1. Introduction
Powerful active-galactic-nuclei (AGN) feature heavily in our understanding of galaxy evolution, with AGN activity thought to promote (e.g. Ishibashi & Fabian Reference Ishibashi and Fabian2012; Silk Reference Silk2013) or suppress (e.g. Croton et al. Reference Croton2006; Davies et al. Reference Davies, Crain, Oppenheimer and Schaye2020; Lammers et al. Reference Lammers2023) star formation in the host galaxy. Meanwhile, radio observations are unaffected by dust obscuration, and so allow such activity to be detected out to higher redshift than is possible at other wavelengths (e.g. Collier et al. Reference Collier2014; Singh et al. Reference Singh2014). This includes finding high-redshift (proto-)clusters, by exploiting the tendency of ‘radio-loud’ AGN to reside in dense environments (Venemans et al. Reference Venemans2007; Wylezalek et al. Reference Wylezalek2013; Hlavacek-Larrondo et al. Reference Hlavacek-Larrondo2015). However, when studying the properties of powerful AGN as a function of redshift and/or environment, detailed research is hindered by small-number statistics.
Currently, the revised Third Cambridge Catalogue of Radio Sources (3CRR; Laing, Riley, & Longair Reference Laing, Riley and Longair1983) is the most-prominent, low-frequency radio-source sample (
$S_{\mathrm{178\,MHz}} \gt 10.9$
Jy) that is optically complete, but this consists of only 173 sources (all in the northern hemisphere). White et al. (Reference White2018, Reference White2020a,b) have created a sample that is over 10 times larger – the GLEAM 4-Jy (G4Jy) Sample – using observations from the Murchison Widefield Array (MWA; Tingay et al. Reference Tingay2013) over the entire southern sky (Dec.
$\lt$
30
$^{\circ}$
). Thanks to these measurements at low radio-frequencies, we can select radio-loud AGN in an orientation-independent way (Barthel Reference Barthel1989). This is because the low-frequency emission of powerful AGN is dominated by the radio lobes, which are not subject to relativistic beaming (Rees Reference Rees1966). The same cannot be said for the radio core, ‘hotspots’, and jets that dominate the emission of sources at high radio-frequencies. As a result of this beaming effect, radio sources selected at high frequencies tend to be biased towards AGN that have their jet axis close to the line-of-sight (e.g. Lister Reference Lister2003). Therefore, with 1 863 of the brightest radio-sources at low frequencies making up the G4Jy Sample, we can test models of powerful AGN more robustly than previously (e.g. Mullin, et al., Reference Mullin, Riley and Hardcastle2008; Wang & Kaiser Reference Wang and Kaiser2008; Best & Heckman Reference Best and Heckman2012; Shabala et al. Reference Shabala2020). However, the paucity of existing optical spectroscopy over the southern sky [e.g. through the 6-degree Field Galaxy Survey, 6dFGS; Jones et al. Reference Jones2009] significantly limits the breadth of science that can be undertaken, hence our follow-up of a large fraction of G4Jy sources (Sejake et al. in preparation) using the Southern African Large Telescope (SALT).
In this paper, our results include some of the very ‘brightest’ of the G4Jy Sample, as determined through their integrated flux-density at
$\sim$
178 MHz. Applying a threshold of 9.0 JyFootnote
a
here allows us to define a subsample that is equivalent to the first revised version of the Third Cambridge catalogue of radio sources (3CR survey; Bennett Reference Bennett1962; Spinrad et al. Reference Spinrad, Djorgovski, Marr and Aguilar1985) of the northern hemisphere. Hence, they are referred to as the G4Jy-3CRE subset (Massaro et al. Reference Massaro and White2023a). To ensure that there is not overlap with the 298 extragalactic sources that comprise the 3CR sample, a criterion is applied with respect to the sky coverage (Dec.
$\lt -$
5
$^{\circ}$
), resulting in a list of 264 G4Jy sources. These sources are being studied in the X-ray (e.g. Massaro et al. Reference Massaro2023b) in order to better-understand how radio jets interact with their surroundings, with additional optical spectroscopy being provided via numerous telescopes (e.g. García-Pérez et al. Reference García-Pérez2024). The intrinsic radio-properties of these G4Jy-3CRE sources will be presented (in more-complete form) in future work.
We also plan to extend X-ray follow-up to the wider G4Jy Sample as a whole. However, to ensure good signal-to-noise ratios we need to have high spectroscopic completeness so that sources are not biased with respect to dust obscuration. This is also necessary for creating appropriate subsamples for observations with the Atacama Large Millimeter/submillimeter Array (ALMA), which would allow us to address questions about the amount of gas that is available for black-hole accretion and star formation. The wide range in optical magnitudes for our sources makes the G4Jy Sample particularly suitable for large-scale follow-up with SALT, offering us a firm foundation for combining with additional spectroscopy in the future.
1.1. Paper outline
In the next section (Section 2), we describe the criteria that were applied to the G4Jy Sample in order to generate the target list for our spectroscopic campaign with SALT. The resulting spectra are presented in Appendix B, and the acquired redshifts are summarised in Section 3. Section 3 also discusses the radio luminosities and linear sizes for G4Jy sources, enabled by the collation of 299 redshifts. J2000 co-ordinates and AB magnitudes are used throughout this work, and we use a
$\Lambda$
CDM cosmology, with
$H_{0} = 70$
km s
$^{-1}$
Mpc
$^{-1}$
,
$\Omega_{m}=0.3$
,
$\Omega_{\Lambda}=0.7$
.
2. Target selection and data acquisition
In creating the G4Jy Sample, White et al. (Reference White2020a,b) used the GaLactic and Extragalactic All-sky MWA (GLEAM) catalogue (Hurley-Walker et al. Reference Hurley-Walker2017) to select 1 863 radio-sources brighter than 4.0 Jy at 151 MHz (with a spatial resolution of
$\sim$
arcmin). They then used higher-resolution images from:
-
1. the TIFR GMRT Sky Survey (TGSS) first alternative data release (ADR1; Intema et al. Reference Intema, Jagannathan, Mooley and Frail2017), at
$\sim$
25 arcsec resolution, -
2. the Sydney University Molonglo Sky Survey (SUMSS) catalogue (Mauch et al. Reference Mauch2003; Murphy et al. Reference Murphy2007), at
$\sim$
45 arcsec, and -
3. the NRAO (National Radio Astronomy Observatory) VLA Sky Survey (NVSS; Condon et al. Reference Condon1998) at
$\sim$
45 arcsec,
to determine the radio morphology of the sources, and help to identify the host galaxies (Fig. 1). The latter was done through careful visual inspection, using W1-band images from AllWISE (Cutri et al. Reference Cutri2013) to avoid being biased against the most dust-obscured sources. The result is that 1 606 of the 1 863 sources have identifications in the (original) G4Jy catalogue (White et al. Reference White2020a,b), 1 253 of which are at Declinations accessible by SALT (
$-76^{\circ} \lt$
Dec.
$\lt 11^{\circ}$
). Considering the 1 606 sources, we remove 136 with spectra from the 6-degree Field Galaxy Survey (6dFGS; Jones et al. Reference Jones2009), and 104 with Sloan Digital Sky Survey (SDSS) spectra (DR12; Alam et al. Reference Alam2015). This illustrates that we still require optical spectroscopy for the vast majority of the sample, primarily in order to obtain robust redshifts and (where possible) determine the accretion modes of these radio galaxies.
An example overlay showing how different sets of radio contours (GLEAM [200 MHz] in red, SUMSS [843 MHz] in blue, and TGSS [150 MHz] in yellow) were used to assess the morphology of a G4Jy source (G4Jy 1628), with the respective beam-sizes of the different radio surveys shown in the bottom left-hand corner. The underlying, inverted-greyscale image is from the W1 band of AllWISE, with green plusses (‘+’) marking AllWISE catalogue positions within 3 arcmin of the radio-centroid position (purple hexagon). This enabled White et al. (Reference White2020a,b) to identify the appropriate host galaxy of the radio emission (white ‘+’), which was followed by thorough checks against published studies before being included the G4Jy catalogue.
When compiling R-band magnitudes for the sample (Fig. 2) in January 2020 (SALT proposal: 2020-1-MLT-008, PI: White), we cross-matched the AllWISE host-galaxy positions with the following datasets, using a radius of 1 arcsec: SuperCOSMOS (Hambly et al. Reference Hambly2001), the National Optical Astronomy Observatory (NOAO) Source Catalog (NSC) DR1 (Nidever et al. Reference Nidever2018), the Dark Energy Spectroscopic Instrument (DESI) Legacy Imaging Surveys (LS) DR8 (Dey et al. Reference Dey2019), SDSS DR12 (Alam et al. Reference Alam2015), SkyMapper DR1.1 (Wolf et al. Reference Wolf2018) and PanSTARRS (Flewelling et al. Reference Flewelling2020) photometry newly-extracted using ProFound (Robotham et al. Reference Robotham2018). (Further details are provided in Appendix A.) The scatter in magnitudes from the different datasets was then assessed and used to convert everything onto the SuperCOSMOS magnitude scale. The aim of our SALT follow-up is to get good completeness for the G4Jy Sample down to
$R = 20.0$
mag (the optical limit of the SuperCOSMOS data), and doing so gives us a subsample of 706 sources (at
$-76^{\circ} \lt$
Dec.
$\lt 11^{\circ}$
, and including the 158 sources with 6dFGS/SDSS spectra in this range). An additional 38 G4Jy sources were added to the SALT target-list, thanks to new host-galaxy identifications provided by Sejake et al. (Reference Sejake2023) via 1300-MHz MeerKAT imaging (of
$\sim$
7-arcsec spatial resolution). For a summary of the number of sources considered, please see Table 1.
(a) Distributions of the R-band magnitudes (Section 2) for different subsets of the G4Jy Sample (
$-76^{\circ} \lt$
Dec.
$\lt 11^{\circ}$
, restricted to
$R\sim 20$
), with the magnitude distribution for 3CRR sources (without restriction) added for comparison (grey histogram). ‘Homogenised’ refers to magnitudes from different surveys being put on the SuperCOSMOS scale (Hambly et al. Reference Hambly2001), and ‘SkyMapper overlap’ refers to the 429 G4Jy-on-SALT targets that appear in the SkyMapper survey. (b) A comparison of the SkyMapper-PSF r-magnitude with the homogenised R-band magnitude for the G4Jy targets that appear in the SkyMapper survey (purple ‘+’). Sources that also belong to the G4Jy-3CRE subset (Massaro et al. Reference Massaro and White2023a) are indicated by red dots.
A summary of how the target list of 586 G4Jy sources was derived from the original catalogue of 1 863 sources (White et al. Reference White2020a,b). This summary also accounts for the 299 redshifts presented in this paper. Note that 12 of the 98 host-galaxy positions provided by Sejake et al. (Reference Sejake2023) were confirmation of existing identifications in the G4Jy catalogue.
Targetting (
${706 - 158 + 38 =}$
) 586 G4Jy sourcesFootnote
b
enables investigation of a subsample that is over 10 times larger than the number of 3CRR sources at
$R \lt 20.0$
. This allows us to determine more-robust statistics for comparing with galaxy-evolution simulations, making this an excellent legacy dataset for studying powerful active galaxies as a function of redshift and environment. The all-sky distribution of our sources means that we have relaxed scheduling restraints with respect to Right Ascension and Declination, whilst the broad range of R-band magnitudes (Fig. 2) means that we can take advantage of SALT time that is available for different moon conditions; ‘Bright’ time is suitable for the 316 targets that are at
$R\lt18.5$
mag, and ‘Grey’ time is selected for the 270 targets that are at
$18.5 \le R\mathrm{/ mag}\lt 20.0$
. Thanks to the sources with brighter magnitudes, we can also accommodate more-difficult seeing conditions (up to 2.5 arcsec).
We observe the sources via the pg0900 grating on the SALT Robert Stobie Spectrograph (RSS), at the standard grating-angle of 15.875
$^{\circ}$
, and with the pc03850 Order-Blocking Filter. This provides the resolution (R =
$\sim$
660–930) and wavelength coverage (
$\sim$
4 500 – 7 500Å) for maximising redshift completeness (
$z\lesssim4.5$
) and broad-line width measurement. Acquisition involves imaging the field using the SALT Imaging CAMera (SALTICAM, also abbreviated to SCAM), and (typically) identifying an acquisition object that allows fainter targets to enter the longslit via the ‘blind-offset’ method (see Appendix B). Also, we split the observations into 2 to 3 exposures per target (in order to enable cosmic-ray removal), and perform wavelength calibrations using the RSSMOSPipeline softwareFootnote
c
(Hilton et al. Reference Hilton2018), facilitated by arc-observations of a Xenon lamp. This Python-based pipeline automatically detects sources that are present along the 2-arcsec-width longslit of the RSS, uses data from the intervening regions for sky subtraction, and performs stacking of spectra from multiple exposures.
3. Results and discussion
The (homogenised, SuperCOSMOS-scale) R-band magnitudes that informed our observations are shown in Fig. 2(a). This dataset not only aids redshift determination but also provides further understanding of host-galaxy properties, which are essential for understanding the role of AGN in galaxy evolution. 124 of the 586 G4Jy sources belong to the G4Jy-3CRE subset (Massaro et al. Reference Massaro and White2023a), and 429 G4Jy sources have a (less than) 1-arcsec crossmatch with DR4 of the SkyMapper survey (Onken et al. Reference Onken2024)Footnote
d
We note that the application of an
$R = 20.0$
mag cut to the homogenised magnitudes corresponds to an
$r \simeq 21.0$
limit in the SkyMapper-PSF magnitudes. A scatter-plot comparison of the magnitudes is shown in Fig. 2(b), and indicates that there is a large degree of scatter, irrespective of source brightness. As such, our multisemester campaign needed to be extended so that underexposed host-galaxies could be reobserved (in order to achieve better signal-to-noise ratios). This is crucial as it facilitates the determination of redshifts for a larger sample of radio sources, which is essential for understanding their evolution and the environments that they inhabit.
3.1. SALT spectra
The first batch of 42 G4Jy sources that already have good signal-to-noise ratios for deriving redshifts are shown in figure B1, with the redshifts obtained being summarised in Table 2. Twenty-six of the sources listed belong to the G4Jy-3CRE subset, and so supplement the 42 (entirely-new) redshifts obtained by García-Pérez et al. (Reference García-Pérez2024) and via ongoing observing campaigns. In addition, we note that six of the observations presented here are made possible by the host-galaxy positions determined through MeerKAT imaging (Sejake et al. Reference Sejake2023).
Spectroscopic redshifts for 42 G4Jy sources, as determined via SALT optical-spectroscopy (Appendix B). Sources belonging to the G4Jy-3CRE subset (Massaro et al. Reference Massaro and White2023a) and the MeerKAT-2019 subset (Sejake et al. Reference Sejake2023) are indicated with a flag of ‘1’ in the respective columns. The point-spread function (PSF) r-band magnitude provided via DR4 of SkyMapper (Onken et al. Reference Onken2024) is also presented, where available. The night of observation is given in the format of YYYY-MM-DD.
If the target spectrum is dominated by emission lines, then we attempt to fit it with a ‘quasar’ templateFootnote
e
; if, instead, absorption lines dominate the spectrum, then we employ an ‘early-type galaxy’ templateFootnote
f
(Yip et al. Reference Yip2004). In sources that appear to have an even mixture of emission- and absorption-lines, we fit the target spectrum with a ‘galaxy’ template that incorporates (for example) both a strong
$\rm {H}\alpha$
emission-line and the Calcium-II H+K doublet
$^{\mathrm{e}}$
. As a reminder, the aim is to obtain redshifts rather than complete emission-/absorption-line characterisation (for which flux calibration would be needed). The error in the redshift is estimated by applying lower and upper limits to the template, and seeing (by eye) how well-aligned the spectral features remain. Such template-shifting allows us to get an idea of the ‘tolerance’ of the fit, and also encompasses any error in the wavelength calibration (which is usually
$\lt 1$
Å).
Like Mauch & Sadler (Reference Mauch and Sadler2007), we see a mixture of emission-line (‘e’), absorption-line (‘a’), and absorption-/emission-line (‘ae’) spectra, and no correspondence between r-band magnitude and the spectroscopic redshift (Table 2). However, we note that the brightness of emission lines allows for the redshift to be more-readily determined than via fitting absorption lines (which are more difficult to differentiate from a noisy continuum), and therefore our first set of results are biased towards the former spectral-type. For the interested reader, Sejake et al. (in prep.) will present the relative fractions for a more-completely-defined subsample of G4Jy sources.
3.2. Comments on individual sources
G4Jy 530, G4Jy 590, and G4Jy 1819 overlap with the sources analysed by García-Pérez et al. (Reference García-Pérez2024). Our redshifts are in agreement for G4Jy 530 and G4Jy 1819, but differ for G4Jy 590 (where they estimate the redshift as
$z = 0.5384 \pm 0.0027$
). Following correspondence with the authors, we provide the corrected spectroscopic-redshift of
$z = 0.529 \pm 0.002$
, in disagreement with the photometric redshift (
$z =0.58$
) provided by Burgess & Hunstead (Reference Burgess and Hunstead2006).
The largest redshift error in the present work is for G4Jy 541 (
$z = 0.200 \pm 0.007 $
), on account of the CaII H+K doublet being tentatively detected (towards the edge of the wavelength coverage) in addition to strong [OIII] emission-lines. We note that there are not emission lines in the template that align with the distinctive peaks at
$\sim$
6 650 and
$\sim$
6 825Å, but caution that the target’s continuum is affected by suboptimal sky-background subtraction.
The ‘double’ radio morphologies of G4Jy 1704 and G4Jy 1705 were first published by White et al. (Reference White2020a,b), based on the PhD thesis of Haigh (Reference Haigh2001), and show extended emission within the Abell-3785 cluster. They are referred to as ‘the dancing ghosts’ by Norris et al. (Reference Norris2021) and Velović et al. (2023), who present ASKAP and MeerKAT imaging of these radio sources, respectively. Our spectroscopic redshifts are in agreement with the photometric redshifts previously acquired (Bilicki et al. Reference Bilicki, Jarrett, Peacock, Cluver and Steward2014), with G4Jy 1704 at
$z = 0.078 \pm 0.001$
and G4Jy 1705 at
$z = 0.076 \pm 0.002$
. These values were determined via fitting their SALT spectra with an ‘early-type galaxy’ template, as shown in figure B1.
Within the SALT spectra we identify the MgII emission-line for 8 of the 42 sources, these being: G4Jy 672, G4Jy 706, G4Jy 901, G4Jy 909, G4Jy 1511, G4Jy 1665, G4Jy 1698, and G4Jy 1709. This is of special interest for follow-up, as the black-hole mass can be estimated via measuring the linewidth of this line (e.g. McLure & Jarvis Reference McLure and Jarvis2002).
We recheck how many G4Jy sources have spectra in SDSS (Appendix C), and find that 53 overlap with SDSS DR16 (Ahumada et al. Reference Ahumada2020), which includes 36 that have had their SDSS-DR12 entries updated. We retain the remaining (
$104 - 36 =$
) 68 SDSS-DR12 sources (Alam et al. Reference Alam2015) that are cross-identified with the G4Jy Sample, and combine these with the 136 redshifts from 6dFGS (Jones et al. Reference Jones2009). The redshift distributions of these different datasets are shown in Fig. 3.
Distributions of the redshifts (Section 3.2) for different subsets of the G4Jy Sample (with no restrictions based on Declination). The redshift distribution for 3CRR sources (Laing et al. Reference Laing, Riley and Longair1983) is added for comparison (grey histogram, scaled by 0.5), and the 6dFGS distribution has been scaled by 0.1.
The highest of these redshifts is
$z=2.17827 \pm 0.00015$
from SDSS DR12, which is corroborated by Hewett & Wild (Reference Hewett and Wild2010), who measure
$z = 2.17920 \pm 0.00055$
. The G4Jy overlay for this source, G4Jy 1065 (also known as 4C +11.45), is shown in Fig. 4. The large MWA beam means that the GLEAM contours are affected by confusion with a nearby unrelated source towards the southwest, whilst the TGSS contours show extended radio morphology at 150 MHz. The latter (‘double’ morphology) was first presented by Barthel et al. (Reference Barthel, Miley, Schilizzi and Lonsdale1988) in a VLA map at 5 GHz.
Radio contours (GLEAM [200 MHz] in red, NVSS [1 400 MHz] in blue, and TGSS [150 MHz] in yellow) for G4Jy 1065, with the respective beam-sizes of the different radio surveys shown in the bottom left-hand corner. The inverted-greyscale image is from the W1 band of AllWISE, with green plusses (‘+’) marking AllWISE catalogue positions within 3 arcmin of the radio-centroid position (purple hexagon). The host galaxy of the radio emission is indicated by a white ‘+’, in close alignment with the radio positions from the different radio surveys (a red square, a blue ‘
$\times$
’, and a yellow diamond, respectively).
3.3. Intrinsic radio properties
The combination of SALT, 6dFGS, and SDSS redshifts allows us to calculate intrinsic radio-properties for a sample of 299 G4Jy sources. For this work, we focus on radio luminosities (based upon measurements at 151 MHz) and projected linear-sizes (based upon angular sizes determined through NVSS and SUMSS catalogue positions, at 1.4 GHz and 843 MHz, respectively; see section 6.3.1 by White et al. Reference White2020a). These values are provided in Table C1 (Appendix C). We note that 49 sources do not have spectral indices because the G4Jy flux-densities (20 measurements across 72–231 MHz, provided in the G4Jy catalogue) are not well-fit by a power-law function (
$S \propto \nu^{\alpha}$
, where
$\alpha$
is the spectral index within the GLEAM band). As a result, we do not calculate their radio luminosities for Fig. 5. This is because these luminosities are K-corrected, assuming a power-law description of the radio emission. Future work will present the broadband spectra of these sources, which show significant spectral curvature in the radio (White et al., in preparation).
Distributions of the 151-MHz radio-luminosities (Section 3.3) for different subsets of the G4Jy Sample (with no restrictions based on Declination). The luminosity distribution for 3CRR sources (Laing et al. Reference Laing, Riley and Longair1983) is added for comparison (grey histogram).
Distributions of the (projected) linear sizes (Section 3.3) for different subsets of the G4Jy Sample (with no restrictions based on Declination). The size distribution for 3CRR sources (Laing et al. Reference Laing, Riley and Longair1983) is added for comparison (grey histogram), with three of these sources (NGC 6251, 3C 326 = G4Jy 1282, and 3C 236) having linear sizes that are beyond the plot range (i.e. 1 900–4 530 kpc).
In order to compare the G4Jy Sample with 3CRR (Laing et al. Reference Laing, Riley and Longair1983), we scale the 178-MHz radio-luminosities to 151-MHz radio-luminosities for the 3CRR sources (with the spectral index,
$\alpha$
, taken from the 3CRR catalogue). We remind the reader that the G4Jy (sub-)sample is not spectroscopically-complete yet, but it is encouraging that new redshifts from SALT are already enabling us to probe a wide parameter space in luminosity (Fig. 5). The five G4Jy sources with the highest luminosities, above
$1.4 \times 10^{29}$
W Hz
$^{-1}$
, are: G4Jy 1511 (SALT,
$z =1.554\pm 0.003$
), G4Jy 1698 (SALT,
$z =1.630 \pm 0.002$
), G4Jy 845 (re-fitted
$z = 1.706 \pm 0.001 $
; see figure B3), G4Jy 1065 (SDSS DR12,
$z=2.17827 \pm 0.00015$
), and G4Jy 682 (SDSS DR16,
$z=1.96818 \pm 0.00016$
). Joining the most-powerful (
$\gtrsim 10^{29}$
W Hz
$^{-1}$
) radio-galaxies in the Universe (e.g. Laing et al. Reference Laing, Riley and Longair1983; Saxena et al. Reference Saxena2018; Capetti & Balmaverde Reference Capetti and Balmaverde2024), we anticipate that the G4Jy Sample will allow more-robust comparisons with cosmological and galaxy-evolution simulations, for a range of redshifts and environments.
The distribution of 151-MHz radio-luminosities against the (projected) linear sizes of the G4Jy Sample (section 3.3), with no restrictions based on Declination. Each set of ‘upper limits’ is with respect to the linear-size value and/or the radio luminosity. Upper limits in the linear size are represented by horizontal arrows, and are the result of the angular size of the source being an upper limit (due to the resolution of SUMSS/NVSS imaging). Vertical arrows represent upper limits in the radio luminosity, which are a consequence of the 151-MHz flux-density being affected by blended emission from unrelated radio sources. (The affected G4Jy sources are demarcated via a ‘confusion flag’ of ‘1’.) Meanwhile, diagonal arrows are used when both the linear size and the radio luminosity are upper limits. The distribution for 3CRR sources (Laing et al. Reference Laing, Riley and Longair1983) is added for comparison (grey circles), with three of these sources (NGC 6251, 3C 326 = G4Jy 1282, and 3C 236) having linear sizes that are beyond the plot range (at 1 900–4 530 kpc). In addition, 49 G4Jy sources do not appear in this figure on account of their lack of a spectral-index fit (White et al. Reference White2020a). (This is required for appropriate K-correction of the radio luminosity.)
Connected to the gas-density distribution of the environment is the linear extent of the radio emission associated with a particular host galaxy. Of course, the 2-dimensional imaging of radio sources means that we can only calculate the projected linear size, when going from the observed frame to the intrinsic frame. As such, all linear sizes are (in effect) lower limits, but for this work we treat these sizes as absolute values (see Fig. 6). We also present the linear-size distributions when upper limits are included (dashed-line histograms in Fig. 6). These correspond to G4Jy sources that are either unresolved in NVSS/SUMSS imaging or have radio emission at 151 MHz that is blended with an unrelated radio-source. (The former are indicated via an inequality sign, ‘<’, associated with the angular size in the G4Jy catalogue, and the latter are assigned a ‘Confusion flag’ of ‘1’; Table C1.).
Again, the linear sizes of 3CRR sources are added for comparison, with these values being obtained from the online 3CRR databaseFootnote
g
. However, they are derived from radio maps at a range of frequencies (typically 1.4–8 GHz) and so the linear sizes will be connected to the different spatial resolutions (and sensitivity to different angular-scales) that are involved. Our linear sizes (for the GJ4y Sample) are more-consistent in that they are from a homogeneous dataset; NVSS or SUMSS, both of which are at 45-arcsec resolution. Either way, another caveat to add, regarding both the G4Jy Sample and the 3CRR sample, is that the linear-size values will be lower limits for FRI (‘centre-brightened’) sources and better estimates for FRII (‘edge-brightened’) sources (Fanaroff & Riley Reference Fanaroff and Riley1974). The scope of this work can be extended to consider FRI and FRII sources separately, once a survey of the entire G4Jy Sample is completed at
$\lesssim$
10 arcsec resolution (allowing such a morphology classification to be performed).
Like White et al. (Reference White2020b), we follow Willis & Strom (Reference Willis and Strom1978) in defining a giant radio galaxy (GRG) as having a projected, linear size of
$\gt1$
Mpc. Amongst the 299 G4Jy sources analysed in this work are three GRGs: G4Jy 1079 (1.271 Mpc), G4Jy 133 (1.388 Mpc), and G4Jy 517 (1.620 Mpc). However, the existence of 6dFGS spectra (Jones et al. Reference Jones2009) means that these are already-known GRGs, as noted in section 4.8 of White et al. (Reference White2020b). If we instead relax the size threshold (for defining a GRG) to 0.7 Mpc (e.g. Malarecki et al. Reference Malarecki2013; Oei et al. Reference Oei2023), then we can identify G4Jy 1665 (SALT,
$z=1.488\pm 0.003$
), G4Jy 1173 (SDSS DR12,
$z=0.05509\pm0.00001$
), G4Jy 604 (SALT,
$z=0.049\pm 0.003$
), G4Jy 619 (6dFGS,
$z=0.05501$
), and G4Jy 1741 (6dFGS,
$z=0.14519$
) as additional sources belonging to this category of radio galaxy.
Of these, G4Jy 604 is a new GRG, with a linear size of 0.846 Mpc. Its absorption-line optical spectrum (Figure B1) may explain why, despite its 6-Jy brightness at 151 MHz, it has not been previously classified as a GRG. Meanwhile, G4Jy 1173 is noted by Oei et al. (Reference Oei2023) as a GRG in the LOw-Frequency ARray (LOFAR) Two-metre Sky Survey (LoTSS), and G4Jy 1741 is recorded in the Rapid ASKAP Continuum Survey (RACS) by Andernach et al. (Reference Andernach, Jiménez-Andrade and Willis2021). As for G4Jy 619 (aka PKS B0634
$-$
205), its giant size has been known since the work of Danziger et al. (Reference Danziger, Goss and Frater1978).
For further interest, in Fig. 7 we present the 151-MHz radio-luminosity against the linear size. It is thought that extended radio-galaxies follow evolutionary tracks across this plane, in what is effectively a power–size (P–D) diagram (Baldwin Reference Baldwin, Heeschen and Wade1982), but this is complicated by various factors, such as: (i) the impact of the environment on the radio morphology (e.g. Vardoulaki et al. Reference Vardoulaki2024), (ii) the strength of magnetic fields (e.g. Miley Reference Miley1980; Jamrozy et al. Reference Jamrozy, Klein, Mack, Gregorini and Parma2004) in and around the radio galaxy (through its connection to the radio luminosity), (iii) the impact of viewing angle on the observed brightness of the radio emission (although this is ameliorated by the low-frequency selection for the G4Jy Sample; Barthel Reference Barthel1989), (iv) assumptions made about the duty cycle of the AGN (e.g. Turner Reference Turner2018), and (v) how quickly hotspots advance through the intergalactic medium (Alexander & Leahy Reference Alexander and Leahy1987). We refer the reader to figure 8 of Hardcastle et al. (Reference Hardcastle2019) and the discussion therein for further details. It is expected that as we gather more redshifts for the G4Jy Sample, we will populate a greater area of the P–D diagram, with follow-up observations (such as those studying X-rays and polarisation) allowing us to break the degeneracy of some of the aforementioned factors.
For now, we note that the G4Jy Sample shows a similar spread in radio luminosity and linear size as the 3CRR sample (Fig. 7). The lack of a G4Jy source larger than 1 620 kpc (Fig. 6) may be a combined effect of current incompleteness of the spectroscopic follow-up, and the relative shallowness of existing optical data (resulting in a bias towards the small volume of the local Universe). We also emphasise that the linear sizes are limited by the 45-arcsec resolution of the SUMSS/NVSS imaging. Combined with the Malmquist bias towards higher radio-luminosities at higher redshifts, this could explain the clustering (and positive trend) of linear-size upper-limits at
$L_{\mathrm{151\ MHz}} \gtrsim 10^{27}$
W Hz
$^{-1}$
. This is because the upper limit of the linear size will become more discrepant with increasing redshift for unresolved sources (in a given radio survey). We conclude with a note that this is being addressed via improved angular sizes for the full sample (White et al., in preparation).
4. Summary
Our ongoing observing campaign with SALT is to obtain optical spectroscopy (PI: White) for a complete sample (SuperCOSMOS
$R2 \lesssim 20.0$
) of powerful radio-galaxies from the G4Jy Sample (White et al. Reference White2020a,b), which are selected at low radio-frequencies and distributed over the entire southern sky. This enables accurate redshift measurements, with new spectra presented in this paper (for 42 G4Jy sources; Fig. B1) and as part of a larger data release (Sejake et al., in preparation). We also re-fit the spectra for five G4Jy sources that appear in SDSS (Alam et al. Reference Alam2015; Ahumada et al. Reference Ahumada2020), and present the corrected redshifts in Appendix C.
For an initial analysis of 299 G4Jy sources, we combine SALT redshifts with those from 6dFGS (Jones et al. Reference Jones2009) and SDSS, and find that the sample spans a wide range in radio luminosities (
$L_{\mathrm{151\,MHz}} = 5.7\times 10^{23}$
–
$3.5\times 10^{29}$
W Hz
$^{-1}$
) and linear sizes (1–1 620 kpc). We expect that the lower flux-density threshold for the G4Jy Sample (
$S_{\mathrm{151\,MHz}} \gt 4.0$
Jy), compared to the famous 3CRR sample (
$S_{\mathrm{178\,MHz}} \gt 10.9$
Jy), will allow us to better-populate the radio-power–physical-size (
$P\!-\!D$
diagram; Fig. 7) and conduct more-detailed investigations of how radio-loud AGN evolve, over a larger fraction of the Universe’s history.Footnote
h
The redshift information collated from multiple sources will also enable us to address several key questions on the interaction of AGN with their environment, without the orientation bias that affects both AGN samples selected at high radio-frequencies (due to relativistic beaming) and those selected in the optical (caused by dust obscuration). Furthermore, the spectroscopy will form a valuable legacy multiwavelength dataset for future detailed active-galaxy studies. This is because it addresses a critical gap in prior work on powerful AGN in the southern sky, aiding research with facilities like the Square Kilometre Array (SKA) and its precursor telescopes.
Acknowledgements
The observations reported in this paper were obtained with the SALT, under program 2020-1-MLT-008 (PI: White). We thank Christian Hettlage for his expertise with SALT schedule-blocks; Lucia Marchetti, Mattia Vaccari, and the SALT Team for helpful comments on the proposal; Christopher White for help with the SALT Finder Charts; and Dan Smith for support. We also thank the anonymous reader for their comments, which improved the scope of the paper, and both Martin Hardcastle and the anonymous referee, regarding the luminosities.
The national facility capability for SkyMapper has been funded through ARC LIEF grant LE130100104 from the Australian Research Council, awarded to the University of Sydney, the Australian National University, Swinburne University of Technology, the University of Queensland, the University of Western Australia, the University of Melbourne, Curtin University of Technology, Monash University and the Australian Astronomical Observatory. SkyMapper is owned and operated by The Australian National University’s Research School of Astronomy and Astrophysics. The survey data were processed and provided by the SkyMapper Team at ANU. The SkyMapper node of the All-Sky Virtual Observatory (ASVO) is hosted at the National Computational Infrastructure (NCI). Development and support of the SkyMapper node of the ASVO has been funded in part by Astronomy Australia Limited (AAL) and the Australian Government through the Commonwealth’s Education Investment Fund (EIF) and National Collaborative Research Infrastructure Strategy (NCRIS), particularly the National eResearch Collaboration Tools and Resources (NeCTAR) and the Australian National Data Service Projects (ANDS). Funding for the Sloan Digital Sky Survey IV has been provided by the Alfred P. Sloan Foundation, the U.S. Department of Energy Office of Science, and the Participating Institutions. SDSS-IV acknowledges support and resources from the Center for High Performance Computing at the University of Utah. The SDSS website is . SDSS-IV is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS Collaboration including the Brazilian Participation Group, the Carnegie Institution for Science, Carnegie Mellon University, Center for Astrophysics | Harvard & Smithsonian, the Chilean Participation Group, the French Participation Group, Instituto de Astrofísica de Canarias, The Johns Hopkins University, Kavli Institute for the Physics and Mathematics of the Universe (IPMU)/University of Tokyo, the Korean Participation Group, Lawrence Berkeley National Laboratory, Leibniz Institut für Astrophysik Potsdam (AIP), Max-Planck-Institut für Astronomie (MPIA Heidelberg), Max-Planck-Institut für Astrophysik (MPA Garching), Max-Planck-Institut für Extraterrestrische Physik (MPE), National Astronomical Observatories of China, New Mexico State University, New York University, University of Notre Dame, Observatário Nacional/MCTI, The Ohio State University, Pennsylvania State University, Shanghai Astronomical Observatory, United Kingdom Participation Group, Universidad Nacional Autónoma de México, University of Arizona, University of Colorado Boulder, University of Oxford, University of Portsmouth, University of Utah, University of Virginia, University of Washington, University of Wisconsin, Vanderbilt University, and Yale University.
Data availability statement
The DOI for the SkyMapper DR4 release is 10.25914/5M47-S621, and the SALT spectra are available through the Zenodo repository for the G4Jy Sample: https://zenodo.org/communities/g4jy/.
Funding statement
This work is based on the research supported in part by the National Research Foundation of South Africa (Grant Number 151060). The financial assistance of the South African Radio Astronomy Observatory (SARAO) towards this research is also hereby acknowledged.
Competing interests
None.
Ethical standards
The research meets all ethical guidelines, including adherence to the legal requirements of the study country.
Appendix A. Collating R-band magnitudes
We collated magnitudes for the (cross-identified) G4Jy sources, as observed in the following filters by various surveys:
• the R2 filter for SuperCOSMOSFootnote i (Hambly et al. Reference Hambly2001),
• the r filter for the National Optical Astronomy Observatory (NOAO) Source Catalog (NSC)Footnote j DR1 (Nidever et al. Reference Nidever2018),
• the r filter for the Dark Energy Spectroscopic Instrument (DESI) Legacy Imaging Surveys (LS)Footnote k DR8 (Dey et al. Reference Dey2019),
• the r filter for SDSS DR12 (Alam et al. Reference Alam2015),
• the Petrosian r filterFootnote l (Bessell et al. Reference Bessell2011) for SkyMapper (SM) DR1.1 (Wolf et al. Reference Wolf2018),
• the r filterFootnote m (Tonry et al. Reference Tonry2012) for PanSTARRS (PS; Flewelling et al. Reference Flewelling2020),
• and Visual Survey Telescope r-band photometry (Kuijken et al. Reference Kuijken2015), newly extracted using ProFound (PF; Robotham et al. Reference Robotham2018).
The greatest coverage was provided by SuperCOSMOS (i.e.
$\sim$
64% of the 586 G4Jy targets), and so this magnitude scale was chosen to be the ‘anchor’ to which the other magnitudes were ‘homogenised’. This was done by applying the following
$y = mx + c$
equations (based upon linear-regression analysis), as appropriate:
\begin{equation} r_{\mathrm{NSC}}^{R2} = (r_{\mathrm{NSC}} - c^{R2}_{\mathrm{NSC}})/m^{R2}_{\mathrm{NSC}} = (r_{\mathrm{NSC}} + 0.493)/1.022\end{equation}
\begin{equation}r_{\mathrm{LS}}^{R2} = (r_{\mathrm{LS}} - c^{R2}_{\mathrm{LS}})/m^{R2}_{\mathrm{LS}}= (r_{\mathrm{LS}} + 0.729)/1.029\end{equation}
\begin{equation}r_{\mathrm{SDSS}}^{R2} = (r_{\mathrm{SDSS}} - c^{R2}_{\mathrm{SDSS}})/m^{R2}_{\mathrm{SDSS}}= (r_{\mathrm{SDSS}} + 1.349)/1.067\end{equation}
\begin{equation}r_{\mathrm{SM}}^{R2} = (r_{\mathrm{SM}} - c^{R2}_{\mathrm{SM}})/m^{R2}_{\mathrm{SM}}= (r_{\mathrm{SM}} - 2.600)/0.845\end{equation}
\begin{equation}r_{\mathrm{PS}}^{R2} = (r_{\mathrm{PS}} - c^{R2}_{\mathrm{PS}})/m^{R2}_{\mathrm{PS}}= (r_{\mathrm{PS}} - 2.908)/0.866\end{equation}
\begin{equation}r_{\mathrm{PF}}^{R2} = (r_{\mathrm{PF}} - c^{R2}_{\mathrm{PF}})/m^{R2}_{\mathrm{PF}}= (r_{\mathrm{PF}} + 0.254)/1.001\end{equation}
Appendix B. SALT spectra and Finder Charts
Within figure B1, we present optical spectra for 42 G4Jy sources (White et al. Reference White2020a,b), including those that belong to the G4Jy-3CRE subset (Massaro et al. Reference Massaro and White2023a) and optical spectra for ‘the dancing ghosts’, G4Jy 1704 and G4Jy 1705.
The corresponding Finder Charts are shown in figure B2, for reference, by way of corroborating the host-galaxy position used for optical follow-up. Note that, for G4Jy 1511, Massaro et al. (Reference Massaro and White2023a) suggest that the AllWISE identification for the host galaxy (White et al. Reference White2020a,b) is ‘confused’ with a (much fainter) optical source nearby. For our observation, SALT is pointed at the optical source that coincides with the AllWISE position (which has ObjectID = 229802842 within SkyMapper DR4Footnote n ).
Additional spectra will be provided to the community as part of a SALT data release that focuses on G4Jy sources at
$-40^{\circ} \lt$
Dec.
$\lt-10^{\circ}$
(Sejake et al. in preparation).
SALT spectra (blue lines) of G4Jy sources (Section 3 and Appendix C). CCD chip-gaps are indicated by the blue line dropping to zero relative-flux, whilst the sky-emission spectra are represented by lighter-blue lines. (The latter is scaled to aid comparison with the target emission, and the scale factor that has been applied is noted in each legend.) The dashed, grey, vertical lines indicate the sky-emission that is used to assess the accuracy of the wavelength calibration, and the target spectrum is fitted with the appropriate template spectrum (red line; Section 3).
SALT Finder Charts, with the target at the centre. A non-zero Position Angle (PA) for the slit indicates that an alignment object was used.
SDSS spectra (blue lines) of G4Jy sources (Section 3 and 7), with sky-emission spectra represented by lighter-blue lines. (The latter is scaled to aid comparison with the target emission, and the scale factor that has been applied is noted in each legend.) The dashed, grey, vertical lines indicate prominent sky-emission, and the target spectrum is re-fitted with the appropriate template spectrum (red lines).
Appendix C. Redshifts and radio properties
Table C1 presents observed and intrinsic radio properties for 299 G4Jy sources, having collated their redshifts from SALT proposal 2020-1-MLT-008 (PI: White, this work), 6dFGS (Jones et al. Reference Jones2009), SDSS DR12 (Alam et al. Reference Alam2015), and SDSS DR16 (Ahumada et al. Reference Ahumada2020). We visually inspect all of the spectra and find that five sources have incorrect redshifts in the SDSS database, mainly on account of emission lines being misidentified. The better fittings of these spectra, and the corrected redshifts, are shown in figure B3 and summarised in Table C2.
In the case of G4Jy 148, the peak emission from the presence of Civ was mislabelled as Lyman-
$\alpha$
, whilst the prominent emission of Ciii] and Mgii went unlabelled. By re-fitting the spectrum (still with a quasar template) we determine a redshift of
$z=1.965$
, and assign an error of 0.005 to accommodate the slight mismatch of the wavelength scaling. For comparison, Osmer, Porter & Green (Reference Osmer, Porter and Green1994) report a redshift of
$z = 1.925$
, and Yao et al. (Reference Yao2019) provide a redshift of
$z=1.972$
. Meanwhile, Steidel & Sargent (Reference Steidel and Sargent1991) measure
$z = 1.9417$
from the Ciii] line and
$z = 1.9635$
from the Mgii line.
For G4Jy 176, the emission lines are correctly labelled by the SDSS data-reduction pipeline but the wavelength identifications are slightly offset. This is most likely due to resonance leading to ‘negative emission’ that affects the ‘weighting’ of the emission line’s central wavelengthFootnote
o
. Therefore, we re-fit the spectrum with a focus on the (non-resonating) Ciii] emission-line, and calculate a spectroscopic redshift of
$z=1.660\pm0.005$
. Again, the relatively large error is to account for the wavelength scaling being slightly different from that of the redshifted template. Whilst the radio emission of this source is well-documented, we cannot find another spectroscopic redshift in published results.
SDSS provided a redshift of
$z =0.24639 \pm 0.00010$
for G4Jy 679 (3C 190), which is primarily based upon strong [Oii] emission being misinterpreted as H
$\alpha$
emission. With what we believe to be the correct interpretation, supported by the Mgii showing a resonant profile, and multiple emission-line identifications, we find that the spectroscopic redshift is
$z = 1.196 \pm 0.001$
. This is in good agreement with
$z = 1.195649 \pm 0.000368$
from optical spectroscopy (Hewett & Wild Reference Hewett and Wild2010), and with
$z=1.1944 \pm 0.0012$
through Hi absorption (Grasha et al. Reference Grasha, Darling, Bolatto, Leroy and Stocke2019).
Like G4Jy 148, G4Jy 845 (3C 243) has its Civ emission mislabelled as Lyman-
$\alpha$
, and the prominent emission of Ciii] and Mgii going unidentified. Our re-fitting of the spectrum leads to a redshift (
$z=1.706\pm0.001$
) that is again in close agreement with that provided by (Hewett & Wild Reference Hewett and Wild2010,
$z=1.70756\pm0.00049$
).
Finally, G4Jy 1240 (3C 316) had originally been identified as the highest-redshift source in the G4Jy Sample, at
$z \sim 5.5$
(!). However, inspection of the SDSS spectrum showed that the [Oiii] line at rest-frame
$\lambda = 5\,007$
Å had been misidentified as a Lyman-
$\alpha$
line. Re-fitting the quasar spectrum gives a redshift of
$z = 0.581 \pm 0.001$
, which is in close agreement with previous spectroscopic redshifts available via prior work (
$z=0.5795$
; Gendre & Wall Reference Gendre and Wall2008; Smith et al. Reference Smith2010). Indeed, via the SDSS helpdesk (with thanks to Joel Brownstein) it was found that, although the lowest-
$\chi^2$
fit in SDSS’s automated pipeline (Bolton et al. Reference Bolton2012) had
$z = 5.49964$
, the third next-best fit has
$z = 0.580673$
. (For interest, we add lower-ranked SDSS redshifts to Table C2.)
Angular sizes, spectral indices, and 151-MHz flux-densities for 299 G4Jy sources, provided in the G4Jy catalogue (White et al. Reference White2020a,b). These G4Jy spectral indices (‘G4Jy_alpha’) are measured across 20 flux-densities, from 72 MHz to 231 MHz. By combining these data with redshift information from SALT (this work), 6dFGS (Jones et al. Reference Jones2009), SDSS DR12 (Alam et al. Reference Alam2015), and SDSS DR16 (Ahumada et al. Reference Ahumada2020), we calculate the radio luminosities and linear sizes of these sources. [The ‘(re-fitted)’ label applies to five sources where we corrected the redshift provided in the SDSS database (see Figure B3).] The ‘Confusion flag’ is also from the G4Jy catalogue, and indicates (via ‘1’) whether the radio emission of the G4Jy source may be blended with another (unrelated) radio source. An inequality (‘<’) in the angular-size column indicates that this value should be treated as an upper limit. See (White et al. Reference White2020a) for further details.
SDSS spectroscopic redshifts for five G4Jy sources, with new redshifts presented as a result of re-fitting the target spectrum (Appendix C and figure B3). The ‘SDSS best-fit’ values are where the reduced-
$\chi^2$
metric is the global minimum (per source). Each of the sources listed are quasars.
Open access
