–density plane (panel (a)) is offset from that of massive galaxies due to their bluer colors and lower overdensities. Clearly, the distinction between the overdensity distribution of AGN hosts and their parent population is due to the fact that AGNs do not reside in red galaxies (panel (b)) that tend to live in denser environments but rather require fuel for accretion that tends to drive subsequent star formation. Therefore, the environments of AGNs hosts are similar to star-forming galaxies (c). We interpret these results as evidence that massive galaxies are more likely to have accreting SMBHs if not subjected to harsher environments that can effectively expel gas, due to processes related to galaxy interactions, thus impacting star formation and subsequent AGN activity.

Zoom In Zoom Out Reset image size

The presence of AGN within well-defined structures such as groups (Georgakakis et al. 2008), clusters (Martini et al. 2007; Kocevski et al. 2009; Lehmer et al. 2009), or large-scale sheets/filaments (Gilli et al. 2003; Silverman et al. 2008a) as compared to the field offer an alternative perspective on the environmental impact on AGN activity. To do so, we utilize the catalog of 800 galaxy groups (Knobel et al. 2009) as described in Section 2.4. As done in previous analyses, we restrict the sample to galaxies with stellar masses above 2.5 × 10¹⁰  M_☉ thus resulting in a sample of 2444 galaxies with a designation as either a "field" or "group" galaxy. In Table 2, we list the numbers of galaxies in each subclass, those that host AGNs (log L_{0.5–10 keV} > 42.48) and the derived AGN fraction as detailed below.

We simply measure the fraction of galaxies hosting AGN for the "group" and "field" populations separately to test whether any environmental dependence exists. Over the full redshift range (0.1 < z < 1.05), we measure the fraction to be 3.28% ± 0.43% for field galaxies. For galaxies in groups, we find an AGN fraction of 2.59% ± 0.55%, based on two or more spectroscopically identified group members, slightly less than that in the field although not statistically significant. To account for redshift-dependent effects and spatial sampling in the group catalog, we further isolate groups having an effective (i.e., corrected) richness (N_eff) defined as the number of galaxies above a given magnitude threshold (see Section 4.2 of Knobel et al. 2009, for details) that can be detected over the full redshift range considered here. This limit is based on the absolute magnitude in the b-band (M_b,lim < 20.5 − z) and includes a redshift term to account for the luminosity evolution of galaxies. In Figure 10(a), we show the distribution of effective richness for all galaxies (log M_* > 10.4) residing in groups that illustrate the scale of the zCOSMOS groups. In addition, the distribution of group mass (Figure 10(b)) is given. Briefly, the group mass is the mean mass of halos in the mock catalog that are associated with groups of a specific effective richness (N_eff). A redshift dependence is inherent given that effective richness is based on an observed quantity (i.e., apparent magnitude). Full details regarding these derived properties can be found in Knobel et al. (2009). Considering a selection on effective richness, we find a similar fraction of galaxies hosting AGN as compared to the field with no obvious dependence on the effective richness (see Table 2) of the group. Furthermore, there is no significant difference in the X-ray luminosity or host-galaxy color (rest-frame U − V) distributions of AGNs in or out of groups (see Georgakakis et al. 2008, for similar analyses). These results are in agreement with our previous findings based on the density field that AGNs in galaxies with M_* > 10^10.4  M_☉ do not show a strong dependence on environment.

Zoom In Zoom Out Reset image size

To determine if any signs of evolution exist, we plot the fraction of group galaxies hosting AGN in two redshift bins (Figure 11). Here, we select groups having N_eff ⩾ 2 as described above. The AGN fraction is remarkably similar between for galaxies in groups and in the field for both redshift intervals. The increase in the AGN fraction in the higher redshift bin for both the field and groups, is most likely due to the overall positive luminosity evolution of the AGN population (e.g., Barger et al. 2005; Silverman et al. 2008b). Therefore, we do not find an enhancement in the AGN content of groups with redshift as strong as reported for clusters (Eastman et al. 2007) above the level expected by the AGN luminosity function. As previously mentioned, we will improve upon the current sample, as justified by the substantial errors, in a future investigation by using the zCOSMOS 20k catalog and deeper Chandra observations (Elvis et al. 2009) that provide additional AGNs of moderate luminosity (42 ≲ L_X ≲ 43) over the redshift range 0.5–1.0. A larger sample will enable us to isolate high-mass galaxies (M_* > 10¹¹  M_☉) as done with the density field.

Zoom In Zoom Out Reset image size

We have assessed whether the environment plays any role in regulating mass accretion onto SMBHs, on scales of a few megaparsecs, using 7234 galaxies (i_ACS < 22.5; 0.1 < z < 1.0) from the zCOSMOS survey ("bright" program). To do so, a parent sample of 2457 galaxies is carefully constructed based on stellar mass (log M_* > 10.4) with no obvious color (U − V) bias. Galaxies undergoing substantial accretion onto a SMBH are identified by the presence of an AGN with X-ray emission detected by XMM-Newton; we find 147 AGN with 101 residing in host galaxies having log M_* > 10.4. With two characterizations of the local environment, we measure the fraction of galaxies as a function of their galaxy overdensity based on the zCOSMOS density field (Kovač et al. 2009) and presence in optically selected galaxy groups (Knobel et al. 2009).

Based on the zCOSMOS density field, we find that the AGNs, with exception of those residing in the most massive host galaxies (log M_* > 11), span a broad range of environments (see Georgakakis et al. 2007 and Montero-Dorta et al. 2009, for similar conclusions), from the field to massive groups, equivalent to galaxies of similar stellar mass. Our findings do appear to be consistent with low redshift studies based on narrow-line AGN from the SDSS (Miller et al. 2003; Kauffmann et al. 2004; Constantin et al. 2008) that taken together demonstrate that the environments depend on the stellar mass of their host galaxy. The lack of an environmental dependence on AGN activity over the lower-mass regime (10.4 <  log M_* <  11) in zCOSMOS is compatible with the results of Miller et al. (2003) since their magnitude-limited sample, dominated by low-luminosity AGN, most likely includes many lower-mass galaxies. An environmental dependence with AGN activity less common in higher-density regions emerges for "strong AGN" residing in galaxies of higher mass (Kauffmann et al. 2004; Constantin et al. 2008) and is also evident in our study based on zCOSMOS galaxies with log M_* > 11. We contend that the discrepant findings based on SDSS studies at low redshift may be due to the differences in the underlying mass distribution of galaxies rather than the differences of the AGN selection.

We find that the presence of young stars in the host galaxies of AGN (Kauffmann et al. 2003; Jahnke et al. 2004; Silverman et al. 2009) is also indicative of their environments. Galaxies hosting AGN in zCOSMOS follow a similar color–density relation to nonactive galaxies (Cucciati et al. 2009, in preparation) with blue hosts (U − V < 1.6) having a lower mean overdensity than red galaxies. An equivalent relation based on SFR or that weighted by stellar mass (specific) supports this assertion. As fully discussed in Cucciati et al. (2009, in preparation), the sSFR is seen to be more strongly dependent on environment than SFR due to the underlying mass dependence discussed above. The majority of massive (log M_* > 11) AGN hosts having colors bluer than the underlying galaxy population agrees with their residence in lower-density environments. Therefore, we conclude that the incidence of AGN activity as a function of environment depends on the properties of its host galaxy namely its mass content both in the form of stars and gas in the interstellar medium.

Furthermore, we carry out a complementary analysis of the environments of AGN by utilizing the optically selected catalog of galaxy groups (Knobel et al. 2009) in zCOSMOS. Such groups have halo masses ∼10¹²–10¹⁴  M_☉ similar to those that host the more-luminous quasars (e.g., Porciani et al. 2004; Pasquali et al. 2008; Bonoli et al. 2008) and less massive than clusters at low redshift. We measure the incidence of AGN activity in galaxies in groups compared to that in the "field." This method effectively enables us to probe smaller physical scales (≲1 Mpc) not possible with the density field due to the spatial resolution of ±1000 km s⁻¹ along the line of sight. An excess signal around quasars, based on either near-neighbor counts (Serber et al. 2006) or the quasar correlation function (Hennawi et al. 2006), highlights the importance of probing scales ∼100 kpc. Although, the environments of 31 X-ray-selected AGN (Waskett et al. 2005) with 0.4 < z < 0.6 from the Canada-France-Redshift Survey fields are indistinguishable from a well-matched control sample of galaxies on scales around 30–500 kpc.

We find in zCOSMOS that AGNs are essentially equally likely to reside in or out of a galaxy group irrespective of its properties (i.e., effective richness). The fraction of group galaxies (log M_* > 10.4) hosting an AGN is similar to "field" galaxies (∼3%). This is in agreement with both our results based on the density field for galaxies spanning the full mass range log M_* > 10.4, and those of Georgakakis et al. (2008) which consider the host-galaxy properties. We note that the current AGN sample in zCOSMOS residing in galaxy groups is too small to investigate if an environmental effect for the most massive galaxies exists as exercised using the density field. Our lack of a strong environmental effect is also consistent with the incidence of X-ray-selected AGN in galaxy clusters (Martini et al. 2007) as compared to the field. Also, an increase in the AGN fraction in the higher of two redshift bins is significantly smaller than the enhancement reported for galaxy clusters (Eastman et al. 2007) and in our case most likely due to the luminosity evolution of the AGN population since both the group and field samples show an increase of equal magnitude.

We conclude that internal processes (Hopkins & Hernquist 2006) play an important role in the growth of SMBHs during an AGN phase of moderate luminosity (L_X ∼ 10⁴³ erg s⁻¹; "Seyfert mode") that accounts for the bulk of the Cosmic X-ray Background (e.g., Gilli et al. 2007). Major mergers of galaxies, possibly relevant for the more-luminous quasar phenomenon, may not be the primary mechanism for fuelling these AGN due to the lack of any enhancement of activity in specific environments likely to be conducive for merging (i.e., galaxy overdensities comparable to the small group scale). In the opposite sense, we do find that for the most massive galaxies (log M_* > 11) the environment plays a role possibly through various physical processes (e.g., tidal stripping, harassment) in higher-density environments that concurrently shuts down star formation and accretion onto supermassive black holes. This highlights the requirements for a galaxy to host an accreting SMBH, as observed by their X-rays, a massive bulge (e.g., Sanchez et al. 2004; Grogin et al. 2005; Silverman et al. 2008a; Gabor et al. 2009), and a sufficient fuel supply as indicated by the young stars usually present in AGN host galaxies (Kauffmann et al. 2003; Silverman et al. 2009).

We thank the referee and Paul Martini for constructive comments that significantly improved the paper. This work is fully based on observations undertaken at the European Southern Observatory (ESO) Very Large Telescope (VLT) under the Large Program 175.A-0839 (P.I.: Simon Lilly).

Facilities: XMM, VLT:Melipal (VIMOS)