The origin of bulges and discs in the CALIFA survey – I. Morphological evolution

MNRAS 504, 3058–3073 (2021) doi:10.1093/mnras/stab1064 Advance Access publication 2021 April 21 The origin of bulges and discs in the CALIFA survey – I. Morphological evolution J. M´ endez-Abreu ,1,2,3,4‹A. de Lorenzo-C´ aceres 1,2,5,6and S. F. S´ anchez7 1Instituto de Astrof´ ısica de Canarias, Calle V´ ıa L´ actea s/n, E-38205 La Laguna, Tenerife, Spain 2Departamento de Astrof´ ısica, Universidad de La Laguna, E-38200 La Laguna, Tenerife, Spain 3Departamento de F´ ısica y del Cosmos, Campus de Fuentenueva, Edificio Mecenas, Universidad de Granada, E-18071 Granada, Spain 4Instituto Carlos I de F´ ısica Te´ orica y Computacional, Facultad de Ciencias, E-18071 Granada, Spain 5Departamento de F´ ısica de la Tierra y Astrof´ ısica, Universidad Complutense de Madrid (UCM), Plaza Ciencias 1, E-28040 Madrid, Spain 6Instituto de F´ ısica de Part´ ıculas y del Cosmos (IPARCOS), Plaza Ciencias 1, E-28040 Madrid, Spain 7Instituto de Astronom´ ıa, Universidad Nacional Aut´ onoma de M´ exico, A.P. 70-264, 04510 M´ exico, D.F., M´ exico Accepted 2021 April 7. Received 2021 March 17; in original form 2020 July 30 ABSTRACT This series of papers aims at understanding the formation and evolution of non-barred disc galaxies. We use the new spectrophotometric decomposition code, C2D, to separate the spectral information of bulges and discs of a statistically representative sample of galaxies from the CALIFA survey. Then, we study their stellar population properties analysing the structureindependent datacubes with the PIPE3D algorithm. We find a correlation between the bulge-to-total (B/T) luminosity (and mass) ratio and galaxy stellar mass. The B/Tmass ratio has only a mild evolution with redshift, but the bulge-to-disc (B/D)mass ratio shows a clear increase of the disc component since redshift z<1 for massive galaxies. The mass–size relation for both bulges and discs describes an upturn at high galaxy stellar masses (log(M/M)>10.5). The relation holds for bulges but not for discs when using their individual stellar masses. We find a negligible evolution of the mass–size relation for both the most massive (log (M,b,d/M)>10) bulges and discs. For lower masses, discs show a larger variation than bulges. We also find a correlation between the S´ ersic index of bulges and both galaxy and bulge stellar mass, which does not hold for the disc mass. Our results support an inside-out formation of nearby non-barred galaxies, and they suggest that (i) bulges formed early-on and (ii) they have not evolved much through cosmic time. However, we find that the early properties of bulges drive the future evolution of the galaxy as a whole, and particularly the properties of the discs that eventually form around them. Key words: galaxies: bulges–galaxies: evolution–galaxies: formation–galaxies: structure. 1 INTRODUCTION Galaxies are complex systems with an intricate combination of different structural components such as bulges, discs, and bars. The relative contribution of these structures to the galaxy luminosity, or mass, define our morphological classifications, which have been the subject of numerous studies along the last century (Hubble 1936; de Vaucouleurs & de Vaucouleurs 1964;Butaetal.2015). Many physical properties of galaxies, such as gas content, stellar age, and star formation rate (SFR), are known to correlate with morphology. Therefore, studying the cosmic evolution of the main structures of galaxies is key to understand the physical mechanisms driving the morphological evolution of galaxies (Clauwens et al. 2018; Tacchella et al. 2019). In the simplest scenario, disc galaxies are made up of a central bulge and an outer disc. Galactic discs are thought to form within dark matter haloes with high angular momentum, and a quiet recent assembly history, as a consequence of angular momentum conservation during the dissipational collapse of gas (Fall & Efstathiou 1980; E-mail: [email protected] Mo,Mao&White1998).Ontheotherhand,bulgesaregenerallyseen as the slowly-rotating result from merger events (Cole et al. 2000). However, more recent, sophisticated simulations are challenging this simple view (e.g. Sales et al. 2012). An early origin of galactic discs, driven by the infall of gas in a rotating dark matter halo, has been the preferred scenario for the formation of this component during decades (Fall & Efstathiou 1980). In this framework, the central regions of discs reach the necessary gas surface mass density to form stars earlier in time than the outer parts, naturally resulting in an inside-out mass growth (Brook et al. 2006). However, this idealized picture is more complex in a hierarchical model of galaxy formation where (i) galaxy mergers can destroy (or thicken) the initial discs (Steinmetz & Navarro 2002) and (ii) gas does not keep all its initial angular momentum, thus producing discs that are too small compared to those observed in nearby galaxies (Navarro & White 1994; Sommer-Larsen, Gelato & Vedel 1999; D’Onghia & Burkert 2004). Therefore, at least for a fraction of disc galaxies, a later formation must be invoked. From the structure formation point of view, alternative dark matter properties (such as Warm Dark Matter) will induce that structure formation occurs later (Sommer-Larsen & Dolgov 2001). More related with the physics of baryons, the effect of stellar or active galactic nuclei C 2021 The Author(s) Published by Oxford University Press on behalf of Royal Astronomical Society Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022 Bulges and discs in the CALIFA survey 3059 (AGN) feedback might also prevent the gas from cooling until relatively late times z<1 (Weil, Eke & Efstathiou 1998; Thacker & Couchman 2001). More recently, hydrodynamical simulations have demonstrated that after a major merger of gas-rich galaxies, a new disc can be formed out of the remaining gas not consumed in the initial starburst (Hopkins et al. 2009). In these delayed formation scenarios, the relation between the initial angular momentum of the halo and that of the disc is expected to be erased. A variety of pathways for bulge formation has been proposed in the literature. At high redshift, when the gas mass fraction in galaxies was higher than at present days, the formation of galaxy bulges was mainly driven by highly dissipative processes. Some of the proposed scenarios are similar for bulges and ellipticals, such as the direct monolithic collapse of protogalactic gas clouds (Eggen, Lynden-Bell & Sandage 1962;Larson1974) or the major mergers of gas-rich galaxies (Kauffmann 1996; Hopkins et al. 2009;Zavala et al. 2012; Avila-Reese, Zavala & Lacerna 2014). Both of these mechanisms imply (in order to not end up in an elliptical galaxy) that the stellar disc is formed after the bulge is already in place. This would be in agreement with the idea of an inside-out mass growth for disc galaxies (Aumer & White 2013; Gonz´ alez Delgado et al. 2015), but some works have questioned that the frequency of major mergers might not be enough to be the primary formation mechanism of bulges and ellipticals (Kitzbichler & White 2008). Other dissipative mechanismshavebeenproposedwherethe disc mightformbefore,or at least concomitantly to, the bulge. At high redshift, primordial gasrich galaxy discs are highly turbulent with star formation occurring in massive clumps (Abraham et al. 1996; Elmegreen, Elmegreen & Hirst 2004; Hinojosa-Go˜ ni, Mu˜ noz-Tu˜ n´ on & M´ endez-Abreu 2016). The coalescence of these giant clumps as they move to the galaxy centre due to dynamical friction has been demonstrated to create new bulges (Noguchi 1999; Immeli et al. 2004; Bournaud, Elmegreen & Elmegreen 2007;Ceverinoetal.2015). However, the role of clumps in bulge formation is still not clear since their survival strongly depends on the feedback implementation (Mandelker et al. 2017; Oklopˇ ci´ cetal.2017). Another bulge formation scenario at high redshift is related to a rapid gas inflow injected to the galaxy centre directly from the surrounding halo (Scannapieco et al. 2009; Zolotov et al. 2015; Tacchella et al. 2016). In this scenario, spheroids tend to form when the spin of newly accreted gas is misaligned with that of the host galaxy, leading to episodic formation of stars with different kinematics that cancel out the net rotation of the galaxy (Sales et al. 2012). At lower redshifts, galaxy bulges might continue to grow from (i) stars already present in their hosting discs through radial migration (Minchev & Famaey 2010) or dissolution of bars (Guo et al. 2020), (ii) ex situ stars accreted from satellites during minor merger events (Aguerri, Balcells & Peletier 2001; Eliche-Moral et al. 2006; Guedes et al. 2013; Rodriguez-Gomez et al. 2017), or (iii) in situ new stars created from the inflow of gas from the outer disc to the galaxy centre due to the gravitational torque exerted by stellar bars (Kormendy & Kennicutt 2004; Athanassoula 2005). All these latter pathways for bulge formation require longer time-scales to modify/create bulges and they are generally referred to as secular processes. The diversity of formation and evolution scenarios for bulges has been generally condensed into two broad observational classes with different characteristics: classical and disc-like bulges (Kormendy & Kennicutt 2004; Athanassoula 2005). Classical bulges are those following surface-brightness distributions with a S´ ersic index n> 2 and bulge-to-total (B/T) luminosity ratio B/T>0.2, they appear rounder than their surrounding discs (M´ endez-Abreu et al. 2010), and their stellar kinematics is dominated by random motions that generally satisfy the Fundamental Plane (FP) correlation (Bender, Burstein & Faber 1992;Falc ´ on-Barroso, Peletier & Balcells 2002; Aguerri et al. 2005). The stellar populations of classical bulges show similarities with those of ellipticals of the same mass. In general, they areoldandmetal-richwithashortformationtime-scale(seeS´ anchezBl´ azquez 2016). On the other hand, disc-like bulges are oblate ellipsoids (Costantin et al. 2017) with apparent flattening similar to their outer discs, surface-brightness distributions well fitted with a S´ ersicprofileofindexn<2andB/T<0.35 (Fisher & Drory 2008). Their kinematics is dominated by rotation in diagrams such as the v/σversus (Kormendy & Kennicutt 2004) and they are identified as low-σoutliersof theFaber–Jacksonrelation (Faber& Jackson1976). Disc-like bulges are also usually dominated by young stars, with the presence of gas and possible recent star formation (Fisher & Drory 2016). Despite this apparently clear separation in their observed properties, a number of studies have demonstrated that (i) several of the previous observables do not show a clear dichotomy, making it difficult to establish the limits between both bulge types, and (ii) using different diagnostics generally lead to different classifications producing heavily contaminated samples (Costantin et al. 2018; M´ endez-Abreu et al. 2018; Costantin et al. 2020). This situation might be caused (or enhanced) by the discovery that bulges in at least some disc galaxies are indeed composite systems, i.e. a single galaxy can host a (kinematically hot, spheroidal) classical bulge and a (kinematically cool, flattened) disc-like bulge (M´ endez-Abreu et al. 2014; Erwin et al. 2015). All previous caveats on bulge classification, together with the fact that the large variety of bulge formation paths envisioned in simulations are only coarsely captured with the current observational division, have prompted us to avoid such a separation in this study. In the last decade, the study of galaxy bulges in particular, and disc galaxies in general, has strongly benefited from (i) developments on photometric techniques to isolate the different stellar components of galaxies, which have become more sophisticated (GASP2D;M ´ endezAbreu et al. 2008;GALFIT,Pengetal.2010;IMFIT,Erwin2015). The generalization of the idea that only detailed multicomponent photometric decompositions are suitable to understand the origin of bulges has contributedto theseimprovementsin recentyears(Gadotti 2009;M ´ endez-Abreu et al. 2017; de Lorenzo-C´ aceres et al. 2019b); (ii) the advent of integral field spectroscopy (IFS), which has given access to the spatially resolved properties of individual components, therefore allowing for a deeper understanding of their kinematic and stellar population properties. However, a general problem hindering our advance on understanding the formation and evolution of bulges and discs galaxies relies on the necessity of new techniques to analyse the data. In particular, there is a strong need for new ways of separating, spectroscopically, the stellar structures shaping the galaxies avoiding issues with contamination due to overlapping. Recently, new algorithms have explored different approaches to work out this problem. We classify them here in three classes: (i) Spectro-photometric decompositions: In this technique, IFS data are understood as a series of two-dimensional images that be decomposed into their structural components using standard photometric decompositions codes (Johnston, Arag´ on-Salamanca & Merrifield 2014; Johnston et al. 2017,2021;M ´ endez-Abreu, S´ anchez & de Lorenzo-C´ aceres 2019a,b;Barsantietal.2021). (ii) Kinematic decompositions: The structural components are assumed to have different kinematic distributions that are directly fitted, or derived, from the spectra (Coccato et al. 2011; Tabor et al. 2017; Coccato et al. 2018; Mehrgan et al. 2019). (iii) Schwarzchild dynamical modelling: The galaxy stellar kinematics is modelled using a variety of stellar orbits. These are then analysed, generally in terms of their angular momentum, to identify structures within the galaxy (Zhu et al. 2018b, MNRAS 504, 3058–3073 (2021) Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022 3060 J. M´ endez-Abreu, A. de Lorenzo-C´ aceres and S. F. S´ anchez a;Pocietal.2019). These new methods are providing more accurate constraints to numerical simulation models of disc galaxy formation. This paper is the first of a series devoted to analyse, with unprecedented detail, a statistically representative sample of bulges and discs in the nearby Universe. To this aim, we have applied our recently developed spectro-photometric decomposition code (C2D;M ´ endezAbreu et al. 2019a) to a sample of photometrically classified bulgeto-disc galaxies observed within the CALIFA IFS survey (S´ anchez et al. 2016b). We have explicitly removed barred galaxies from our sample to simplify the interpretation of our results. Nonetheless, we appreciate they constitute a important channel for bulge formation, and have an strong impact on the evolution of galactic discs, so they will be thoroughly studied in a separate paper. As stated before, in this series of papers, we refrain from a pre-defined separation of bulge types and study both bulges and discs spectro-photometric properties as a continuous population. In this first paper, we focus on understanding the morphological evolution of disc galaxies, to this aim, we explore the relations between the main photometric properties of bulges and discs, as derived from typical photometric decompositions, their stellar masses obtained from spectral synthesis modelling and, whenever possible, their time evolution using the derived star formation histories (SFHs). This paper is organized as follows. Section 2 describes the sample of photometrically classified bulge-to-disc galaxies from the CALIFA survey. Section 3 highlights the main features of our new methodology to extract the spectro-photometric properties of bulges and discs using IFS data and its analysis to derive their stellar population properties. Section 4 shows the main relations between photometric properties and galaxy mass, as well as their time evolutionwheneverpossible. Sections 5and 6summarizeour mainconclusions. Throughout this paper, we assume a flat cosmology with m =0.27, λ=0.73, and a Hubble constant H0=71 km s−1Mpc−1. 2 CALIFA SAMPLE OF BULGE AND DISC GALAXIES The sample of galaxies analysed in this work is drawn from the CALIFA data release 3 (DR3; S´ anchez et al. 2016a). This data release comprises 667 galaxies covering a wide range of stellar masses and Hubble types. M´ endez-Abreu et al. (2017) carried out a multicomponent multiband (g,r,andibands) photometric decomposition of 404 galaxies present in the CALIFA DR3 using SDSS imaging. This includes all galaxies but those with high inclination (i>70◦)orininteraction(seeM ´ endez-Abreu et al. 2017, for more details). From this study, we take the 70 galaxies that were successfully fitted using only a central bulge (represented with a S´ ersic profile) and an outer disc (modelled with a single exponential profile), as well as the 64 galaxies that were fitted using a central bulge and an outer broken exponential disc, i.e. discs profiles of types II (down-bending) and III (up-bending) following the definition of Erwin, Beckman & Pohlen (2005). In order to avoid possible issues with the fit around the break radius, we discard those galaxies with rbreak <25 arcsec, therefore remaining with 57 galaxies with a wellbehaved bulge and disc inside the CALIFA field of view (FoV). We finally include in the sample the 32 early-type galaxies classified as unknown in M´ endez-Abreu et al. (2018) and already analysed using C2Din M´ endez-Abreu et al. (2019a). This latter sample represents early-type galaxies for which our photometric approach is not able to classify them in either a simple S´ ersic or S´ ersic +Exponential fit, as both options return the same statistical solution. These galaxies are analysed here using their two-component (bulge and disc) best fit. We run C2D+PIPE3D over this sample of 159 galaxies finding that for 30 of them, the code does not converge, mainly due to their low B/Tratio. Therefore, the final sample used in this paper comprises 129 non-barred disc galaxies with a photometric bulge: 58 bulge-todisc, 40 bulge-to-disc with a break, and 31 early-type galaxies. In addition, we also use 41 photometrically classified elliptical galaxies for comparison (see M´ endez-Abreu et al. 2019a, for details). The CALIFA survey (S´ anchez et al. 2016c) observed all galaxies using the PMAS/PPaK instrument with an FoV of 74×64arcsec2. After a three pointing dithering pattern, the reconstructed datacubes cover up to 2.5 ×re, g (galaxy effective radius) for 90 percent of the sample (Walcher et al. 2014). The wavelength range and spectroscopic resolution for the adopted V500 setup (3745–7500 Å, R∼850) are perfectly suited to study the properties of stellar populations and ionized-gas emission lines. The typical spatial resolution of the datacubes is full width at half-maximum (FWHM) ∼2.5 arcsec, corresponding to ∼1 kpc at the average redshift of the survey (Garc´ ıa-Benito et al. 2015). 3C2D+PIPE3D ANALYSIS OF THE SAMPLE The spectro-photometric decomposition of the CALIFA galaxies into abulgeanddiscisperformedusingC2D(M´ endez-Abreuetal.2019a). This new methodology allows us to separate the spectral contribution of each structural component providing an independent datacube for both bulge and disc. An extensive description of the code and its reliability is presented in M´ endez-Abreu et al. (2019a). In brief, the application of C2DtotheCALIFAdataisbasedontheideathat datacubes can be worked out as a sequence of quasi-monochromatic 2D images at different wavelengths. Thus, standard 2D photometric decomposition techniques are able to isolate the photometric contribution of both bulge and disc. In C2D, the photometric decomposition engine is provided by GASP2D(M´ endez-Abreu et al. 2008,2014), a code that has been extensively tested on different galaxy samples with multiple structures (e.g. de Lorenzo-C´ aceres et al. 2019a,b, 2020). Moreover, GASP2Dwas used in M´ endez-Abreu et al. (2017) to perform a multiband photometric decomposition of the CALIFA galaxies using SDSS imaging. This is of particular importance in our case since the structural parameters of bulges and discs in C2Dare anchored to those derived using SDSS due to the coarse spatial resolutionof theCALIFA datacubes.In fact, wenotehere thatperforming a completely free bulge-to-disc fitting directly on the CALIFA datacubes should be avoided (see M´ endez-Abreu et al. 2018,for the effect of spatial resolution on bulge parameters). Rearranging the best-fitting intensity values for each quasi-monochromatic image into a datacube we are able to recover the characteristic spectrum for each component. In addition, C2Dprovides an independent datacube (with spatial and spectral information) for each component. To do this, for each quasi-monochromatic image, the B/Tand the disc-tototal (D/T=1−B/T) ratios are computed spaxel-wise. Each fraction is then multiplied by the observed CALIFA datacube in that spaxel and wavelength producing independent bulge and disc datacubes. It is worth reminding that our spectro-photometric decomposition relies on the common assumptions that (i) bulges and discs can be well described using a S´ ersic and exponential analytical function, respectively (but see Breda, Papaderos & Gomes 2020 for possible issues when dealing with late-type galaxies); and (ii) galaxy structure changes smoothly with wavelength. Further details on the specific application of C2Dto CALIFA data are presented in M´ endez-Abreu et al. (2019a). The second step of our analysis consists of deriving the stellar population and ionized gas properties of bulges and discs in our sample. We use the PIPE3D pipeline (S´ anchez et al. 2016a), which MNRAS 504, 3058–3073 (2021) Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022 Bulges and discs in the CALIFA survey 3061 Figure 1. Left-hand panels: NGC 2730 (log(M/M)=9.8 and B/T=0.17). Right-hand panels: NGC 5513 (log(M/M)=11.1 and B/T=0.61). Upper panels: integrated spectrum over one effective radius of the galaxy (re, gal) for the whole galaxy (black), bulge (red), and disc (blue). Bottom panels: luminosity (dashed lines) and mass (solid lines) fraction of stars contributing to a given stellar population age. Different colours as in the upper panels. Fractions are also computed within 1re, gal. is specifically designed to extract the stellar population and ionizedgas properties from IFS data and it has been extensively tested on CALIFA data (e.g. S´ anchez et al. 2015,2016b). PIPE3D adopts the GSD156 library of simple stellar populations from Cid Fernandes et al. (2013), which comprises 156 templates covering 39 stellar ages (from 1 Myr to 13 Gyr) and four metallicities (Z/Z=0.2, 0.4, 1, and 1.5 dex). The best-fitting stellar-population model spectra to the galaxy continuum is subtracted from the original cube to create a gas-pure cube including the ionized-gas emission lines only. Emission lines are then measured using a set of Gaussian functions and subtracted to the original spectra to perform the analysis in an iterative sequence. The main data products obtained from PIPE3D include luminosity/mass-weighted ages and metallicities, SFHs, and intensity maps of strong emission lines for both components. A Salpeter initial mass function (IMF) was adopted in the stellar population analysis (Salpeter 1955). We refer the reader to the presentation paper of PIPE3D (S´ anchez et al. 2016b) for a detailed description of its application to CALIFA. Fig. 1shows the spectra (upper panels) and the fraction of light/mass (bottom panels) contributed by stars of different ages (marginalized over all possible metallicities) for the galaxy, bulge, and disc spectra of both NGC 2730 and NGC 5513. All these quantities are integrated within an ellipse with semimajor axis of one effective radius of the galaxy (re, gal;seeTable1) and oriented with the ellipticity and position angle of the outer disc. They are shown as an example of a typical low-mass and low-B/Tgalaxy (NGC 2730) and a high-mass and high-B/Tgalaxy (NGC 5513) in our sample. 4 RESULTS In this section, we first describe the general properties of our sample and define the stellar mass ranges where it can be considered statistically representative of the general population of local galaxies. We then focus on the main relations obtained between the photometric properties of our bulges (re,n) and discs (h) with the stellar mass (M) derived for both the whole galaxy and the individual components. We provide the cosmic evolution of these properties for our sample when possible. 4.1 Spectro-photometric properties of the sample Fig. 2shows the distribution of the main spectro-photometric propertiesof thesample discussed inthis paper. Theindividual values for each galaxy, and each component (bulge and disc), are provided in Table 1. The stellar population properties will be provided in tabular form in a forthcoming paper. Throughout this paper, we use the photometric properties (mainly re,n,andB/T) obtained from the multicomponent photometric decompositions described in M´ endezAbreu et al. (2017). They were performed using the Sloan Digital Sky Survey imaging (SDSS-DR7; Abazajian et al. 2009)intheg,r, and ibands. The values used in this paper correspond to the r-band results, but the conclusions are not altered if we use any other band instead. The galaxy properties such as global effective radii and Hubble types are obtained from Walcher et al. (2014). The stellar masses are derived from the stellar population analysis carried out applying PIPE3D to the original CALIFA datacubes (global stellar mass) and individual component datacubes obtained from C2D(bulge and disc stellar masses). Thetypicaleffectiveradiiforbulgesaresmallerthan for discs,with thoseofthewholegalaxiesshowingintermediatevalues(Fig.2,panel a). The effective radii for the disc component are computed from their fitted exponential scale length such as re=1.678×h. Since we use broken profiles to describe the surface brightness profile of some galaxy discs in M´ endez-Abreu et al. (2017), it is worth mention that allvaluesusedinthisworkrefertotheinnerdisc.Wefindmeanvalues of 1.9, 11.5, and 5.3 kpc for bulges, discs, and galaxies, respectively. The distribution of the bulge S´ ersic index (panel b) is relatively constant for n<3, dropping quickly for higher values. We obtain a mean S´ ersic index of 2.2 for our sample bulges. The B/Tluminosity ratio (panel c), described in detail in Section 4.3, spans 0.02 <B/T< 0.73 with a mean value of 0.26, therefore covering the whole range of possibilities from almost bulgeless galaxies to nearly ellipticals (a B/T>0.8 is generally used to classify a galaxy as a single-component elliptical; M´ endez-Abreu et al. 2018). The wide range of bulges and discs covered in this study is also represented by their Hubble types (panel d). There is a drop for very late galaxies due to the lack of photometrically detected bulges in many of those galaxies. This is also reflected on the decreasing number of galaxies with low masses shown in panel (d). The mean stellar masses (in logarithmic scale) MNRAS 504, 3058–3073 (2021) Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022 3062 J. M´ endez-Abreu, A. de Lorenzo-C´ aceres and S. F. S´ anchez Table 1. Spectro-photometric properties of the sample. Galaxy re, b re, d re, gal nB/THT log (M,b/M)log(M,d/M)log(M,gal/M) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) IC0159 0.91 5.14 3.99 0.80 0.07 Sdm 8.93 9.68 9.77 IC0208 0.52 6.41 4.25 0.57 0.02 Sbc 8.87 9.98 10.04 IC0307 0.89 10.29 5.38 1.81 0.21 Sab 10.47 10.60 10.87 IC0944 1.36 11.46 4.96 0.80 0.15 Sab 10.38 10.82 10.96 IC1151 0.45 5.28 3.56 0.59 0.02 Scd 8.39 9.62 9.66 Columns: (1) Galaxy name; (2)–(4) effective radius in the rband for the bulge, disc (1.678×h), and the galaxy in kpc; (5) S´ ersic index of the bulge; (6) bulge-to-total luminosity ratio in the rband; (7) Hubble type obtained from Walcher et al. (2014); (8)–(10) stellar mass computed using the stellar population analysis for the bulge, disc, and whole galaxy. We show only the first five galaxies of the sample; the remaining are available in the online version. (a) (b) (c) (d) (e) Figure 2. Panel (a): distribution of effective radii for bulges (red), discs (blue), and the whole galaxy (black). The values for bulges and discs are from the r-band photometric decompositions of M´ endez-Abreu et al. (2017). The effective radii of the galaxies are obtained from the r-band analysis performed in Walcher et al. (2014). Panel (b): distribution of S´ ersic indices for our bulges and (panel c) bulge-to-total light ratio in the rband from photometric decompositions of M´ endez-Abreu et al. (2017). Panel (d): visual Hubble-type classification of our galaxies. Panel (e): stellar mass for bulges (red), discs (blue), and the whole galaxy (black) computed in this work. for our bulges, discs, and whole galaxies are 10.0, 10.1, and 10.5 dex, respectively. Despite the limited number of galaxies in our sample, we discuss in Section 4.2 how it is statistically representative of the local Universe in the mass range 9.5 <log(M/M)<12. It is well known that some physical properties of bulges, discs, and (global) galaxies are correlated. In particular, the Hubble type correlates (with more or less scatter) with the S´ ersic index, B/T luminosity ratio, and galaxy mass (Laurikainen et al. 2007; Weinzirl et al. 2009; Laurikainen et al. 2010;M ´ endez-Abreu et al. 2017; Gao et al. 2019). Fig. 3shows these correlations for the sample used in this study. We note that, due to these correlations and for the sake of clarity, we will only show (and discuss) in this paper those relations with galaxy mass. None the less, whenever we discuss trends with stellar mass they might be also interpreted as trends with Hubble type unless otherwise stated. 4.2 Stellar mass functions of bulges and discs The CALIFA-DR3 sample of galaxies is selected from the SDSSDR7 based on an angular diameter and redshift selection, thus it is not strictly complete in either mass or volume. Nevertheless, Walcher et al. (2014) demonstrated that these constraints do not strongly bias the sample, and that it represents quite well the whole population of nearby galaxies. Therefore, they computed a volume correction for each individual galaxy that can be used to correct for the selection function. Therefore, although the final observed sample in the CALIFA-DR3 is not complete in volume, it is possible to reconstruct volume corrected sample properties using the CALIFADR3 (see S´ anchez et al. 2016c) within the completeness limits Figure 3. Distribution of the bulge S´ ersic index (top panel), B/Tluminosity ratio, and galaxy stellar mass (bottom panel) as a function of the Hubble type. Grey circles show the values for individual galaxies and black circles represent the mean values for each Hubble type. MNRAS 504, 3058–3073 (2021) Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022 Bulges and discs in the CALIFA survey 3063 Figure 4. Top left-hand panel: luminosity function for the CALIFA-DR3 sample of galaxies (blue circles) and a subsample of 105 galaxies from this work (blue stars). This subsample represents those galaxies extracted from the CALIFA mothersample (see the text for details). The dashed line shows the SDSS luminosity function given by Blanton et al. (2003). The vertical lines indicate the magnitude limits where our sample is complete. Top right-hand, bottom left-hand, and bottom right-hand panels show the galaxy (black), bulge (red), and disc (blue) mass functions, respectively. The dashed lines show the comparison with the mass function derived by Thanjavur et al. (2016). The dotted line in the galaxy mass function was obtained from Kelvin et al. (2014) and in the bulge and disc mass functions from Moffett et al. (2016). Error bars show the Poissonian uncertanties. described in Walcher et al. (2014), that is, −19 >Mr>−23.1. Fig. 4 shows the luminosity function in the r−band for both our sample of bulge-to-disc galaxies and the CALIFA-DR3. We note that volume corrections are not applicable for all galaxies in the CALIFA-DR3, but only for those in the CALIFA mothersample (see S´ anchez et al. 2016c), so the luminosity function of our sample galaxies (Fig. 4, top left-hand panel) includes only 105 galaxies. The remaining 24 galaxies were drawn from the so-called extended sample included in the CALIFA-DR3 (see S´ anchez et al. 2016c). There is good agreement between the luminosity functions of CALIFA-DR3 and our sample, as well as for the SDSS luminosity function given by Blanton et al. (2003), within the magnitude limits −19.5 >Mr> −23. This range is slightly shorter that the one found in Walcher et al. (2014), but it demonstrates that our sample of bulge-to-disc galaxies is statistically representative of the whole nearby galaxies population within these limits. The stellar masses used in this study are obtained from the stellar population analysis described in Section 3 and integrated over the whole FoV of CALIFA. In order to understand the mass limits where our sample is representative, we computed the stellar masses of those galaxies within the complete luminosity range −19.5 >Mr>−23. We find that they correspond to a lower limit in galaxy stellar mass log(M/M)≃9.5. This is in agreement with the typical stellar mass where bulge-to-disc galaxies dominate over pure-disc systems in the CALIFA survey (see fig. 9 in M´ endez-Abreu et al. 2017). Therefore, we consider that our sample is statistically representative of bulgeto-disc galaxies for masses with log(M/M)>9.5. Our sample consists of 121 galaxies within this stellar mass with 104 included in the original CALIFA mothersample. The galaxy, bulge, and disc stellar mass functions shown in Fig. 4have been computed using the latter subsample. Due to the relative small number statistics of our sample we do not attempt to fit the stellar mass functions, but instead we compare them with those derived in other surveys such as GAMA(Moffettetal.2016) andSDSS(Thanjavuretal.2016).These previous studies based their galaxy, bulge, and disc mass estimations on either a colour empirical calibration or SED fitting to broad-band imaging data, while our masses are derived from a full spectrum fitting technique to the CALIFA datacubes. However, we find a general good agreement with both works. Our galaxy mass function (Fig. 4, top right-hand panel) does not help to solve the discrepancy between the previous works at low masses (log (M/M)<9.5), however it seems to favour the Schechter modelling of Thanjavur et al. (2016) at the high-mass end. This difference was discussed in Moffett et al. (2016) as due to the smaller volume covered by MNRAS 504, 3058–3073 (2021) Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022 3064 J. M´ endez-Abreu, A. de Lorenzo-C´ aceres and S. F. S´ anchez GAMA with respect to SDSS. Since the CALIFA sample is based on SDSS, we consider this might also be the reason for our better match with Thanjavur et al. (2016). In contrast, our bulge mass function is better represented using the Schechter modelling of the GAMA data from Moffett et al. (2016). They used a visual classification to separate single or multicomponent galaxies, which is similar to our human-supervised multicomponent decompositions performed in M´ endez-Abreu et al. (2017), and more accurate than the automatic classification carried out by Thanjavur et al. (2016). As previously stated, at stellar masses log (M/M)<9.5 single-component, puredisc systems start to dominate the mass function and this might create the flat slope in the SDSS low-mass spheroids. The disc mass function is similar in both studies and close to our values except for our lowest mass bin. As already indicated, the analysis of the luminosity and mass functions in this section suggest that our sample of bulge-to-disc galaxies is statistically representative of the whole population for galaxies in the magnitude range −19.5 >Mr>−23, which corresponds to galaxies in the stellar mass range 9.5 <log (M/M) <12. Regarding the separated components, bulges and discs are well represented in the mass ranges 8 <log (M,b/M)<11.5and 9.5<log(M,d/M)<11, respectively. For the sake of clarity, we mark these limits whenever possible in forthcoming figures. Using the previous global galaxy mass completeness limits (9.5 < log(M/M)<12) and the volume correction for each galaxy, we compute the fraction of mass in bulges and discs for our sample. We find that 49 and 51 percent of the galaxy mass is in bulges and discs, respectively. These values are quite different from previous fractions reported in Weinzirl et al. (2009) and Driver et al. (2007). The former used a sample of 143 bright spirals measuring that ∼70 per cent of the stellar mass is in discs, ∼10 per cent is in stellar bars and ∼20 per cent is in bulges. Driver et al. (2007) used the Millennium Galaxy Catalog finding that 68.6 per cent of the stellar mass to be in discs, and 32.6 percent in bulges. However, our results agree with the more recent study of Moffett et al. (2016) using the GAMA survey. They derivedthat 50per cent ofthe localstellar massdensity isin spheroids and 48 per cent in discs. However, they computed lower stellar mass densities in spheroids (1.24±0.49 ×108MMpc−3) and discs (1.20±0.45 ×108MMpc−3) compared to our 2.02 ×108and 1.9 ×108MMpc−3mass density for bulges and discs, respectively. We argue that most of the differences with previous studies rely on the different samples. On one side, we targeted only bulge-todisc galaxies discarding barred systems. On the other hand, the mass fraction of bulges and discs is strongly dependent on the mass range of the galaxy sample. Fig. 5shows this trend for our sample using only those galaxies in the mass regime where our sample is complete and correcting for their available volume. We find a transition mass at log (M/M)∼10.75, where high-mass galaxies are mass dominated by the bulges and low-mass galaxies by discs. This transition value is slightly lower than the one found in Moffett et al. (2016,log(M/M)∼10.9) and slightly larger than that measured by Thanjavur et al. (2016,log(M/M)∼10.5) but, taken into account our bin size (0.5 dex), is in good agreement with those works. 4.3 B/T luminosity and mass ratio and its relation to galaxy mass Fig. 6shows the B/Tdistribution of our sample galaxies as a function of the global galaxy mass. We show the standard B/Tlluminosity ratio obtained from the photometric decompositions performed in the SDSS rband (orange circles and red stars; M´ endez-Abreu et al. Figure 5. Volume-corrected stellar-mass fraction of galactic bulges (red) and discs (blue) as a function of the global galaxy mass. Only the 104 galaxies in the mass range 9.5 <log (M/M)<12 and belonging to the CALIFA mothersample are used in this analysis. Figure 6. Distribution of the bulge-to-total (B/T) ratio as a function of galaxy mass. Small grey and orange circles show the results for individual galaxies using either mass or r-band light, respectively. Large black and red stars represent mean values using mass and r-band light, respectively. Errors bars show 1σdeviations. Mean values are only computed in the complete global galaxy mass range (9.5 <log (M/M)<12). The shaded area shows the stellar mass range where our sample is incomplete. 2017), but also the B/Tmmass ratio (grey circles and black stars). The latter was derived using the results from our spectrophotometric decomposition, and computing the stellar mass of the bulges and discs through our stellar population analysis on the individual datacubes. The mean values are listed in Table 2. There is a clear trend between both B/Tland B/Tmand the galaxy stellar mass, with more massive galaxies being more dominated by the bulge luminosity or stellar mass. This is consistent with the results shown in Fig. 5, and with previous results in the literature (Weinzirl et al. 2009;M ´ endez-Abreu et al. 2017). We also show that on average B/Tm>B/Tlat all galaxy masses. This holds even when using the B/Tlin the SDSS iband. Therefore, the simple hypothesis that, for a given galaxy, both the bulge and disc luminosities can be transformed into stellar masses assuming the same M/Lratio is not accurate, otherwise B/Tm∼B/Tl. Fig. 7shows the evolution of the B/Tmand bulge-to-disc (B/Dm) mass ratio as a function of cosmic time and redshift. We compute the amount of mass in each component (bulge and disc), for each MNRAS 504, 3058–3073 (2021) Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022 Bulges and discs in the CALIFA survey 3065 Table 2. Mean luminosity and mass B/Tratios. log(M/M)B/TlB/Tm (1) (2) (3) 9.5–10.0 0.09 ±0.08 0.19 ±0.19 10.0–10.5 0.19 ±0.18 0.33 ±0.24 10.5–11.0 0.27 ±0.15 0.47 ±0.23 11.0–11.5 0.45 ±0.11 0.76 ±0.17 11.5–12.0 0.50 ±0.24 0.76 ±0.24 Columns: (1) Mass interval in log units; (2) mean and standard deviation values of the B/Tluminosity ratio in the SDSS rband; (3) mean and standard deviation values of the B/Tmass ratio. Only galaxies in the complete mass range have been used. time-step, using their integrated SFHs across the FoV of CALIFA (see S´ anchez et al. 2018, for details). This fossil approach has the caveatthatour spectro-photometric definitionofbulgeanddisc might be different at z=0 than at higher redshift, but it is otherwise very useful to provide a direct comparison with simulations. It is worth nothing that our definition does not take into account any difference between either in situ and ex situ formation of the stars in our bulges and disc or any stellar population mixing due to radial migration. It is alsoimportant tonoticethat the timeresolution for stellarpopulations older than 8 Gyr is scarce and therefore any evolution for z1 should be taken carefully. Numerical simulations show that the ex situ stellar mass fraction in galaxies is only relevant at high masses (log(M/M)>11; Rodriguez-Gomez et al. 2017; Tacchella et al. 2019) and exclusively affect only the growth of bulges (Clauwens et al. 2018). Still, the dynamical redistribution of in situ formed stars is not fully understood. The evolution of the B/Tmass ratio with redshift is almost constant for a given galaxy stellar mass, showing typical values that are higher (at all redshifts) for more massive galaxies (similar to those shown in Fig. 6). Therefore, the relative growth of the bulge mass with respect to the global galaxy does not evolve significantly with time. None the less, despite Fig. 7showing a fairly constant behaviour of B/T with redshift, we find that this parameter can become insensitive to variations when one of the components overly dominates. Indeed, the redshift evolution of the B/Dmass ratio (Fig. 7, right-hand panel) demonstrates that, for the most massive galaxies (10.5 < log(M/M)<12) and even if the B/Tdoes not change, the actual relative growth of bulges and disc can vary over orders of magnitude in mass. The derived time evolution of the B/Tseems to be in contradiction with the evolution predicted in some the IllustrisTNG numerical simulations. Tacchella et al. (2019) found a significant evolution of the spheroid-to-total (S/T) mass ratio with redshift, in particular for their low-mass galaxies with log(M/M)<10.5. However, the evolution is milder in the EAGLE simulations (Clauwens et al. 2018). The convergent point in both simulations is that there is evidence for an epoch of disc formation for galaxies with log(M/M)>10. This is compatible with our D/Tevolution, which is also dependent on the mass and redshift. We further explore the mass growth of bulges and discs in a companion paper (M´ endez-Abreu et al., in preparation). 4.4 Mass–size relations for bulges and discs Fig. 8shows the mass–size relation for the bulges and discs in our galaxy sample. The size of the galaxy components is obtained from the r-band photometric decompositions of M´ endez-Abreu et al. (2017). The bulge effective radius (re), the disc effective radius (1.678×h), and the global galaxy effective radius (re, gal obtained from a single S´ ersic fit) are therefore used as measure of the galaxy/component size. The stellar masses of the bulges, discs, and for the global galaxies are computed from the stellar population analysis of their associated CALIFA datacubes. Fig. 8also shows the mass–size relation for the galaxies in our sample as if they were considered a single component (grey isocontours) and for a comparison sample of photometrically defined ’pure’ elliptical galaxies (golden contours; see Section 2). We first focus on the global galaxy stellar mass versus component size (left-hand panel in Fig. 8). We find that, for a given galaxy mass, discs are always larger than bulges at all masses ranging from 8<log(M/M)<12. This is consistent with the general picture that spheroids and discy galaxies show distinct trends (Kauffmann et al. 2003). Both ’pure’ elliptical galaxies and complete galaxies (i.e. without any decomposition) show an intermediate behaviour between bulges and discs. The different behaviour of ellipticals and bulges has also been pointed out in previous studies (Gadotti 2009; Laurikainen et al. 2010). Our global galaxy stellar mass versus component size relations show an upturn at log (M/M)∼10.5–11 for all the systems used in this study: bulges, discs, ellipticals (even if they only cover the massive end), and global galaxies. To describe the shape of the relation we perform a fit to the M–reusing a simple power law (Shen et al. 2003; Lange et al. 2016). However, we find this simple modelling is not a good representation of the relations due to the aforementioned upturn. Therefore, we divide the sample into two mass bins: 8 <log(M/M)<10.5 and 10.5 <log (M/M)< 12 and perform two different fits to capture the change in the slope. The best-fitting values are shown in Table 3and they are also shown in Fig. 8. For the sake of comparison we also plot the results from Lange et al. (2016) using their sample of z=0 discs and spheroids analysed in the rband for galaxies with log(M/M)>9. We find a good agreement in the slopes of the mass–size relation for bulges and discs when using our low-mass galaxy sample. The zero-point of the relation for the bulges is also in good agreement, but our discs are systematically larger than those analysed in Lange et al. (2016). In addition, previous studies of the M–rehave found that, when divided into components, they are typically less curved than the global galaxy relation (Bernardi et al. 2014; Lange et al. 2016) except for the discs in late-type galaxies, that cannot be fit with a single power law. This is in contradiction with our results since we find the upturn in all the systems analysed in this study. Another piece of information that can be obtained from analysing the mass–size relation using the stellar masses of the individual components (bulges and discs). This is shown in the right-hand panel of Fig. 8. The best-fitting values to the power law obtained for the bulge distribution show a similar behaviour to those derived using the global galaxy mass, even if the actual values of both slopes and zero-points are different. In addition, we find that the scatter around the best fit is lower when considering only the mass of the bulge component than with the global galaxy mass. The situation is completely different for galaxy discs where (i) there are only a few systems with log (M,d/M)>10.5 and (ii) both the bestfitting values and the scatter are less constrained than when using the global galaxy mass. Fig. 8also includes colour-coded information about the B/T luminosity ratio for each galaxy. The global galaxy stellar mass versus component size (left-hand panel) shows the trend between B/Tand stellar mass discussed in Section 4.1, where more massive galaxies host more prominent bulges. We find that the upturn in the mass–size relation happens at log (M/M)∼10.5 and B/T∼ 0.2. This result holds for bulges when using only their stellar mass (right-hand panel), but not for discs. MNRAS 504, 3058–3073 (2021) Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022 3066 J. M´ endez-Abreu, A. de Lorenzo-C´ aceres and S. F. S´ anchez Figure 7. Time evolution of the bulge-to-total (B/T) mass ratio (left-hand panel) and bulge-to-disc (B/D) mass ratio (right-hand panel) for galaxies in different mass bins (see colours in the legend). The mass of every component has been computed using the stellar population analysis. The redshift is shown in the upper x-axis. The shaded area represents stellar ages (or redshifts) that are not well resolved in our stellar population analysis (see the text for details). Figure 8. Mass–size relation for the bulges (circles) and discs (stars) of our galaxy sample. The leftand right-hand panels show the distributions as a function of the global galaxy mass and the mass of each component, respectively. All stellar masses are derived from our stellar population analysis. The effective radii of both bulges and discs (1.678×h) are obtained from the r-band photometric decompositions of M´ endez-Abreu et al. (2017). Grey isocontours represent the mass–size relation of our galaxy sample without separating their structures. Golden isocontours show the mass–size relation of the photometricallydefinedpure ellipticals in the CALIFA sample (see M´ endez-Abreu et al. 2018, for details). The B/Tluminosity ratio for each galaxy is coloured in reddish (bulge) and bluish (disc) colours according to the colourbars. The blue and red solid lines show the best fit obtained by Lange et al. (2016) for their sample of discs and spheroids, respectively. The blue and red dotted lines represent the best fits (see Table 3) obtained for our sample of discs and bulges, respectively. To perform our fit to the mass–size relation we divided the sample into two mass bins: 8 <log(M/M)<10.5 and 10.5 <log(M/M)<12 to capture the clearly appreciated change in the slope. The grey and blue shaded areas show the global and disc stellar mass range where our sample is incomplete, respectively. Table 3. Best-fitting values to the M–rerelation. log(M/M)ab (1) (2) (3) Bulge 8.0–10.5 0.88 ±0.48 0.20 ±0.05 Bulge 10.5–12.0 −6.2 ±0.39 0.87 ±0.04 Disc 8.0–10.5 0.43 ±0.41 0.33 ±0.04 Disc 10.5–12.0 −2.49 ±0.33 0.61 ±0.03 Bulge 8.0–10.5 −0.39 ±0.37 0.34 ±0.04 Bulge 10.5–12.0 −4.22 ±0.39 0.71 ±0.04 Disc 8.0–10.5 −2.4 ±0.60 0.63 ±0.06 Columns: (1) Mass interval in log units used for the bulge and disc fits; (2) and (3) aand bbest fitted coefficients to the relation logre=a+blog(M/M), respectively. Above and below the line show results for global galaxy mass and individual component mass, respectively 4.4.1 Mass–size for bulges and discs and the Hubble type Themass–size relationof bulgesand discsasa functionof theHubble typefollowsa similartrend aswith theB/Tmass ratio shown in Fig. 8. Galaxies with earlier Hubble types are generally more massive and they have larger bulges and discs with respect to later types. This is expected due to the known relations between the Hubble types, B/T,andS ´ ersic index shown in Section 4.1. Due to the limited size of our sample, and their low number of very late-type galaxies, we cannot create individual mass–size relations for different Hubble types. Nevertheless, Table 4shows the median sizes of our bulges and discs in different bins of mass and Hubble type (only for bins with more than two galaxies). We find that, independently of the Hubble type of the galaxy, bulges and discs are always larger in size as the stellar mass of the galaxy increases. Therefore the mass–size relation holds for any MNRAS 504, 3058–3073 (2021) Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022 Bulges and discs in the CALIFA survey 3073 Weinzirl T., Jogee S., Khochfar S., Burkert A., Kormendy J., 2009, ApJ, 696, 411 Zavala J., Avila-Reese V., Firmani C., Boylan-Kolchin M., 2012, MNRAS, 427, 1503 Zhu L., van de Ven G., M´ endez-Abreu J., Obreja A., 2018b, MNRAS, 479, 945 Zhu L. et al., 2018a, MNRAS, 473, 3000 Zolotov A. et al., 2015, MNRAS, 450, 2327 SUPPORTING INFORMATION Supplementary data are available at MNRAS online. suppl data Pleasenote:OxfordUniversity Pressisnotresponsibleforthecontent or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article. This paper has been typeset from a TEX/L A TEX file prepared by the author. MNRAS 504, 3058–3073 (2021) Downloaded from https://academic.oup.com/mnras/article/504/2/3058/6244233 by Universidad de Granada - Historia de las Ciencias user on 01 April 2022