Is there a redshift cutoff for submillimetre galaxies?
aa r X i v : . [ a s t r o - ph ] J u l Mon. Not. R. Astron. Soc. , 1–11 (Submitted) Printed 22 October 2018 (MN L A TEX style file v2.2)
Is there a redshift cutoff for submillimetre galaxies?
G. Raymond ⋆ , S. A. Eales , S. Dye , R. Carlberg and M. Sullivan Cardiff University, School of Physics & Astronomy, Queens Buildings, The Parade, Cardiff, CF24 3AA, U.K. Department of Astronomy and Astrophysics, University of Toronto, Toronto, ON M5S 3H4, Canada Department of Physics (Astrophysics), University of Oxford, Denys Wilkinson Building, Keble Road, Oxford, OX1 3RH, U.K.
ABSTRACT
We present new optical and infrared photometry for a statistically complete sample ofseven 1.1 mm selected sources with accurate coordinates. We determine photometricredshifts for four of the seven sources of 4.64, 4.54, 1.49 and 0.18. Of the other threesources two are undetected at optical wavelengths down to the limits of very deepSubaru and Canada-France-Hawaii Telescope images ( ∼
27 mag AB, i band) and theremaining source is obscured by a bright nearby galaxy. The sources with the highestredshifts are at higher redshifts than all but one of the ∼
200 sources taken from thelargest recent 850 µ m surveys, which may indicate that 1.1 mm surveys are moreefficient at finding sources at very high redshifts than 850 µ m surveys.We investigate the evolution of the number density with redshift of our sampleusing a banded V e /V a analysis and find no evidence for a redshift cutoff, although thenumber of sources is very small. We also perform the same analysis on a statisticallycomplete sample of 38 galaxies selected at 850 µ m from the GOODS-N field and findevidence for a drop-off in the number density beyond z ∼ Key words: galaxies:distances and redshifts, galaxies:evolution, galaxies:high-redshift, submillimetre, infrared:galaxies, galaxies:statistics
Submillimetre (submm) galaxies (SMGs), first detected at850 µ m with the Submillimetre Common User Bolome-ter Array (SCUBA) (Holland et al. 1999), are a signifi-cant population of high redshift star forming galaxies (e.g.,Hughes et al. 1998; Blain et al. 2002). They are believed tobe dust enshrouded galaxies undergoing prodigious levels ofstar formation (e.g., Hughes et al. 1998; Eales et al. 1999)in which the optical/UV radiation emitted by the stars isabsorbed by the dust and then re-emitted in the submm.Star formation rates in excess of 1000 M ⊙ yr − have beeninferred (Scott et al. 2002), much higher than locally. Thegalaxies in these samples have been found to account forup to one tenth of the total far-infrared/submm energeticbackground (e.g., Dye et al. 2007) and many authors haveargued that these galaxies are the progenitors for the el-liptical galaxies we see in the local Universe (Eales et al.1999; Scott et al. 2002; Dunne, Eales & Edmunds 2003).Thus understanding the nature of these sources is of great ⋆ E-mail: [email protected] importance for the understanding of galaxy formation andevolution as a whole.Observations of SMGs at ∼ c (cid:13) Submitted RAS
G. Raymond et al. has been useful for both identifying the counterparts andestimating redshifts. Due to the low surface density of ra-dio sources on the sky, the probability of the radio counter-part being coincidental with the submm source by chanceis small. Due to the high positional accuracy of radio obser-vations, it is then possible to identify the optical counter-part and retrieve a spectroscopic redshift. It is also possibleto estimate the redshift using the ratio of radio to submmflux (e.g., Hughes et al. 1998; Carilli & Yun 1999, 2000;Smail et al. 2000).Chapman et al. (2005), using the Low ResolutionImaging Spectrograph (LRIS) (Oke et al. 1995) on theKeck I telescope, managed to obtain spectroscopic redshiftsfor a total of 73 radio-identified SMGs with a median 850 µ m flux of 5.3 mJy. The galaxies in this sample were foundto lie at a median redshift of z = 2 . z max = 3 .
6. However, the K-correction which allows us todetect high-redshift SMGs does not similarly benefit theirradio fluxes and so radio identified SMGs are subjected toa radio selection effect which limits redshifts to z < ∼ µ m selected SMGs thathas close to 100% redshifts. The sample consists of 35 galax-ies, 21 with secure optical counterparts and 12 with tentativeoptical counterparts, and its completeness means that unlikeprevious surveys it is not biased towards low-z sources. Themedian redshift determined for this sample is z ∼ .
2. Us-ing this sample, Wall, Pope & Scott (2008) examined theepoch dependency of the number density of SMGs. Theyfound an apparent redshift cutoff at z > µ m to 1200 µ m fluxes compared to that expected from a low redshiftgalaxy. One possible explanation is that these sources areat very high redshifts. If this is true, then observations at1.1 mm would be better at detecting SMGs at the highestredshifts than observations at 850 µ m. A new complete sam-ple of 1.1 mm selected SMGs located in the COSMOS field(Scoville et al. 2007) has been compiled by Younger et al.(2007). The sources were selected initially at 1.1 mm withthe AzTEC camera (Scott et al. 2008; Wilson et al. 2008)on the JCMT. The resultant catalogue consists of 44 sourceswith S/N > . σ , 10 of which are robust with S/N > σ .Follow up observations by Younger et al. (2007) were thenmade with the Submillimetre Array (SMA) at 890 µ m forthe 7 highest significance AzTEC sources, allowing their po- sitions to be determined with an accuracy of ∼ . HyperZ photometric redshiftpackage (Bolzonella, Miralles & Pell´o 2000). Throughoutthis work we employ a concordance cosmological model withΩ total = 1, Ω m = 0 .
3, Ω Λ = 0 . H = 75 kms − Mpc − .All magnitudes quoted are AB. We searched for optical counterparts and measured new pho-tometry using deep Subaru , CFHT and IRAC images ofthe AzTEC sources. The IRAC and Subaru images are thepublicly available COSMOS images taken by the COSMOSteam (Scoville et al. 2007). The CFHT images are takenfrom the CFHT Deep Legacy Survey. The images we usedwere taken using the CFHT g M , r M , i M , z M , Subaru B j , V j ,r+, i+, z+ and IRAC channel 1 and 2 filters to average 3 σ depths of approximately 28.4, 27.9, 27.6, 26.5, 29.0, 28.2,28.3, 27.7, 26.4, 24.1 and 23.6 mags respectively.We searched the i-band images (figure 2) at the SMA co-ordinates. We find bright i-band counterparts for AzTEC1,3 and 7, all of which were previously known. We also finda faint i-band counterpart for AzTEC5 at the SMA coordi-nates. We find no objects directly at the SMA coordinatesfor AzTEC2, but there is a bright object offset from thisposition by 3”, meaning that the magnitude limits of thisSMG are not useful.We find no optical counterparts directly at the SMA co-ordinates for AzTEC4 and 6 in the Subaru and CFHT imag-ing. For the latter source, however, there is a bright i-bandcounterpart offset from the SMA position by ∼ ∼ σ )which could be AzTEC6’s counterpart or the true counter-part may be too faint to see. There is also a faint i-bandsource, offset from AzTEC4’s SMA position by ∼ ∼ σ ),in the Subaru imaging. For these two sources we added thei band CFHT and Subaru images, inversely weighting theimages by the square of the noise, in order to try and detectany very faint possible counterparts. The stacked i-band im-ages for AzTEC4 and 6 are shown in figure 2. We still do notfind counterparts at the SMA positions for AzTEC4 and 6 An additional uncertainty of 0.3 mags in the Subaru B j bandmagnitudes is taken into account in this photometry due to thepossibility of a red leak or a shift in the blue cutoff of this filter.c (cid:13) Submitted RAS, MNRAS , 1–11 s there a redshift cutoff for submillimetre galaxies? and given the good coincidence between the SMA and op-tical positions for the other AzTEC sources we tentativelyconclude that the true counterparts have not yet been de-tected.The typical full-width half-maximum (FWHM) of theoptical and IRAC channel 1 and 2 point spread functions(PSFs) are ∼ - J095942.86+022938.2 - AzTEC1 is thebrightest submm source in the sample with fluxes of F µm = 15 . ± . F . mm = 10 . ± . ± AzTEC2 - J100008.05+022612.2 - AzTEC2 is detectedin the submm with fluxes F µm = (12 . ± .
0) mJy and F . mm = (9 . ± .
3) mJy. No objects are found directlyat the SMA coordinates, but there is a bright object offsetfrom SMA position by 3”. Thus the limit on the magnitudeof the optical counterpart is not very useful.
AzTEC3 - J100020.70+023520.5 - AzTEC3 is detectedin the submm with fluxes F µm = 8 . ± . F . mm = 7 . ± . ∼ µ m image is ∼ µ m emissionis associated only with the central object. Optical fluxeswere measured using a aperture of diameter 1.26” and thesource is detected in the Subaru i+ band at 26.18 ± AzTEC4 - J095931.72+023044.0 - AzTEC4 is detectedin the submm with fluxes F µm = 14 . ± . F . mm = 6 . ± . ∼ σ ),in the Subaru image with a magnitude of 27.43 ± ± AzTEC5 - J100019.75+023204.4 - AzTEC5 is detectedin the submm with fluxes F µm = 9 . ± . F . mm = 7 . ± . ± AzTEC6 - J100006.50+023837.7 - AzTEC6 is detectedin the submm with fluxes F µm = 8 . ± . F . mm = 7 . ± . ∼ ∼ σ ). This could therefore be the opticalcounterpart, or the true counterpart may be too faint todetect. The source offset from the SMA position has aSubaru i+ magnitude of 25.38 ± c (cid:13) Submitted RAS, MNRAS000
AzTEC3 - J100020.70+023520.5 - AzTEC3 is detectedin the submm with fluxes F µm = 8 . ± . F . mm = 7 . ± . ∼ µ m image is ∼ µ m emissionis associated only with the central object. Optical fluxeswere measured using a aperture of diameter 1.26” and thesource is detected in the Subaru i+ band at 26.18 ± AzTEC4 - J095931.72+023044.0 - AzTEC4 is detectedin the submm with fluxes F µm = 14 . ± . F . mm = 6 . ± . ∼ σ ),in the Subaru image with a magnitude of 27.43 ± ± AzTEC5 - J100019.75+023204.4 - AzTEC5 is detectedin the submm with fluxes F µm = 9 . ± . F . mm = 7 . ± . ± AzTEC6 - J100006.50+023837.7 - AzTEC6 is detectedin the submm with fluxes F µm = 8 . ± . F . mm = 7 . ± . ∼ ∼ σ ). This could therefore be the opticalcounterpart, or the true counterpart may be too faint todetect. The source offset from the SMA position has aSubaru i+ magnitude of 25.38 ± c (cid:13) Submitted RAS, MNRAS000 , 1–11
G. Raymond et al.
Figure 1.
Subaru i+ band cutouts for AzTEC1 to 7, with the exception of AzTEC4 and 6, which are the combined CFHT and Subarui-band images. Each image has a field of view of 15.3”x15.3” and a scale of 0.15”/pixel. The SMA coordinates of each source arehighlighted by cross-hairs and the optical counterpart (including the objects offset from AzTEC4’ and 6s SMA coordinates by ∼ ∼ ∼ other. IRAC magnitudes were measured using an apertureof diameter 5.88”. A correction of +0.13 mag is appliedto the IRAC magnitudes in channels 1 and 2. Because ofthe good agreement between the SMA the optical positionsfor the other AzTEC sources we tentatively conclude thatthis is not the true counterpart, although we do estimate aphotometric redshift for it. AzTEC7 - J100018.06+024830.5 - AzTEC7 is detectedin the submm with fluxes F µm = 12 . ± . F . mm = 8 . ± . ± c (cid:13) Submitted RAS, MNRAS , 1–11 s there a redshift cutoff for submillimetre galaxies? Figure 1 – continued Photometric redshifts were determined by apply-ing the photometric redshift package,
HyperZ (Bolzonella, Miralles & Pell´o 2000), to our 11 bandphotometry (Subaru:B, V, r+, i+, z+; CFHT: g M , r M , i M , z M ; IRAC: 3.6 µ m, 4.5 µ m). The spectra used forfitting in this work are taken from the set compiled byDye et al. (2008), which is optimized for the determinationof photometric redshifts when including filters in thenear/mid-infrared. Dye et al. (2008) compared the photo-metric redshifts determined using these spectral templateswith those determined using synthetic spectra constructedfrom the best-fit star formation history for their sample of60 SCUBA sources. Since these methods are completelyindependent and the redshifts found using both sets of templates were found to be in good agreement, we assumethat our template set is adequate.We varied the redshift in the range z = 0 to 10. We em-ployed the reddening regime of Calzetti et al. (2000), with A V allowed to vary in the range A V = 0 to 5 in steps of0.2. We used a minimum photometric error of 0.05 magni-tudes for each band. For wavebands in which we have nodetection we took the flux of the source to be zero witha 1 σ error equal to the sensitivity of the detector in thatwaveband. The photometric redshifts obtained are listed intable 2.The median redshift of the sample is 2.7 which is some-what higher than the median redshift, 2.2, of the sample pre-sented by Chapman et al. (2005). The maximum redshiftfound is 4.64 and the minimum redshift found is 0.18. Com-paring the redshift distribution of this sample to that of thesamples presented in Chapman et al. (2005), Pope et al. c (cid:13) Submitted RAS, MNRAS , 1–11
G. Raymond et al.
AzTEC1 AzTEC2 AzTEC3 AzTEC4 AzTEC5 AzTEC6 AzTEC 7RA 09:59:42.86 10:00:08.05 10:00:20.70 09:59:31.72 10:00:19.75 10:00:06.50 10:00:18.06Dec +02:29:38.2 +02:26:12.2 +02:35:20.5 +02:30:44.0 +02:32:04.4 +02:38:37.7 +02:48:30.5Optical Ap. Size 1.94” ... 1.26” 2.57” 1.68” ...(1.59”) 2.87” m B > > > ± > ± ± m V ± ± > ± > ± ± m r + ± ± > ± > ± ± m i + ± ± ± ± > ± ± m z + ± ± > ± > ± ± g M > > > > > ± r M ± ± > ± > ± i M ± ± > ± > ± z M ± ± > ± > ± m . µm ± ± ± ± ± ± m . µm ± ± ± ± ± ± Table 1.
Photometry for the AzTEC sources, given in AB magnitudes. First two rows give the SMA co-ordinates. Aperture sizes arethe diameters used for measuring optical and IRAC magnitudes. The IRAC magnitudes are corrected to take into account the differencein the seeing and aperture sizes for the IRAC and optical imaging (see text). No optical counterparts were found for AzTEC2. The onlynearby optical counterparts for AzTEC4 and 6 are offset from their SMA positions by ∼ σ and ∼ σ respectively. We give the photometryfor these objects in parentheses.ID z χ min NotesAzTEC1 4 . ± .
06 1.537 ...AzTEC2 ... ... No optical counterpart.AzTEC3 4 . ± .
10 2.196 There is a secondary chi-squared minimum at the lowerredshift of z ∼ . ∼ . ± .
10 1.488 There is a secondary chi-squared minimum at the higherredshift of z ∼ ∼ . ± .
01) (6.172) The redshift and chi-squared values are for the optical sourceoffset from AzTEC6’s SMA position. The chi-squared fit tothis source is much poorer compared to the others in the sample.This may further imply that the nearby optical counterpart we haveselected is not the true counterpart to AzTEC6 and that the IRACemission is unassociated with the optical emission.AzTEC7 0 . ± .
01 7.021 CFHT data not available. There are several other possible redshiftswith chi-squared fit values of ∼
10 up to z ∼
2. Even the bestchi-squared fit is still relatively poor however, which may be dueto the unusual nature of the source.
Table 2.
The best photometric redshift fits for the sources with their minimum reduced χ value, χ min . Notes of interest on thephotometric redshifts, including secondary fits, for each source are also given. We do not give the best fit SED type as typically for eachsource there are several SED types which fit equally well. Note that reduced chi-squared values given here are not those directly outputby HyperZ , which takes the number of degrees of freedom as being the (number of filters − A v , SEDs type and the normalization to vary. Thus the correct number of degrees offreedom is given by (number of filters − (2006), Dye et al. (2008) and Clements et al. (2008), wenote that only one of the sources in this combined sample of ∼
200 850 µ m selected sources is at a comparably high red-shift as our two highest redshift sources, although this dif-ference is not significant when analyzed with a Kolmogorov-Smirnov test. However two of the other AzTEC sources areundetected to very faint limits in the i-band, and these factsmay indicate that 1.1 mm surveys find more sources at veryhigh redshifts than 850 µ m surveys. V E / V A ANALYSIS
Wall, Pope & Scott (2008) examined a sample of 38 SMGs in the GOODS-N field and found evidence for a diminutionin the space density of SMGs at redshifts z >
3. They alsofound evidence for two separately evolving sub-populationsseparated by luminosity. In this paper we present the resultsof our re-examination of this result using a banded V e /V a analysis and a range of empirical SEDs rather than the the-oretical SED used by Wall et al.The most well known method of investigating the evo-lution of the space density of galaxies with redshift is the h V /V max i test (Schmidt 1968; Rowan-Robinson 1968). V is the co-moving volume enclosed by the galaxy (that vol-ume which the field of view traces out in moving from aredshift of z = 0 out to the galaxy) and V max is the vol-ume that would be enclosed by the galaxy were it pushed c (cid:13) Submitted RAS, MNRAS , 1–11 s there a redshift cutoff for submillimetre galaxies? Figure 2.
The left hand column shows the photometric data points for the AzTEC sources with optical counterparts. The best spectralfits for the sources are overlaid. The right hand column shows the marginalized reduced χ distribution as a function of redshift. TheAzTEC6 plots correspond to the nearby optically bright object. to the redshift at which its flux drops to the survey limit.This method encounters problems when a survey enclosestwo galaxy populations, one undergoing positive evolution,and the other negative. If we have a uniform distribution ofgalaxies in space, then we expect the value of h V /V max i tobe 0.5 ± (12 N ) − . , where N is the number of sources in thesample. A value of h V /V max i > . h V /V max i < . h V /V max i may still beclose to 0.5, incorrectly implying zero evolution. This problem can be solved by implementing instead a h V e /V a i test (Dunlop & Peacock 1990). This is effectivelya banded version of the h V /V max i test. V e , the effective vol-ume, is the volume enclosed between a minimum redshift z low and the redshift of the galaxy. V a , the accessible vol-ume, is the volume enclosed between z low and the redshiftat which the galaxy’s flux drops below the sensitivity of thesurvey. By investigating the variation of h V e /V a i with z low we can distinguish between a positively evolving and a neg-atively evolving population.We investigated the evolution of the space density ofthe sample with redshift through the implementation of a h V e /V a i test. Wall et al. based the k-correction necessary c (cid:13) Submitted RAS, MNRAS , 1–11
G. Raymond et al.
Figure 2 – continued to calculate accessible volume on a single theoretical SED,whereas real galaxies have a range of SEDs. To investigatethis, we carried out the h V e /V a i analysis using two differentassumptions about SEDs. We used the two extreme two-component dust models of Dunne & Eales (2001), who pro-vided fits to the hottest and coldest local SMGs. The coldSED, based on NGC 958, contains dust at temperatures of20 and 44 K with a cold-to-hot dust mass ratio of 186:1. Thehot SED, based on IR1525+36, contains dust at tempera-tures of 19 and 45 K with a cold-to-hot dust mass ratio of15:1. Figure 3, which shows the predicted flux versus redshiftplot for the different models, shows the effect of using dif-ferent SED templates on the flux-redshift relation. The twoSED types are normalized such that they produce a flux of1 mJy at a redshift of z = 1.We took the limiting flux of each source in the GOODSsample to be 3 . σ and measured h V e /V a i for z low = 0 to 4 insteps of 0.1. We also separated sources into two samples ofequal size according to luminosity. In doing this we are ableto determine whether there are differences in the evolutionof the two sub-populations.Our results for the 38 SMGs of Wall, Pope & Scottare shown in figure 4. We find good evidence for the exis-tence of a redshift cutoff at z > z > Figure 3.
Flux versus redshift for both the cold (solid line) andhot (dashed line) SEDs. Both SEDs are normalized such that theyproduce a flux of 1 mJy at a redshift of z = 1. is far more marginal. Thus we find evidence to support theconclusions given in Wall, Pope & Scott (2008): there is aredshift cutoff for the sample and that there is evidence fortwo separately evolving sub-populationsAn additional uncertainty about this results is thatPope et al. (2006) claim that only 60% of their identifica-tions are reliable. Therefore we also performed the h V e /V a i analysis only on sources with reliable identifications, the re-sults of which are shown in figure 5. Using these sources only,we still find good evidence for a redshift cutoff at z > c (cid:13) Submitted RAS, MNRAS , 1–11 s there a redshift cutoff for submillimetre galaxies? Figure 4.
The distribution of the values of h V e /V a i with z low for the GOODS-N sample. Figures in the left hand column are for hotSEDs and figures in the right hand column are for cold SEDs. The sample is also separated into high and low luminosity sources. Thedashed line denotes the position of h V e /V a i = 0 . σ error. z > z = 4and repeating the analysis (figure 6). Doing this, we findthat for hot SEDs our results are largely unaffected, with arelatively clear cutoff at redshifts higher than z = 1. How-ever for the cold SEDs we find that our results are stronglyaffected, with no clear redshift cutoff up to a redshift of z ∼ c (cid:13) Submitted RAS, MNRAS , 1–11 G. Raymond et al.
Figure 5.
The distribution of the values of h V e /V a i with z low for the GOODS-N sample, using only the sources with reliable identifica-tions. Figures in the left hand column are for hot SEDs and figures in the right hand column are for cold SEDs. The dashed line denotesthe position of h V e /V a i = 0 . σ error. Figure 6.
The distribution of the values of h V e /V a i with z low for the GOODS-N sample, where four (roughly half)of the unreliableidentifications have been pushed to redshifts of z = 4. Figures in the left hand column are for hot SEDs and figures in the right handcolumn are for cold SEDs. The dashed line denotes the position of h V e /V a i = 0 . σ error. We also performed a banded h V e /V a i analysis on oursample of AzTEC sources (excluding the AzTEC6 counter-part) the results of which are shown in figure 7, but oursample is too small to find any clear evidence of a redshiftcutoff. We give new Subaru, CFHT and IRAC photometry for anumber of sources in the AzTEC / COSMOS survey withaccurate coordinates from SMA imaging. We have estimatedphotometric redshifts for four of the seven galaxies in thesample. We find a median redshift of z mean ∼ .
57 anda maximum of z max = 4 .
50. Of the sources in the com-bined 850 µ m surveys presented in Chapman et al. (2005),Pope et al. (2006), Dye et al. (2008) and Clements et al. (2008), consisting of ∼
200 sources, only one is at a redshiftgreater than our two highest redshift sources. This in ad-dition to the fact that we are unable to detect two of oursources in the optical bands down to very faint magnitudesmay indicate that 1.1 mm surveys are more efficient at de-tecting very high-redshift sources than 850 µ m surveys.Re-investigating the space density evolution of asample of 38 GOODS-N sources (Pope et al. 2006;Wall, Pope & Scott 2008) with more realistic SEDs we finda redshift cutoff at z ∼ z ∼ c (cid:13) Submitted RAS, MNRAS , 1–11 s there a redshift cutoff for submillimetre galaxies? Figure 7.
The distribution of the values of h V e /V a i with z low for the AzTEC sample. The panels on the left hand side uses the hot SEDand the panels on the right hand side uses the cold SED. The dashed line denotes the position of h V e /V a i = 0 . σ error. We performed a similar test on the AzTEC sources butwere unable to draw any reliable conclusions as the sampleis too small. The GOODS-N sample is also relatively small,and therefore any evidence for redshift cutoffs and differentlyevolving sub-populations must be treated with caution. Inorder to harden our conclusions in general we require largersurveys with accurate redshifts. We would also need surveystaken over larger areas of sky in order to take into accountthe effects of cosmic variance. Future, larger surveys (e.g.with Herschel, SCUBA2) therefore will enable us to morerobustly determine the nature of the number density evolu-tion of SMGs in the Universe.
Acknowledgements
G. Raymond, S. Eales and S. Dye acknowledge support fromthe Science and Technologies Facilities Council.Based on observations obtained withMegaPrime/MegaCam, a joint project of CFHT andCEA/DAPNIA, at the Canada-France-Hawaii Telescope(CFHT) which is operated by the National ResearchCouncil (NRC) of Canada, the Institut National des Sciencede l’Univers of the Centre National de la Recherche Sci-entifique (CNRS) of France, and the University of Hawaii.This work is based in part on data products produced atTERAPIX and the Canadian Astronomy Data Centre aspart of the Canada-France-Hawaii Telescope Legacy Survey,a collaborative project of NRC and CNRS.
REFERENCES
Blain, A. W., Smail, I., Ivison, R. J., Kneib, J.-P., Frayer,D. T., 2002, PhR, 369, 111Bolzonella, M., Miralles, J.-M., Pell´o, R., 2000, A&A, 363,476Calzetti, D, Armus, L., Bohlin, R. C., Kinney, A. L., Koorn-neef, J., Storchi-Bergmann, T., 2000, ApJ, 533, 682Carilli, C. L., Yun, M. S., 1999, ApJ, 513, L13Carilli, C. L., Yun, M. S., 2000, ApJ, 530, 618 Chapman, S. C.,Blain, A. W., Smail, I., Ivison, R. J., 2005,ApJ, 622, 772Clements et al., 2008, MNRAS, 387, 247Dunlop, J. S., Peacock, J. A., 1990, MNRAS, 247, 19-+Dunne, L., Eales, S. A., 2001, MNRAS, 327, 697Dunne, L., Eales, S. A., Edmunds, M. G., 2003, MNRAS,341, 589Dye, S., Eales, S. A., Ashby, M. L. N., Huang, J.-S., Egami,E., Brodwin, M., Lilly, S., Webb, T., 2007, MNRAS, 375,725Dye, S., et al., 2008, MNRAS, 386, 1107Eales, S. A, Lilly, S., Gear, W., Dunne, L., Bond, J. R.,Hammer, F., Le F`evre, O., Crampton, D., 1999, ApJ, 515,518Eales, S., Bertoldi, F., Ivison, R., Carilli, C., Dunne, L.,Owen, F., 2003, MNRAS, 344, 169Holland, W. S., et al., 1999, MNRAS, 303, 659Hughes, D. H., et al., 1998, Nature, 394, 241Ivison, R. J.,Smail, I., Le Borgne, J.-F., Blain, A. W.,Kneib, J.-P., Bezecourt, J., Kerr, T. H., Davies, J. K.,1998, MNRAS, 298, 583Koekemoer et al., 2007, ApJ supplemental, 172, 196Morgan, H. L., Edmunds, M. G., 2003, MNRAS, 343, 427Oke, J. B., et al., 1995, PASP, 107, 375-+Pope et al., 2006, MNRAS, 370, 1185Priddey, R. S., McMahon, R. G., 2001, MNRAS, 342, L17Rowan-Robinson, M., 1968, MNRAS, 138, 445-+Schmidt, M., 1968, ApJ, 151, 393-+Scott, S. E., 2002, MNRAS, 331, 817Scott, K. S. et al., 2008, MNRAS, 385, 2225Scoville, N., et al., 2007, ApJ supplemental, 172, 150Smail, I., Ivison, T. J., Owen, F. N., Blain, A. W., 2000,ApJ, 528, 612Vlahakis, C., Eales, S., Dunne, L., 2007, MNRAS, 379, 1042Wall, J. V., Pope, A., Scott, D., 2008, MNRAS, 383, 435Younger, J. D., et al., 2007, ApJ, 671, 1531Wilson, G. W., et al., 2008, MNRAS, 386, 807 c (cid:13) Submitted RAS, MNRAS000