A SYSTEMATIC SURVEY OF PROTOCLUSTERS AT z ∼ 3–6 IN THE CFHTLS DEEP FIELDS

We made use of publicly available data from the CFHTLS (T0007: Gwyn 2012), which was obtained with MegaCam mounted at the prime focus of the CFHT. The Deep Fields of the CFHTLS were used in this study, which consist of four independent fields of about 1 deg² area each (∼4 deg² area in total) observed in the u*, g', r', i', and z' bands. The field center and limiting magnitudes of each field are summarized in Table 1. The seeing size (FWHM) and pixel scale of all the images are ∼07 and 0186, respectively. Although data at other wavelengths, such as near- or mid-infrared imaging, are available in a part of the CFHTLS Deep Fields, the depth and coverage are significantly different from field to field. Therefore, our protocluster search is conducted only based on the optical data to make a uniform survey.

Table 1.  Photometric Data and the Number of Dropout Galaxies

Field	R.A.	decl.	Area	u*a	g'a	r'a	i'a	z'a	Nub	Ngb	Nrb	Nib
	(J2000)	(J2000)	(arcmin2)	(mag)	(mag)	(mag)	(mag)	(mag)
D1	02:25:59	−04:29:40	3063	28.12	28.32	27.77	27.30	26.39	17110	10416	2433	148
D2	10:00:28	+02:12:30	2902	28.07	28.19	27.70	27.30	26.45	14515	11160	2539	231
D3	14:19:27	+52:40:56	3161	28.14	28.38	27.91	27.48	26.43	21454	14896	2579	232
D4	22:15:31	−17:43:56	3035	27.96	28.19	27.67	27.17	26.26	10484	11288	1926	188

Notes.

^a3σ limiting magnitude in a 14 aperture. ^bNumber of u-, g-, r-, or i-dropout galaxies.

Download table as:  ASCIITypeset image

We created two multicolor catalogs with SExtractor (version 2.8.6; Bertin & Arnouts 1996), in which i'- and z'-band images were used as detection images. The detection images were first smoothed with a Gaussian function; then objects were detected by requiring a minimum of three adjacent pixels each above 1σ of the sky background rms noise. Then, the magnitudes and several other photometric parameters were measured in the other band images at exactly the same positions and with the same aperture of 1.4 arcsec as in the detection-band images using the "double image mode." The Galactic extinction was removed for each field based on the measurement of Schlafly & Finkbeiner (2011). The individual catalogs were masked to remove the regions where the detection and photometric measurements of objects may have been seriously affected. These regions are around bright stars and diffraction and bleed spikes from bright stars. The regions near the frame edges, whose depth is systematically shallow, were also excluded. As noted in Gwyn (2012), bright stars with ≲9 mag produce a large halo whose radius is ∼3.5 arcmin. The total masked regions were ∼15%–20% of the FoV, and the effective areas of our analysis are shown in Table 1. Finally, ∼330,000–420,000 and 230,000–270,000 objects were detected down to the 3σ limiting magnitudes, defined as the magnitude corresponding to three times the standard deviation in the sky flux measured in empty 1.4 arcsec apertures, of the i' and z' bands in each field, respectively. To estimate the detection completeness of each i'- and z'-band image, we used the IRAF task mkobjects to create artificial objects on the original images. Artificial objects, which were given a Gaussian profile with a FWHM the same as the seeing size, were randomly distributed on the real image outside of twice the FWHM of the real objects to avoid blending artificial objects with real objects. We generated 50,000 artificial objects in the 20–30 mag range and extracted them using SExtractor with the same parameter set. This procedure was repeated 10 times, and the detection completeness was 70%–50% at the 3σ limiting magnitudes of i' and z' bands in each field. At fainter than 3σ limiting magnitude, the detection completeness drops sharply to ∼10%. We confirmed that the results of our completeness tests are consistent with the values described in the CFHTLS data release (we find ∼84% completeness at the same magnitude at which ∼80% is expected). It should be noted that detection completeness at bright magnitudes depends on blending with neighbor objects, and we carefully masked out bright objects.

We selected z ∼ 3–6 galaxy candidates using the Lyman-break technique (u-, g-, r-, and i-dropout galaxies). The i'-band detection catalog was used for the selection of u-, g-, and r-dropout galaxies, and i-dropout galaxies were selected from the z'-band detection catalog, based on the following color-selection criteria (van der Burg et al. 2010; Toshikawa et al. 2012):

$$\begin{eqnarray*}u\mbox{--}\mathrm{dropouts} & : & 1.0\lt ({u}^{* }-g^{\prime} )\wedge -1.0\lt (g^{\prime} -r^{\prime} )\\ & & \lt 1.2\wedge 1.5(g^{\prime} -r^{\prime} )\lt ({u}^{* }-g^{\prime} )-0.75,\\ g\mbox{--}\mathrm{dropouts} & : & 1.0\lt (g^{\prime} -r^{\prime} )\wedge -1.0\lt (r^{\prime} -i^{\prime} )\\ & & \lt 1.0\wedge 1.5(r^{\prime} -i^{\prime} )\lt (g^{\prime} -r^{\prime} )\\ & & -0.80\wedge {u}^{* }\gt {m}_{\mathrm{lim},2\sigma },\\ r\mbox{--}\mathrm{dropouts} & : & 1.2\lt (r^{\prime} -i^{\prime} )\wedge -1.0\lt (i^{\prime} -z^{\prime} )\\ & & \lt 0.7\wedge 1.5(i^{\prime} -z^{\prime} )\lt (r^{\prime} -i^{\prime} )\\ & & -1.00\wedge {u}^{* },g^{\prime} \gt {m}_{\mathrm{lim},2\sigma },\\ i\mbox{--}\mathrm{dropouts} & : & (i^{\prime} -z^{\prime} )\gt 1.5\wedge {u}^{* },g^{\prime} ,r^{\prime} \gt {m}_{\mathrm{lim},2\sigma },\end{eqnarray*}$$

where m_lim,2σ is a 2σ limiting magnitude. In redder bands than Lyman break (e.g., g' and r' bands for u-dropout galaxies), we only used objects detected with more than 2σ significance in order to accurately estimate their color; on the other hand, 2σ limiting magnitude was used as the color limit if objects were detected at less than 2σ significance in bluer bands (e.g., u* band for u-dropout galaxies). The results do not significantly change even if we use 1σ or 3σ significance as the limiting magnitude. We estimated the redshift distribution resulting from these criteria by using the population synthesis model code GALAXEV (Bruzual & Charlot 2003) and the absorption of the intergalactic medium (IGM) (Madau 1995). In GALAXEV, we simulated a large variety of galaxy spectral energy distributions (SEDs) using the simple stellar population model. We assumed a Salpeter (1955) initial mass function with lower and upper mass cutoffs m_L = 0.1 M_☉ and m_U = 100 M_☉, two metallicities (0.2 and 0.02 Z_☉), and two SFHs of constant and instantaneous star formation. We extracted model spectra with ages between 5 Myr and the age of the universe at that redshift and applied the reddening law of Calzetti et al. (2000) with E(B − V) between 0.00 and 1.50. The model magnitudes were estimated by convolving these simulated SEDs with the filter transmission curves. We then added the photometric noise, which is typically Δm = 0.04 mag and 0.13 mag at the 10σ and 3σ limiting magnitudes, respectively. In this process, we assumed that the magnitude distribution of the simulated galaxies and the observed dropout galaxies were the same. Then, the redshift distributions of the u-, g-, r-, and i-dropout galaxies are estimated by applying the same color-selection criteria of dropout galaxies to these simulated SEDs (Figure 2). The expected redshift ranges of the dropout selections are z ∼ 2.8–3.7, 3.3–4.3, 4.3–5.1, and 5.7–6.5 for u-, g-, r-, and i-dropout galaxies, respectively. It should be pointed out that these estimates rely on some assumptions (e.g., the model of IGM absorption, SFH). Although we used the IGM model of Madau (1995), other models have been proposed (Meiksin 2006; Inoue et al. 2014). Furthermore, Thomas et al. (2014) found that IGM absorption can vary significantly among different lines of sight, but there is a degeneracy between IGM absorption and dust extinction. However, as long as the properties of the galaxies in overdense regions and in the field are not too different, both populations will be affected in the same manner, implying that we can still search for relative overdensities as a tracer of protoclusters. The numbers of selected dropout galaxies detected in each field are shown in Table 1. Note that the limiting magnitude of this study corresponds to about ${M}_{\mathrm{UV}}^{* }+2.6$, ${M}_{\mathrm{UV}}^{* }+2.4$, ${M}_{\mathrm{UV}}^{* }+1.7$, and ${M}_{\mathrm{UV}}^{* }$ (where ${M}_{\mathrm{UV}}^{* }$ is the characteristic magnitude of the Schechter functions fitted to the dropout galaxies at each redshift) for u-, g-, r-, and i-dropout galaxies, respectively (Bouwens et al. 2007; van der Burg et al. 2010). Thus, the survey depth reaches at least the typical brightness of dropout galaxies, even for i-dropout galaxies.

We evaluated the contamination rate for these color-selection criteria by comparing the dropout selection regions with the positions of contamination on a two-color diagram (Figure 1). The major sources of the contamination are dwarf stars and old elliptical galaxies, the latter possibly able to satisfy the color criteria due to the 4000 Å/Balmer break. To estimate the contamination rate of dwarf stars, we use the TRILEGAL code (Girardi et al. 2005), which can simulate number count and broadband photometry of stars in any Galaxy field. Since this model enables us to set up various structural parameters of thin disk, thick disk, halo, and bulge, we used an exponential disk model with default values of scale length and height, and a Chabrier IMF was applied. The galactic latitudes were set to be the same as those of the observations ($| b| =40-60^\circ$). Then, photometry of the simulated dwarf stars was calculated for the filter set for CFHT/MegaCam. Next, we simulated old galaxy SEDs using the GALAXEV, assuming two relatively high metallicities (Z_☉ and 2.5Z_☉), and we extracted model spectra with ages of 1.0–10.0 Gyr. The redshift tracks of old galaxies are away from all dropout selection regions. And, although a few dwarf stars are located only within the r- and i-dropout selection regions, the main locus of dwarf stars lies far from these regions. Actually, the contamination rate of dwarf stars in the r-dropout samples is expected to be 2.2%–7.8% depending on the galactic latitude, and the contamination rate in i-dropout samples is 3.4%–6.4% in the CFHTLS Deep Fields. Based on these simulations of dwarf stars and old galaxies, the dropout selection criteria used in this study are confirmed to be able to separate high-redshift galaxies from contaminations.

Zoom In Zoom Out Reset image size

We have estimated the local surface number density by counting the number of dropout galaxies within a fixed aperture in order to determine the overdensity significance quantitatively. Chiang et al. (2013) presented a useful definition of the characteristic radius of protoclusters based on cosmological simulations, which encloses 65% of the mass, based on a combination of N-body dark matter simulations and semianalytical galaxy-formation models. Although their result was based on the three-dimensional distribution of protocluster galaxies, it is still a useful guide for constructing a map of the projected local surface number density when searching for protoclusters. According to the characteristic radius of protoclusters having a descendent halo mass of 1–3 × 10¹⁴ M_☉ at z = 0, the radius of 0.75 physical Mpc was used for u-, g-, and r-dropout galaxies, which correspond to 1.6, 1.8, and 1.9 arcmin, respectively. Although the characteristic radius is expected to be larger for protoclusters having more massive descendent masses, these still show a clear overdensity even on 0.75 physical Mpc scales; thus, the aperture size of 0.75 physical Mpc should be effective to find protoclusters with a descendent halo mass of >1 × 10¹⁴ M_☉. However, we still have to consider the projection effects resulting from the large redshift uncertainty of our dropout selection. We discuss the effectiveness of our protocluster search more quantitatively by using a theoretical model in the following subsection. It should be noted that protoclusters have a nearly constant physical size at z ≳ 3 (Chiang et al. 2013; Muldrew et al. 2015). We used a slightly larger radius of 1.0 physical Mpc (2.9 arcmin) for i-dropout galaxies in order to reduce the large Poisson error resulting from a too-small aperture. The apertures were distributed over the CFHTLS Deep Fields in a grid pattern at intervals of ∼20''. We measured the mean and the dispersion, σ, of the number of galaxies in an aperture over the field. The surface number density in masked regions was assumed to be the same as the mean surface number density. Apertures in which more than 5% area is masked are not used in the following analysis. Using the mean and σ of the number of dropout galaxies in an aperture, surface number density contours of u-, g-, r-, and i-dropout galaxies were calculated and are plotted in Figures 3–6, respectively. We note that this is the same procedure that was applied to the i-dropout galaxies in the Subaru Deep Field (SDF) to draw their surface number density contour, which led to the discovery of the protocluster at z = 6.01 from Toshikawa et al. (2012, 2014). In order to study the effects of incompleteness, we performed the same overdensity measurement but for u-, g-, and r-dropout galaxies brighter than 26.0 mag in the i' band (for i-dropout galaxies, it is hard to select a brighter subset because of the small sample size). From this analysis, we confirmed that the change of limiting magnitude does not have a significant effect on our protocluster selection.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Although we can clearly see some overdense regions in the sky distribution maps presented in Figures 3–6, it is not straightforward to find plausible protocluster candidates since the large redshift uncertainty of the dropout technique, which is Δz ∼ 1 (∼230 − 60 physical Mpc at z ∼ 3.1–6.0), hampers the identification of clustering structure in three-dimensional space. On one hand, overdensities associated with real protoclusters could be weakened by fore- or background galaxies; on the other hand, chance alignments of the large-scale structure or superpositions of filaments could erroneously enhance the surface overdensity. Therefore, we will make use of predictions from simulations to understand the relation between surface overdensity significance and the probability of finding real protoclusters.

To this end, we used a set of light-cone models constructed by Henriques et al. (2012). A brief outline of the light-cone models is presented below. First, the assembly history of the dark matter halos was traced using an N-body simulation (Springel et al. 2005), in which the length of the simulation box was 500 h⁻¹ Mpc and the particle mass was 8.6 × 10⁸ h⁻¹ M_☉. The distributions of dark matter halos were stored at discrete epochs. Next, the processes of baryonic physics were added to dark matter halos at each epoch using a semianalytic galaxy-formation model (Guo et al. 2011). Based on the intrinsic parameters of galaxies predicted by the semianalytic model, such as stellar mass, SFH, metallicity, and dust content, the photometric properties of simulated galaxies were estimated from the stellar population synthesis models developed by Bruzual & Charlot (2003). Then, these simulated galaxies in boxes at different epochs were projected along the line of sight, and intergalactic medium (IGM) absorption was applied (Madau 1995) in order to mimic a pencil-beam survey, as described in Overzier et al. (2013). Finally, 24 light-cone models with 1.4 × 1.4 deg² FoV were extracted using these procedures.

The simulated u-, g-, r-, and i-dropout galaxy catalogs were made by matching the expected redshift distribution of each dropout galaxy sample (Figure 2).⁷ We also applied the same limiting magnitude cut only for the detection band used for the observations to the simulated catalogs (i band for u-, g-, r-dropout galaxies, and z band for i-dropout galaxies). The average stellar mass of these simulated dropout galaxies is ∼2 × 10⁹ M_⊙, and the 90% quantile range is ∼2 × 10⁸ to 1 × 10¹⁰ M_⊙. This is consistent with that observed (e.g., Stark et al. 2009), implying that simulated dropout galaxies trace similar structures. Based on these simulated catalogs, we calculated local number density maps as in Section 3.1 and selected overdense regions in the same way as in the CFHTLS Deep Fields. For each overdense region of dropout galaxies, we selected the strongest spike in the redshift distribution, and the dominant dark matter structure was defined by the most massive halo in that redshift spike. The descendent halos at z = 0 for each overdense region were then easily identified by tracing the halo merger tree of those halos. As shown in Figure 7, although there is a large scatter, the significance of the overdensity at high redshift is clearly correlated with the descendent halo mass at z = 0, whose probability of no correlation is <0.01 based on the Spearman rank correlation test. Thus, it is possible to select reliable protocluster candidates through the surface overdensity. In this study, we set the criterion of protocluster candidates at >4σ overdensity significance in order to obtain a high purity of real protoclusters. Based on this criterion, 76 (90)% of these candidates of u-dropout (i-dropout) galaxies are expected to be in real protoclusters. However, the completeness is very small (∼5% and 10% for u- and i-dropout galaxies), mainly due to the projection effect of the dropout technique. It should be noted that the average descendent mass of protoclusters with >4σ overdensity significance is ∼5 × 10¹⁴ M_⊙ (Figure 7). A total of 21 candidates were identified from z ∼ 3 to z ∼ 6 (five, five, six, and five candidates for the maps of u-, g-, r-, and i-dropout galaxies, respectively). The coordinates and overdensity of the protocluster candidates are listed in Table 2. Since these numbers of protocluster candidates are consistent with the model prediction, in which ∼2.9–6.4 candidates per observed area (∼4 deg²) are found in each redshift bin from z ∼ 3 to z ∼ 6, most of the candidates are expected to be real protoclusters. To summarize, from the wide-field imaging of the CFHTLS Deep Fields, we have made a large protocluster sample at z ∼ 3–6 without the aid of any special probes, such as QSOs, RGs, or SMGs; thus, this sample is not only large but also complementary with previous studies targeting QSO, RG, or SMG fields.

Zoom In Zoom Out Reset image size

We carried out spectroscopic observations using Subaru/FOCAS (Kashikawa et al. 2002), Keck II/DEIMOS (Faber et al. 2003), and Gemini-N/GMOS (Hook et al. 2004). In these observations, eight protocluster candidates from z ∼ 3 to z ∼ 6 were observed in total (two at each redshift). The target protocluster candidates and the configuration of spectroscopic observations are described in Tables 2 and 3, respectively. All these observations were conducted with Multi-Object Spectroscopy (MOS) mode. The slits typically had a length of 6–8 arcsec and a width of 0.8–1.0 arcsec. The grisms used were selected in order to have the highest efficiency at the wavelength of the redshifted Lyα line of targeted dropout galaxies and the spectral resolution of $\lt 2.8(1+{z}_{[{\rm{O}}{\rm{II}}]})\mathring{\rm A}$, where 2.8 Å is the wavelength separation of the [O ii] doublet ($\lambda =3726.0,3728.8\,\mathring{\rm A}$) in the rest frame. Therefore, our spectroscopic observations were set up to have a resolution that is sufficient to resolve the [O ii] emission into the doublet to check for contamination by foreground interlopers. The wavelength coverage is also wide enough to cover the expected redshift range of the dropout galaxies. In the FOCAS observations, the telescope was dithered along the slit to enable more accurate sky subtraction between exposures, and we used nod-and-shuffle mode, which allows an increase in the accuracy of sky subtraction by real-time flipping to the sky position in the GMOS observation. Although higher priority was given to brighter galaxies, we designed slit masks so as to allocate as many objects as possible. Furthermore, a slit of each mask was allocated for a bright star (∼20 mag) to monitor the time variations of seeing size or atmospheric transmission between exposures. Although the sky condition was good and stable during all observing nights, we removed only a few poor-quality frames by checking the bright star. Long slit exposures of one of the spectroscopic standard stars HZ44, BD+28d4211, and G191-B2B were taken each night with all configurations used in the night, and we corrected the difference of air mass between science targets and standard stars in the flux calibration. The data taken by FOCAS and GMOS were reduced in a standard manner with IRAF, and the pipeline spec2d⁸ was used for the reduction of the data taken by DEIMOS. The 3σ detection limits of the emission lines are typically 5.0 × 10⁻¹⁸, 4.0 × 10⁻¹⁸, 1.0 × 10⁻¹⁸, and 1.3 × 10⁻¹⁸ erg s⁻¹ cm⁻² for u-, g-, r-, and i-dropout galaxies, respectively, assuming the line width of FWHM = 5.0 Å.

Table 3.  Overview of Our Spectroscopic Observations

Date	Instrument	Target	Grism	Resolution (Å)	Coverage (Å)	texp (minutes)	Nmask	Seeing
2012 May 13 and 14	GMOS	D3ID01	R600	4.5	7500–10000	330	1	05
2012 Oct 21	FOCAS	D1ID01	VPH900	5.7	7500–10100	220	1	09
2014 Aug 24	DEIMOS	D1GD01	600ZD	3.5	5000–9300	120	1	07
		D4GD01	600ZD	3.5	5000-9300	120	1	07
2014 Oct 20 and 21	FOCAS	D1RD01	VPH650	5.5	6000–8300	280	1	07
		D1GD01	VPH520	2.5	4900–6500	100	1	09
		D1UD01	VPH520	2.5	4300–5900	60	4	06
		D4GD01	VPH520	2.5	4900–6500	120	2	07
		D4UD01	VPH520	2.5	4300–5900	60	3	08
2014 Oct 24 and 25	FOCAS	D4RD02	VPH650	5.5	6000–8300	120	1	08

Download table as:  ASCIITypeset image

All emission lines we detected are single emission lines, which are not likely to be Hβ or [O iii] emission lines because the wavelength coverage of our observation is wide enough to detect all these multiple lines simultaneously. Only [O iii]λ5007 emission, which is generally the strongest emission among them, might resemble a single emission line if the other lines are too faint to be detected. However, according to the typical line flux ratio of [O iii]λ5007 and [O iii]λ4959 (${f}_{\lambda 5007}\sim 3\times {f}_{\lambda 4959}$), even [O iii]λ4959 should be detected in our spectroscopic observations since the signal-to-noise ratio (S/N) of detected emission lines is ∼10 on average (and always >3). Therefore, we investigated the possibility that Hα and [O ii] emission lines contaminate dropout galaxy samples based on both imaging and spectroscopic data.

Only r- and i-dropout galaxies can be contaminated by Hα-emitting galaxies according to its wavelength (λ_Hα = 6562.8 Å). Since higher-redshift dropout galaxies are selected using a redder color criterion to detect the strong Lyman break, it is almost impossible to mimic this color by a Balmer break object at lower redshift based on the expected color of passive galaxies, as described in Section 2.2. Even if Balmer breaks of passive and old galaxies were strong enough to satisfy the dropout color criteria, these stellar populations are unlikely to have Hα emission, as a diagnostic of the star-formation activity. On the other hand, dusty starburst galaxies with strong Balmer breaks could be contaminated because they would show a Hα emission line as well as a very red color, resulting from the combination of Balmer breaks and dust reddening. However, their Hα emissions can be discriminated from Lyα emissions, which appear right at the Lyman break. Therefore, we consider the possibility of finding foreground Hα emission negligible.

Regarding [O ii] doublet emission lines, it is possible to distinguish between Lyα and [O ii] emission lines based on the line profile. The spectral resolution of most of our spectroscopic observations was set high enough to resolve [O ii] emission lines as doublets (${\rm{\Delta }}\lambda =3.8\mbox{--}6.3\,\mathring{\rm A}$ at $z\sim 0.3\mbox{--}1.3$), although it would be practically difficult to resolve these in most cases because of low S/N. In this case, the [O ii] emission line is typically skewed blueward, while the Lyα emission line from high-redshift galaxies is skewed redward. Therefore, the skewness of the line profile allows us to distinguish between Lyα and [O ii] emission lines; however, it should be noted that [O ii] emission is sometimes skewed redward when assuming any exotic physical properties of the H ii region (e.g., election density). To improve the way they are distinguished, Kashikawa et al. (2006) introduced "weighted" skewness, which makes use of line width as well as line profile. Lyα emission usually has a larger line width than [O ii] emission because Lyα emission typically emerges in an outflow or galactic wind. Therefore, we calculated the weighted skewness of all spectroscopically detected galaxies. The asymmetric emission lines with S_w > 3 are clear evidence of Lyα emission from high-redshift galaxies, though it would be more difficult to distinguish them from nearby emission-line galaxies at z ∼ 3, where the IGM attenuation is weaker than at higher redshifts. As shown in Table 4, most of the emission lines of this study have large S_w. However, 17% of all identified emission lines have S_w < 3; although this would be caused by strong sky line residuals and low S/N data, we could not rule out the possibility of [O ii] emission lines. In order to make the line profile measurement more accurate by reducing the effect of sky noise and low S/N data, we made a composite spectrum of all 24 emission lines with S_w < 3 by taking a median in the rest frame, assuming they were Lyα, normalized by the peak flux. The weighted skewness of the composite spectrum was found to be S_w = 8.1 ± 1.2, indicating that most of the emission lines even with S_w < 3 in individual spectra are real Lyα emission lines from high-redshift galaxies. In addition to the line profile, the possibility of [O ii] emission can further be reduced by taking into account photometric data. Although [O ii] emission is closer to Balmer break than Hα emission, it will still be difficult to find a sharp break near the emission line, except for peculiar galaxies such as dusty starburst galaxies.

Table 4.  Observed Properties of All Spectroscopically Confirmed Dropout Galaxies

ID	R.A.	decl.	ma	Redshift	MUV	fLyα	LLyα	EW0	Sw
	(J2000)	(J2000)	(mag)		(mag)	(10−18 erg s−1 cm−2)	(1042 erg s−1)	(Å)	(Å)
D1ID01 (three galaxies)
1	02:27:18.8	−04:50:08.3	25.45 ± 0.06	${5.966}_{-0.004}^{+0.002}$	−21.57 ± 0.06	2.66 ± 0.41	1.05 ± 0.16	2.41 ± 0.46	2.25 ± 1.21
2	02:27:21.0	−04:50:49.3	25.97 ± 0.10	${6.044}_{-0.002}^{+0.002}$	−20.85 ± 0.13	19.55 ± 0.72	7.92 ± 0.29	40.05 ± 6.13	5.21 ± 0.72
3	02:27:19.0	−04:53:48.0	26.30 ± 0.13	${6.325}_{-0.003}^{+0.002}$	−20.76 ± 0.24	31.73 ± 0.46	14.29 ± 0.21	81.79 ± 24.95	7.85 ± 0.59
D3ID01 (two galaxies)
1	14:19:22.5	+52:57:22.5	25.21 ± 0.05	${5.749}_{-0.002}^{+0.002}$	−21.49 ± 0.05	5.80 ± 0.88	2.09 ± 0.32	3.79 ± 0.84	5.41 ± 1.16
2	14:19:17.2	+52:56:14.4	25.74 ± 0.08	${5.756}_{-0.002}^{+0.002}$	−20.94 ± 0.08	8.16 ± 1.10	2.95 ± 0.40	10.84 ± 2.14	5.38 ± 3.85
D1RD01 (six galaxies)
1	02:24:45.467	−04:58:52.83	26.37 ± 0.06	${4.431}_{-0.002}^{+0.002}$	−19.85 ± 0.14	1.60 ± 0.39	0.31 ± 0.08	4.09 ± 1.14	3.08 ± 5.41
2	02:24:45.957	−04:56:57.69	26.14 ± 0.05	${4.602}_{-0.002}^{+0.002}$	−20.14 ± 0.12	2.75 ± 0.46	0.59 ± 0.10	5.84 ± 1.17	0.72 ± 4.11
3	02:24:42.586	−04:58:36.00	26.81 ± 0.09	${4.742}_{-0.003}^{+0.002}$	−19.49 ± 0.21	5.22 ± 0.46	1.20 ± 0.10	21.69 ± 5.06	6.73 ± 2.10
4	02:24:38.212	−04:57:15.05	26.16 ± 0.05	${4.840}_{-0.002}^{+0.002}$	−20.21 ± 0.13	5.67 ± 0.73	1.37 ± 0.18	12.79 ± 2.28	15.74 ± 5.69
5	02:24:45.964	−04:54:34.80	26.12 ± 0.05	${4.890}_{-0.002}^{+0.002}$	−20.35 ± 0.12	1.73 ± 0.31	0.43 ± 0.08	3.52 ± 0.74	3.11 ± 2.06
6	02:24:43.757	−04:54:31.19	26.32 ± 0.05	${4.894}_{-0.002}^{+0.002}$	−20.19 ± 0.14	1.56 ± 0.29	0.39 ± 0.07	3.69 ± 0.85	0.58 ± 6.76
D4RD02 (three galaxies)
1	22:16:46.722	−17:28:02.00	26.00 ± 0.05	${4.630}_{-0.002}^{+0.002}$	−20.27 ± 0.12	1.27 ± 0.17	0.28 ± 0.04	2.44 ± 0.43	3.47 ± 21.52
2	22:16:39.959	−17:31:34.58	25.94 ± 0.04	${4.865}_{-0.002}^{+0.002}$	−20.41 ± 0.12	10.31 ± 0.45	2.52 ± 0.11	19.51 ± 2.48	14.12 ± 2.55
3	22:16:45.765	−17:29:19.89	26.07 ± 0.05	${4.952}_{-0.002}^{+0.004}$	−20.36 ± 0.14	9.98 ± 0.35	2.54 ± 0.09	20.67 ± 2.93	15.00 ± 2.88
D1GD01 (36 galaxies)
1	02:25:28.536	−04:17:14.12	26.93 ± 0.07	${3.435}_{-0.001}^{+0.001}$	−18.85 ± 0.15	12.50 ± 1.49	1.34 ± 0.16	43.82 ± 8.47	3.32 ± 8.74
2	02:25:30.408	−04:15:56.70	26.92 ± 0.07	${3.623}_{-0.001}^{+0.001}$	−18.92 ± 0.16	6.94 ± 0.92	0.84 ± 0.11	25.90 ± 5.31	2.74 ± 2.46
3	02:25:32.014	−04:17:03.56	27.17 ± 0.09	${3.705}_{-0.001}^{+0.001}$	−18.66 ± 0.21	6.85 ± 1.01	0.88 ± 0.13	34.18 ± 8.92	8.91 ± 4.93
4	02:25:25.565	−04:17:12.58	27.15 ± 0.09	${3.717}_{-0.001}^{+0.001}$	−18.73 ± 0.20	4.51 ± 0.90	0.58 ± 0.12	21.26 ± 6.04	1.08 ± 1.34
5	02:26:12.550	−04:18:41.23	26.69 ± 0.06	${3.733}_{-0.001}^{+0.001}$	−18.94 ± 0.17	21.41 ± 1.71	2.79 ± 0.22	83.76 ± 15.69	−0.32 ± 3.13
6	02:25:17.290	−04:14:02.23	25.40 ± 0.02	${3.738}_{-0.001}^{+0.001}$	−20.49 ± 0.04	21.57 ± 1.23	2.82 ± 0.16	20.45 ± 1.43	6.11 ± 0.86
7	02:25:51.739	−04:14:37.26	26.07 ± 0.03	${3.744}_{-0.001}^{+0.001}$	−19.91 ± 0.07	3.07 ± 0.79	0.40 ± 0.10	4.97 ± 1.32	0.23 ± 2.27
8	02:25:39.708	−04:14:20.73	25.22 ± 0.01	${3.754}_{-0.001}^{+0.001}$	−20.76 ± 0.04	9.32 ± 1.58	1.23 ± 0.21	6.97 ± 1.20	9.41 ± 9.42
9	02:26:11.563	−04:19:21.65	25.88 ± 0.03	${3.755}_{-0.001}^{+0.001}$	−19.82 ± 0.08	39.82 ± 2.21	5.27 ± 0.29	70.40 ± 6.78	8.13 ± 1.98
10	02:26:10.246	−04:18:18.50	26.81 ± 0.06	${3.759}_{-0.001}^{+0.001}$	−18.69 ± 0.21	25.61 ± 1.08	3.40 ± 0.14	128.41 ± 28.60	5.91 ± 0.54
11	02:25:33.011	−04:14:45.24	25.28 ± 0.02	${3.766}_{-0.001}^{+0.001}$	−20.53 ± 0.04	47.32 ± 2.23	6.30 ± 0.30	43.78 ± 2.74	12.15 ± 1.04
12	02:26:07.202	−04:17:12.22	26.73 ± 0.06	${3.793}_{-0.001}^{+0.001}$	−19.17 ± 0.15	9.89 ± 1.63	1.34 ± 0.22	32.77 ± 7.27	3.60 ± 4.24
13	02:25:59.907	−04:15:45.42	26.23 ± 0.04	${3.797}_{-0.001}^{+0.001}$	−19.74 ± 0.09	10.03 ± 2.29	1.36 ± 0.31	19.73 ± 4.83	7.09 ± 4.96
14	02:25:57.460	−04:18:10.27	26.01 ± 0.03	${3.799}_{-0.001}^{+0.001}$	−19.99 ± 0.07	7.82 ± 1.38	1.06 ± 0.19	12.24 ± 2.32	8.82 ± 4.36
15	02:25:42.923	−04:15:38.74	26.41 ± 0.04	${3.800}_{-0.001}^{+0.001}$	−19.44 ± 0.12	17.61 ± 1.24	2.40 ± 0.17	45.41 ± 6.08	9.85 ± 1.19
16	02:25:44.405	−04:14:11.81	25.32 ± 0.02	${3.803}_{-0.001}^{+0.001}$	−20.64 ± 0.04	24.22 ± 2.34	3.30 ± 0.32	20.78 ± 2.16	10.09 ± 2.61
17	02:25:34.147	−04:14:21.23	26.40 ± 0.04	${3.809}_{-0.001}^{+0.001}$	−19.59 ± 0.11	8.27 ± 1.39	1.13 ± 0.19	18.65 ± 3.66	6.15 ± 4.06
18	02:25:56.529	−04:17:27.85	26.77 ± 0.06	${3.818}_{-0.001}^{+0.001}$	−19.15 ± 0.16	10.76 ± 1.85	1.48 ± 0.25	36.77 ± 8.52	3.17 ± 8.69
19	02:25:30.087	−04:15:15.84	25.85 ± 0.03	${3.827}_{-0.001}^{+0.001}$	−20.20 ± 0.06	6.02 ± 0.94	0.83 ± 0.13	7.83 ± 1.30	8.59 ± 1.02
20	02:25:49.845	−04:14:53.42	26.57 ± 0.05	${3.829}_{-0.001}^{+0.001}$	−19.35 ± 0.13	13.03 ± 1.37	1.81 ± 0.19	37.33 ± 6.22	13.66 ± 4.39
21	02:25:41.772	−04:16:06.53	25.70 ± 0.02	${3.843}_{-0.001}^{+0.002}$	−20.30 ± 0.06	21.31 ± 1.34	2.98 ± 0.19	25.65 ± 2.14	−0.62 ± 0.96
22	02:25:56.593	−04:15:15.20	26.89 ± 0.07	${3.859}_{-0.001}^{+0.001}$	−18.85 ± 0.21	19.31 ± 1.22	2.73 ± 0.17	88.80 ± 19.46	5.15 ± 0.78
23	02:25:39.324	−04:14:40.82	26.79 ± 0.06	${3.866}_{-0.001}^{+0.001}$	−19.12 ± 0.17	13.16 ± 1.11	1.87 ± 0.16	47.68 ± 8.84	5.37 ± 0.92
24	02:26:11.555	−04:17:39.51	26.38 ± 0.04	${3.886}_{-0.001}^{+0.001}$	−19.35 ± 0.14	33.92 ± 2.51	4.87 ± 0.36	100.23 ± 15.93	1.63 ± 1.82
25	02:25:27.362	−04:16:43.59	27.28 ± 0.10	${3.891}_{-0.001}^{+0.001}$	−18.37 ± 0.32	17.16 ± 1.63	2.47 ± 0.24	125.83 ± 44.70	2.87 ± 0.77
26	02:25:25.816	−04:16:38.94	27.61 ± 0.13	${3.927}_{-0.001}^{+0.001}$	−18.04 ± 0.43	13.14 ± 1.83	1.93 ± 0.27	133.94 ± 67.59	−1.16 ± 6.97
27	02:26:01.853	−04:14:41.24	27.73 ± 0.15	${4.000}_{-0.001}^{+0.001}$	−17.65 ± 0.61	15.71 ± 2.24	2.41 ± 0.34	238.11 ± 183.28	5.40 ± 1.71
28	02:25:31.239	−04:15:49.81	26.37 ± 0.04	${4.054}_{-0.001}^{+0.001}$	−19.76 ± 0.12	18.39 ± 1.82	2.92 ± 0.29	41.22 ± 6.28	5.63 ± 2.24
29	02:25:35.470	−04:14:15.68	26.25 ± 0.04	${4.119}_{-0.001}^{+0.001}$	−19.94 ± 0.11	23.18 ± 1.90	3.82 ± 0.31	45.59 ± 6.16	3.35 ± 4.05
30	02:25:34.433	−04:15:05.46	26.37 ± 0.04	${4.185}_{-0.001}^{+0.001}$	−19.83 ± 0.13	25.45 ± 1.46	4.35 ± 0.25	57.91 ± 8.23	8.31 ± 0.57
31	02:25:39.830	−04:14:53.33	26.20 ± 0.04	${4.236}_{-0.002}^{+0.001}$	−20.19 ± 0.11	23.32 ± 2.01	4.11 ± 0.35	39.12 ± 5.27	6.86 ± 1.12
32	02:25:11.525	−04:16:20.17	25.91 ± 0.03	${4.276}_{-0.001}^{+0.001}$	−19.10 ± 0.29	107.86 ± 2.43	19.42 ± 0.44	505.39 ± 153.02	2.40 ± 0.33
33	02:25:57.659	−04:14:24.90	26.54 ± 0.05	${4.385}_{-0.001}^{+0.001}$	−18.83 ± 0.40	61.69 ± 2.19	11.79 ± 0.42	393.36 ± 177.53	5.26 ± 0.81
34	02:25:21.473	−04:16:01.50	27.04 ± 0.08	${4.391}_{-0.002}^{+0.001}$	−19.20 ± 0.30	23.76 ± 2.33	4.55 ± 0.45	107.90 ± 36.20	5.27 ± 1.87
35	02:25:08.716	−04:15:24.74	27.10 ± 0.08	${4.395}_{-0.001}^{+0.001}$	−19.34 ± 0.27	52.76 ± 2.69	10.13 ± 0.52	210.28 ± 60.76	7.74 ± 0.99
36	02:25:28.097	−04:14:54.46	27.68 ± 0.14	${4.442}_{-0.001}^{+0.001}$	−18.01 ± 0.78	22.59 ± 2.06	4.45 ± 0.41	316.27 ± 334.36	4.13 ± 1.99
D4GD01 (42 galaxies)
1	22:16:55.191	−17:25:51.91	24.53 ± 0.01	${3.568}_{-0.001}^{+0.001}$	−21.32 ± 0.02	11.47 ± 1.23	1.34 ± 0.14	4.54 ± 0.50	7.39 ± 2.56
2	22:17:00.190	−17:25:06.37	26.39 ± 0.05	${3.569}_{-0.001}^{+0.001}$	−19.45 ± 0.11	9.05 ± 1.27	1.06 ± 0.15	20.01 ± 3.48	4.83 ± 1.81
3	22:16:55.670	−17:20:49.98	27.16 ± 0.10	${3.581}_{-0.001}^{+0.001}$	−18.67 ± 0.21	6.39 ± 0.65	0.76 ± 0.08	29.05 ± 6.79	1.78 ± 1.80
4	22:16:49.872	−17:21:53.02	25.42 ± 0.02	${3.622}_{-0.001}^{+0.001}$	−20.44 ± 0.05	14.01 ± 1.24	1.70 ± 0.15	12.85 ± 1.26	5.97 ± 4.29
5	22:17:01.326	−17:20:52.55	26.81 ± 0.07	${3.624}_{-0.001}^{+0.001}$	−19.04 ± 0.15	5.40 ± 1.15	0.66 ± 0.14	17.95 ± 4.70	−0.73 ± 2.60
6	22:16:54.811	−17:28:37.91	26.93 ± 0.08	${3.626}_{-0.001}^{+0.001}$	−18.89 ± 0.18	10.59 ± 1.15	1.29 ± 0.14	40.75 ± 8.47	3.86 ± 1.26
7	22:16:58.872	−17:28:33.27	26.33 ± 0.04	${3.628}_{-0.001}^{+0.001}$	−19.53 ± 0.10	6.92 ± 1.32	0.84 ± 0.16	14.68 ± 3.15	5.65 ± 3.60
8	22:17:07.296	−17:28:45.15	26.22 ± 0.04	${3.654}_{-0.001}^{+0.001}$	−19.61 ± 0.10	16.61 ± 1.61	2.06 ± 0.20	33.43 ± 4.49	4.80 ± 3.23
9	22:16:51.756	−17:24:57.97	26.26 ± 0.04	${3.666}_{-0.001}^{+0.001}$	−19.59 ± 0.10	11.90 ± 0.89	1.49 ± 0.11	24.67 ± 3.00	1.03 ± 2.79
10	22:16:42.993	−17:15:53.36	26.96 ± 0.08	${3.669}_{-0.001}^{+0.001}$	−18.90 ± 0.18	4.98 ± 0.89	0.62 ± 0.11	19.56 ± 4.98	7.49 ± 2.83
11	22:16:50.981	−17:18:49.87	26.71 ± 0.06	${3.670}_{-0.001}^{+0.001}$	−19.10 ± 0.15	11.16 ± 2.01	1.40 ± 0.25	36.23 ± 8.51	5.17 ± 4.24
12	22:16:53.509	−17:19:06.60	25.74 ± 0.03	${3.671}_{-0.001}^{+0.001}$	−20.08 ± 0.06	23.63 ± 1.76	2.96 ± 0.22	31.08 ± 2.99	5.67 ± 2.85
13	22:16:49.716	−17:17:00.96	26.45 ± 0.05	${3.671}_{-0.001}^{+0.001}$	−19.41 ± 0.12	8.69 ± 1.12	1.09 ± 0.14	21.37 ± 3.67	7.45 ± 1.80
14	22:16:53.576	−17:19:07.20	24.96 ± 0.01	${3.671}_{-0.001}^{+0.001}$	−20.92 ± 0.03	17.97 ± 1.17	2.25 ± 0.15	10.92 ± 0.77	14.77 ± 4.05
15	22:16:51.410	−17:17:50.44	26.02 ± 0.03	${3.672}_{-0.001}^{+0.001}$	−19.83 ± 0.08	14.83 ± 1.31	1.86 ± 0.16	24.75 ± 2.90	4.22 ± 1.98
16	22:16:54.326	−17:18:34.98	25.95 ± 0.03	${3.675}_{-0.001}^{+0.001}$	−19.95 ± 0.07	6.62 ± 0.81	0.83 ± 0.10	9.85 ± 1.39	5.83 ± 2.25
17	22:16:57.890	−17:21:51.88	26.42 ± 0.05	${3.675}_{-0.001}^{+0.001}$	−19.37 ± 0.12	18.62 ± 1.19	2.34 ± 0.15	47.33 ± 6.39	4.28 ± 0.83
18	22:16:51.591	−17:18:12.00	26.30 ± 0.04	${3.681}_{-0.001}^{+0.001}$	−19.61 ± 0.10	4.78 ± 0.95	0.60 ± 0.12	9.83 ± 2.17	4.43 ± 2.62
19	22:16:55.554	−17:20:14.08	26.66 ± 0.06	${3.681}_{-0.001}^{+0.001}$	−19.14 ± 0.15	12.33 ± 1.31	1.55 ± 0.17	38.80 ± 7.05	5.73 ± 2.17
20	22:16:48.909	−17:15:31.09	26.51 ± 0.05	${3.685}_{-0.001}^{+0.001}$	−19.35 ± 0.12	8.96 ± 1.38	1.13 ± 0.17	23.36 ± 4.59	−3.55 ± 6.95
21	22:16:55.005	−17:21:00.75	25.78 ± 0.03	${3.717}_{-0.001}^{+0.001}$	−20.06 ± 0.07	22.37 ± 1.51	2.89 ± 0.19	31.09 ± 2.93	9.26 ± 1.83
22	22:16:46.962	−17:21:06.42	25.93 ± 0.03	${3.719}_{-0.001}^{+0.001}$	−19.74 ± 0.09	41.34 ± 1.66	5.35 ± 0.21	77.34 ± 7.55	5.17 ± 0.88
23	22:16:46.961	−17:17:10.24	27.19 ± 0.10	${3.720}_{-0.001}^{+0.001}$	−18.64 ± 0.24	6.30 ± 0.94	0.81 ± 0.12	32.32 ± 9.19	3.09 ± 2.67
24	22:16:42.903	−17:17:35.09	25.47 ± 0.02	${3.721}_{-0.001}^{+0.001}$	−20.44 ± 0.05	14.86 ± 1.46	1.92 ± 0.19	14.53 ± 1.58	2.22 ± 2.25
25	22:16:50.522	−17:18:22.62	26.00 ± 0.03	${3.723}_{-0.001}^{+0.001}$	−19.87 ± 0.08	15.39 ± 1.44	2.00 ± 0.19	25.62 ± 3.13	11.86 ± 2.51
26	22:16:49.533	−17:16:44.13	26.38 ± 0.05	${3.728}_{-0.001}^{+0.001}$	−19.50 ± 0.11	8.94 ± 1.05	1.16 ± 0.14	20.82 ± 3.35	10.33 ± 3.45
27	22:17:09.126	−17:28:52.31	26.73 ± 0.06	${3.730}_{-0.001}^{+0.001}$	−19.07 ± 0.17	11.89 ± 1.54	1.55 ± 0.20	41.26 ± 8.65	3.25 ± 1.57
28	22:16:56.467	−17:17:20.11	26.68 ± 0.06	${3.831}_{-0.001}^{+0.001}$	−19.12 ± 0.17	19.18 ± 0.72	2.66 ± 0.10	67.69 ± 11.99	4.13 ± 0.77
29	22:17:01.475	−17:23:58.97	27.01 ± 0.08	${3.837}_{-0.001}^{+0.001}$	−18.82 ± 0.23	13.81 ± 1.17	1.92 ± 0.16	64.73 ± 16.19	5.90 ± 1.67
30	22:17:04.114	−17:29:22.86	26.57 ± 0.06	${3.839}_{-0.001}^{+0.001}$	−19.29 ± 0.16	19.06 ± 1.23	2.66 ± 0.17	58.23 ± 9.70	3.59 ± 1.20
31	22:16:59.785	−17:26:15.22	25.50 ± 0.02	${3.852}_{-0.003}^{+0.001}$	−20.57 ± 0.05	12.90 ± 1.35	1.81 ± 0.19	12.16 ± 1.40	4.41 ± 1.16
32	22:17:00.167	−17:27:32.72	26.70 ± 0.06	${3.854}_{-0.001}^{+0.001}$	−19.03 ± 0.19	23.89 ± 1.47	3.36 ± 0.21	93.40 ± 19.22	3.68 ± 2.03
33	22:16:44.680	−17:17:48.48	25.93 ± 0.03	${3.856}_{-0.001}^{+0.001}$	−19.49 ± 0.13	75.34 ± 1.29	10.62 ± 0.18	192.60 ± 24.90	1.53 ± 0.42
34	22:16:49.846	−17:17:16.49	26.41 ± 0.05	${4.026}_{-0.001}^{+0.001}$	−19.11 ± 0.22	48.03 ± 2.06	7.50 ± 0.32	193.21 ± 43.99	8.03 ± 0.93
35	22:16:51.997	−17:26:10.95	25.85 ± 0.03	${4.076}_{-0.001}^{+0.001}$	−20.35 ± 0.08	27.36 ± 1.87	4.40 ± 0.30	36.05 ± 3.71	5.22 ± 1.24
36	22:16:53.458	−17:20:03.45	27.06 ± 0.09	${4.093}_{-0.001}^{+0.001}$	−19.19 ± 0.22	7.49 ± 0.66	1.22 ± 0.11	29.18 ± 6.98	7.24 ± 1.49
37	22:16:52.593	−17:29:00.63	26.89 ± 0.08	${4.109}_{-0.001}^{+0.001}$	−19.17 ± 0.23	17.48 ± 1.22	2.86 ± 0.20	70.02 ± 17.27	5.76 ± 1.73
38	22:16:59.778	−17:22:16.93	25.75 ± 0.03	${4.126}_{-0.001}^{+0.001}$	−20.12 ± 0.11	68.15 ± 2.28	11.28 ± 0.38	115.08 ± 12.30	4.28 ± 0.67
39	22:17:03.102	−17:25:52.33	25.23 ± 0.02	${4.170}_{-0.001}^{+0.001}$	−20.63 ± 0.07	120.44 ± 3.00	20.43 ± 0.51	130.16 ± 9.74	11.81 ± 0.64
40	22:16:48.708	−17:15:41.17	26.41 ± 0.05	${4.182}_{-0.001}^{+0.002}$	−19.79 ± 0.15	25.01 ± 1.35	4.27 ± 0.23	59.09 ± 9.28	1.45 ± 0.88
41	22:16:49.635	−17:15:26.63	27.50 ± 0.13	${4.220}_{-0.001}^{+0.001}$	−18.24 ± 0.54	17.35 ± 1.48	3.03 ± 0.26	173.91 ± 113.30	6.64 ± 1.53
42	22:16:56.050	−17:24:57.12	26.65 ± 0.06	${4.258}_{-0.001}^{+0.001}$	−19.59 ± 0.19	22.63 ± 1.57	4.03 ± 0.28	66.48 ± 13.79	6.99 ± 1.76
D1UD01 (30 galaxies)
1	02:24:33.775	−04:22:05.64	27.48 ± 0.08	${2.730}_{-0.001}^{+0.001}$	−17.14 ± 0.33	36.73 ± 4.16	2.26 ± 0.26	357.30 ± 132.78	6.12 ± 1.92
2	02:24:24.047	−04:19:30.14	26.75 ± 0.04	${2.936}_{-0.001}^{+0.001}$	−18.82 ± 0.09	11.49 ± 1.98	0.84 ± 0.15	28.27 ± 5.46	5.96 ± 3.75
3	02:24:38.501	−04:19:31.91	25.96 ± 0.02	${2.954}_{-0.001}^{+0.001}$	−19.63 ± 0.04	24.14 ± 1.82	1.80 ± 0.14	28.80 ± 2.48	5.79 ± 1.33
4	02:24:32.251	−04:20:05.64	26.36 ± 0.03	${2.961}_{-0.001}^{+0.001}$	−19.18 ± 0.07	24.91 ± 1.67	1.87 ± 0.13	45.24 ± 4.18	10.26 ± 1.40
5	02:24:30.247	−04:20:25.53	24.45 ± 0.01	${2.977}_{-0.001}^{+0.001}$	−21.21 ± 0.01	52.64 ± 3.69	3.99 ± 0.28	14.85 ± 1.05	11.87 ± 1.47
6	02:24:35.414	−04:20:32.25	26.00 ± 0.02	${3.124}_{-0.001}^{+0.001}$	−19.61 ± 0.05	58.84 ± 2.09	5.01 ± 0.18	81.48 ± 5.02	8.38 ± 1.17
7	02:24:32.181	−04:18:52.41	27.00 ± 0.05	${3.127}_{-0.001}^{+0.001}$	−18.75 ± 0.12	11.77 ± 2.07	1.01 ± 0.18	36.17 ± 7.54	12.53 ± 5.35
8	02:24:26.931	−04:18:09.40	25.10 ± 0.01	${3.130}_{-0.001}^{+0.001}$	−20.77 ± 0.02	16.28 ± 1.97	1.39 ± 0.17	7.81 ± 0.95	1.18 ± 1.96
9	02:24:32.111	−04:19:01.04	26.73 ± 0.04	${3.131}_{-0.001}^{+0.001}$	−19.09 ± 0.09	9.38 ± 1.71	0.80 ± 0.15	21.03 ± 4.21	7.64 ± 2.41
10	02:24:32.361	−04:18:33.93	27.32 ± 0.07	${3.132}_{-0.001}^{+0.001}$	−18.45 ± 0.15	8.49 ± 1.34	0.73 ± 0.12	34.32 ± 7.41	0.35 ± 3.73
11	02:24:38.052	−04:17:50.69	25.76 ± 0.02	${3.150}_{-0.001}^{+0.001}$	−20.08 ± 0.04	18.81 ± 2.10	1.64 ± 0.18	17.21 ± 2.00	6.70 ± 2.73
12	02:24:36.424	−04:20:40.01	27.41 ± 0.08	${3.193}_{-0.001}^{+0.001}$	−18.27 ± 0.18	17.26 ± 1.44	1.55 ± 0.13	86.42 ± 17.51	−1.63 ± 4.53
13	02:24:39.007	−04:17:25.43	26.14 ± 0.02	${3.200}_{-0.001}^{+0.001}$	−19.77 ± 0.05	14.54 ± 1.65	1.31 ± 0.15	18.37 ± 2.25	6.27 ± 1.62
14	02:24:35.609	−04:19:31.99	27.32 ± 0.07	${3.220}_{-0.001}^{+0.001}$	−18.46 ± 0.16	14.64 ± 1.81	1.34 ± 0.17	62.97 ± 12.76	3.07 ± 2.45
15	02:24:36.250	−04:19:11.89	25.81 ± 0.02	${3.258}_{-0.001}^{+0.001}$	−20.04 ± 0.04	56.13 ± 2.21	5.29 ± 0.21	57.99 ± 3.23	12.49 ± 1.98
16	02:24:36.988	−04:18:09.47	25.44 ± 0.01	${3.274}_{-0.001}^{+0.002}$	−20.53 ± 0.03	42.95 ± 2.36	4.10 ± 0.23	28.53 ± 1.73	14.83 ± 2.07
17	02:24:27.725	−04:17:48.50	27.03 ± 0.06	${3.284}_{-0.001}^{+0.001}$	−18.83 ± 0.13	20.07 ± 1.68	1.93 ± 0.16	64.53 ± 9.60	5.80 ± 1.54
18	02:24:35.157	−04:17:00.64	26.61 ± 0.04	${3.344}_{-0.001}^{+0.001}$	−19.48 ± 0.08	8.56 ± 1.33	0.86 ± 0.13	15.65 ± 2.68	7.98 ± 2.43
19	02:24:28.399	−04:20:01.41	26.27 ± 0.03	${3.351}_{-0.001}^{+0.001}$	−19.77 ± 0.06	23.60 ± 1.80	2.38 ± 0.18	33.30 ± 3.16	10.49 ± 1.71
20	02:24:38.367	−04:17:16.11	27.34 ± 0.07	${3.357}_{-0.001}^{+0.001}$	−18.71 ± 0.15	7.92 ± 1.12	0.80 ± 0.11	29.73 ± 6.09	3.94 ± 1.62
21	02:24:41.996	−04:18:59.15	27.01 ± 0.05	${3.400}_{-0.001}^{+0.001}$	−19.11 ± 0.11	12.78 ± 1.57	1.33 ± 0.16	34.43 ± 5.69	10.15 ± 2.55
22	02:24:37.488	−04:19:20.22	26.12 ± 0.02	${3.426}_{-0.001}^{+0.001}$	−20.02 ± 0.05	34.92 ± 1.95	3.71 ± 0.21	41.50 ± 3.12	9.26 ± 0.97
23	02:24:28.416	−04:21:30.12	26.78 ± 0.04	${3.435}_{-0.001}^{+0.001}$	−19.32 ± 0.10	23.86 ± 1.70	2.55 ± 0.18	54.13 ± 6.44	3.95 ± 0.73
24	02:24:36.548	−04:18:26.31	26.56 ± 0.04	${3.454}_{-0.001}^{+0.001}$	−19.61 ± 0.08	21.02 ± 1.16	2.28 ± 0.13	37.07 ± 3.42	7.23 ± 1.34
25	02:24:35.602	−04:16:54.03	27.31 ± 0.07	${3.455}_{-0.001}^{+0.001}$	−18.60 ± 0.19	26.89 ± 1.59	2.92 ± 0.17	119.80 ± 23.53	3.58 ± 1.71
26	02:24:44.626	−04:19:35.65	26.24 ± 0.03	${3.463}_{-0.001}^{+0.001}$	−19.95 ± 0.06	30.63 ± 1.60	3.34 ± 0.17	39.74 ± 3.03	7.64 ± 2.19
27	02:24:38.037	−04:22:12.46	27.49 ± 0.08	${3.529}_{-0.001}^{+0.001}$	−18.67 ± 0.19	20.88 ± 1.47	2.38 ± 0.17	91.63 ± 19.14	9.51 ± 2.56
28	02:24:29.531	−04:21:43.03	26.83 ± 0.05	${3.550}_{-0.001}^{+0.001}$	−19.45 ± 0.10	29.51 ± 2.19	3.41 ± 0.25	64.39 ± 8.00	3.44 ± 2.08
29	02:24:35.608	−04:21:10.87	27.10 ± 0.06	${3.551}_{-0.001}^{+0.001}$	−19.25 ± 0.12	14.71 ± 1.57	1.70 ± 0.18	38.72 ± 6.21	11.86 ± 3.78
30	02:24:24.653	−04:19:31.71	26.99 ± 0.05	${3.555}_{-0.001}^{+0.001}$	−19.40 ± 0.11	11.07 ± 1.24	1.29 ± 0.14	25.39 ± 3.88	3.50 ± 3.09
D4UD01 (16 galaxies)
1	22:14:03.642	−18:00:09.90	26.61 ± 0.04	${2.973}_{-0.001}^{+0.001}$	−19.05 ± 0.08	6.08 ± 1.27	0.46 ± 0.10	12.47 ± 2.79	3.65 ± 6.22
2	22:13:58.570	−17:59:30.91	27.05 ± 0.06	${3.008}_{-0.001}^{+0.001}$	−18.60 ± 0.13	7.66 ± 1.28	0.60 ± 0.10	24.50 ± 5.14	0.09 ± 1.53
3	22:14:11.265	−17:59:56.51	25.84 ± 0.02	${3.037}_{-0.001}^{+0.001}$	−19.70 ± 0.05	57.83 ± 2.77	4.61 ± 0.22	68.72 ± 4.69	7.67 ± 0.97
4	22:13:53.488	−17:56:54.74	26.56 ± 0.04	${3.046}_{-0.001}^{+0.001}$	−18.95 ± 0.10	33.35 ± 2.43	2.67 ± 0.19	80.04 ± 9.72	5.06 ± 1.72
5	22:13:51.171	−17:57:18.84	26.66 ± 0.04	${3.138}_{-0.001}^{+0.001}$	−19.15 ± 0.09	11.14 ± 1.56	0.96 ± 0.13	23.74 ± 3.92	4.81 ± 1.76
6	22:13:53.773	−17:57:40.48	27.09 ± 0.06	${3.210}_{-0.001}^{+0.001}$	−18.79 ± 0.13	9.29 ± 1.75	0.84 ± 0.16	29.27 ± 6.72	4.17 ± 1.95
7	22:13:54.597	−17:59:06.04	26.79 ± 0.05	${3.241}_{-0.001}^{+0.001}$	−19.12 ± 0.10	13.84 ± 1.33	1.29 ± 0.12	32.89 ± 4.58	2.97 ± 4.01
8	22:13:55.114	−17:59:55.62	26.26 ± 0.03	${3.242}_{-0.001}^{+0.001}$	−19.68 ± 0.06	16.00 ± 1.44	1.49 ± 0.13	22.77 ± 2.46	3.70 ± 2.06
9	22:14:04.835	−17:57:44.20	26.93 ± 0.06	${3.243}_{-0.001}^{+0.001}$	−18.81 ± 0.14	27.86 ± 1.96	2.60 ± 0.18	88.14 ± 13.39	9.82 ± 1.41
10	22:14:04.154	−18:00:05.58	26.72 ± 0.05	${3.243}_{-0.001}^{+0.001}$	−19.11 ± 0.11	23.82 ± 1.90	2.22 ± 0.18	56.94 ± 7.36	7.18 ± 1.45
11	22:14:03.430	−17:59:22.71	26.03 ± 0.02	${3.249}_{-0.001}^{+0.001}$	−19.90 ± 0.05	24.84 ± 1.88	2.32 ± 0.18	28.98 ± 2.62	5.89 ± 1.17
12	22:14:09.396	−17:57:58.56	26.98 ± 0.06	${3.336}_{-0.001}^{+0.001}$	−19.12 ± 0.11	3.32 ± 0.73	0.33 ± 0.07	8.44 ± 2.09	4.28 ± 10.94
13	22:14:16.330	−17:57:22.22	27.54 ± 0.10	${3.341}_{-0.001}^{+0.001}$	−18.46 ± 0.21	9.19 ± 1.56	0.92 ± 0.16	43.24 ± 11.65	6.27 ± 2.88
14	22:14:09.371	−17:58:15.61	26.99 ± 0.06	${3.341}_{-0.001}^{+0.001}$	−19.02 ± 0.13	13.93 ± 1.74	1.39 ± 0.17	39.00 ± 6.87	4.65 ± 2.26
15	22:13:58.117	−17:59:46.75	26.92 ± 0.06	${3.560}_{-0.001}^{+0.001}$	−19.48 ± 0.11	12.41 ± 1.54	1.44 ± 0.18	26.50 ± 4.40	5.63 ± 1.58
16	22:14:07.173	−18:00:24.05	26.48 ± 0.04	${3.635}_{-0.001}^{+0.001}$	−20.03 ± 0.08	37.94 ± 2.08	4.65 ± 0.26	51.22 ± 4.73	8.40 ± 1.04

Note.

^aThe apparent aperture magnitude of detection band is $i^{\prime}$ band for u-, g-, and r-dropout, and $z^{\prime}$ band for i-dropout galaxies.

Download table as:  ASCIITypeset images: 1 2 3 4

From these considerations, it is unlikely that Hα or [O ii] emission lines contaminate our dropout samples, and Lyα is the most plausible interpretation to explain both photometric and spectral features. Since the major contamination in the photometric selection is completely different from that in the spectroscopic observation, the combination of photometric and spectroscopic observations enables us to select a clean sample of high-redshift galaxies. We can regard all single emission lines detected from our dropout sample as Lyα emission lines.

Our protocluster confirmation completely depends on the detection of Lyα emission; therefore, we might only select a part of the galaxy populations in these protoclusters. We might miss protoclusters, if they were mainly composed of passive or dusty galaxies without Lyα emission. This may lead to a possible selection bias in our protocluster search. However, protoclusters at z ∼ 2–3 have been found to mainly contain star-forming galaxies. It is worth noting that known protoclusters that include a large number of older or dustier galaxies, like the SSA22 (z = 3.1) and the spiderweb (z = 2.2) protoclusters, also show a significant overdensity in LAEs (Steidel et al. 2000; Kuiper et al. 2010; Kubo et al. 2013). These results suggest that the possible bias introduced by tracing protoclusters only by LAEs is probably not significant.

The number of dropout galaxies located in each protocluster candidate region is listed in Table 2, and the number of spectroscopically observed galaxies is shown in Table 5. From these numbers, about half, or at least 35%, of dropout galaxies in protocluster candidate regions were observed by our follow-up spectroscopic observations. We carefully discriminated real emission lines from sky lines or noise by examining both two-dimensional and one-dimensional spectra, and all emission lines identified in this study are shown in Figure 9. We estimate the observed properties of the spectroscopically confirmed galaxies, such as UV absolute magnitude at 1300 Å in the rest frame (M_UV), Lyα luminosity (L_Lyα), and rest-frame Lyα equivalent width (EW₀), shown in Table 4. The redshifts were derived by the peak wavelength of the Lyα emission line, assuming the rest wavelength of Lyα to be 1215.6 Å. These measurements could be overestimated if there was a galactic outflow. When emission lines are located near strong sky lines, the position of the peak could be shifted. These effects of sky lines and the wavelength resolution are taken into account when estimating the error. Observed line flux, f_Lyα, corresponds to the total amount of the flux within the line profile. The slit loss was corrected based on the ratio of slit width and seeing size for each observation, and its flux error was estimated from the combination of the line width and the noise level per 1 Å at wavelengths blueward of Lyα. Since continuum flux was too faint to be detected in the observed spectra, M_UV was estimated from the broadband magnitude (g', r', i', and z' bands for u-, g-, r-, and i-dropout galaxies). The M_UV is defined as the absolute magnitude at 1300A in the rest-frame. It is derived from the broadband photometry after correcting the contribution of IGM absorption and the Lyα emission to the broadband photometry based on the spectroscopic. In this calculation, we have assumed a UV slope β (f_λ ∝ λ^β) to be −2, which is consistent with the previous observations (e.g., Bouwens et al. 2012). The broadband magnitude has a negligible dependence on UV slope because the broadband width is only ∼1000–1500 Å. We have confirmed that M_UV changes only a few percent at maximum when the UV slope was varied from −3.0 to −1.0. In addition, EW₀ was estimated by combining f_Lyα and M_UV. The results found for each region are described in the following subsections.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.3.1. The i-Dropout Protocluster Candidate in the D1 Field

We have observed eight i-dropout galaxies in the D1ID01 region out of 10 candidates. Almost all i-dropout galaxies in the D1ID01 region were spectroscopically observed, as shown in Figure 10. Three galaxies clearly have single emission lines, which can be identified as Lyα emission lines of z ∼ 6 galaxies. Their photometric and spectroscopic properties are summarized in Table 4. Two of three galaxies (ID = 1 and 2) have close redshifts with a difference of Δz = 0.08, corresponding to the radial distance of 4.7 Mpc in the physical scale. From our selection criteria, we can expect an ∼0.2–0.4 galaxy in a Δz = 0.1 bin if they were homogeneously distributed in redshift space. The possibility to have two galaxies within Δz < 0.1 is 16%. Although their distribution is more concentrated than homogeneous, these two galaxies are unlikely to merge into a single halo by z = 0 based on our analysis of the possible separations of protocluster galaxies at z ∼ 6 (Figure 8). For now, we cannot conclude that there is a real protocluster in the D1ID01 region because of the small number of statistics.

Zoom In Zoom Out Reset image size

4.3.2. The i-Dropout Protocluster Candidate in the D3 Field

As for the D3ID01 region, eight i-dropout galaxies were observed out of 16 candidates. The completeness of our spectroscopic observation is smaller (∼50%) than for the D1ID01 region, which has fewer protocluster member candidates. Many faint i-dropout galaxies are still to be observed because we assigned higher priorities to the brighter i-dropout galaxies. Lyα emission lines were detected from two of the eight spectroscopic targets. The sky distribution of the targets is shown in Figure 11. Table 4 describes the properties of the spectroscopically confirmed galaxies. These two galaxies have almost the same redshift with a difference of Δz < 0.01 (<0.5 Mpc in physical scale). The possibility that two galaxies have this small redshift separation is only 1.2%, and these two galaxies can certainly be expected to be in the same halo at z = 0 based on this small separation. While we could not make a clear conclusion because of the small number of confirmed galaxies, the discovery of a close galaxy pair at z ∼ 6 could imply the existence of a protocluster. Interestingly, Toshikawa et al. (2014) have found that galaxies tend to be in close pairs in another protocluster at z = 6.01.

Zoom In Zoom Out Reset image size

The relatively small number of confirmed candidates can be attributed to the observational limit since our spectroscopic samples are biased to brighter galaxies (${M}_{\mathrm{UV}}\lt -20.75$). We compared the D3ID01 protocluster candidate with the z = 6.01 protocluster in the SDF (Toshikawa et al. 2014), which was also identified by the combination of dropout selection and follow-up spectroscopy targeting Lyα emission like this study. Applying the same magnitude limit of M_UV < −20.75 to the SDF protocluster, the number of remaining protocluster members was only two out of a total of 10. Furthermore, Ouchi et al. (2005) reported the discovery of two protoclusters at z ∼ 5.7. These were discovered from a narrow-band survey, and six and four LAEs are included in each protocluster. Although LAE selection is different from our dropout selection, it is useful to check the distribution of the UV continuum and the Lyα luminosity of protocluster galaxies. Based on our observational limits of UV continuum and Lyα luminosity, only about two LAEs would be identified for these protoclusters. Therefore, it is not unreasonable to expect only two confirmed member galaxies in this study even if there is a real protocluster.

4.3.3. The r-Dropout Protocluster Candidate in the D1 Field

We have spectroscopically observed 15 r-dropout galaxies in the D1RD01 region, and we detected single emission lines from six galaxies. The sky distribution of the observed galaxies is shown in Figure 12. In the >1σ overdense region, there are ∼40 galaxies; thus, only ∼38% r-dropout galaxies were observed by the follow-up spectroscopy. Two galaxies (ID = 5 and 6) out of six are clustering both in spatial (Δsky = 33 arcsec) and redshift space (Δz = 0.004) at z = 4.89 and whose three-dimensional separation is 0.7 Mpc in the physical scale. Considering the observed volume (3 arcmin radius and Δz ∼ 0.8), it is unlikely (<1%) that the close pair is reproduced by a uniform random distribution of six galaxies in three-dimensional space. However, it is unclear whether this galaxy pair will grow into a cluster at z = 0 because of the small number of confirmed galaxies; at least, these two galaxies are expected to merge into a single halo. Since there are many spectroscopically unobserved galaxies, further follow-up observations will enable us to clarify whether there is a protocluster or not.

Zoom In Zoom Out Reset image size

4.3.4. The r-Dropout Protocluster Candidate in the D4 Field

In the D4RD02 region, the total integration time of follow-up spectroscopic observation was only 2 hr, which was half of that in the D1RD01 region. Thus, although 12 r-dropout galaxies were observed, Lyα emission lines were detected from only three galaxies. The sky distribution of the observed galaxies is shown in Figure 13. These three galaxies are largely separated in redshift space. Since about 20 r-dropout galaxies remain to be spectroscopically observed, further follow-up observations will be necessary to draw a conclusion.

Zoom In Zoom Out Reset image size

4.3.5. The g-Dropout Protocluster Candidate in the D1 Field

Combining the DEIMOS and FOCAS follow-up observations, 123 g-dropout galaxies were observed, and the redshifts of 36 galaxies were determined by detecting Lyα emission lines. The sky distribution of the observed galaxies is shown in Figure 14. Figure 15 shows the redshift distribution of confirmed galaxies. Although galaxies seem to be clustering at z ∼ 3.8, these galaxies are spread over a large projected area, as shown in Figure 14. Since the DEIMOS has a wide FoV (∼16.7 × 5.0 arcmin²), which is larger than the area of the D1GD01 region, some g-dropout galaxies outside the overdense region were also observed. When we focus only on galaxies in the overdense region, shown by the red-line histogram in Figure 15, the peak around z = 3.8 becomes lower, and only three galaxies are clustering within the expected redshift range of the protocluster candidate (Δz < 0.01). Given the total number of confirmed galaxies in the overdense region, the group of three galaxies can be reproduced even from a random homogeneous distribution with a probability of 21%. Based on this probability, it cannot be dismissed that the observed redshift distribution is drawn from a uniform random distribution. Hence, we cannot currently conclude that there is a protocluster in the D1GD01 region without further observations. Hereafter, we regard the D1GD01 region as not being a protocluster. The high surface overdensity observed in this field could be attributed to a coincidental alignment of large-scale structure on a scale of Δz ∼ 0.1–0.2, which is too large to grow into a single halo by z = 0.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.3.6. The g-Dropout Protocluster Candidate in the D4 Field

Combining the DEIMOS and FOCAS follow-up observations, 144 g-dropout galaxies were spectroscopically observed in the D4GD01 protocluster region, and the redshifts of 42 galaxies were determined by detecting Lyα emission lines. The sky distribution of the observed galaxies is shown in Figure 16. The redshift distribution is shown in Figure 17. There is a clear excess at z = 3.67, and, as contrasted with the D1GD01 region, 11 galaxies are clustered in a narrow redshift range of Δz = 0.016, corresponding to 2.6 Mpc in the physical scale. Since it is almost impossible (<0.01%) to explain this clustering with a random homogeneous distribution, we concluded that there is a protocluster at z = 3.67 that includes 11 member galaxies (ID = 10–20).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

It should also be noted that an AGN was found in this region at (ΔR.A., Δdecl.) = (−1.9, 6.8) arcmin by our spectroscopy, as shown in Figure 16. The redshift was derived to be z = 3.72 based on its He ii and C iii] emission lines (Figure 18). According to this estimate, the redshift separation between the AGN and the center of the protocluster is Δz = 0.05, which corresponds to the radial distance of ∼8 physical Mpc. As the redshift separation is too large, it is unlikely that this AGN is a part of the protocluster members that will merge into a single halo by z = 0.

Zoom In Zoom Out Reset image size

4.3.7. The u-Dropout Protocluster Candidate in the D1 Field

We have spectroscopically observed 95 u-dropout galaxies in the D1UD01 region, and 30 galaxies have single emission lines. The sky distribution of the observed galaxies is shown in Figure 19. The redshift distribution is shown in Figure 20. There is a excess at z = 3.13, including five galaxies within Δz = 0.008. The probability to reproduce this excess by drawing from a uniform random distribution of 30 galaxies is only 0.9%; thus, the five galaxies were found to be significantly clustered, though the absolute excess is only five. The spatial and redshift separations among these five galaxies are small enough to merge into a single halo by z = 0 compared with the model prediction; therefore we confirmed a protocluster at z = 3.13, which includes five member galaxies (ID = 6–10).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.3.8. The u-Dropout Protocluster Candidate in the D4 Field

We have spectroscopically observed 57 u-dropout galaxies in the D4UD01 region, and 16 galaxies have single emission lines. The sky distribution of the observed galaxies is shown in Figure 21. The redshift distribution is shown in Figure 22. There is a peak at z = 3.24, consisting of five galaxies within Δz = 0.008. The probability to reproduce this excess by a uniform random distribution of 16 galaxies was found to be less than 0.1%. These five galaxies are expected to merge into a single halo by z = 0 compared with the model prediction. Therefore, we confirmed a protocluster at z = 3.24, which includes five member galaxies (ID = 7–11).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.3.9. Summary of Protocluster Confirmation

We investigated the three-dimensional distribution of protocluster galaxies in the D4GD01 region, as shown in Figure 23. In this analysis, the distances are simply estimated from the measured redshift, including any possible peculiar velocity component. However, we have checked that this does not significantly affect our estimates given the typical size of protoclusters.⁹ This protocluster seems to have a region where galaxies are strongly concentrated, reminiscent of a cluster core. To discuss the internal structure quantitatively, we calculated the spatial separation of galaxies with respect to the Nth nearest neighbor. In Figure 24, the distributions of the separation from the first to sixth nearest galaxies of individual protocluster galaxies are shown, and the red-line histograms indicate the expected distribution if 11 galaxies are randomly located inside the protocluster. The distribution of protocluster galaxies from the third to fifth nearest are significantly different from the random distribution, whose significances are p < 0.05 based on the Kolmogorov–Smirnov (KS) test; especially, the significance of fifth nearest is less than 0.01. In contrast, there are no significant differences in the distribution of sixth or higher nearest galaxies. These results suggest that the protocluster has a subgroup consisting of six galaxies; therefore, the subgroup cannot be seen in the distribution of the sixth or higher nearest galaxies, which are consistent with the random distribution. The six galaxies constituting the subgroup are defined by having a shorter separation than 1.0 physical Mpc from the fifth nearest galaxies (ID = 11–16). The distributions of first and second nearest galaxies could also be divided into two groups of shorter and longer separations, though they are not so significant (p ∼ 0.2), due to focusing on too small a scale. These six galaxies are indicated by red points in Figure 23 and are located near the center of the protocluster. There are several galaxies in the region surrounding the core, which could assemble into the core to form a rich cluster. This is in clear contrast to what we saw in the protocluster at z ∼ 6, which was found to have several small subgroups, like galaxy pairs (Toshikawa et al. 2014). Although we have only one protocluster at each redshift, if they are the progenitor and descendant of each other, the transformation of protocluster internal structure from z ∼ 6 to z ∼ 4 may be indicative of the virializing process over cosmic time, whereby protoclusters dynamically evolve into a more and more concentrated structure. At z ∼ 2–3, the virializing process would not be completed yet, though there are richer protoclusters where some massive and passive galaxies have already appeared. The internal structures of some protoclusters have been more closely investigated using extensive spectroscopy, and some protoclusters were found to have significant substructure (Kuiper et al. 2011, 2012). Therefore, even at the same redshift, protoclusters could have a large variety of internal structures. In this study based on a small number of protoclusters, the difference of internal structure found in the z ∼ 6 and z ∼ 4 protoclusters is assumed to result from the evolutionary phase of the representative protoclusters. However, larger samples at each redshift will be required to statistically match progenitors and descendants, allowing us to study cluster formation over cosmic time (e.g., Chiang et al. 2013).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We compared several galaxy properties between protocluster members and coeval field galaxies to investigate whether there are any differences that are due to their environment. The D4GD01 protocluster consists of 11 members, while the number of field g-dropout galaxies available is 67 by combining the galaxies in the D1 and D4 fields. The protocluster and field galaxies were identified by the same imaging and spectroscopic observations; thus, some sample bias, if any, would affect both samples in the same way. This should allow us to make a fair comparison between protocluster and field galaxies. The average of L_Lyα, M_UV, and EW₀ are described in Table 6. Since the D4GD01 protocluster has a core-like substructure, as mentioned in Section 5.1, the 11 members were divided into two groups of six galaxies in the core and five galaxies in the outskirts. Table 6 also shows the average properties of these two subgroups. In Figure 25, the properties of individual protocluster and field galaxies are plotted on the M_UV and EW₀ diagram. We found that protocluster galaxies have significantly smaller EW₀ or L_Lyα luminosity than field galaxies (KS p-value is <0.05); especially, fainter protocluster members in M_UV are strongly suppressed in their Lyα emissions compared with field galaxies. No significant difference between the core and the outskirts in the EW₀ and M_UV distribution was found.

Zoom In Zoom Out Reset image size

Previous studies have found little evidence for significant differences between the properties of galaxies inside and outside protoclusters, at least at z ≳ 4 (Overzier et al. 2008, 2009b; Toshikawa et al. 2014). It is therefore interesting that we are finding a difference in the EW₀ or L_Lyα distributions between field and protocluster in the D4GD01 system at z = 3.67. This is perhaps an indication that the situation is changing around z ∼ 4. One simple mechanism that could reduce the Lyα EW₀ would be dust, which traps Lyα photons. If dust is a major reason for the small EW₀, M_UV is expected to be systematically smaller in the protocluster because UV flux can also be easily attenuated by dust, as in the case with Lyα emission. However, we did not find any systematic difference in M_UV between the protocluster and field galaxies (see Table 6). Although dust attenuation does not seems to be the reason for this result, it is still possible that the amount of UV flux attenuated by dust is being compensated for by higher SFR or other effects related to UV emission; in that case, we would not find any difference in the distribution of apparent M_UV between protocluster and field galaxies even if there were a clear difference of dust attenuation. Therefore, in order to more directly estimate the dust attenuation, we compared the UV slope between protocluster and field galaxies. The UV slope of each g-dropout galaxy was determined from the i − z color. However, since the i'-band image was significantly deeper (∼1 mag) than the z'-band image, some g-dropout galaxies were not detected in the z' band. For the estimate of the UV slope, we therefore only used g-dropout galaxies that were detected in the z'-band image with >2σ significance (<26.70 mag). Although this discordant depth between the i'- and z'-band images will lead to a bias in the estimate of UV slope, the protocluster and field galaxies would be equally biased because they were selected by the same criteria and from the same data set. The number of g-dropout galaxies used in the estimate was nine for the protocluster and 50 for the field. Note that there is no difference in the fraction of galaxies detected in the z' band between protocluster (9/11) and field (50/67). The UV slope was calculated from

$$\begin{eqnarray}&&\beta =-0.4\times \displaystyle \frac{{m}_{i}-{m}_{z}}{\,{\mathrm{log}}_{10}\,{\lambda }_{\mathrm{eff},i}-\,{\mathrm{log}}_{10}\,{\lambda }_{\mathrm{eff},z}}-2.0,\end{eqnarray} \tag{ 1 }$$

where λ_eff is the effective wavelength. The average β of the protocluster galaxies was β = −1.88 ± 0.38, and that of the field galaxies was β = −1.92 ± 0.17; thus, it would be hard to explain the difference in EW₀ simply by dust attenuation. It should be noted that the radiative process of Lyα photons to escape from a galaxy is very complicated and affected by many other quantities, such as dust geometry, gas kinematics, and outflow (e.g., Verhamme et al. 2008; Duval et al. 2014).

Neutral hydrogen gas within a protocluster is another possible reason for the small EW₀. Cucciati et al. (2014) found a large amount of neutral hydrogen gas (∼10¹²–10¹³ M_☉) in the intracluster space of a protocluster by examining spectra of background galaxies of a protocluster at z = 2.9, which showed absorption at the same wavelength as the observed Lyα of the protocluster. If the same is true in the D4GD01 protocluster, though our follow-up spectroscopy is not deep enough to check this even by stacking, a small EW₀ could be explained as being a result of resonant scattering by the intracluster neutral hydrogen gas. While the UV photons can penetrate neutral hydrogen gas, Lyα emission is scattered and diffused, consistent with our results. Suppose that a nearly mature protocluster, such as the D4GD01 protocluster, had already accumulated significant cold intracluster gas at z = 3.67; the intracluster gas would come either from the outside of the protocluster drawn in by the strong gravitational potential of the protocluster or could be brought in with the evolved member galaxies themselves through inflows and outflows.

Although it is difficult to identify the cause, we have found that the average EW₀ of z = 3.67 protocluster members is significantly smaller than that of field galaxies. However, it is still under debate how Lyα EW₀ depends on environments. Actually, in contrast with our study, Yamada et al. (2012) have found larger EW₀ in the SSA22 protocluster at z = 3.09. Kuiper et al. (2012) reported a protocluster being composed of two subgroups at z = 3.13, where the one subgroup has larger EW₀, but the other has smaller. Furthermore, there are some protoclusters that have no significant difference of Lyα EW₀ from that of field galaxies (Mawatari et al. 2012). Since the number of known protoclusters is limited even around z ∼ 3 so far, a large-enough sample will be required to address a general feature. This may have consequences for the measurement of the Lyα fraction (Ono et al. 2012; Treu et al. 2013), which is one of the ways to probe reionization at high redshifts. A detailed study of galaxy properties will be made in the future by combining multiwavelength data.