ASSEMBLY OF MASSIVE GALAXIES IN A HIGH-z PROTOCLUSTER

Using the multi-band photometric data, we obtained the photometric redshift of the K-selected sources using the hyperz code (Bolzonella et al. 2000). The SED fitting is performed with the redshift, spectral type, age, and dust extinction as the free parameters. The best-fit SED is determined by minimizing the χ² value. The template spectra we used are those for the simple stellar population with the fixed (solar) metallicity from GALAXEV by Bruzual & Charlot (2003).

Figure 4 shows the photometric redshifts for the spectroscopic redshifts obtained in the previous literature and the NASA/IPAC Extragalactic Database (NED). The relative errors, (z_photo − z_sp)/(1 + z_sp), are also shown in the bottom panel. The errors are similar if compared with those in the previous studies (e.g., Ichikawa et al. 2007). The uncertainty of Δz  ∼  0.5 at z  ∼  3 remains since the Balmer/4000 Å break is shifted in the middle of H band.

Zoom In Zoom Out Reset image size

The median photometric redshift of 26 LBGs with 3.06  <  z_spec  <  3.13 is z_phot = 3.11 and the standard deviation is 0.08. Ninety-two percent of the LBGs is in the range of 2.6  <  z_phot  <  3.6. We therefore pick up the objects with 2.6  <  z_phot  <  3.6 as the candidate galaxies at z = 3.1. We finally selected 474 objects brighter than K = 24.

The red J − K color of DRGs is due to the Balmer or 4000 Å break of galaxies at z  ≳  2, or to the dust extinction (Franx et al. 2003; Reddy et al. 2005; Förster Schreiber et al. 2004). Wuyts et al. (2009) confirmed by spectroscopic observation that DRGs with K_Vega  <  22.5 (K_AB  <  24.35) are dominated by z  >  2 galaxies. In the GOODS-North field, 86% of DRGs are in the range of 2  <  z_phot  <  4 (Kajisawa et al. 2011), and the median photometric redshift is z_phot = 2.5.

We used the same criteria as in Kajisawa et al. (2006), which corresponds to J − K_AB  >  1.4 in our photometric system. In our data, we sampled 356 DRGs with K  <  24 in the whole observed area. The distribution of the photometric redshifts for DRGs is shown in Figure 5. We find that 70% of DRGs in the SSA22 area are located at 2  <  z_phot  <  4. The median photometric redshift is z_phot = 2.6.

Zoom In Zoom Out Reset image size

We compared the current sample of DRGs with the previously known sample of LBGs (Steidel et al. 2003). In the area of SSA22a, which has a substantial overlap with our MOIRCS fields, they listed in total of 171 LBGs including those without spectroscopic redshift (29 objects have the redshift between z = 3.06 and 3.12). Of them, only the four objects (SSA22a-C54, SSA22a-M14, SSA22-MD39, and SSA22a-aug96C20) meet the J − K color criteria. The K-band counterpart of SSA22a-M14a is, however, displaced from the optical source and may not be the same object or the same region in a galaxy. The spectroscopic redshift of SSA22a-aug96C20 is 1.357. No spectroscopic redshift is available for SSA22a-C54 and SSA22a-MD39.

We also checked the optical colors of our sample of DRGs to see how many have the colors match the LBG criteria. Among the 356 objects with K  <  24, 7 objects with R  <  25.5, magnitude limit for the spectroscopic sample of Steidel et al. (2003), and 23 objects with R  >  25.5 satisfies the LBG criteria adopted here, u* − V  >  2.9(V − R) + 0.37, V − R  <  0.5, and u* − V  >  1.4. These numbers should be referred carefully as the color selection only properly works for those objects with high photometric accuracy (>10σ in V & R). The optical colors of those optically faint red objects are largely scattered due to the photometric errors. In conclusion, we found that the overlap between the current samples of DRG and LBG is very small.

We also sampled hyper extremely red objects (HEROs; Totani et al. 2001) with J − K_AB  >  2.1 as a subset of DRGs. Totani et al. (2001) suggested that the extremely red color in J − K is well explained by primordial elliptical galaxies reddened by dust and still in the starburst phase at z  ∼  3. Indeed, in the GOODS-North field, all of the HEROs are in the range of 2 < z_phot < 4 (Kajisawa et al. 2011), and the median photometric redshift of HEROs is z_phot = 3.0.

The number of HEROs in the observed field is 31. We find that 94% (29/31) of HEROs in our field are classified as 2  <  z_phot  <  4. The median photometric redshift of HEROs is z_phot = 2.7.

It should be noted that the result of the SED fitting for the HEROs tends to include relatively large uncertainty due to their extreme red color. Fifty-five percent of the HEROs are not detected in J band at 2σ level (see Figures 2 and 3). The number of the HEROs which are detected only in K band is four while the rest of the objects are detected at least in the relatively deeper R-band image.

The distributions of the K-selected objects with z_phot = 2.6–3.6 and DRGs on the sky are shown in Figure 6, where LABs (Matsuda et al. 2004) and LAEs (Hayashino et al. 2004) are also plotted. The figure shows the samples with the magnitude limit of K  <  24.0. The dotted lines are the region observed with MOIRCS. We note that the noisy area at the center of the SSA22-M6 field is removed for the DRGs selection to avoid the spurious detection. The solid contours delineate the high-density region of the LAEs in Hayashino et al. (2004).

Zoom In Zoom Out Reset image size

In order to investigate whether there is an excess in number density of the galaxies in the SSA22 protocluster compared with the blank fields, we first investigate the surface number density of DRGs. The cumulative and differential number counts of DRGs in the whole field are shown in Figure 7. The DRG counts show the excess at K > 23 compared to those in all the three blank fields, namely, the GOODS-North field (MOIRCS Deep Survey; Kajisawa et al. 2006, 2011), the GOODS-South field, and the HDF-South field (Grazian et al. 2006). The completeness fraction and the 90% limits in SSA22-M1 (deepest) and M4 (shallowest) are also plotted. The cumulative number density of DRGs in the entire observed field is 3.1 arcmin² down to K = 23.85, which is ∼ 33% larger than that in GOODS-N.

Zoom In Zoom Out Reset image size

Figure 8 shows the DRG number counts in each FOV. The number density in GOODS-N and the 90% completeness fraction are also plotted in the figure. The densest fields are SSA22-M3 and M6, which overlap with each other, followed by SSA22-M4, M2, M5, and M1. The number density of the DRGs in M6 is 2.1–1.8 times larger than that in GOODS-N at K = 23–23.5. In addition, the excess of HEROs is found in this field (see Sections 5.1 and 5.3). Our observed fields cover the highest density peak of LAEs observed by Matsuda et al. (2011) and Yamada et al. (2012) in the 1.38 deg² survey area (Figure 1). The outstanding density peak of LAEs with the size of ≈3 arcmin lies in SSA22-M3 and M6, whose location is shown by the cross symbol in Figures 1 and 6. The fact that the largest overdensity of DRGs appears at around the highest density peak of LAEs suggests that the stellar mass distribution traces the structure of LAEs in the scale of several arcminutes, or a few Mpc at z = 3.1.

Zoom In Zoom Out Reset image size

We also investigated the surface number density of photo-z selected objects and found that the counts of the objects at 2.6  <  z_phot  <  3.6 in the SSA22 fields are larger than those in the GOODS-N field (see Figure 9 below). The direct comparison of the two samples should be discussed carefully, however, since the depth or the photometric uncertainty of them is different.

Zoom In Zoom Out Reset image size

These results for the K-selected galaxies can be compared with the surface density of LBGs in the SSA22 fields. Steidel et al. (2003) found the average surface density of the color-selected LBG candidates with R  <  25.5 to be 1.72 arcmin⁻². While the surface density of the 11 fields observed with the Palomar 200 inch telescope ranges from 1.10 to 2.15 arcmin⁻², the density in SSA22a and SSA22b are 1.88 and 1.15 arcmin⁻², respectively. The higher contrast of the protocluster in the K-selected galaxies is due to the fact that the more stellar massive galaxies traced by the K-band sample have intrinsically much stronger clustering properties and may be easier to be dominated by a single large structure.

As seen in Figure 6, photo-z selected objects and DRGs are frequently seen around the position of LABs. We investigate the spatial correlation between these K-selected objects and rest UV-selected galaxies (LABs and LAEs) in the field. Figure 9 shows the surface number densities of K-selected objects as a function of a radius from LABs and LAEs. As shown in the figure, we found that the K-selected objects are correlated with LABs within a radius of 6'', which corresponds to 50 kpc scale at z = 3.1. This implies that these objects are likely to associate with LABs. On the other hand, no significant correlation between LABs and K-selected objects at the larger scale is detected.

In contrast with LABs, LAEs are not correlated with K-selected objects in scale of less than 1 arcmin while the large-scale density peaks of LAEs and K-selected objects coincide (Section 4.1). Most of the LAEs are not detected in the K-selected images to the limit except for some HEROs (see Section 5.3.1). The expected typical luminosity of LAEs in the SSA22 protocluster is K ≃ 26.3, which is estimated from the stacked image of LAEs that are not detected in K band individually. The result is consistent with that of LAEs in a field environment (Ono et al. 2010).

Here, we describe the detailed near-infrared properties of LABs. The number of the LABs (Matsuda et al. 2004) in our observed fields is 20, and the K-band counterparts are detected for 75% (15/20) of them. Figure 10 show the images of the 15 LABs, which have plausible K-band counterparts. Each panel has a length of 234 on a side, which corresponds to 180 kpc at z = 3.1. Circles are photo-z selected objects and DRGs with z_phot = 2.6–3.6 are colored red. Red squares are DRGs with z_phot < 2.6 or z_phot  >  3.6. The contours are isophotal area of Lyα emission with surface brightness of 28.0 mag arcmin⁻² obtained by the narrowband observation in Matsuda et al. (2004).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Since the expected average surface number of DRGs or photo-z selected objects is only ∼0.03 in the median Lyα isophotal area of 23 arcsec², or 0.0012 arcsec⁻² in surface density, it is interesting that a significant number of K-selected objects are detected around the LABs in our field. Table 3 shows K-band magnitude, J − K color, z_phot, and z_spec of the K-selected objects in Figure 10. Spectroscopic redshifts of 11 photo-z selected objects including 9 LBGs listed in Table 3 have been confirmed to be z_spec = 3.1.

On the other hand, we detect no K-band counterparts for five LABs, LAB9, 19, 25, 26, and 35, as shown in Figure 11. Please note that we plot all photo-z selected objects and DRGs even below the nominal magnitude limit. Detailed description for each object is given below.

Zoom In Zoom Out Reset image size

4.2.1. LABs with K-band Counterparts

4.2.2. LABs with No K-counterparts

In order to determine whether massive galaxies have already been formed in the high-redshift protocluster, we investigate the distribution of the stellar mass of the K-band selected objects in the SSA22-M1 to M6 fields. The stellar mass is derived from the SED fitting of the BVRi'z'JHK photometric data. Spectral types, ages, and dust extinction are treated as free parameters. The redshifts are fixed at the photometric redshifts described in Section 3.1, while the spectroscopic redshifts are used for LBGs if available. The redshifts of the K-selected objects in LABs described in Section 4.2.1 are fixed at z = 3.1. We used the template spectra obtained from GALAXEV (Bruzual & Charlot 2003) with star formation histories with e-folding time τ =0, 1, 9 Gyr. The best-fit SED is determined from the minimum χ² value. It should be noted that only the objects fitted with χ² < 1 (85% of the sample) are used for the following discussion.

Figure 12 shows the sky distribution of the K-selected galaxies with M_* > 10¹¹ M_☉. These red massive galaxies are supposed to be the representative population in the SSA22 protocluster. While the number density of the galaxies with M_* > 10¹¹ M_☉ in GOODS-N is 0.16 ± 0.04 arcmin⁻², it was 0.37 arcmin⁻² in SSA22, 2.3 times larger.

Zoom In Zoom Out Reset image size

In Figure 12, the surface number density of the massive galaxies is the highest near the peak of the LAE density (cross) although a little bit shifted to the south. There are 14 galaxies within 4' × 7' (≈MOIRCS FOV) area, which gives the surface density of 0.5 arcmin⁻² or 3.1 times of the average of the GOODS-N field.

On the other hand, we note that the number densities of DRGs and massive galaxies in SSA22-M1 are the smallest among the observed six fields. This is interesting as the SSA22-M1 field is the densest region of LABs and the two largest LABs (LAB1 and LAB2) locate in the area. The field is ∼10 Mpc (in comoving scale) away from the density peak of LAEs.

Massive galaxies in protoclusters at z  >  2 have been identified in some other fields mostly around the radio galaxies. Steidel et al. (2005) identified a protocluster at z = 2.3 in the HS 1700+643 field, which contains a significant number of old and massive galaxies. From their spectroscopic sample in the ∼70 arcmin² field, there are at least three member galaxies with >10¹¹ M_☉. Kodama et al. (2007) observed the field around the high-z radio galaxies at z = 2–3 with MOIRCS. They reported that there is an excess of galaxies with >10¹¹ M_☉ on the bright end of the NIR color–magnitude sequence in the fields of MRC 1138−262 at z = 2.2 and USS 1558−003 at z = 2.5 while there are few massive galaxies on the red sequence in the protoclusters at z = 2.9 and 3.1. For the MRC 1138−262 field, two of the massive red galaxies are identified as the protocluster members by MOIRCS spectroscopy (Doherty et al. 2010). Zirm et al. (2008) also investigated the HST NICMOS images of the field and found the significant overdensity of the red faint galaxies in the color–magnitude diagram. In addition, Hatch et al. (2009) derived the stellar mass of 19 galaxies in the Lyα halo of the radio galaxy MRC 1138−262, which is called as the Spiderweb galaxy (Miley et al. 2006), and found that there are no galaxies with >10¹¹ M_☉ in the 0.3 arcmin² field except for the radio galaxy itself, which has the stellar mass of 10¹² M_☉. Compared to these previous studies, thanks to our wide field coverage, our results clearly identified the overdensity of massive (>10¹¹ M_☉) red galaxies in a protocluster at z = 3.1. Since this SSA22 protocluster is found as the largest high-density region of star-forming galaxies known so far, the stellar mass assembly is likely to be more progressed.

Figure 13 shows the histogram of the stellar mass distribution for the photo-z selected galaxies. The number density in the SSA22 peak as well as that in GOODS-N is shown with the solid and dotted lines, respectively. Both the samples are selected with the same selection criterion with z_phot = 2.6–3.6 and K  <  24.0. The turnoff of the distribution at the less mass end (log M_*/M_☉ ≲ 10.5) is due to the magnitude cutoff. The stellar mass distribution in GOODS-N without the cutoff is also indicated with the dotted thin line. It is found that the stellar mass distributions in both fields are not much different, while the number density in SSA22 is larger than that in GOODS-N. The median values of the stellar mass with K  <  24.0 in SSA22 and GOODS-N are 2.7 × 10¹⁰ M_☉ and 2.2 × 10¹⁰ M_☉, respectively.

Zoom In Zoom Out Reset image size

The excess in the surface number density of SSA22 to that of GOODS-N, which is regarded as the cluster component, is shown in the bottom panel of Figure 13. It might be interesting to compare this with the stellar mass distributions in nearby typical rich clusters. As an example, the stellar mass distribution for the Coma cluster is also shown in the figure. We focus on the massive end (log M_*/M_☉ ≳ 10.5) of the mass function because of the magnitude cutoff mentioned above. The stellar mass function of the Coma cluster galaxies is derived from the H-band luminosity function (LF) obtained by de Propris et al. (1998; corresponds to M*_K = −24.0, α = −0.78 for the bright end) and the typical mass-to-light ratio of local elliptical galaxies (M_*/L_K ≃ 0.6). The number density is averaged in the observed area of 292 × 225 (0.8 × 0.6 Mpc) in de Propris et al. (1998) and converted to the physical density at z = 3.1. It should be noted that the weight for the brightest galaxy in the Coma cluster in fitting LF seems low (Figure 2 in de Propris et al. 1998) and its contribution is not well reproduced by the obtained Schechter parameters. Therefore, we added it to the brightest bin manually. As shown in the figure, it is suggested that the equivalent of ≈10%–15% of the galaxies with >10^10.5 M_☉ in the Coma cluster is already seen in the SSA22 field. In practice, however, the central region of the Coma cluster is supposed to be extended several times larger at z  ∼  3. For example, Suwa et al. (2006) showed that a present-day cluster of galaxies with several Mpc can be extended to 20–40 Mpc in the comoving scale at z = 4–5 in their simulation. By assuming the spherical model for the nonlinear collapse of the overdense region, the surface number density at the epoch of turnaround is four times lower than that of the collapsed object. If the protocluster does not reach the maximum expansion, it may be slightly larger. As for crude estimation, the stellar mass distribution in the Coma cluster divided by three is shown in the bottom panel of Figure 13. The figure suggests that the number of the >10^10.5 M_☉ galaxies in SSA22 is equivalent to ∼ 30%–40% of that in the Coma cluster.

As shown in Section 4.2, most of the LABs are associated with the K-selected objects. The significant number excess around LAB that these galaxies are formed in LABs. It is also interesting to see a notable fraction of them consists of multiple K-selected components. The formation of early-type galaxies was generally understood by either "Monolithic collapse" (Eggen et al. 1962) or "(Major) Merger" (Searle & Zinn 1978) scenario. In the realistic cosmological scenario, however, early-type galaxies may have experienced hierarchical gaseous multiple merging in their early evolutionary phase, which may be observed as the combination of the extended gaseous halo and the multiple stellar components.

Indeed, such phenomena is discussed using the cosmological numerical simulations for the formation of early-type galaxies (Meza et al. 2003; Naab et al. 2007). The scenario can be called as "on-going assembly" or "hierarchical multiple merging." The most outstanding examples observed in our observational data are the two giant objects, LAB1 and LAB2, which have six and five K-selected objects, respectively, some of which are DRGs, within their Lyα halos. As shown in the model simulations, these multiple components will finally merge into a single galaxy, which may evolve into a massive elliptical in the local universe.

It is interesting to note that the SSA22-M1 field where the LABs are the most populate is a few Mpc (in physical scale) separated from the center of the clustering of the K-selected galaxies or LAEs. The many (5/8) of the LABs in the SSA22-M1 field, LAB1, LAB2, LAB7, LAB16, and LAB30, in fact have multiple K-band components. While the galaxy formation in the central region of the cluster statistically proceeds earlier, we may see a more dynamically young phase of massive galaxy formation in the region a little separated from the center.

We investigate the stellar mass of the photo-z selected objects associated with the LABs in order to investigate how much stellar mass in the LABs has already been formed. The method is the same as that in the previous section, while the redshift is fixed to be z = 3.1. The result is listed in Table 4. We here assume that all the broadband emission is due to the stellar light and ignored the contribution of the nebula emission lines, which will be constrained by the future spectroscopic observations.

The stellar mass of the individual components of the LAB counterparts ranges from 3.8 × 10⁹ M_☉ to 1.7 × 10¹¹ M_☉. Figure 14 also shows the distributions of the stellar mass of individual components and the integrated stellar mass associated with the LABs. Figure 15 shows the color and the stellar mass of each individual component. LBGs are shown with the blue dots. As seen in the figure, low-mass components in the LABs are dominated by LBGs. LAB7 and LAB11 have multiple LBGs with <10¹⁰ M_☉, which are close to each other.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

With respect to the integrated total stellar mass in each LAB, 70% of the LABs (14/20) have >10¹⁰ M_☉, and the median value is 3.1 × 10¹⁰ M_☉. Twenty percent of the LABs have the stellar mass M_*  >  10¹¹ M_☉. Typical LBGs in the general field have M_* ≃ 2 × 10¹⁰ M_☉ (Shapley et al. 2001), which is comparable or slightly smaller than the median stellar mass of the LABs of the current sample.

The total stellar mass in each LAB versus the properties of the Lyα emission are plotted in Figure 16. The horizontal axes show the integrated stellar mass of photo-z selected objects associated with each LAB, i.e., the objects in Table 4. The vertical axes in Figures 16(a), (b), and (c) show the luminosity, size, and velocity width of the LABs, respectively. Note that there is a correlation between the luminosity and the isophotal area of the Lyα emission of the LABs (Matsuda et al. 2004) and these are not totally independent. In Figure 16(a), or Figure 16(b), we see that some massive galaxies with M_*  ∼  10¹¹ M_☉ are associated with the brighter and larger LABs (LAB1, LAB2, LAB3). As also suggested in Uchimoto et al. (2008), this implies that the origin of the Lyα emission in these LABs is related to the previous (or recent) star formation activity represented by the accumulated stellar mass. On the other hand, LAB12 and 14 have large stellar mass of M_*  ∼  10¹¹ M_☉ in spite of their low luminosity in Lyα. Both LABs are associated with X-ray sources, and they are detected in 8 μm and 24 μm (Geach et al. 2009), suggesting that they suffer from dust obscuration. LAB14 is also detected as a submillimeter-bright galaxy (SMG; Geach et al. 2005). In these cases, Lyα emission is supposed to be suppressed by the dust.

Zoom In Zoom Out Reset image size

In Figure 16(c), we can see that the total stellar mass and the velocity width of Lyα emission for nine LABs obtained by Matsuda et al. (2006; Y. Matsuda 2011, private communication) have notable correlation. Here we should note that the velocity widths are estimated by assuming a single Gaussian profile, and the stellar mass estimation includes the uncertainty of photo-z. Despite such uncertainty, the correlation suggests not only that the photo-z selected objects are indeed associated with the LABs but also that the estimated velocity width represents their physical quantities.

Although the correlation can be explained by some different scenarios (Matsuda et al. 2006), we first discuss the case in which we assume that the Lyα gas clouds are gravitationally bounded. With this assumption, their dynamical mass can be evaluated from the velocity width and the radius which is represented by the extent of the Lyα emission. Then, our result suggests that the dynamical mass of LABs is correlated with their stellar mass. It is known that the dynamical mass of local spheroidal galaxies at z < 1 correlates with the stellar mass by a factor of ∼5–10 (Drory et al. 2004; Bundy et al. 2007). On the other hand, the stellar masses of most of our sample are in the range of 10^10.3–10^11.5 M_☉, and dynamical mass of LABs estimated by Matsuda et al. (2006) ranges over 10^11.7–10^13.3 M_☉. They are different by a factor of ∼50–80. As the ratio seems significantly larger than the local value, the stellar mass assembly may be still in progress in LABs.

Of course, LABs may not be virialized, and the virial radius may not be equal to the size of the Lyα halo.

As an alternative case, if we assume that the Lyα emission is due to the outflow from galactic superwinds, the characteristic timescales of star formation can be estimated by their spatial extents and typical outflow speeds. Matsuda et al. (2006) evaluated the timescale of ∼10⁷ − 10⁸ yr for the SSA22 LABs. Assuming the star formation rate of 10³ M_☉ yr⁻¹ as inferred from the stacked submillimeter flux, the estimated amount of the formed stellar mass, ∼10¹⁰ M_☉, is consistent with the timescale. In this case, it is suggested that the observed stellar components is formed from the star formation activities which cause the outflow.

K-band selection is also useful to detect the highly dust obscured starbursting galaxies. They are an important population that dominates the stellar mass of the protocluster. Here, we focus on the dusty starbursts in our sample, namely, HEROs which are the objects with extremely red color of J − K > 2.1 (see Section 3.3). We also discuss the AzTEC sources detected in Tamura et al. (2009), which are claimed to have the spatial correlation with LAEs.

5.3.1. Overdensity of HEROs in the SSA22 Field

We detected 31 HEROs down to K = 24.0 in our observed field. The left and middle panels of Figure 17 show the sky distribution of the HEROs. The figure also shows the 2σ density contour of LAEs. It is found that the number density of HEROs in SSA22-M6 is the highest (12 HEROs) among the six fields in SSA22. In the right panel of Figure 17, the surface density of HEROs is shown as a function of a radial distance from LABs, LAEs, and AzTEC sources (see the next section). The mean number density of 0.28 arcmin⁻² in the SSA22 field is 2.3 times larger than that in GOODS-N (0.12 ± 0.03 arcmin⁻²) while those in SSA22-M6 is the 3.1 times larger. In addition, the marginal excesses of HEROs are found around LABs/LAEs, while there are no HEROs around LAB1 and LAB2.

Zoom In Zoom Out Reset image size

Out of the 23 HEROs detected in R band (see Section 3.3), 11 have z_phot = 2.6–3.6. The result of the SED fitting for the 11 HEROs shows that the stellar mass range from 2.7 × 10¹⁰ M_☉ to 3.7 × 10¹¹ M_☉, and the median is 1.6 × 10¹¹ M_☉. The 55% (6/11) of the HEROs are massive galaxies with >10¹¹ M_☉. It is likely that HEROs are among the most massive populations in the protocluster.

Of all the 31 HEROs, 2 objects are associated with Lyα emission, which are expected to be indeed located at z = 3.1. They are found in LAB14 and the extended LAE in the vicinity of LAB35 (the object A; see Section 4.2.2). Both of them are located in the 2σ high-density contour of LAEs. The object A in vicinity of LAB35 has J − K = 2.45 and z_phot = 3.06 (Figure 11). The stellar mass of this HERO is 2.65 × 10¹¹ M_☉, which is one of the most massive galaxies in this region.

5.3.2. Correlation between K-selected Objects and AzTEC Sources

The SSA22 protocluster is observed at 1100 μm using AzTEC camera mounted on ASTE (Tamura et al. 2009). They reported the concentration of SMGs in this region. Figure 18 shows the sky distribution of the K-selected objects as well as the AzTEC sources in Tamura et al. (2009). The left panel indicates the distribution of the photo-z selected objects, and the middle one indicates that of DRGs. As shown in the right panel of Figure 18, the association between K-selected objects and the AzTEC sources is notable.

Zoom In Zoom Out Reset image size

Indeed, Tamura et al. (2010) reported the multi-wavelength identification of the brightest source, SSA22-AzTEC1. The object is identified with the Very Large Array radio source, whose counterpart is detected in K- and H-band images. On the other hand, no counterpart is detected in J band and optical bands. The object is classified as a HERO. Tamura et al. (2010) estimated the redshift as z_phot = 3.19 by their radio–submillimeter photo-z.

In the right panel of Figure 17, we show the correlation between HEROs and the AzTEC sources. It is, however, found to be dominated by the contribution of K-selected objects around SSA22-AzTEC20 as shown in Figure 19. They are supposed to be the counterparts of AzTEC20 and also in the "hierarchical multiple merging" phase. It is interesting to note that significant Lyα emission is not detected around the objects.

Zoom In Zoom Out Reset image size

Thus, the statistical evidence (clustering) as well as a few notable individual examples implies that at least a fraction of the AzTEC sources are indeed associated with the SSA22 protocluster.