MNRAS 433, 2706–2726 (2013) doi:10.1093/mnras/stt835 Advance Access publication 2013 June 27 Lyman break and ultraviolet-selected galaxies at z∼1 – I. Stellar populations from the ALHAMBRA survey I. Oteo,1,2‹´ A. Bongiovanni,1,2 J. Cepa,1,2 A. M. P´ erez-Garc´ ıa,1,2,3 A. Ederoclite,4 M. S´ anchez-Portal,3,5 I. Pintos-Castro,1,2,15 R. P´ erez-Mart´ ınez,6J. Polednikova,1,2 J. A. L. Aguerri,6E. J. Alfaro,7T. Aparicio-Villegas,7,16 N. Ben´ ıtez,7T. Broadhurst,8 J. Cabrera-Ca˜ no,9F. J. Castander,10 M. Cervi˜ no,7D. Cristobal-Hornillos,7,4 A. Fernandez-Soto,11,17 R. M. Gonzalez-Delgado,7C. Husillos,7L. Infante,12 V. J. Mart´ ınez,13,14 I. M´ arquez,7J. Masegosa,7I. Matute,7M. Moles,7,4 A. Molino,7 A. del Olmo,7J. Perea,7M. Povi´ c,7F. Prada,7J. M. Quintana7and K. Viironen4 1Instituto de Astrof´ ısica de Canarias (IAC), E-38200 La Laguna, Tenerife, Spain 2Departamento de Astrof´ ısica, Universidad de La Laguna (ULL), E-38205 La Laguna, Tenerife, Spain 3Asociaci´ on ASPID. Apartado de Correos 412, La Laguna, Tenerife, Spain 4Centro de Estudios de F´ ısica del Cosmos de Arag´ on, Plaza San Juan 1, Planta 2, E-44001 Teruel, Spain 5Herschel Science Centre (ESAC), Villafranca del Castillo, E-28692 Madrid, Spain 6XMM/Newton Science Operations Centre (ESAC), Villafranca del Castillo, Spain 7Instituto de Astrof´ ısica de Andaluc´ ıa (CSIC), Glorieta de la Astronom´ ıa s/n, E-18008 Granada, Spain 8School of Physics and Astronomy, Tel Aviv University, 69978 Tel Aviv, Israel 9Facultad de F´ ısica, Departamento de F´ ısica At´ omica, Molecular y Nuclear, Universidad de Sevilla, Sevilla, Spain 10Institut de Ciencies de lEspai, IEEC-CSIC, Barcelona, Spain 11Instituto de F´ ısica de Cantabria (CSIC-UC), E39005, Santander, Spain 12Departamento de Astronom´ ıa, Ponticia Universidad Catolica, 7820436 Macul, Santiago, Chile 13Departament d’ Astronom´ ıa i Astrof´ ısica, Universitat de Valencia, Valencia, Spain 14Observatori Astronomic de la Universitat de Valencia, Valencia, Spain 15Centro de Astrobiolog´ ıa, INTA-CSIC, PO Box – Apdo. de correos 78, Villanueva de la Ca˜ nada, E-28691 Madrid, Spain 16Observatrio Nacional-MCT, Rua Jos Cristino, 77, CEP 20921-400, Rio de Janeiro-RJ, Brazil 17Unidad Asociada Observatorio Astronmico (Universitat de Valncia / IFCA-CSIC), Parc Cientfic UV, E-46980 Paterna, Spain Accepted 2013 May 9. Received 2013 May 9; in original form 2012 November 16 ABSTRACT We take advantage of the exceptional photometric coverage provided by the combination of GALEX data in the ultraviolet (UV) and the ALHAMBRA survey in the optical and near-infrared to analyse the physical properties of a sample of 1225 GALEX-selected Lyman break galaxies (LBGs) at 0.8z1.2 that are located in the COSMOS field. This is the largest sample of LBGs studied in this redshift range to date. According to a spectral energy distribution (SED) fitting with synthetic stellar population templates, we find that LBGs at z∼1 are mostly young galaxies with a median age of 341Myr and have intermediate dust attenuation, Es(B−V)∼0.20. Owing to the selection criterion, LBGs at z∼1 are UV-bright galaxies and have a high dust-corrected total star formation rate (SFR), with a median value of 16.9 Myr−1. Their median stellar mass is log (M∗/M)=9.74. We find that the dustcorrected total SFR of LBGs increases with stellar mass and that the specific SFR is lower for more massive galaxies (downsizing scenario). Only 2 per cent of the galaxies selected through the Lyman break criterion have an active galactic nucleus nature. LBGs at z∼1 are located mostly over the blue cloud of the colour–magnitude diagram of galaxies at their redshift, with only the oldest and/or the dustiest deviating towards the green valley and red E-mail: [email protected] C 2013 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

Lyman break and UV-selected galaxies at z∼1 2707 sequence. Morphologically, 69 per cent of LBGs are disc-like galaxies, with the fractions of interacting, compact, or irregular systems being much lower, below 12 per cent. LBGs have a median effective radius of 2.5 kpc, and larger galaxies have a higher total SFR and stellar mass. Compared with their high-redshift analogues, we find evidence that LBGs at lower redshifts are larger, redder in the UV continuum, and have a major presence of older stellar populations in their SEDs. However, we do not find significant differences in the distributions of stellar mass or dust attenuation. Key words: galaxies: evolution – galaxies: photometry–galaxies: high-redshift– galaxies: star formation – cosmology: observations – ultraviolet: galaxies. 1 INTRODUCTION Much effort has been devoted over the last decades to searching for high-redshift star-forming (SF) galaxies. Different selection criteria select distinct kinds of galaxies. Among these criteria, the most successful and most commonly used are the Lyman-alpha and the Lyman break techniques, which pick up the so-called Lyman-alpha emitters (LAEs) and Lyman break galaxies (LBGs), respectively. The Lyman-alpha technique is based on looking for a Lyman-alpha emissionin theredshiftedopticalspectrum ofgalaxiesbyemploying a combination of narrowand broad-band filters. Specifically, the narrow-band filter is used to isolate the Lyαline, and the broadband one(s) to constrain its nearby continuum (Cowie & Hu 1998; Gronwall et al. 2007; Gawiser et al. 2006; Ouchi et al. 2008, 2010; Bongiovanni et al. 2010; Shioya et al. 2009). The choice of the central wavelength of the narrow-band filter determines the redshift of the selected LAEs, which have been searched for, found, and analysed from z∼2.0 up to z∼7 and beyond (Gawiser et al. 2006; Murayama et al. 2007; Nilsson et al. 2007, 2009; Pirzkal et al. 2007; Ouchi et al. 2008; Finkelstein et al. 2009b; Guaita et al. 2011; Hibon et al. 2011, 2012; Oteo et al. 2011, 2012a,b). LBGs are found by employing a combination of broad-band filters that sample the red-ward and blue-ward zones of the redshifted Lyman break of galaxies, located at 912Å in the rest-frame (Madau et al. 1996; Steidel et al. 1996, 2003). The choice of the red-ward and blue-ward filters determines the location in wavelength of the Lymanbreakand,consequently,theredshift of the selected galaxies. Many samples of LBGs have been found and examined at different redshifts, mostly at z3 (Madau et al. 1996; Steidel et al. 1996, 1999, 2003; Stanway, Bunker & McMahon 2003; Bunker et al. 2004; Giavalisco et al. 2004; Iwata et al. 2007; Verma et al. 2007). At z3 the number of LBGs reported and studied is much lower than that at higher redshifts (Burgarella et al. 2006, 2007; Ly et al. 2009, 2011; Basu-Zych et al. 2011; Hathi et al. 2010, 2013; Nilsson et al. 2011; Haberzettl et al. 2012; Chen et al. 2013), despite the fact that this redshift range is quite important as it is thought that the peak of the cosmic star formation of the Universe took place in that epoch. Apart from the Lyαand Lyman break techniques, various other methods have been used for selecting high-redshift galaxies in the literature. Adelberger et al. (2004) defined various colour selection criteria that employ several combinations of optical colours to select galaxies at different redshifts: GRi for sources within 0.85  z1.15, GRz for 1.0z1.5, and UnGR for 1.4z2.1and 1.9z2.7. The galaxies selected in this way have been traditionally called BM/BX galaxies. Another ground-based optical colour selection criterion is the BzK method, which aims to find galaxies in the redshift range 1.4z2.5 and to classify them as SF or passively evolving systems (Daddi et al. 2004). Both kinds of galaxies are often associated with LBGs. Haberzettl et al. (2012) found that near-ultraviolet (NUV) data provide a greater efficiency for selecting SF galaxies. Furthermore, they reported that, although the BM/MX and BzK techniques are very efficient for detecting sources within 1.0z3.0, they are biased against those SF galaxies that are more massive and contain a noticeable amount of red stellar populations. Haberzettl et al. (2012) argue that, therefore, a NUV-based LBG selection criterion is more appropriate for comparisons with the populations found at z3.0. The physical properties of high-redshift SF galaxies have been traditionallyanalysedbyfittingtheirobservedspectralenergydistributions (SEDs) built from their photometric data1to SED templates obtained from stellar population models (such as Bruzual & Charlot 2003, hereafter BC03; Lai et al. 2008; Gawiser et al. 2007; Nilsson et al. 2007, 2009; Yabe et al. 2009; Finkelstein et al. 2008, 2009a,c, 2010a; Magdis et al. 2010). This procedure, in principle, enables the determination of age, dust attenuation, star formation rate (and history), metallicity and stellar mass. This is because the SED obtained in stellar population models depends on (among others) all these parameters. In practice, however, this procedure has several limitations. For example, metallicity does not have a strong influence on the shape of the rest-frame optical SEDs and, therefore, its determination from SED-fitting suffers from large uncertainties. On the other hand, the degeneracy between dust attenuation and age and that between star-formation history (SFH) and age mean that the three parameters are difficult to constrain accurately at the same time. With a good wavelength coverage of the UV continuum and the 4000-Å Balmer break it is feasible to improve the determination of the SED-derived dust attenuation and stellar age. However, dust attenuation would still suffer from uncertainties, and the only way to obtain accurate values is by employing direct far-infrared (FIR) detections around the dust emission peak (Burgarella et al. 2011; Oteo et al. 2013a). Despite these caveats, many previous works have analysed the physical properties of LBGs at different redshifts by employing an SED-fitting method, as it is the only way to analyse their properties with large samples of galaxies. At z∼5, LBGs have been reported to be much younger (<100 Myr) and to have lower stellar masses (109M) than their analogues at z∼2.0–3.0 in a similar rest-frame UV luminosity range (Verma et al. 2007; Yabe et al. 2009; Haberzettl et al. 2012). Most previous works have focused on LBGs that are located at z3, where the Lyman break is shifted to the optical and can be sampled with filters in ground-based telescopes. At z2, the Lyman break is located in the UV, and LBGs can be found only through observations from space, for example with GALEX 1In this work we use the term SED to refer to a set of photometric points. However, it should be noted that SED is also applied to spectroscopic data in many works. Some of the limitations quoted here for the SED-fitting technique apply only to photometric SEDs, and not to spectroscopic ones. Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

2708 I. Oteo et al. (Burgarella et al. 2006, 2007; Haberzettl et al. 2012), HST and UVIS filters (Hathi et al. 2013), or with the Swift satellite (BasuZych et al. 2011). In this work, we aim to analyse the physical properties of a sample of 1225 GALEX-selected LBGs at z∼1 located in the COSMOS field by using data from the Advanced Large, Homogeneous Area Medium Band Redshift Astronomical (ALHAMBRA) survey (Moles et al. 2008), which covers the optical range with 20 medium-band filters (width about 300 Å) and the NIR with the traditional JHKs broad-band filters. The combination of the ALHAMBRA survey with observations in other wavelengths (GALEX and IRAC) allows an unprecedented coverage of the UV continuum and optical Balmer break. This can disrupt some of the degeneracies outlined above and provide more accurate results for the SED-derived physical properties. Because we study LBGs at z∼1, their observed fluxes are high enough that the photometry has a good signal-to-noise ratio. These two facts (exceptional coverage of the SED and the high signal-to-noise ratio) are not usually achieved at higher redshifts. This emphasizes the importance of studying intermediate-redshift LBGs. The paper is organized as follows. In Section 2 we present the data sets employed both in the UV with GALEX and in the opticalto-NIR with the ALHAMBRA survey. In Section 3 we combine these two data sets to build a general sample of UV-selected galaxies at 0 z2. In Section 3 we also explain how we carry out the SED fits using BC03 templates for these UV-selected sources with the aim of obtaining their photometric redshift, rest-frame UV luminosity, and other physical properties such as dust attenuation, age and stellar mass. In Section 4 we define the selection criterion adopted in this work to look for LBGs at z∼1. The SED-derived physical properties of the selected LBGs are discussed in Section 5. The morphology and physical sizes of the LBGs studied are analysed in Section 6, and in Section 7 we show their location in the colour–magnitude diagram (CMD). In Section 8 we compare the properties of our GALEX-selected LBGs with those reported in previous works for LBGs at higher redshifts. Finally, we summarize the main conclusions of the work in Section 9. Throughout this paper we assume a flat universe with (m, ,h 0)=(0.3,0.7,0.7), and all magnitudes are listed in the AB system (Oke & Gunn 1983). 2 DATA SETS On the UV side we use data coming from observations of the COSMOS field with the GALEX satellite (Martin et al. 2005) in both the NUV and far-ultraviolet (FUV) bands as part of the Deep Imaging Survey (PI: D. Schiminovich). GALEX catalogues were created by using the EM-algorithm (Guillaume et al. 2006), aimed at resolving blended objects in the farand the near-UV using the information (position and shape) available from existing, well-resolved catalogues on the visible range. In the concrete case of the COSMOS field, the prior optical photometric information corresponds to a u*-band mosaic (and its SEXTRACTOR-derived catalogue) based on CFHT-u* observations. With a list of optical prior positions, the algorithm measures their UV fluxes on the GALEX images by adjustingaGALEX pointspreadfunction model. The algorithm wasrun on the four NUV and the four FUV GALEX images covering the COSMOS field obtained as a product of the GALEX pipeline processing, version 1.61. On the optical and NIR side we use the ALHAMBRA survey (Moles et al. 2005, 2008), which employs a set of 20 equal-width (∼300-Å) medium-band filters covering the optical range from 3500 to 9700 Å plus the traditional JHKs broad-band NIR filters to observe a region of 4 square degrees distributed into eight distinct fields. Among them, we focus our work on the COSMOS field owing to the wealth of photometric and spectroscopic data in the UV, optical, NIR and other wavelengths. The observations were carried out with the 3.5-m telescope of the Calar Alto Observatory using the wide-field camera LAICA in the optical and the OMEGA-2000 camera in the NIR. The optical filter system adopted in ALHAMBRA was set with the aim of optimizing the output of the survey in terms of the zphot accuracy (Ben´ ıtez et al. 2009). The simulations performed in Ben´ ıtez et al. (2009) relating the image depth, zphot accuracy, and number of filters indicate that the filter system of ALHAMBRA enables a zphot precision, for normal SF galaxies, that is three times better than that for traditional 4–5 broad-band filter sets. In addition, the complementary usage of NIR data improves the determination of photometric redshifts. In this work we have used the catalogues coming from the Internal Data Release 3. The data reduction and the construction of the catalogues were carried out by the ALHAMBRA team. Because the ALHAMBRA survey performs observations in 23 filters, it is necessary to work with care when defining the detection of objects. With the aim of not biasing the detection to any particular kind of object as a consequence of the selection in a single band, a special technique was employed. It is based on creating a deep detection image built as the sum of the individual frames with the highest efficiencies. This includes the filters centred between 457 and 829 nm. The photometry of the sources was obtained by running SEXTRACTOR in its dual mode. The deep image is used for source detection, and the photometry is then extracted in each individual frame. As a result, the average depth (for 3σdetections) is 24.5 and 22 mag in the optical and NIR, respectively. In this work we employ the SEXTRACTOR-derived AUTO MAG as the best approximation to the total magnitude for all calculations. The characterization in the optical range of the ALHAMBRA photometric system can be found in Aparicio Villegas et al. (2010), and the NIR number counts of one of the fields are presented in Crist´ obal-Hornillos et al. (2009). Further details on the quality of the data, the reduction process, the depth, etc. will be published in Husillos et al. (in preparation). We note that the characterization of ALHAMBRA filters (complete wavelength coverage with almost no overlapping filters) provides an SED that can be considered as a low-resolution optical spectrum of the observed sources. It should be noted that the Lyman break selection that will be employed in this work is based purely on UV GALEX data, whereas the ALHAMBRA survey is used in the analysis of their SED-derived physical properties. 3GALEX AND ALHAMBRA DATA: UV-SELECTED GALAXIES AND SED FITTING The LBGs that will be studied in this work are taken from a multiwavelength catalogue of UV-selected SF galaxies that we build by combining the GALEX observations with the data coming from the ALHAMBRA survey: we look for GALEX detections around 2arcsec of the optical position of the sources in the ALHAMBRA survey and retain only those galaxies that have, at least, a detection in the NUV band. This produces a sample of 39 734 UV-selected sources for which we have photometric information from the UV to the NIR. With the aim of obtaining the photometric redshifts and the physical properties of those UV-selected sources we fit their observed fluxes to BC03 templates with the Zurich Extragalactic Bayesian Redshift Analyzer (ZEBRA, Feldmann et al. 2006) code, which, in its maximum-likelihood mode, employs a χ2minimization algorithm Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

Lyman break and UV-selected galaxies at z∼1 2709 over the templates to find the one that best fits the observed SED of each input object. We build a set of BC03 templates associated with various physical properties of galaxies by using the software GALAXEV. In this process we adopt a Salpeter (1955) initial mass function (IMF) distributing stars from 0.1 to 100Mand select a fixed value for the metallicity of Z=0.2 Z. We consider values of age from 1 Myr to 7 Gyr, in steps of 10 Myr from 1 Myr to 1 Gyr and in steps of 100Myr from 1 Gyr to 7 Gyr. Dust attenuation is included in the templates via the Calzetti et al. (2000) law and parametrized through the colour excess in the stellar continuum, Es(B−V). We select values for Es(B−V) ranging from 0 to 0.7 in steps of 0.05. We include intergalactic medium (IGM) absorption, adopting the Madau (1995) prescription. Regarding the SFR, we adopt time-constant models. In this case, different values of the SFR do not change the shape of the templates, and the SFR can be obtained using the Kennicutt (1998) calibration: SFRUV,uncorrected[Myr−1]=1.4×10−28L1500,(1) where L1500 is the rest-frame UV luminosity at 1500 Å. The L1500 is obtained for each galaxy by convolving its best-fitting template with a top-hat filter (300 Å width) centred in the 1500 Å rest-frame. It should be noted that, throughout this work, we distinguish between LUV defined in a νLνway (units of ergs−1)andL1500 considered in Lνunits (ergs−1Hz−1). The SFR derived from equation (1) is uncorrected for the attenuation that dust produces in the SEDs of galaxies. In order to obtain an estimation of the dust-corrected total SFR we have to introduce into equation (1) the dust-corrected L1500.Itis obtained from L1500 by multiplying it by the dust correction factor 100.4A1500 ,whereA1500 is the dust attenuation at 1500 Å. The values of A1500 are obtained from the SED-derived Es(B−V) assuming the Calzetti et al. (2000) law. Once both the age and dust-corrected total SFR are known for each source, and according to the assumed time-independent SFH, the stellar mass can be obtained from the product of the two quantities. In this work we also analyse the UV continuum slope, β, of our UV-selected galaxies (see for example Calzetti, Kinney & StorchiBergmann 1994). This parameter is important in the study of how galaxies form/grow, as it is related to age, metallicity, stellar IMF, and, most importantly, dust attenuation. Furthermore, UV colours seem to be related to the UV luminosities of SF galaxies (Bouwens et al. 2009, 2010a, 2011) and are easier to measure than optical restframe colours in high-redshift galaxies, for which IRAC detections would be mandatory. The combination of the GALEX photometry and the bluest optical bands of the ALHAMBRA survey provides a good sampling of the UV continuum at z∼1, the redshift range in which our LBGs are located, as can be seen in the SED fits shown in Fig. 1. Different works employ different methods to obtain the UV continuum slope of galaxies in different redshift ranges, the most popular and traditionally used being that in which βis quantified by using two broad-band filters that sample two zones of the observed UV continuum (Meurer et al. 1997; Kong et al. 2004; Hathi, Malhotra & Rhoads 2008; Overzier et al. 2008; Bouwens et al. 2010b; Finkelstein et al. 2010b; Dunlop et al. 2012). In other works, βis obtained by using a power-law fit to the observed fluxes of the studied galaxies, using filters that sample the same zone of the SED at different redshifts (Bouwens et al. 2012). In our work, we obtain βfor each galaxy by fitting the UV continuum of its best-fitting template with a power law in the form fλ∼λβ (Calzetti et al. 1994). In this process we employ the rest-frame wavelength range 1300 Å λ3000 Å. This range contains all the windows defined in Calzetti et al. (1994) in their definition of the UV continuum slope. This approach has the advantage of using all the available fluxes of each source, resulting in more robust signal-to-noise ratio determinations. The method employed here is similar to that used by Finkelstein et al. (2012) in their study of the redshift evolution of the UV continuum slope from z∼8toz∼ 4. In that work, they present some illustrative examples showing the differences in the UV continuum slope when using the different techniques. In an SED-fitting procedure, the reliability of the results, that is, the similarity between the observed SED and that represented by its best-fitting template, is related to the χ2value of the fits. Here we define the reduced χ2,χ2 rof each best-fitting template as the ratio between its χ2and the number of filters minus one employed in the fit, χ2 r=χ2/(N−1) (see for example de Barros, Schaerer & Stark 2012). From a visual inspection of the SED-fitting results, we consider that the BC03 templates truly represent the observed SED for each galaxy when χ2 r<10 (for some examples of χ2values and the quality of the fittings see Fig. 1). Imposing χ2 r<10 to the fittings of the whole sample of 39 754 UV-selected galaxies, we end up with a robust sample of 35 810 galaxies. From now on, only sources with χ2 r<10 are considered. The χ2 rvalues depend on the observed fluxes and also on their photometric uncertainties. In this way, if a galaxy has a photometry with high photometric errors, the χ2 rmight be low even when its best-fitting template does not represent its observed SED properly. Therefore, a low value of χ2 rcan be due either to a good SED fit or to a bad SED fit with a photometry with high uncertainties. Thus, we should check the typical photometric errors of the ALHAMBRA photometry of our sources to analyse whether the low values of χ2 rare due to truly accurate fits or are the consequence of high photometric errors. As an example, we show in the left panel of Fig. 2 the photometric errors in the ALHAMBRA filter centred at 613 nm of our sample of UV-selected sources for 0≤z≤2 as a function of their observed magnitude in the same band. As expected, the photometric errors increase with the observed magnitude. If we consider that a fit is reliable for galaxies with typical photometric errors below 0.4 mag, we can trust only those SED-fitting results for galaxies typically brighter than about 25 mag. In the right panel of Fig. 2 we represent the observed magnitudes of the GALEX-selected LBGs that will be studied in this work. It can be seen that most of them meet the previous criterion, and therefore we can consider that the low values of χ2 rare statistically attributable to good SED fits rather than to high photometric errors. It should be noted that in this work we employ BC03 templates associated with a time-independent SFR. Other kinds of SFHs can be used, such as those that are exponentially declining or composed of bursts of star formation. In the first case, the SFR is characterized by the decaying time-scale, τSFR, which would be another parameter to obtain in the SED fitting, increasing the degrees of freedom in the process. Distinguishing between different kinds of SFHs is very challenging, even with a good photometric coverage of the SED of galaxies. Therefore, the results reported in this work should be understoodasthosederivedwiththatchoiceoftheSFR,butdifferent values of the SED-derived parameters might be obtained if other temporal variations of the SFH were assumed. The analysis of the differences in the SED-fitting results depending on the assumption of the SFH will be studied in Section 5.3. 3.1 Photometric redshifts and their accuracy The good coverage of the observed UV-to-NIR SEDs of galaxies provided by the ALHAMBRA survey in combination with Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

2710 I. Oteo et al. Figure 1. Examples of SED-fitting results with GALEX+ALHAMBRA data for nine GALEX-selected LBGs randomly selected from the whole sample. The blue points are the observed GALEX and ALHAMBRA fluxes, and the black curves are the best-fitting Bruzual & Charlot (2003) (BC03) templates of each object. The BC03 templates were build by assuming a constant SFR, a Salpeter IMF, and a fixed metallicity of Z=0.2Z. In each panel we indicate the SED-derived redshift, age, dust attenuation, and reduced χ2associated with each best-fitting template. It can be seen that the combination of GALEX and ALHAMBRA data provides a very good sampling of the rest-frame UV continuum and the 4000-Å Balmer break of our UV-selected galaxies. GALEX data is expected to give accurate determination of photometric redshifts (zphot) at the expected redshift range of GALEXselected LBGs, namely z∼1. This is because at z∼1, GALEX+ALHAMBRA data cover the rest-frame UV continuum and the Balmer break, which are two of the most important features to fit in the SED of galaxies for determining photometric redshifts. In Fig. 3 we compare the obtained zphot with spectroscopic redshifts (zspec) for those sources in the whole sample of 35 810 UV-selected galaxies with χ2 r<10 that have available spectra from the zCOSMOS survey (Lilly et al. 2007). Here we define the accuracy of zphot as σz =z/(1 +zspec), with z =|zphot − zspec|. See also Matute et al. (2012) for a discussion of the photometric redshift accuracy of the ALHAMBRA survey. It can be seen in Fig. 3 that within the redshift range 0.8z1.2there is a good agreement between the photometric and spectroscopic redshifts, namely σz less than 0.05 for most galaxies (see the horizontal dashed lines). It should be noted that this accuracy applies only to galaxies that are as bright as the sources in the spectroscopic survey. All the galaxies with spectroscopic redshift from the zCOSMOS survey shown in Fig. 3 have r-band observed magnitudes typically brighter than 23.5 mag, and therefore the reliability of the photometric redshifts can be guaranteed up to that limit. Hereafter, photometric redshifts are used for the sources without a zCOSMOS counterpart. For those sources with a zCOSMOS spectrum, we redo the SED fits and employ the results based on zspec. Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

Lyman break and UV-selected galaxies at z∼1 2711 Figure 2. Left: Photometric errors against the observed magnitude in the ALHAMBRA filter centred at 613 nm for our sample of UV-selected galaxies with ALHAMBRA measurements and χ2 r<10 in the SED-fitting results. The red curve corresponds to the median value of the error distribution for each value of the observed magnitude. Horizontal lines represent photometric errors of 0.1 and 0.2 mag. Right: Distribution of the observed magnitude in the ALHAMBRA filter centred at 613 nm for LBGs and a sample of UV-selected galaxies with ALHAMBRA counterparts at the same redshift range as LBGs. Histograms have been normalized to their maxima in order to clarify the representation. Figure 3. Accuracy of the photometric redshift determination with the combination of GALEX and ALHAMBRA data for the whole sample of 35 810 UV-selected galaxies with ALHAMBRA measurements and χ2 r<10 in the SED-fitting results. In these plots, only galaxies with available spectroscopic redshift from the zCOSMOS survey (Lilly et al. 2007) are considered. Red dots represent our GALEX-selected LBGs, and grey open triangles are the remaining galaxies in the sample. In the left panel, the vertical and horizontal dashed straight lines represent the photometric redshift locus where most GALEX-selected LBGs are expected to be located according to their UV colour selection, 0.8z1.2. In the right panel, the vertical dashed straight lines represent the photometric redshift locus where most GALEX-selected LBGs are expected to be located according to their UV colour selection. The horizontal dashed straight lines indicate the values of the photometric redshift accuracy, σz =z/(1 +zspec), equal to ±0.05. The horizontal continuous straight line represents where spectroscopic and photometric redshifts would agree. 4 UV-SELECTED GALAXIES AT z∼1 As noted in Section 1, the choice of the red-ward and blue-ward filters determines the wavelength at which the Lyman break is located and, therefore, the redshifts of the selected galaxies. In this work we aim to analyse the physical properties of those LBGs whose Lyman break is located between the GALEX FUV and NUV filters, which are centred at 1528 and 2271 Å (in terms of their effective wavelengths) and have bandwidths of 1344–1786 and 1771–2831Å, respectively. These wavelengths imply that the redshifts of these GALEX-selected LBGs are expected to be around z∼0.95 considering an intermediate wavelength of 1780Å between the two filters. In order to formulate an analytic selection criterion to segregate our LBGs we use a large set of BC03 templates associated with a metallicity of Z=0.2Z, a constant SFR, and various values of age and dust attenuation. We study the location of those templates in a colour–colour diagram as a function of redshift. To do that, each Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

2712 I. Oteo et al. Figure 4. Colour selection of the GALEX-selected LBGs studied in this work. Left: synthetic FUV–NUV colour tracks as a function of redshift according to a set of Bruzual & Charlot (2003) (BC03) stellar population templates associated with different values of age and dust attenuation. The horizontal dashed red line indicates the colour cut employed in this work, which, according to the tracks shown in black, is expected to select galaxies at z≥0.7 (vertical red dashed line). Right: Transmission curves of the FUV (orange) and NUV (red) GALEX filters. We also show the location of the Lyman break at z=0.8 with a vertical dashed black line and the BC03 stellar population templates associated with two different values of age and dust attenuation (black curves). It can be seen that, although the colour selection shown in the left panel is expected to segregate galaxies at z≥0.7, the Lyman break does not almost completely pass the FUV channel until z∼0.8. template is redshifted from z=0uptoz=2, and then we apply the corresponding absorption in the intergalactic medium following the Madau et al. (1996) prescription, and obtain the FUV −NUV synthetic observed colours by convolving the templates with the transmission curves of the GALEX filters. The results are shown in Fig. 4. As a general trend and as could be expected by the location of the Lyman break as a function of redshift, the FUV −NUV colour increases with redshift up to z∼1. Looking at the different tracks represented in Fig. 4, we decided to impose a colour cut of 1.5 (red dashed horizontal line), and therefore our colour selection for GALEX-selected LBGs is FUV −NUV > 1.5.(2) It is important to note that the application of this criterion requires the detection of each galaxy in both the FUV and the NUV channel. The left panel of Fig. 4 indicates that in imposing such a colour selection criterion we segregate galaxies located at z≥0.7 (this threshold is represented by the red dashed vertical line). At 0.7 ≤ z≤0.8, however, the FUV flux is strongly affected by the photons of the Lyman continuum (Lyc); that is, by those UV photons whose wavelengths are lower than the wavelength of the Lyman break. Therefore, if we really want to sample the Lyman break between the FUV and NUV without a significant contamination of Lyc photons in the FUV filter we have to limit the redshift of the galaxies to z≥ 0.8. This situation is schematized in the right panel of Fig. 4. Thus, we define, as a first approximation, GALEX-selected LBGs as those galaxies that are detected in both the FUV and the NUV channel and whose fluxes in each band satisfy equation (2) and are located at z≥0.8. This sample comprises 475 galaxies. It is worth noting that there is a difference between the selection criterion that we apply here and those applied to look for highredshift LBGs. At z2, LBGs are usually found by employing not only the difference in colour that characterizes the Lyman break (equation 2) but also a difference in colour at redder wavelengths (see for example Steidel et al. 2003; Madau etal. 1996). This is done in order to rule out lower-redshift interlopers. This is important because at high redshifts the photometric redshifts might suffer from large uncertainties, and then it is not always possible to select galaxies in a specific redshift range based on zphot. However, in our case, as noted in Section 3.1 and shown in Fig. 3, we have accurate values of the photometric redshift for our UV-selected galaxies at z∼1, and therefore that supplementary condition is not needed. Barger, Cowie & Wang (2008) selected LBGs at 0.6 ≤z≤1.4 by employing a double colour selection criterion combining FUV − NUV and NUV −U. If we limit our sample in NUV magnitude to their limit, NUV <23.75, all but one of our GALEX-selected LBGs satisfy the double colour selection criterion of Barger et al. (2008). This is schematized in Fig. 5. The U-band data for the galaxies in the panel have been taken from the broad-band photometric catalogue in the COSMOS field (Capak et al. 2007). Conversely, if we trust the photometric redshifts obtained from the combination of GALEX and ALHAMBRA data we find that with the selection criterion of Barger et al. (2008) we would miss a population of GALEXselected LBGs fainter than NUV =23.5 mag whose NUV −U colour is typically redder than those of GALEX-selected LBGs with NUV <23.5 mag. Theapplicationofequation(2)requires the detection of the galaxies in both the FUV and the NUV GALEX channels so that the amplitude of the break can be measured. However, it is possible that a galaxy has such a strong Lyman break that is is detected in the NUV but undetected in the FUV channel. In order to include these FUV-undetected galaxies in the sample of GALEX-selected LBGs we have to ensure that the non-detection in FUV is caused by a strong Lyman break. FUV observations in the COSMOS field have a limiting magnitude of FUV ∼26.5mag. Galaxies brighter than that value in the wavelength range covered by the FUV filters Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

Lyman break and UV-selected galaxies at z∼1 2713 Figure 5. Location of our GALEX-selected LBGs in a colour–colour diagram. The window enclosed by the orange solid lines is the selection region for LBGs at 0.6 <z<1.4 with NUV <23.75. The subsample of our GALEX-selected LBGs that satisfy NUV <23.75 are represented with red filled dots, while the remaining fainter LGBs are plotted with black symbols. should have been detected. Because we select LBGs with a FUV − NUV colour cut of 1.5, that limiting magnitude would imply a limit of 25 mag in the NUV channel. In this way, we include in the previous sample of LBGs those galaxies that are brighter than 25 mag in the NUV channel, are at 0.8 ≤z≤1.2, and are undetected in the FUV channel. This method for selecting FUV-undetected LBGs has also been applied in, for example, Burgarella et al. (2007). With these galaxies included, we end up with an initial sample of 1246 GALEX-selected LBGs. A visual inspection of the galaxies with available ACS information (see Section 6 for more details) reveals that the contamination due to the low spatial resolution of the GALEX images is lower than 5 per cent. In addition, from this visual inspection we check that there is no stellar contamination in the derived LBG sample. 4.1 X-ray counterparts and AGN contamination In this work we are interested only in those LBGs that are SF galaxies, and we therefore need to rule out the active galactic nucleus (AGN) contribution. With this aim, we look for Chandra X-ray detections (Elvis et al. 2009) around 3 arcsec (Povi´ cetal. 2009, 2012) of the ALHAMBRA-based spatial coordinates of our GALEX-selected LBGs. The area where our LBGs are located is almost totally covered by the Chandra footprint. We find that only 21 LBGs are detected in the X-ray and, therefore, probably have an AGN nature. When using the catalogue of AGNs in the COSMOS field (Salvato et al. 2011) we do not find any extra AGN identification. The only AGN-LBGs represent an AGN contamination of about 2 per cent. Fig. 6 represents (black histogram) the distribution of FUV − NUV colours of the galaxies in the whole sample of UV-selected sources with measurements in the FUV and NUV channels that are detected in the X-ray, have GALEX and ALHAMBRA counterparts, Figure 6. Distribution of the FUV −NUV colour for X-ray-detected galaxies at 0.8 ≤z≤1.2 with GALEX and ALHAMBRA counterparts (black histogram). The red histogram represents the distribution of the FUV − NUV colour of the galaxies at 0.8 ≤z≤1.2 spectroscopically classified as AGNs in Cowie, Barger & Hu (2010) via emission-line diagnosis. The vertical dashed line indicates the colour threshold for selecting LBGs in this work. In this plot, only galaxies with detection in both the FUV and the NUV channel are included. and are at 0.8 ≤zphot ≤1.2 (the redshift range within which our GALEX-selected LBGs are located). It can be seen that most X-raydetected galaxies, and therefore galaxies with an AGN nature, have FUV −NUV coloursbelow the colour threshold utilized in this work for selecting LBGs (see equation 2). This UV colour distribution for AGNs at z∼1 explains the low percentage of AGNs among the GALEX-selected LBGs. We also plot in Fig. 6 the FUV −NUV colour distribution of the galaxies spectroscopically classified as AGNs via emission-line diagnosis in Cowie et al. (2010) and which are located in the same redshift range as our GALEX-selected LBGs. It can be seen again that most AGNs at z∼1haveFUV −NUV colours below the colour threshold considered in this work for selecting LBGs, reinforcing the fact that the LBG colour selection does not tend to segregate galaxies with an AGN nature. Low values of the AGN contribution in samples of LBGs at different redshifts have also been reported. Lehmer et al. (2005) found AGN fractions of 1.2, 0.4, 0.3 and 0.4 per cent in their sample of U-, B435-, V606-andi775-dropouts, respectively. Basu-Zych et al. (2011) reported an AGN fraction for their sample of LBGs at 0.5 < z<2.0 of 5–6percent, and Nandra et al. (2002) found an AGN contribution of about 3 per cent in their sample of LBGs at z∼3. In the subsequent analysis we do not take into consideration the GALEX-selected LBGs with an AGN nature. We thus end up with a sample of 1225 SF GALEX-selected LBGs. This is the largest sample of LBGs studied at z∼1 to date. Fig. 1 shows the UVto-NIR SEDs of nine of the GALEX-selected LBGs in our final sample. This small subsample is representative of the whole sample of LBGs. It can clearly be seen that the combination of GALEX and ALHAMBRA provides an excellent coverage of the rest-frame UV continuum, Balmer break, and NIR SEDs of these galaxies. Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

2714 I. Oteo et al. Figure 7. Distribution of the NUV (blue histogram), ALH-706 (green histogram), and Ks (red histogram) apparent magnitudes for our GALEXselected LBGs at z∼1. Fig. 7 represents the distributions of the apparent brightness of our GALEX-selected LBGs in the NUV channel, optical ALH-706 ALHAMBRA filter, and in the NIR Ks band. These distributions should be taken into account when comparing the results for our LBGs with those reported in other published studies that employ different photometric information. Our GALEX-selected LBGs at z∼1 have NUV magnitudes around 24.5–25.0 mag and optical magnitudes typically between 22 and 24.5 mag. The median value of their Ks magnitude is 22 mag. Furthermore, it can be seen that the spread in the magnitudes increases with the central wavelength of the filters. Whereas the NUV magnitudes are distributed mostly within a range of 1 mag width, the Ks-band magnitude spans from 20 to 24 mag. 4.2 High-redshift analogues In this work, we also aim to compare the SED-derived physical properties of LBGs at different redshifts. LBGs at high redshift (i.e. z>3) tend to be intrinsically more luminous than those selected in the present work owing to an observational bias. If we want to compare LBGs at different redshifts, and therefore galaxies that are selected with similar selection criteria, we must limit the rest-frame UV luminosity of our GALEX-selected LBGs at z∼1 to the same range as that for LBGs at z>3, which is typically log (LUV/L)≥ 10.2. In this sense, we define UV-bright LBGs as those LBGs at 0.8z1.2 that have log (LUV/L)≥10.2. This subsample is formed by 181 galaxies. 4.3 UV-faint galaxies At the redshift range of our GALEX-selected LBGs there are many SF galaxies that are not selected through the dropout technique either because they do not have a strong break between the FUV and NUV filters or because they are undetected in the FUV channel and are not bright enough in the NUV filter to ensure a strong Lyman break between the two filters. All these galaxies will be termed UVfaint galaxies. This sample will not be studied in the present work but it will be used in a forthcoming work (Oteo et al. 2013b) in which FIR observations will be used to constrain the FIR SED of both GALEX-selected LBGs and UV-faint galaxies. In that case, the comparison between FIR-detected LBGs and UV-faint galaxies will help us to understand the galaxies that are selected under the dropout technique in opposition to other UV-fainter SF galaxies and to place LBGs in a more general scenario of SF galaxies at 0.8z1.2. 5 SED-DERIVED STELLAR POPULATIONS 5.1 Physical properties of LBGs at z∼1 Fig. 8 shows with red shaded histograms the distributions of photometric/spectroscopic redshift, rest-frame UV luminosity, age, dust attenuation, dust-corrected total SFR, stellar mass, and UV continuum slope for our GALEX-selected LBGs. As a consequence of their selection criterion, our GALEX-selected LBGs are located at 0.8z1.2 and have rest-frame UV luminosities log(LUV/L)>9.6. The median values of the SED-derived physical properties of our GALEX-selected LBGs are summarized in Table 1. It can be seen that they are blue and young galaxies with moderate dust attenuation. Owing to their brightness in the rest-frame UV, they have relatively high values of the UV-derived and dust-corrected total SFRs. In its maximum-likelihood mode, ZEBRA gives not only the bestfitting templates but also the probability that any of the other nonbest-fitting templates can represent the photometric SED of a given galaxy. This probability can be used for deriving the uncertainties of the SED-derived parameters. With this aim, we define the weighted average (WA) of a given SED-derived physical property as WA = N iPifi/N,wherePiis the probability that a given template, i, can represent the observed SED of a given galaxy, fiis the value of one of the physical properties associated with the ith template, and Nis the number of templates. If the best-fitting template of a given galaxy has a much higher probability of representing its observed SED than any of the other templates, the average WA of a given physical property would be quite similar to the best-fitting one and the uncertainty of that property should be low. On the other hand, if several templates associated with very different values of a given property have a similar probability of representing the observed SED of a given galaxy, the WA would be dissimilar to the best-fitting value, and the uncertainty should be high. Following this idea, we definetheuncertainty of a givenparameter asthe differencebetween the best-fitting value and its corresponding weighted average. The physical parameters intrinsically related to the BC03 templates considered in this work are the age and the dust attenuation. The UV-derived dust-uncorrected SFR is obtained from the normalization of each observed template to the observed SED, and the stellar mass is obtained from the values of age, dust attenuation, and dust-uncorrected SFR. Thus, the procedure outlined above for deriving the uncertainties should be first applied to age and dust attenuation. As a result, we obtain that the median values of the uncertainties of age and dust attenuation for our LBGs are Age = 390 Myr and Es(B−V)=0.05, respectively. The typical uncertainty of the SED-derived age is of the same order as the median age of our GALEX-selected LBGs. This implies that, even with the exceptional photometric coverage of the ALHAMBRA survey, which samples quite well the Balmer break of SF galaxies at z∼ 1, the age is a parameter difficult to determine accurately with an Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

Lyman break and UV-selected galaxies at z∼1 2721 Figure 15. Distribution of the effective radii (left) and the S´ ersic indices (right) for our GALEX-selected LBGs with available ACS images. Histograms have been normalized to their maxima in order to clarify the representations. make this definition true, the dependent variable κis coupled to n (Penget al.2010).Foreachinputgalaxy, GALFIT providesitseffective radius(in pixels)and the S´ ersicindex,alongwith their uncertainties. In order to convert the effective radius in pixels into the physical size in kiloparsecs we employ the ACS pixel scale and the assumed cosmology for calculating the arcsec/pixel at the redshift of each galaxy. Fig. 15 shows the distribution of the effective radius and the S´ ersic index for our GALEX-selected LBGs. The median effective radius for our LBGs is 2.48 kpc. The values of the S´ ersic indices for LBGs are compatible with most of them being disc-like galaxies. This is in agreement with the results of the visual morphological analysis. Shown in Fig. 16 are the relations between the physical sizes of our GALEX-selected LBGs and their dust-corrected total SFR and stellar mass. It can be seen that there is a trend between the physical size and both SFR and stellar mass: larger galaxies tend to form stars faster, and to have higher stellar masses. The correlation between the total SFR and the effective radius exists whatever the dust attenuation method employed. The size–stellar mass relation has also been reported to occur in LBGs at a similar and higher redshift ranges. Mosleh et al. (2012) found that the stellar mass– size relation for LBGs persists up to z∼5. 7 COLOUR–MAGNITUDE DIAGRAM An important tool with which to analyse the properties of our GALEX-selected LBGs is their location in a colour–magnitude diagram (CMD). Traditionally, this kind of diagram has been used to separate local galaxies between non-SF galaxies earlier than the Sa morphological type and SF galaxies later than Sb in morphological type. In a colour space, the former tend to populate a red sequence and the latter are located in the so-called blue cloud (Strateva et al. 2001; Hogg et al. 2002). This behaviour translates into a bimodal distribution of the colour of galaxies, which allows one both to study the nature of different samples of galaxies by looking at their position in the colour space and to look for galaxies with different SF natures, by imposing conditions on their location in such a diagram. Furthermore, this bimodality in the local Universe has been proved to apply at higher redshifts, at least up to z∼1.6 (Blanton et al. 2003; Bell et al. 2004; Weiner et al. 2005; Cirasuolo et al. 2007; Franzetti et al. 2007; Williams et al. 2009; Taylor et al. 2009; Nicol et al. 2011). Following this idea, we plot in Fig. 17 the location of our GALEX-selected LBGs in a CMD associated with the magnitudes in the uand rbroad-band filters of the SDSS survey. The apparent uand rand absolute rmagnitudes are obtained by convolving the best-fitting template of each galaxy with the transmission of the uand rSDSS filters shifted in wavelength according to the redshift of each source. Along with those points we also represent a sample of SDSS local galaxies taken from the DR7 (Abazajian et al. 2009) and a sample of zphot-selected galaxies at z∼ 1 taken from the ALHAMBRA survey. The sample of local galaxies Figure 16. Left: Effective radius against the dust-corrected total SFR for our GALEX-selected LBGs at z∼1 with available ACS images. As indicated in the legend, we include points associated with the total SFR obtained with the SED-derived dust attenuation (red symbols) and with the UV continuum slope and the application of the Overzier et al. (2011) (orange symbols) and Takeuchi et al. (2012) (purple symbols) laws. Black, dark grey, and light grey squares represent the median values of the dust-corrected total SFR obtained with the SED-derived dust attenuation, the Overzier et al. (2011), and the Takeuchi et al. (2012) laws, respectively, for each considered bin of effective radius. Right: Effective radius against the SED-derived stellar mass for our GALEX-selected LBGs at z∼1 (red dots) with available ACS images. Grey squares represent the median values of the SED-derived stellar mass for each considered bin of effective radius. The grey straight line represents a linear fit to the grey squares. In both plots, the blue error bars represent the typical uncertainty in the determinations of the SED-derived dust-corrected total SFR and stellar mass (see Section 5.1) and effective radius. Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

2722 I. Oteo et al. Figure 17. Left: Locus of our GALEX-selected LBGs (red dots) in a colour–magnitude diagram (CMD). For comparison, we represent with blue dots the locations in such a diagram of a sample of local galaxies selected from the SDSS survey. Furthermore, we represent with black contours the typical CMD for galaxies at z∼1 obtained from a general population of galaxies at that redshift taken from the ALHAMBRA survey. These contours show the location of the blue cloud, green valley, and red sequence at that redshift and clarify the discussions given in the text. Right: As for the left panel, but segregating old-LBGs (brown) and dusty-LBGs (orange). Old-LBGs are those LBGs older than 1200Myr, while dusty-LBGs are those LBGs whose dust attenuation is higher than Es(B−V)=0.4. comprises all the galaxies in the SDSS whose spectroscopic redshifts are below 0.035. In this case, the magnitudes plotted are those that we extract from the photometric catalogues of the SDSS survey. At such low redshifts there is no need for K-correction. In order to build the sample of galaxies at z∼1 we select all the galaxies in the ALHAMBRA survey (in all the already observed fields) whose photometric redshifts are around that value and whose observed Ks-band magnitudes are brighter than 22 mag, similar to the limits employed in Williams et al. (2009) and Taylor et al. (2009). By using the optical and NIR photometry of the ALHAMBRA survey we fit their SEDs with BC03 templates and obtain their u−rcolours and absolute r-band magnitudes in the same way as for LBGs. It can be seen in the left panel of Fig. 17 that the bimodality that is seen in the local Universe is also clearly present at z∼1. This result also indicates the power of the ALHAMBRA survey in characterizing the CMD of galaxies at different redshifts. The majority of our GALEX-selected LBGs are located in the blue cloud of galaxies at their redshift, indicating that these kinds of galaxies are blue and active SF galaxies, as expected from their selection criteria in the UV. A minority of LBGs are shifted towards the red sequence or are located between the blue cloud and the red sequence, the so-called green valley. This position does not necessarily indicate that these galaxies are non-SF. Actually, it can be the case that these galaxies have redder optical colours either because of a significant amount of dust that is attenuating their bluest emitted light, and/or because there is an important contribution of old stellar populations in their SEDs, being more evolved systems. To clarify this issue we show in the right panel of Fig. 17 the position of our GALEX-selected LBGs in the CMD as a function of SED-derived dust attenuation and age. We arbitrarily consider two subclasses within the LBGs: those with age 1200 Myr (old-LBGs) and those with age 1200 Myr and Es(B−V)0.4 (dusty-LBGs). It can be clearly seen from Fig. 17 that those LBGs that are located over the green valley or near the red sequence are old-LBGs and dustyLBGs, whereas those LBGs whose ages are younger than 1200Myr and have low/intermediate [Es(B−V)<0.4] dust attenuation are located over the blue cloud. 8 COMPARISON WITH HIGH-REDSHIFT LBGS Inthis section we analysethe differences/similaritiesbetween LBGs at z∼1andz∼3 in order to study whether the Lyman break selection criterion selects different kinds of galaxies at different redshifts. As noted in Section 4.2, high-redshift LBGs tend to be intrinsically brighter than those studied in this work as a consequence of the use of magnitude-limited observations. If we want to compare galaxies that are selected at different redshifts with a similar selection criterion we must limit the rest-frame UV luminosities of the samples to the same range. We use the rest-frame UV luminosities, as LBGs at any redshift are selected in the rest-frame UV. This sample was defined in Section 4.2 as UV-bright LBGs and it comprises 65 galaxies. According to their SEDs, UV-bright LBGs are less dusty, have a higher SFR, are more massive, and have a bluer UV continuum slope than the whole population of LBGs at z∼1. However, there is no significant difference in their ages, which are mostly below 500 Myr for both populations. Fig. 18 shows the distributions of the SED-derived age, dust attenuation, and stellar mass for our UV-bright LBGs (orange histograms) and for a sample of LBGs at z∼3 studied in Papovich, Dickinson & Ferguson (2001) (green histograms). UV-bright LBGs at z∼1 have ages distributed mainly between 1 and 400 Myr with a median value of 171 Myr, whereas LBGs at z∼3 are younger galaxies with a median age of 36Myr. The difference in the median values is of the same order as the uncertainties of the SED-derived age at z∼1. However, as can be seen in the histogram shown in Fig. 18, at z∼1 there is a presence of older stellar populations (with ages mainly between 150 and 400 Myr) than at z∼3. A Kolmogorov–Smirnov test applied to both histograms gives a very low probability that they represent similar distributions. This could indicate that the galaxies selected with the Lyman break criterion at Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

Lyman break and UV-selected galaxies at z∼1 2723 Figure 18. DistributionoftheSED-derivedage,dust attenuation, andstellar mass for our UV-bright LBGs (orange histograms) and high-redshift LBGs (green histograms) taken from Papovich et al. (2001). Histograms have been normalized to their maxima in order to clarify the representations. z∼1 are at a later evolutionary stage than those at z∼3. It should be noted that, as pointed out in many previous works and in Section 5, the uncertainties of the SED-derived stellar age are usually high and, furthermore, the age evolution found can be the consequence of diverse factors: (i) using different SFHs in the analysis of the SED of the galaxies; (ii) employing photometric information with different wavelength coverages; (iii) the degeneracy between dust attenuation, metallicity, and age, etc. Therefore, the previous evidence of an evolution of the age of LBGs with redshift is not conclusive and should be confirmed in further studies with a more detailed study of the rest-frame UV-to-NIR SED of these galaxies. The dust attenuation distribution of LBGs at z∼1seemstocontain lower values than the distribution at high redshift, although both have median values of Es(B−V)=0.25. The typical uncertainty of the SED-derived dust attenuation in our work is Es(B−V)= 0.1 (see Section 5). This value along with the similar median values of the distributions at z∼1andz∼3 prevent us from constraining any evolution of the median values of dust attenuation of LBGs with redshift. This can be an effect of the procedure employed. An SED-fitting technique is not precise enough to constrain an evolution of dust attenuation with redshift, and other techniques should be employed. The direct measurement of dust emission of LBGs in the FIR could give clues for addressing this issue. Regarding stellar masses, it can be seen that the distributions at z∼1andz∼3 span a similar range. The median values of the stellar mass of our UV-bright LBGs and LBGs at z∼3are log(M∗/M)=10.0 and log (M∗/M)=9.7, respectively. This difference is similar to the typical uncertainties of the stellar mass determinations performed in this work, and therefore we cannot constrain any evolution in the stellar mass of LBGs with redshift either. The median value of the stellar mass found in the present work is between those reported in Magdis et al. (2010) for IRAC8µm detected LBGs and IRAC-8 µm faint LBGs at z∼3, namely logM∗/M=11 and log M∗/M=9, respectively. Regarding the UV continuum slope, UV-bright LBGs have a median value of β=−1.44. This value is larger (redder) than those reported at higher redshifts (Lehnert & Bremer 2003; Bouwens et al. 2006; Hathi et al. 2008; Bouwens et al. 2009; Wilkins et al. 2011; Bouwens et al. 2011; Castellano et al. 2012), indicating that LBGs at lower redshifts tend to be redder in the UV continuum than those at higher redshifts. The UV continuum slope is not a parameter associated directly with the BC03 templates, but it is obtained once the best-fitting BC03 template for each galaxy is known. Consequently, this parameter is more insensitive to the various SFHs adopted for building the BC03 template, and therefore is a good and accurate indicator of the evolution of LBGs with redshift. Mosleh et al. (2011) studied the redshift evolution of the physical sizes of samples of LBGs and other UV/submm-selected galaxies at differentredshiftsandfoundthattheirsizeincreaseswithdecreasing redshift. Mosleh et al. (2012) studied the redshift evolution of LBGs from z∼1uptoz∼7 and found that the median size of LBGs at a given stellar mass increases towards lower redshifts. The median size of our sample of UV-bright LBGs is 2.92 kpc. Mosleh et al. (2012) studied the redshift evolution of LBGs considering galaxies in two bins of stellar mass: 8.6 <log (M∗/M)<9.5 and 9.5 < log (M∗/M)<10.4. As indicated above, the median value of the stellar mass of our UV-bright LBGs is log(M∗/M)=10.0. The median value of the size of UV-bright LBGs is slightly lower than those presented in Mosleh et al. (2012) and Mosleh et al. (2011) for the corresponding stellar mass range. This small difference is probably because, although these authors also work with GALEXselected LBGs, they consider galaxies located within 0.6 <z< 1.4, whereas we limit the redshift of our sample to z>0.8. The inclusion of galaxies at lower redshifts might increase the median value of the size, explaining the difference found between the two works. It is important to remark again that there is a fundamental difference in the selection criteria for LBGs at different redshifts. Given that the combination of GALEX and ALHAMBRA (including IRAC for a subsample) provides very accurate photometric redshifts for our galaxies, we do not need any extra condition to rule out interlopers. However, this is not the case at high redshifts. At z2, the photometric redshifts of the sources are not accurate enough to ensure a proper cleaning of the sample from interlopers. As a consequence, additional criteria should be applied. These extra criteria usually involve limits in the observed optical colours of the samples. For example, as discussed in Madau et al. (1996) for their sample of F300W dropouts, the application of extra criteria rules out interlopers that are in another redshift range, but they also missed galaxies at the proper redshift at the same time. The missing galaxies tend to be redder than those included in the final sample, either because they are older or because they are attenuated by dust. Therefore, it is clear that the additional criteria employed at high redshift discard of certain kinds of subclasses of galaxies at each redshift. In contrast, in our work, as in Burgarella et al. (2006, 2007), we do not apply any extra observed optical colour criteria and therefore we include in the sample all kinds of galaxies that have a break between the FUV and NUV filters, regardless of their age, dust attenuation or optical colour. All previous results seem to indicate that LBGs at z∼1tend to be older, to be bluer in their UV continuum, and to have larger sizes than those at higher redshifts. Therefore, SED fitting and morphological studies indicate that LBGs at lower redshifts are at a later evolutionary stage than those at higher redshifts. It should be noted that these differences between LBGs at different redshifts were found by comparing the results obtained in this work with those from previous studies performed by other authors. Therefore, this comparison might suffer from uncertainties arising from the Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

2724 I. Oteo et al. different methods employed in each work or from slightly different selection criteria, for example the use of BC03 templates associated with different SFHs or different photometric coverage of the SEDs of galaxies at different redshifts. Thus, in order to properly characterize the evolution of LBGs with redshift a more precise work should be undertaken in which the photometric SEDs of the galaxies are sampled over the same rest-frame wavelength range, and the selection criteria and the procedures employed for the analysis of the physical properties of the galaxies are as similar as possible. 9 CONCLUSIONS In this work we have analysed the physical properties derived from the SED-fitting of a sample of 1225 GALEX-selected LBGs at 0.8z1.2 by using a combination of UV and optical/NIR data coming from GALEX observations and the ALHAMBRA survey, respectively. ALHAMBRA uses a set of 20 medium-band (width ∼300Å) optical and the three classical NIR JHKs filters to cover the observed optical SED of galaxies in an unprecedented way. This provides a good sampling of both the UV continuum slope and the Balmer break, increasing significantly the accuracy of the results of the SED-fitting technique. We defined LBGs as those galaxies that have a difference of colour greater than 1.5 mag between the FUV and NUV filters of the GALEX satellite. Our main conclusions are as follows. (i) According to the SED-fitting with BC03 templates built assuming a constant SFR, Salpeter IMF, and metallicity Z=0.2 Z, GALEX-selected LBGs at z∼1 are young galaxies with ages mostly younger than 300 Myr, with a median dust attenuation of Es(B−V) of 0.20, and a median UV continuum slope of −1.53. As a consequence of the selection criteria used they are UV-bright objects with UV-uncorrected SFRs of about 2.0 Myr−1. When dust-correcting their rest-frame UV luminosity, their total SFR turns out to have a median value of 46.4 Myr−1. Combining the total SFRs and ages, we find that GALEX-selected LBGs have a median stellar mass of log(M∗/M)=9.74. Only 2 per cent of the galaxies selected with the Lyman break selection criterion have an AGN, according to their X-ray emission. (ii) LBGs with higher stellar masses have higher total SFRs and lower values of the specific SFR. The anticorrelation between the specific SFR and stellar mass supports the downsizing scenario, whereby more massive galaxies formed their stars earlier and faster than galaxies with a lower stellar mass. (iii) Morphologically,LBGsatz∼1aremostlydisc-likegalaxies (about 69per cent), while those remaining are interacting, compact or irregular systems in much lower percentages. This is confirmed by their S´ ersic indices, which are typically below 0.5. The median effective radius for our GALEX-selected LBGs at z∼1 is 2.48 kpc. Larger galaxies tend to have higher total SFRs and stellar masses. (iv) In a CMD, most GALEX-selected LBGs are located over the blue cloud at their redshift, which indicates that they are active SF galaxies. Some LBGs are located over the green valley or near the red sequence. They turn out to be the dustiest and/or oldest galaxies in the samples, signs that they are more evolved systems. (v) Comparing with their high-redshift analogues, we find that the galaxies selected through the Lyman break criterion at z∼1 seem to be at a later evolutionary stage than those at high redshifts. However, the uncertainties in the SED-derived age are typically significant, and consequently the age evolution should be confirmed with a more detailed study of the rest-frame UV-to-NIR SEDs of LBGs at different redshifts. We do not find any significant difference in the distributions of stellar mass or dust attenuation for LBGs at high and intermediate redshifts. LBGs at lower redshifts are larger, have a greater contribution of older stellar population to their SEDs, and are redder in their UV continuum than their high-redshift analogues. ACKNOWLEDGEMENTS The authors would like to thank the referee for the careful reading of the manuscript and for valuable feedback that has improved the presentation of our results. I. Oteo would also like to thank Professor Tsutomu T. Takeuchi for kindly providing useful comments. This research has been supported by the Spanish Ministerio de Econom’a y Competitividad (MINECO) under grant AYA2011-29517-C0301. Some/all of the data presented in this paper were obtained from the Multimission Archive at the Space Telescope Science Institute (MAST). STScI is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555. Support for MAST for non-HST data is provided by the NASA Office of Space Science via grant NNX09AF08G and by other grants and contracts. This work is based on observations made with the European Southern Observatory telescopes obtained from the ESO/ST-ECF Science Archive Facility and from zCOSMOS observations carried out using the Very Large Telescope at the ESO Paranal Observatory under Programme ID: LP175.A-0839. Funding for the SDSS and SDSS-II was provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, the US Department of Energy, the National Aeronautics and Space Administration, the Japanese Monbukagakusho, the Max Planck Society, and the Higher Education Funding Council for England. The SDSS web site is http://www.sdss.org/. The SDSS is managed by the Astrophysical Research Consortium for the Participating Institutions. The Participating Institutions are the American Museum of Natural History, the Astrophysical Institute Potsdam, the University of Basel, the University of Cambridge, the Case Western Reserve University, the University of Chicago, Drexel University, Fermilab, the Institute for Advanced Study, the Japan Participation Group, Johns Hopkins University, the Joint Institute for Nuclear Astrophysics, the Kavli Institute for Particle Astrophysics and Cosmology, the Korean Scientist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos National Laboratory, the Max-Planck-Institute for Astronomy (MPIA), the Max-Planck-Institute for Astrophysics (MPA), New Mexico State University, Ohio State University, the University of Pittsburgh, the University of Portsmouth, Princeton University, the United States Naval Observatory, and the University of Washington. Financial support from the Spanish grant AYA2010-15169 and from the Junta de Andalucia through TIC-114 and the Excellence Project P08-TIC03531 is acknowledged. REFERENCES Abazajian K. N. et al., 2009, ApJS, 182, 543 Adelberger K. L., Steidel C. C., Shapley A. E., Hunt M. P., Erb D. K., Reddy N. A., Pettini M., 2004, ApJ, 607, 226 Aparicio Villegas T. et al., 2010, AJ, 139, 1242 Barger A. J., Cowie L. L., Wang W.-H., 2008, ApJ, 689, 687 de Barros S., Schaerer D., Stark D. P., 2012, preprint (arXiv:1207.3663) Basu-Zych A. R., Hornschemeier A. E., Hoversten E. A., Lehmer B., Gronwall C., 2011, ApJ, 739, 98 Bell E. F. et al., 2004, ApJ, 608, 752 Ben´ ıtez N. et al., 2009, ApJ, 692, L5 Blanton M. R. et al., 2003, ApJ, 594, 186 Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

Lyman break and UV-selected galaxies at z∼1 2725 Bongiovanni A. et al., 2010, A&A, 519, L4 Bouwens R. J., Illingworth G. D., Blakeslee J. P., Franx M., 2006, ApJ, 653, 53 Bouwens R. J. et al., 2009, ApJ, 705, 936 Bouwens R. J. et al., 2010a, ApJ, 708, L69 Bouwens R. J. et al., 2010b, ApJ, 708, L69 Bouwens R. J. et al., 2011, ApJ, 737, 90B Bouwens R. J. et al., 2012, ApJ, 754, 83B Bruzual G., Charlot S., 2003, MNRAS, 344, 1000 ( BC03) Bunker A. J., Stanway E. R., Ellis R. S., McMahon R. G., 2004, MNRAS, 355, 374 Burgarella D. et al., 2006, A&A, 450, 69 Burgarella D., Le Floc’h E., Takeuchi T. T., Huang J. S., Buat V., Rieke G. H., Tyler K. D., 2007, MNRAS, 380, 986 Burgarella D. et al., 2011, ApJ, 734, L12 Calzetti D., Kinney A. L., Storchi-Bergmann T., 1994, ApJ, 429, 582 Calzetti D., Armus L., Bohlin R. C., Kinney A. L., Koornneef J., StorchiBergmann T., 2000, ApJ, 533, 682 Capak P. et al., 2007, ApJS, 172, 99 Castellano M. et al., 2012, A&A, 540, A39 Chen Z., Shu C. G., Burgarella D., Buat V., Huang J.-S., Luo Z. J., 2013, MNRAS, 431, 2080 Cirasuolo M. et al., 2007, MNRAS, 380, 585 Cowie L. L., Hu E. M., 1998, AJ, 115, 1319 Cowie L. L., Barger A. J., Hu E. M., 2010, ApJ, 711, 928 Crist´ obal-Hornillos D. et al., 2009, ApJ, 696, 1554 Daddi E., Cimatti A., Renzini A., Fontana A., Mignoli M., Pozzetti L., Tozzi P., Zamorani G., 2004, ApJ, 617, 746 Daddi E. et al., 2007, ApJ, 670, 156 DamenM.,F¨ orsterSchreiberN. M., FranxM.,Labb´ eI.,Toft S.,vanDokkum P. G., Wuyts S., 2009, ApJ, 705, 617 DunlopJ. S.,McLure R. J.,Robertson B.E., EllisR.S., StarkD. P.,Cirasuolo M., de Ravel L., 2012, MNRAS, 420, 901 Dunne L. et al., 2009, MNRAS, 394, 3 Elbaz D. et al., 2007, A&A, 468, 33 Elbaz D. et al., 2011, A&A, 533, A119 Elmegreen B. G., Elmegreen D. M., Fernandez M. X., Lemonias J. J., 2009, ApJ, 692, 12 Elvis M. et al., 2009, ApJS, 184, 158 Erb D. K., Steidel C. C., Shapley A. E., Pettini M., Reddy N. A., Adelberger K. L., 2006, ApJ, 647, 128 Feldmann R. et al., 2006, MNRAS, 372, 565 Feulner G., Goranova Y., Drory N., Hopp U., Bender R., 2005, MNRAS, 358, L1 Finkelstein S. L., Rhoads J. E., Malhotra S., Grogin N., Wang J., 2008, ApJ, 678, 655 Finkelstein S. L., Cohen S. H., Malhotra S., Rhoads J. E., 2009a, ApJ, 700, 276 Finkelstein S. L., Rhoads J. E., Malhotra S., Grogin N., 2009b, ApJ, 691, 465 Finkelstein S. L., Rhoads J. E., Malhotra S., Grogin N., 2009c, ApJ, 691, 465 Finkelstein S. L., Papovich C., Giavalisco M., Reddy N. A., Ferguson H. C., Koekemoer A. M., Dickinson M., 2010a, ApJ, 719, 1250 Finkelstein S. L., Papovich C., Giavalisco M., Reddy N. A., Ferguson H. C., Koekemoer A. M., Dickinson M., 2010b, ApJ, 719, 1250 Finkelstein S. L. et al., 2012, ApJ, 756, 164 Franzetti P. et al., 2007, A&A, 465, 711 Gawiser E. et al., 2006, ApJ, 642, L13 Gawiser E. et al., 2007, ApJ, 671, 278 Giavalisco M. et al., 2004, ApJ, 600, L103 Gonz´ alez V., Labb´ e I., Bouwens R. J., Illingworth G., Franx M., Kriek M., Brammer G. B., 2010, ApJ, 713, 115 Gronwall C. et al., 2007, ApJ, 667, 79 Guaita L. et al., 2011, ApJ, 733, 114 Guillaume M., Llebaria A., Aymeric D., Arnouts S., Milliard B., 2006, in Dougherty E. R., Astola J. T., Egiazarian K. O., Nasrabadi N. M., Rizvi S. A., eds, Society of Photo-Optical Instrumentation Engineers (SPIE) Conf.Ser.Vol.6064,Image Processing:Algorithmsand Systems,Neural Networks, and Machine Learning. SPIE, p. 332 Haberzettl L., Williger G., Lehnert M. D., Nesvadba N., Davies L., 2012, ApJ, 745, 96 Hathi N. P., Malhotra S., Rhoads J. E., 2008, ApJ, 673, 686 Hathi N. P. et al., 2010, ApJ, 720, 1708 Hathi N. P. et al., 2013, ApJ, 765, 88 Hibon P., Malhotra S., Rhoads J., Willott C., 2011, ApJ, 741, 101 Hibon P., Kashikawa N., Willott C., Iye M., Shibuya T., 2012, ApJ, 744, 89 Hogg D. W. et al., 2002, AJ, 124, 646 Iwata I., Ohta K., Tamura N., Akiyama M., Aoki K., Ando M., Kiuchi G., Sawicki M., 2007, MNRAS, 376, 1557 Karim A. et al., 2011, ApJ, 730, 61 Kennicutt R. C., Jr, 1998, ARA&A, 36, 189 Kong X., Charlot S., Brinchmann J., Fall S. M., 2004, MNRAS, 349, 769 Lai K. et al., 2008, ApJ, 674, 70 Lehmer B. D. et al., 2005, AJ, 129, 1 Lehnert M. D., Bremer M., 2003, ApJ, 593, 630 Lilly S. J. et al., 2007, ApJS, 172, 70 Ly C. et al., 2009, ApJ, 697, 1410 Ly C., Malkan M. A., Hayashi M., Motohara K., Kashikawa N., Shimasaku K., Nagao T., Grady C., 2011, ApJ, 735, 91 Madau P., 1995, ApJ, 441, 18 Madau P., Ferguson H. C., Dickinson M. E., Giavalisco M., Steidel C. C., Fruchter A., 1996, MNRAS, 283, 1388 Magdis G. E., Rigopoulou D., Huang J.-S., Fazio G. G., 2010, MNRAS, 401, 1521 Martin D. C. et al., 2005, ApJ, 619, L1 Matute I. et al., 2012, A&A, 542A, 20M Meurer G. R., Heckman T. M., Lehnert M. D., Leitherer C., Lowenthal J., 1997, AJ, 114, 54 Meurer G. R., Heckman T. M., Calzetti D., 1999, ApJ, 521, 64 Moles M. et al., 2005, preprint (astro-ph/0504545) Moles M. et al., 2008, AJ, 136, 1325 Mosleh M., Williams R. J., Franx M., Kriek M., 2011, ApJ, 727, 5 Mosleh M. et al., 2012, ApJ, 756, L12 Murayama T. et al., 2007, ApJS, 172, 523 Nandra K., Mushotzky R. F., Arnaud K., Steidel C. C., Adelberger K. L., Gardner J. P., Teplitz H. I., Windhorst R. A., 2002, ApJ, 576, 625 Nicol M.-H., Meisenheimer K., Wolf C., Tapken C., 2011, ApJ, 727, 51 Nilsson K. K. et al., 2007, A&A, 471, 71 Nilsson K. K., Tapken C., Møller P., Freudling W., Fynbo J. P. U., Meisenheimer K., Laursen P., ¨ Ostlin G., 2009, A&A, 498, 13 Nilsson K. K., M¨ oller-Nilsson O., Rosati P., Lombardi M., K¨ ummel M., Kuntschner H., Walsh J. R., Fosbury R. A. E., 2011, A&A, 526, A10 Noeske K. G. et al., 2007, ApJ, 660, L43 Oke J. B., Gunn J. E., 1983, ApJ, 266, 713 Oteo I. et al., 2011, ApJ, 735, L15 Oteo I. et al., 2012a, A&A, 541, A65 Oteo I. et al., 2012b, ApJ, 751, 1390 Oteo I. et al., 2013a, A&A, 554L, 3O Oteo I. et al., 2013b, MNRAS, preprint (arXiv:1306.1121) Ouchi M. et al., 2008, ApJS, 176, 301 Ouchi M. et al., 2010, ApJ, 723, 869 Overzier R. A. et al., 2008, ApJ, 673, 143 Overzier R. A. et al., 2011, ApJ, 726, L7 Pannella M. et al., 2009, ApJ, 698, L116 Papovich C., Dickinson M., Ferguson H. C., 2001, ApJ, 559, 620 Peng C. Y., Ho L. C., Impey C. D., Rix H., 2010, AJ, 139, 2097 Pirzkal N., Malhotra S., Rhoads J. E., Xu C., 2007, ApJ, 667, 49 Povi´ c M. et al., 2009, ApJ, 706, 810 Povi´ c M. et al., 2012, A&A, 541A, 118P Rix H.-W. et al., 2004, ApJS, 152, 163 Rodighiero G. et al., 2010, A&A, 518, L25 Salim S. et al., 2007, ApJS, 173, 267 Salpeter E. E., 1955, ApJ, 121, 161 Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018

2726 I. Oteo et al. Salvato M. et al., 2011, ApJ, 742, 61 Sersic J. L., 1968, in Sersic J. L., ed., Atlas de Galaxias Australes. Observatorio Astronomico, Cordoba, Argentina Shioya Y. et al., 2009, ApJ, 696, 546 Stanway E. R., Bunker A. J., McMahon R. G., 2003, MNRAS, 342, 439 Steidel C. C., Giavalisco M., Dickinson M., Adelberger K. L., 1996, AJ, 112, 352 Steidel C. C., Adelberger K. L., Giavalisco M., Dickinson M., Pettini M., 1999, ApJ, 519, 1 Steidel C. C., Adelberger K. L., Shapley A. E., Pettini M., Dickinson M., Giavalisco M., 2003, ApJ, 592, 728 Strateva I. et al., 2001, AJ, 122, 1861 Takeuchi T. T., Yuan F.-T., Ikeyama A., Murata K. L., Inoue A. K., 2012, ApJ, 755, 144 Taylor E. N. et al., 2009, ApJ, 694, 1171 Verma A., Lehnert M. D., F¨ orster Schreiber N. M., Bremer M. N., Douglas L., 2007, MNRAS, 377, 1024 Weiner B. J. et al., 2005, ApJ, 620, 595 Wilkins S. M., Bunker A. J., Stanway E., Lorenzoni S., Caruana J., 2011, MNRAS, 417, 717 Williams R. J., Quadri R. F., Franx M., van Dokkum P., Labb´ e I., 2009, ApJ, 691, 1879 Wolf C., Meisenheimer K., R¨ oser H.-J., 2001, A&A, 365, 660 Wolf C. et al., 2004, A&A, 421, 913 Wolf C. et al., 2005, ApJ, 630, 771 Yabe K., Ohta K., Iwata I., Sawicki M., Tamura N., Akiyama M., Aoki K., 2009, ApJ, 693, 507 This paper has been typeset from a T EX/L A TEX file prepared by the author. Downloaded from https://academic.oup.com/mnras/article-abstract/433/4/2706/1079761 by Universidad de Sevilla user on 26 March 2018