With an integral-field spectrograph, it is also possible to in- crease the sample of high-redshift galaxies by identifying objects with equivalent widths lower than our detection threshold of 80 8 (20 8 inthe LAE rest frame). However, the gain in doing so is likely to be small: according to Figure 9, Ly rest-frame equivalent widths e-fold with a scale length of 75 8 . If this law extrapolates to weaker lined systems, as suggested by the models of Le Delliou et al. ( 2006), then most LAEs are already being detected, and pushing the observations to lower equivalent widths will only increase the number counts by 20%. Further- more, as the data of Hogg et al. ( 1998) demonstrate, contami- nation by foreground [O ii ] objects increases rapidly once the equivalent width cutoff drops below 50 8 inthe observers frame (or 12 8 inthe rest frame of Ly). Unless one can accept a large increase inthe fraction of contaminants, surveys for high-redshift galaxies need to either stay above this threshold or extend to the near-IR (to detect H and [O iii ] k5007 inthe interlopers).
sources with few counts can be explained considering that these two spectral parameters have a significant degree of degeneracy. The problem is further complicated by the redshifted absorption cut- off, which moves outside theChandra bandpass atz3 even for significant degrees of obscuration, introducing further uncertainties. To better constrain the column density, we fixed the photon index to = 1.8 and repeated the spectral analysis on all the sources. This photon index is a widely used value for sources with low counting statistics, being considered the typical slope for AGN power-law emissionin X-rays (e.g. Turner et al. 1997; Tozzi et al. 2006). The best-fitting parameters, fluxes and intrinsic luminosities derived using this spectral model are reported in Table 1. Similar results are obtained fitting simultaneously the source and background spectra. All the X-ray spectra of the sources, fitted using this model, are shown in Appendix B. The effects of assuming a flatter photon index ( = 1.6) will be discussed in Sections 3.3 and 4.
EW > 20 Å (Hogg et al. 1998) and the small volume available for z 0 objects. Local universe [O ii ] emitters would be several arcseconds across, so would stand out clearly in our catalogs. In any case, the exclusion of GALEX-detected sources should rule out this contribution. 4. Obscured AGN, which are capable of triggering a narrow- band excess through their narrow emission lines. Since we found 10 AGNs as X-ray sources intheChandra catalog and some of those may be obscured or Compton thick, and most models predict a roughly equal number of obscured and unobscured AGN at this redshift (Treister et al. 2004), we set an upper limit on residual AGN contamination of 10 ± 10 objects. This will be probed via follow-up spectroscopy. Note that heavily obscured AGN may not show any emission lines at all and therefore would not be found in our sample; this reinforces confidence in our upper limit. We stacked 66 LAEs in our sample with coverage inthe 2 Ms CDF-S image (Luo et al. 2008) and found 3σ upper limits for the soft-band (hard-band) stacked flux of 4 × 10 −18 (2 × 10 −17 ) erg s −1 cm −2 , corresponding to a luminosity of 1.3 ×10 41 (6.7×10 41 ) erg s − 1atz = 2.1. The observed soft-band implies a 3σ upper limit on the average SFR of 30 M yr − 1 (Ranalli et al. 2003). Compared to our typical rest-UV SFR of 4 M yr −1 , this implies that the dust correction must be less than a factor of 7. Because individual X-ray detections above the 2 Ms flux limit of 2 × 10 −17 erg s −1 cm −2 were removed from our LAE sample in this region, any remaining AGN must have soft- band luminosity below 7×10 41 erg s −1 . Inthe extreme case,
jects). The median simulated flux inthe 10–30 keV band ( ≈ 3 × 10 −16 erg cm −2 s −1 ) is about two orders of magnitude below the sensitivity limit. This result is a natural conse- quence of the requirement that the sources are not individu- ally detected inthe 2–8 keV band of the 4 Ms Chandra expo- sure; the extraordinarily high sensitivity of the 4 Ms CDF-S places a tight constraint on the intrinsic luminosities of the ISX sources and prevents them from being detected by NuS- TAR. Therefore, it is not likely that NuSTAR or ASTRO-H will detect any of the ISX sources presented here. However, they will probably detect some of the X-ray selected CT AGN can- didates (e.g., Tozzi et al. 2006; Comastri et al. 2011) in hard X-rays; such detections will be useful for a clear determina- tion of the intrinsic spectral shape and power of these sources. One other approach to identify heavily obscured or CT AGNs is via X-ray spectroscopy at relatively low energies (< 10 keV) complemented by multiwavelength data (e.g., Polletta et al. 2006; Tozzi et al. 2006; Alexander et al. 2008, 2011; Comastri et al. 2011). The X-ray emission of these objects is characterized by a flat continuum and often a strong rest-frame 6.4 keV iron Kα fluorescent line (e.g., Della Ceca et al. 2008; Murphy & Yaqoob 2009). For the ISX sources presented here, spectroscopic analyses are prob- ably not feasible due to the small number of counts expected. However, it is worth performing a further study of the X-ray selected CT AGN candidates in Tozzi et al. (2006), which were previously studied using only the1 Ms CDF-S data. With the current 4 Ms CDF-S data and the3 Ms XMM-Newton observations on the CDF-S, we will be able to constrain bet- ter their nature (e.g., Comastri et al. 2011). Inthe case of the CDF-S receiving further Chandra exposure (e.g., 10 Ms to- tal), some ( ≈ 15%) of the heavily obscured AGNs inthe ISX sample could be detected inthe HB, given the simulated prop- erties and expected 10 Ms sensitivity.
Compton-thick fraction. Most prior studies have found that R ( usually probed indirectly via the strength of the iron K-line ) is inversely correlated with luminosity and generally weak ( i.e., R 1 ) for moderately luminous AGNs ( e.g., Iwasawa & Taniguchi 1993; Nandra et al. 1997; Page et al. 2005; Ricci et al. 2011 ) . Spectral analysis of a single source in our NuSTAR sample ( NuSTAR J033202-2746.8 inthe ECDFS: Del Moro et al. 2014 ) found a moderate re ﬂ ection component ( R ≈ 0.55 ) for this high luminosity ( L 10 40 keV - » 6.4 ´ 10 44 erg s − 1 ) source. However, Ballantyne ( 2014 ) found that strong re ﬂ ec- tion ( R ≈ 1.7 ) at all luminosities is needed to reconcile different measurements of the local XLF across a wide range of X-ray energies (∼ 0.5 – 200 keV ) . Our measurements of the 10 – 40 keV XLF indicate that moderate-to-strong re ﬂ ection ( R ∼ 1 − 2 ) is required to describe the average spectral characteristics of L 10 40 keV - ~ 10 43 46 - erg s − 1 AGNs atz ∼ 0.1 – 3. The extent, strength, and spectral characteristics of re ﬂ ection provide insights into the physical nature of the obscuring material and the accretion disk ( e.g., García et al. 2013; Falocco et al. 2014; Brightman et al. 2015 ) . Strong re ﬂ ection could also indicate a substantial population of rapidly spinning black holes inthe detected sample; however, a relatively small intrinsic fraction of high-spin sources (∼ 7% ) can potentially dominate the observed number counts at a given ﬂ ux limit ( Brenneman et al. 2011; Vasudevan et al. 2015 ) . Accurately measuring the distribution of re ﬂ ection is thus an important challenge for future statistical studies of AGN populations.
IFRSs are a class of radio sources recently discovered inthe Australia Telescope Large Area Survey by Norris et al. (2006). These are radio sources brighter than a few hundred μJy at 1.4 GHz which have no observable infrared counterpart inthe Spitzer Wide-area Infrared Extragalactic Survey (SWIRE; Lonsdale et al. 2004). Most have flux densities of a few hundred μJy at 1.4 GHz, but some are as bright as 20 mJy, resulting in extreme infrared to radio ratios. Recent very long baseline interferometry detections have constrained the radio sources sizes to less than 0.03 arcsec, suggesting IFRSs are compact AGNs (Norris et al. 2007; Middelberg et al. 2008). Deeper Spitzer legacy survey data intheextendedChandraDeepFieldSouth yielded two IRAC detections of the four IFRSs in that field, and SED modeling of these sources with the new constraints showed that they are consistent with z > 1 radio- loud AGNs (Huynh et al. 2010).
of a galaxy’s formation and their size evolution should be much less steep than that of the overall galaxy popula- tion. If, on the other hand, Lyα emission is able to escape from a large fraction of galaxies with previous genera- tions of stars already in place, they should more closely trace the Ferguson et al. (2004) law. The heterogeneity seen inthe present samples suggests that the LAE pop- ulation contains both types of objects, but more work is needed to elucidate whether this division is an actual bimodality or simply a continuous range of properties. Thez = 2.1 LAE sample, in particular, may contain up to ∼ 15% contamination from low-redshift galaxies (see Section 3.1) - spectroscopic follow-up is needed to ac- curately estimate the contamination fraction and isolate the types of objects that contribute to the contamina- tion. Furthermore, analysis of thedeep rest-frame opti- cal (observed-NIR) imaging obtained as part of the Wide Field Camera 3 Early Release Science (Windhorst et al. 2010) and the Cosmic Assembly Near-infrared Deep Ex- tragalactic Legacy Survey would allow us to determine where most of the stellar mass in these objects lies and whether it coincides spatially with the rest-UV emission from the young stars.
Figure 3. Composite spectral energy distribution of the X-ray detected LABs in SSA 22. Where shown, upper limits are atthe 3σ level, and where relevant we have shown the range of luminosities inthe sample to indicate variations from source-to-source. As a guide, we show the SED of Arp 220 (Silva et al. 1998) redshifted to z = 3.09 and normalized to our observed 4.5 μm luminosity. For comparison, we also show the radio quiet quasar (RQQ) template of Elvis et al. (1994) redshifted and scaled to our average X-ray flux. The UV luminosity predicted by the RQQ template is in good agreement with the X-ray/UV power-law extrapolation of Steffen et al. (2006) which we indicate as a dotted line and point at λ = 2500 Å. Inthe inset we show a fit to the optical–near-IR photometry using hyperz . The fit is a moderately reddened (A V ∼ 1.5 mag) continuous star formation history of age ∼ 100 Myr. This is to be compared with the intrinsic UV luminosity from the AGN
properties of a large sample of AGNs at moderately high red- shifts selected through their X-ray emission. Inthe observed 3.6–8.0 m band, over half of these X-ray-selected AGNs show infrared colors consistent with those of galaxies. Across the nar- row wavelength range of the IRAC filters, most galaxies are also consistent with power-law spectral shapes. As a result, IRAC data alone are not sufficient for a complete or reliable selection of (moderate-luminosity) AGNs. At 24 m, dust from both star- forming galaxies and AGNs is detected. Unfortunately, observed IRAC to 24 m colors do not effectively separate galaxies pow- ered by AGNs from those experiencing star formation. Previous studies with IRAS have shown the far-infrared to be valuable for discriminating active from inactive galaxies (Sanders & Mirabel 1996); to apply similar techniques to present deep surveys re- quires longer wavelength data, not just at 24 m but at 70 m and, if possible, 160 m as well.
We present reliable multiwavelength identifications and high-quality photometric redshifts for the 462 X-ray sources inthe ≈2 Ms ChandraDeepField-South (CDF-S) survey. Source identifications are carried out using deep optical-to-radio multiwavelength catalogs, and are then combined to create lists of primary and secondary counterparts for the X-ray sources. We identified reliable counterparts for 442 (95.7%) of the X-ray sources, with an expected false-match probability of ≈ 6.2%; we also selected four additional likely counterparts. The majority of the other 16 X-ray sources appear to be off-nuclear sources, sources associated with galaxy groups and clusters, high-redshift active galactic nuclei (AGNs), or spurious X-ray sources. A likelihood-ratio method is used for source matching, which effectively reduces the false-match probability at faint magnitudes compared to a simple error-circle matching method. We construct a master photometric catalog for the identified X-ray sources including up to 42 bands of UV-to-infrared data, and then calculate their photometric redshifts (photo-z’s). High accuracy inthe derived photo-z’s is accomplished owing to (1) the up-to-date photometric data covering the full spectral energy distributions (SEDs) of the X-ray sources, (2) more accurate photometric data as a result of source deblending for ≈10% of the sources inthe infrared bands and a few percent inthe optical and near-infrared bands, (3) a set of 265 galaxy, AGN, and galaxy/AGN hybrid templates carefully constructed to best represent all possible SEDs, (4) the Zurich Extragalactic Bayesian Redshift Analyzer used to derive the photo-z’s, which corrects the SED templates to best represent the SEDs of real sources at different redshifts and thus improves the photo-z quality. The reliability of the photo-z’s is evaluated using the subsample of 220 sources with secure spectroscopic redshifts. We achieve an accuracy of |Δ z | /(1 + z) ≈ 1% and an outlier [with |Δ z | /(1 + z) > 0.15] fraction of ≈ 1.4% for sources with spectroscopic redshifts. We performed blind tests to derive a more realistic estimate of the photo-z quality for sources without spectroscopic redshifts. We expect there are ≈ 9% outliers for the relatively brighter sources (R 26), and the outlier fraction will increase to ≈ 15%–25% for the fainter sources (R 26). The typical photo-z accuracy is ≈6%–7%. The outlier fraction and photo-z accuracy do not appear to have a redshift dependence (for z ≈ 0–4). These photo-z’s appear to be the best obtained so far for faint X-ray sources, and they have been significantly (50%) improved compared to previous estimates of the photo-z’s for the X-ray sources inthe ≈2 Ms ChandraDeepField-North and ≈1 Ms CDF-S. Key words: cosmology: observations – galaxies: active – galaxies: distances and redshifts – galaxies: photometry – X-rays: galaxies
We present results from a new ultra-deep ≈ 400 ks Chandra observation of the SSA22 protocluster atz = 3.09. We have studied the X-ray properties of 234 z ∼ 3 Lyman Break Galaxies (LBGs; protocluster and field) and 158 z = 3.09 Lyα Emitters (LAEs) in SSA22 to measure the influence of the high-density protocluster environment on the accretion activity of supermassive black holes (SMBHs) in these UV-selected star-forming populations. We detect individually X-ray emission from active galactic nuclei (AGNs) in six LBGs and five LAEs; due to small overlap between the LBG and LAE source population, ten of these sources are unique. At least six and potentially eight of these sources are members of the protocluster. These sources have rest-frame 8–32 keV luminosities inthe range of L 8 − 32 keV = (3–50) ×10 43 ergs s −1 and an average observed-frame 2–8 keV to 0.5–2 keV band ratio (BR)
We present 0.5–2 keV, 2–8 keV, 4–8 keV, and 0.5–8 keV (hereafter soft, hard, ultra-hard, and full bands, respectively) cumulative and differential number-count (log N –log S) measurements for the recently completed ≈4 Ms ChandraDeepField-South (CDF-S) survey, the deepest X-ray survey to date. We implement a new Bayesian approach, which allows reliable calculation of number counts down to flux limits that are factors of ≈1.9–4.3 times fainter than the previously deepest number-count investigations. Inthe soft band (SB), the most sensitive bandpass in our analysis, the ≈4 Ms CDF-S reaches a maximum source density of ≈27,800 deg −2 . By virtue of the exquisite X-ray and multiwavelength data available inthe CDF-S, we are able to measure the number counts from a variety of source populations (active galactic nuclei (AGNs), normal galaxies, and Galactic stars) and subpopulations (as a function of redshift, AGN absorption, luminosity, and galaxy morphology) and test models that describe their evolution. We find that AGNs still dominate the X-ray number counts down to the faintest flux levels for all bands and reach a limiting SB source density of ≈14,900 deg −2 , the highest reliable AGN source density measured at any wavelength. We find that the normal-galaxy counts rise rapidly near the flux limits and, atthe limiting SB flux, reach source densities of ≈12,700 deg −2 and make up 46% ± 5% of the total number counts. The rapid rise of the galaxy counts toward faint fluxes, as well as significant normal-galaxy contributions to the overall number counts, indicates that normal galaxies will overtake AGNs just below the ≈4 Ms SB flux limit and will provide a numerically significant new X-ray source population in future surveys that reach below the ≈ 4 Ms sensitivity limit. We show that a future ≈10 Ms CDF-S would allow for a significant increase in X-ray-detected sources, with many of the new sources being cosmologically distant (z 0.6) normal galaxies.
Figure 13. Stellar identiﬁcation using Bz K colors. In each panel, we show the Bz K diagram for sources inthe MUSYC ECDFS catalog (black points), and compare our Bz K star selection (dashed line) to other complimentary stellar classiﬁcations: viz., SED-classiﬁed “stars” from the COMBO-17 survey (Wolf et al. 2004, open stars), GEMS point sources (H¨aussler et al. 2007, open circles), and spectrally classiﬁed stars (open squares/red symbols). The left panel shows the agreement between Bz K selection and these other indicators; inthe right panel we show where Bz K selection disagrees with other indicators. So, for example, circles inthe left panel show all GEMS point sources, whereas inthe right panel they show those Bz K-selected “stars” that are not GEMS point sources. Inthe either panel, the stellar sequence in Bz K color space can be seen to be isolated by 0.1 mag in (z − K ) from deep ﬁeld galaxies. This includes QSOs, which can be seen inthe left panel as GEMS point sources scattered throughout the galaxy population. Although there are a handful of spectrally classiﬁed stars lying well outside the Bz K stellar selection region (open squares inthe left panel), these objects are neither COMBO-17 “stars” nor GEMS point sources (stars and circles inthe right panel); i.e., the spectral classiﬁcation is wrong. Of the Bz K-selected stars which are not GEMS point sources (circles inthe right panel), roughly half are faint stars superposed over a diffuse background galaxy, and roughly half are faint galaxies whose photometry is signiﬁcantly affected by a bright, nearby star.
Six of the SMGs have multiple robust counterparts; of these four SMGs (LESS 2, LESS 27, LESS 49 and LESS 74) have two counterparts, which as we will show in Section 4.1, having pho- tometric redshifts consistent with them being atthe same distance and thus possibly physically associated. The choice of the precise counterpart for the SMG is therefore irrelevant for these sources as their physical interpretation is not dependent upon this. However, the other two SMGs (LESS 10 and LESS 49) each has two robust counterparts with photometric redshifts and SEDs that suggest they are not physically associated. In these cases, from the information currently available, it is not possible to determine which of the two counterparts is the source of the submillimetre flux, or whether the LABOCA detection is a blend of theemission from two galaxies. To avoid bias, we have included all of the multiple counterparts in our analysis, but we note that their small number means that their inclusion does not significantly affect our results.
Multi-object spectroscopy of 92 LAE candidates, along with other MUSYC targets, was performed with Magellan- Baade+IMACS on Oct. 26-27, 2003, Oct. 7-8, 2004, Feb. 4-7, 2005, Nov. 2-3, 2005, Oct. 25-27, 2006, Nov. 21-22, 2006 and Feb. 18-20, 2007. The 300 line/mm grism was used with 1.2 00 slitlets to cover 4000 − 9000Å at a resolution of R = 640 i.e., 470 km s −1 , atthe wavelength of Lyα emis- sion. Mask exposure times ranged from 2 to 5 hours, with the longer exposures sufficient to detect Lyα emission lines down to our completeness limit of ∼ 1.5 × 10 −17 ergs cm −2 s −1 , as- suming clear conditions and minimal slit losses. Details of our spectroscopy will be given in P. Lira et al. (in prep). Red- shifts were confirmed to lie at 3.08 < z < 3.14 for 61 of the LAEs, with 1 interloping AGN atz = 1.60 where [C III]λ1909 falls inthe narrowband filter, and the other 30 objects lacking sufficient S/N to yield redshifts. Our success rate for the slit- masks with the highest S/N was 90%, setting an upper limit of 10% on possible contamination of our LAE sample. The rate of non-detections was higher in masks with shorter exposure times resulting from weather or instrument challenges, con- sistent with the reduced S/N. Our spectroscopy shows that the sample is not contaminated by z = 0.34 [O II] emission-linegalaxies, which are the typical interlopers for narrow-band- selected LAE samples. These have been eliminated by re- quiring observer’s-frame EW> 80Å which eliminates all but the rarest [O II] emitters (Terlevich et al. 1991,Hogg et al. 1998,Stern et al. 2000).
Inthe course of an investigation of the diffuse intergalactic light in X-ray emitting clusters atin- termediate redshifts (Melnick et al. 1999), we detected a puzzling S-shaped arc-like structure inthe ROSAT cluster RX J0054.0–2823 (Faure et al. 2007), which we tentatively identified as the gravi- tationally lensed image of a background galaxy at a redshift between z = 0.5 and z = 1.0. The cluster, however, is characterized by having three dominant D or cD galaxiesinthe center, two of which are clearly interacting. We designed an observing strategy that allowed us to simultaneously observe the arc, the diffuse Intra-Cluster Light (ICL), and a magnitude limited sample of individual galaxiesinthefield by taking advantage of the multi-object spectroscopic mode of the FORS2 in- strument on Paranal. By optimizing the mask design (see below), we were able to obtain: (a) very deep observations of the arc; (b) very deep long-slit observations of the ICL; and (c) redshifts and flux distributions for 654 galaxies, of which 550 are inthe pencil beam and at 0.275 ≤ z ≤ 1.05 .
An alternative possible explanation for the suppressed vari- ability at low luminosities is a change in accretion structure. Ptak et al. (1998) found a similar drop in variability strength below L 2−10 keV ≈ 2×10 41 erg s − 1in a sample of LLAGNs and LINERs observed with ASCA on variability timescales of less than a day. The authors hypothesized that a radiatively ineffi- cient accretion flow (RIAF; e.g., Yuan & Narayan 2004) could be responsible for suppressed short-timescale variability at low luminosities due to the larger extent of the X-ray source. This scenario would not obviously explain the reduced variability on ∼month–year timescales seen here. RIAF models also predict a hard X-ray photon index due to the lack of an optically thick ac- cretion disk, which provides the soft X-ray photons. The stacked X-ray photon index for variable galaxies (Γ stack ≈ 1.93 ± 0.13;
Most of our objects lie below λ ∼ 0.1 which means that the range of high Eddington ratios is not fully probed by the CDF AGN. To sample the area of λ & 0.1 requires that we look atthe regions probed by investigations of the rarer luminous quasars (Fig. 7). Kollmeier et al. (2006) and Steinhardt & Elvis (2010) show that the most luminous objects inthe SDSS have λ ∼ 0.2 − 0.3, but do not investigate any absorption properties. Greene et al. (2009) look at a large sample of narrow line AGN, and find that they have high Eddington ratios (∼ 0.2). Although these sources are obscured, there are no available values for their hydrogen column densities. Just et al. (2007) do study the X-ray properties of 59 of the most op- tically luminous quasars and find that each is consistent with no in- trinsic absorption and collectively are obscured by a mean column density < 2 × 10 21
log( X ) ~ 43.3 erg s − 1atz ~ 1.5 for the 2 – 10 keV band for theChandra observations ) . Essentially all of X-ray detections atz > 0.5 have L X > 10 42 erg s - 1 , a luminosity level typically due to an AGN and a higher X-ray luminosity than would be expected from star formation using the relation of Mineo et al. ( 2014 ) . It should be noted that our observing program speci ﬁ cally targeted X-ray AGNs whenever possible so the total fraction of X-ray AGNs among our ( U ) LIRGs is high. Therefore, the total AGN fractions quoted here should not be considered as absolutes, but rather the comparison between AGNs identi ﬁ ed via the different methods is what we are interested in. For comparison, ∼ 15% – 30% of ( U ) LIRGs at this redshift in COSMOS are X-ray-detected AGNs ( Kartaltepe et al. 2010a ) . Overplotted on each panel are the lower-limit abundance sequence, the redshift-dependent AGN classi ﬁ ca- tion line, and the starburst – AGN mixing sequence for scenarios 3 and 4 from Kewley et al. ( 2013b ) . We identify all galaxies above the AGN classi ﬁ cation line as “ BPT-selected AGNs, ” although this new dividing line is uncertain and likely does not select all AGNs ( especially those in composite systems ) . The mixing sequences range from Scenarios 1 – 4 and span both normal and extreme ISM conditions and metal-rich and metal- poor AGN narrow-line regions ( NLRs ) at high redshift. Here, we plot scenario 3 ( extreme ISM conditions and metal-rich AGN NLRs ) and scenario 4 ( extreme ISM conditions and metal-poor AGN NLRs ) since they appear to be the best match for our high-redshift data points. The percentage of sources that fall within the bounds for scenarios 1 – 4 are 62%, 52%, 66%, and 64% at intermediate redshift and 70%, 28%, 73%, and 53% at high redshift. Scenario 2 ( normal ISM conditions and metal- poor AGN NLRs ) has the lowest fraction, while the fraction inthe other three scenarios is comparable.
to represent an important phase inthe evolution of galaxies as they are linked to the formation of massive galaxies via gas-rich star- bursting mergers followed by an AGN-driven quenching of the star formation (e.g. Sanders et al. 1988a,b). Recent studies (Dey et al. 2008; Bussmann et al. 2009, 2012) have suggested a similar evolu- tionary sequence where DOGs are an important intermediate phase between gas-rich major mergers (traced by SMGs) and quasars atz ∼ 2. These studies describe an evolutionary scenario in which the starbursting nature of SMGs evolves into the composite nature of DOGs as an underlying AGN grows; this is followed by a quasar phase that terminates star formation, leading to the formation of a passive, massive elliptical galaxy. Within this context, DOGs could provide a key insight to an extremely dusty stage inthe evolution of galaxiesatz ∼ 2, where both AGN and star formation activity coex- ist. Their composite nature was until relatively recently inaccessible prior to the availability of sensitive mid- to far-infrared data.